feat: add 7.x source code

This commit is contained in:
Julien Neuhart
2021-08-22 12:52:44 +02:00
parent e457155950
commit 0f5e8fd314
111 changed files with 31188 additions and 0 deletions

494
pkg/modules/api/api.go Normal file
View File

@@ -0,0 +1,494 @@
package api
import (
"context"
"errors"
"fmt"
"net/http"
"os"
"regexp"
"sort"
"strconv"
"strings"
"time"
"github.com/alexliesenfeld/health"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/gc"
"github.com/labstack/echo/v4"
flag "github.com/spf13/pflag"
"go.uber.org/multierr"
"go.uber.org/zap"
)
func init() {
gotenberg.MustRegisterModule(API{})
}
// API is a module which provides an HTTP server. Other modules may add
// "multipart/form-data" routes, middlewares or health checks.
type API struct {
port int
readTimeout time.Duration
processTimeout time.Duration
writeTimeout time.Duration
rootPath string
traceHeader string
disableHealthCheckLogging bool
webhookAllowList *regexp.Regexp
webhookDenyList *regexp.Regexp
webhookErrorAllowList *regexp.Regexp
webhookErrorDenyList *regexp.Regexp
webhookMaxRetry int
webhookRetryMinWait time.Duration
webhookRetryMaxWait time.Duration
disableWebhook bool
multipartFormDataRoutes []MultipartFormDataRoute
externalMiddlewares []Middleware
healthChecks []health.CheckerOption
logger *zap.Logger
srv *echo.Echo
}
// MultipartFormDataRouter is a module interface which adds
// "multipart/form-data" routes to the API.
type MultipartFormDataRouter interface {
Routes() ([]MultipartFormDataRoute, error)
}
// MultipartFormDataRoute represents a "multipart/form-data" route. All routes
// uses the HTTP POST method.
type MultipartFormDataRoute struct {
// Path is the sub path of the route. Must start with a slash.
// Required.
Path string
// Handler is the function which handles the request.
// Required.
Handler func(ctx *Context) error
}
// MiddlewareProvider is a module interface which adds middlewares to the API.
type MiddlewareProvider interface {
Middlewares() ([]Middleware, error)
}
// MiddlewarePriority is a type which helps to determine the execution order of
// middlewares provided by the MiddlewareProvider modules.
type MiddlewarePriority uint32
const (
VeryLowPriority MiddlewarePriority = iota
LowPriority
MediumPriority
HighPriority
VeryHighPriority
)
// Middleware is a middleware which can be added to the API's middlewares
// chain.
//
// middleware := &Middleware{
// Handler: func() echo.MiddlewareFunc {
// return func(next echo.HandlerFunc) echo.HandlerFunc {
// return func(c echo.Context) error {
// rootPath := c.Get("rootPath").(string)
// healthURI := fmt.Sprintf("%shealth", rootPath)
//
// // Skip the middleware if health check URI.
// if c.Request().RequestURI == healthURI {
// // Call the next middleware in the chain.
// return next(c)
// }
//
// // Your middleware process.
// // ...
//
// // Call the next middleware in the chain.
// return next(c)
// }
// }
// }(),
// }
type Middleware struct {
// RunBeforeRouter tells if the middleware should run before the router
// process an HTTP request.
// Optional.
RunBeforeRouter bool
// Priority tells if the middleware should be positioned high or not in
// the middlewares chain.
// Default to VeryLowPriority.
// Optional.
Priority MiddlewarePriority
// Handler is the function of the middleware.
// Required.
Handler echo.MiddlewareFunc
}
// HealthChecker is a module interface which allows adding health checks to the
// API.
//
// See https://github.com/alexliesenfeld/health for more details.
type HealthChecker interface {
Checks() ([]health.CheckerOption, error)
}
// Descriptor returns an API's module descriptor.
func (API) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "api",
FlagSet: func() *flag.FlagSet {
fs := flag.NewFlagSet("api", flag.ExitOnError)
fs.Int("api-port", 3000, "Set the port on which the API should listen")
fs.String("api-port-from-env", "", "Set the environment variable with the port on which the API should listen - override the default port")
fs.Duration("api-read-timeout", time.Duration(30)*time.Second, "Set the maximum duration allowed to read a complete request, including the body")
fs.Duration("api-process-timeout", time.Duration(30)*time.Second, "Set the maximum duration allowed to process a request")
fs.Duration("api-write-timeout", time.Duration(30)*time.Second, "Set the maximum duration before timing out writes of the response")
fs.String("api-root-path", "/", "Set the root path of the API - for service discovery via URL paths")
fs.String("api-trace-header", "Gotenberg-Trace", "Set the header name to use for identifying requests")
fs.Bool("api-disable-health-check-logging", false, "Disable health check logging")
fs.String("api-webhook-allow-list", "", "Set the allowed URLs for the webhook feature using a regular expression")
fs.String("api-webhook-deny-list", "", "Set the denied URLs for the webhook feature using a regular expression")
fs.String("api-webhook-error-allow-list", "", "Set the allowed URLs in case of an error for the webhook feature using a regular expression")
fs.String("api-webhook-error-deny-list", "", "Set the denied URLs in case of an error for the webhook feature using a regular expression")
fs.Int("api-webhook-max-retry", 4, "Set the maximum number of retries for the webhook feature")
fs.Duration("api-webhook-retry-min-wait", time.Duration(1)*time.Second, "Set the minimum duration to wait before trying to call the webhook again")
fs.Duration("api-webhook-retry-max-wait", time.Duration(30)*time.Second, "Set the maximum duration to wait before trying to call the webhook again")
fs.Bool("api-disable-webhook", false, "Disable the webhook feature")
return fs
}(),
New: func() gotenberg.Module { return new(API) },
}
}
// Provision sets the module properties.
func (a *API) Provision(ctx *gotenberg.Context) error {
flags := ctx.ParsedFlags()
a.port = flags.MustInt("api-port")
a.readTimeout = flags.MustDuration("api-read-timeout")
a.processTimeout = flags.MustDuration("api-process-timeout")
a.writeTimeout = flags.MustDuration("api-write-timeout")
a.rootPath = flags.MustString("api-root-path")
a.traceHeader = flags.MustString("api-trace-header")
a.disableHealthCheckLogging = flags.MustBool("api-disable-health-check-logging")
a.webhookAllowList = flags.MustRegexp("api-webhook-allow-list")
a.webhookDenyList = flags.MustRegexp("api-webhook-deny-list")
a.webhookErrorAllowList = flags.MustRegexp("api-webhook-error-allow-list")
a.webhookErrorDenyList = flags.MustRegexp("api-webhook-error-deny-list")
a.webhookMaxRetry = flags.MustInt("api-webhook-max-retry")
a.webhookRetryMinWait = flags.MustDuration("api-webhook-retry-min-wait")
a.webhookRetryMaxWait = flags.MustDuration("api-webhook-retry-max-wait")
a.disableWebhook = flags.MustBool("api-disable-webhook")
// Port from env?
portEnvVar := flags.MustString("api-port-from-env")
if portEnvVar != "" {
val, ok := os.LookupEnv(portEnvVar)
if !ok {
return fmt.Errorf("environment variable '%s' does not exist", portEnvVar)
}
if val == "" {
return fmt.Errorf("environment variable '%s' is empty", portEnvVar)
}
port, err := strconv.Atoi(val)
if err != nil {
return fmt.Errorf("get int value of environment variable '%s': %w", portEnvVar, err)
}
a.port = port
}
// Get routes from modules.
mods, err := ctx.Modules(new(MultipartFormDataRouter))
if err != nil {
return fmt.Errorf("get multipart/form-data routers: %w", err)
}
routers := make([]MultipartFormDataRouter, len(mods))
for i, router := range mods {
routers[i] = router.(MultipartFormDataRouter)
}
for _, router := range routers {
routes, err := router.Routes()
if err != nil {
return fmt.Errorf("get routes: %w", err)
}
a.multipartFormDataRoutes = append(a.multipartFormDataRoutes, routes...)
}
// Get middlewares from modules.
mods, err = ctx.Modules(new(MiddlewareProvider))
if err != nil {
return fmt.Errorf("get middleware providers: %w", err)
}
middlewareProviders := make([]MiddlewareProvider, len(mods))
for i, middlewareProvider := range mods {
middlewareProviders[i] = middlewareProvider.(MiddlewareProvider)
}
for _, middlewareProvider := range middlewareProviders {
middlewares, err := middlewareProvider.Middlewares()
if err != nil {
return fmt.Errorf("get middlewares: %w", err)
}
a.externalMiddlewares = append(a.externalMiddlewares, middlewares...)
}
// Sort middlewares by priority.
sort.Slice(a.externalMiddlewares, func(i, j int) bool {
return a.externalMiddlewares[i].Priority > a.externalMiddlewares[j].Priority
})
// Get health checks from modules.
mods, err = ctx.Modules(new(HealthChecker))
if err != nil {
return fmt.Errorf("get health checkers: %w", err)
}
healthCheckers := make([]HealthChecker, len(mods))
for i, healthChecker := range mods {
healthCheckers[i] = healthChecker.(HealthChecker)
}
for _, healthChecker := range healthCheckers {
checks, err := healthChecker.Checks()
if err != nil {
return fmt.Errorf("get health checks: %w", err)
}
a.healthChecks = append(a.healthChecks, checks...)
}
loggerProvider, err := ctx.Module(new(gotenberg.LoggerProvider))
if err != nil {
return fmt.Errorf("get logger provider: %w", err)
}
logger, err := loggerProvider.(gotenberg.LoggerProvider).Logger(a)
if err != nil {
return fmt.Errorf("get logger: %w", err)
}
a.logger = logger
return nil
}
// Validate validates the module properties.
func (a API) Validate() error {
var err error
if a.port < 1 || a.port > 65535 {
err = multierr.Append(err,
errors.New("port must be more than 1 and less than 65535"),
)
}
if !strings.HasPrefix(a.rootPath, "/") {
err = multierr.Append(err,
errors.New("root path must start with /"),
)
}
if !strings.HasSuffix(a.rootPath, "/") {
err = multierr.Append(err,
errors.New("root path must end with /"),
)
}
if len(strings.TrimSpace(a.traceHeader)) == 0 {
err = multierr.Append(err,
errors.New("trace header must not be empty"),
)
}
if err != nil {
return err
}
routesMap := make(map[string]MultipartFormDataRoute, len(a.multipartFormDataRoutes))
for _, route := range a.multipartFormDataRoutes {
if route.Path == "" {
return errors.New("route with empty path cannot be registered")
}
if !strings.HasPrefix(route.Path, "/") {
return fmt.Errorf("route %s does not start with /", route.Path)
}
if route.Handler == nil {
return fmt.Errorf("route %s has a nil handler", route.Path)
}
if _, ok := routesMap[route.Path]; ok {
return fmt.Errorf("route %s is already registered", route.Path)
}
routesMap[route.Path] = route
}
for _, middleware := range a.externalMiddlewares {
if middleware.Handler == nil {
return errors.New("a middleware has a nil handler")
}
}
return nil
}
// Start starts the HTTP server.
func (a *API) Start() error {
a.srv = echo.New()
a.srv.HideBanner = true
a.srv.HidePort = true
a.srv.Server.ReadTimeout = a.readTimeout
a.srv.Server.WriteTimeout = a.writeTimeout
a.srv.HTTPErrorHandler = httpErrorHandler(a.traceHeader)
a.srv.Pre(
latencyMiddleware(),
rootPathMiddleware(a.rootPath),
traceMiddleware(a.traceHeader),
loggerMiddleware(a.logger, a.disableHealthCheckLogging),
)
for _, externalMiddleware := range a.externalMiddlewares {
if externalMiddleware.RunBeforeRouter {
a.srv.Pre(externalMiddleware.Handler)
continue
}
a.srv.Use(externalMiddleware.Handler)
}
hardTimeout := a.processTimeout + (time.Duration(5) * time.Second)
a.srv.GET(
fmt.Sprintf("%shealth", a.rootPath),
func() echo.HandlerFunc {
checks := append(a.healthChecks, health.WithTimeout(a.processTimeout))
checker := health.NewChecker(checks...)
return func(echoCtx echo.Context) error {
health.NewHandler(checker).ServeHTTP(echoCtx.Response().Writer, echoCtx.Request())
return nil
}
}(),
timeoutMiddleware(hardTimeout),
)
formsGroup := a.srv.Group(
fmt.Sprintf("%sforms", a.rootPath),
contextMiddleware(
contextMiddlewareConfig{
traceHeader: a.traceHeader,
timeout: struct {
process time.Duration
write time.Duration
}{
process: a.processTimeout,
write: a.writeTimeout,
},
webhook: struct {
allowList *regexp.Regexp
denyList *regexp.Regexp
errorAllowList *regexp.Regexp
errorDenyList *regexp.Regexp
maxRetry int
retryMinWait time.Duration
retryMaxWait time.Duration
disable bool
}{
allowList: a.webhookAllowList,
denyList: a.webhookDenyList,
errorAllowList: a.webhookErrorAllowList,
errorDenyList: a.webhookErrorDenyList,
maxRetry: a.webhookMaxRetry,
retryMinWait: a.webhookRetryMinWait,
retryMaxWait: a.webhookRetryMaxWait,
disable: a.disableWebhook,
},
},
),
timeoutMiddleware(hardTimeout),
)
// Add routes from other modules.
for _, route := range a.multipartFormDataRoutes {
formsGroup.POST(
route.Path,
func(route MultipartFormDataRoute) echo.HandlerFunc {
return func(c echo.Context) error {
ctx := c.Get("context").(*Context)
err := route.Handler(ctx)
if err != nil {
return fmt.Errorf("handle request: %w", err)
}
return nil
}
}(route),
)
}
// As the listen method is blocking, run it in a goroutine.
go func() {
err := a.srv.Start(fmt.Sprintf(":%d", a.port))
if !errors.Is(err, http.ErrServerClosed) {
a.logger.Fatal(err.Error())
}
}()
return nil
}
// StartupMessage returns a custom startup message.
func (a API) StartupMessage() string {
return fmt.Sprintf("server listening on port %d", a.port)
}
// Stop stops the HTTP server.
func (a API) Stop(ctx context.Context) error {
return a.srv.Shutdown(ctx)
}
// GraceDuration updates the expiration time of files and directories parsed by
// the gc.GarbageCollector.
func (a API) GraceDuration() time.Duration {
duration := a.readTimeout + a.processTimeout + a.writeTimeout
if a.disableWebhook {
return duration
}
for i := 0; i < a.webhookMaxRetry; i++ {
// Yep... Golang does not allow int * time.Duration.
duration += a.webhookRetryMaxWait
}
return duration
}
// Interface guards.
var (
_ gotenberg.Module = (*API)(nil)
_ gotenberg.Provisioner = (*API)(nil)
_ gotenberg.Validator = (*API)(nil)
_ gotenberg.App = (*API)(nil)
_ gc.GarbageCollectorGraceDurationModifier = (*API)(nil)
)

793
pkg/modules/api/api_test.go Normal file
View File

@@ -0,0 +1,793 @@
package api
import (
"bytes"
"context"
"errors"
"mime/multipart"
"net/http"
"net/http/httptest"
"os"
"reflect"
"testing"
"time"
"github.com/alexliesenfeld/health"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/labstack/echo/v4"
"go.uber.org/zap"
)
type ProtoModule struct {
descriptor func() gotenberg.ModuleDescriptor
}
func (mod ProtoModule) Descriptor() gotenberg.ModuleDescriptor {
return mod.descriptor()
}
type ProtoValidator struct {
ProtoModule
validate func() error
}
func (mod ProtoValidator) Validate() error {
return mod.validate()
}
type ProtoMultipartFormDataRouter struct {
ProtoValidator
routes func() ([]MultipartFormDataRoute, error)
}
func (mod ProtoMultipartFormDataRouter) Routes() ([]MultipartFormDataRoute, error) {
return mod.routes()
}
type ProtoMiddlewareProvider struct {
ProtoValidator
middlewares func() ([]Middleware, error)
}
func (mod ProtoMiddlewareProvider) Middlewares() ([]Middleware, error) {
return mod.middlewares()
}
type ProtoHealthChecker struct {
ProtoValidator
checks func() ([]health.CheckerOption, error)
}
func (mod ProtoHealthChecker) Checks() ([]health.CheckerOption, error) {
return mod.checks()
}
type ProtoLoggerProvider struct {
ProtoModule
logger func(mod gotenberg.Module) (*zap.Logger, error)
}
func (factory ProtoLoggerProvider) Logger(mod gotenberg.Module) (*zap.Logger, error) {
return factory.logger(mod)
}
func TestAPI_Descriptor(t *testing.T) {
descriptor := API{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(API))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestAPI_Provision(t *testing.T) {
for i, tc := range []struct {
ctx *gotenberg.Context
setEnv func(i int)
expectPort int
expectMiddlewares []Middleware
expectErr bool
}{
{
ctx: func() *gotenberg.Context {
fs := new(API).Descriptor().FlagSet
err := fs.Parse([]string{"--api-port-from-env=FOO"})
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: fs,
},
nil,
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
fs := new(API).Descriptor().FlagSet
err := fs.Parse([]string{"--api-port-from-env=PORT"})
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: fs,
},
nil,
)
}(),
setEnv: func(i int) {
err := os.Setenv("PORT", "")
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
},
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
fs := new(API).Descriptor().FlagSet
err := fs.Parse([]string{"--api-port-from-env=PORT"})
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: fs,
},
nil,
)
}(),
setEnv: func(i int) {
err := os.Setenv("PORT", "foo")
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
},
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
fs := new(API).Descriptor().FlagSet
err := fs.Parse([]string{"--api-port-from-env=PORT"})
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: fs,
},
nil,
)
}(),
setEnv: func(i int) {
err := os.Setenv("PORT", "1337")
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
},
expectPort: 1337,
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoMultipartFormDataRouter }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error {
return errors.New("foo")
}
mod.routes = func() ([]MultipartFormDataRoute, error) {
return nil, nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoMiddlewareProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error {
return errors.New("foo")
}
mod.middlewares = func() ([]Middleware, error) {
return nil, nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoMiddlewareProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error {
return nil
}
mod.middlewares = func() ([]Middleware, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoMultipartFormDataRouter }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error {
return nil
}
mod.routes = func() ([]MultipartFormDataRoute, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoHealthChecker }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error {
return errors.New("foo")
}
mod.checks = func() ([]health.CheckerOption, error) {
return nil, nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoHealthChecker }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error {
return nil
}
mod.checks = func() ([]health.CheckerOption, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoLoggerProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.logger = func(_ gotenberg.Module) (*zap.Logger, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod1 := struct{ ProtoMultipartFormDataRouter }{}
mod1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod1 }}
}
mod1.validate = func() error {
return nil
}
mod1.routes = func() ([]MultipartFormDataRoute, error) {
return []MultipartFormDataRoute{{}}, nil
}
mod2 := struct{ ProtoMiddlewareProvider }{}
mod2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "bar", New: func() gotenberg.Module { return mod2 }}
}
mod2.validate = func() error {
return nil
}
mod2.middlewares = func() ([]Middleware, error) {
return []Middleware{
{
Priority: VeryLowPriority,
},
{
Priority: LowPriority,
},
{
Priority: MediumPriority,
},
{
Priority: HighPriority,
},
{
Priority: VeryHighPriority,
},
}, nil
}
mod3 := struct{ ProtoHealthChecker }{}
mod3.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "baz", New: func() gotenberg.Module { return mod3 }}
}
mod3.validate = func() error {
return nil
}
mod3.checks = func() ([]health.CheckerOption, error) {
return []health.CheckerOption{health.WithDisabledAutostart()}, nil
}
mod4 := struct{ ProtoLoggerProvider }{}
mod4.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "qux", New: func() gotenberg.Module { return mod4 }}
}
mod4.logger = func(_ gotenberg.Module) (*zap.Logger, error) {
return zap.NewNop(), nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(API).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod1.Descriptor(),
mod2.Descriptor(),
mod3.Descriptor(),
mod4.Descriptor(),
},
)
}(),
expectMiddlewares: []Middleware{
{
Priority: VeryHighPriority,
},
{
Priority: HighPriority,
},
{
Priority: MediumPriority,
},
{
Priority: LowPriority,
},
{
Priority: VeryLowPriority,
},
},
},
} {
if tc.setEnv != nil {
tc.setEnv(i)
}
mod := new(API)
err := mod.Provision(tc.ctx)
if tc.expectPort != 0 && mod.port != tc.expectPort {
t.Errorf("expected port %d but got %d", tc.expectPort, mod.port)
}
if !reflect.DeepEqual(mod.externalMiddlewares, tc.expectMiddlewares) {
t.Errorf("expected %+v, but got: %+v", tc.expectMiddlewares, mod.externalMiddlewares)
}
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestAPI_Validate(t *testing.T) {
for i, tc := range []struct {
port int
rootPath string
traceHeader string
routes []MultipartFormDataRoute
middlewares []Middleware
expectErr bool
}{
{
port: 0,
expectErr: true,
},
{
port: 65536,
rootPath: "foo",
expectErr: true,
},
{
port: 10,
rootPath: "/foo/",
traceHeader: "foo",
routes: []MultipartFormDataRoute{
{
Path: "",
},
},
expectErr: true,
},
{
port: 10,
rootPath: "/foo/",
traceHeader: "foo",
routes: []MultipartFormDataRoute{
{
Path: "foo",
},
},
expectErr: true,
},
{
port: 10,
rootPath: "/foo/",
traceHeader: "foo",
routes: []MultipartFormDataRoute{
{
Path: "/foo",
},
},
expectErr: true,
},
{
port: 10,
rootPath: "/foo/",
traceHeader: "foo",
routes: []MultipartFormDataRoute{
{
Path: "/foo",
Handler: func(_ *Context) error { return nil },
},
{
Path: "/foo",
Handler: func(_ *Context) error { return nil },
},
},
expectErr: true,
},
{
port: 10,
rootPath: "/foo/",
traceHeader: "foo",
middlewares: []Middleware{
{
Priority: HighPriority,
},
},
expectErr: true,
},
{
port: 10,
rootPath: "/foo/",
traceHeader: "foo",
routes: []MultipartFormDataRoute{
{
Path: "/foo",
Handler: func(_ *Context) error { return nil },
},
},
middlewares: []Middleware{
{
Priority: HighPriority,
Handler: func() echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
return next(c)
}
}
}(),
},
},
},
} {
mod := API{
port: tc.port,
rootPath: tc.rootPath,
traceHeader: tc.traceHeader,
multipartFormDataRoutes: tc.routes,
externalMiddlewares: tc.middlewares,
}
err := mod.Validate()
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestAPI_Start(t *testing.T) {
mod := new(API)
mod.port = 3000
mod.rootPath = "/"
mod.multipartFormDataRoutes = []MultipartFormDataRoute{
{
Path: "/foo",
Handler: func(ctx *Context) error {
ctx.outputPaths = []string{
"/tests/test/testdata/api/sample1.txt",
}
return nil
},
},
{
Path: "/bar",
Handler: func(_ *Context) error { return errors.New("foo") },
},
}
mod.externalMiddlewares = []Middleware{
{
RunBeforeRouter: true,
Handler: func() echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
return next(c)
}
}
}(),
},
{
Handler: func() echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
return next(c)
}
}
}(),
},
}
mod.logger = zap.NewNop()
err := mod.Start()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
// health request.
recorder := httptest.NewRecorder()
healthRequest := httptest.NewRequest(http.MethodGet, "/health", nil)
mod.srv.ServeHTTP(recorder, healthRequest)
if recorder.Code != http.StatusOK {
t.Errorf("expected %d status code but got %d", http.StatusOK, recorder.Code)
}
// "multipart/form-data" request.
multipartRequest := func(URL string) *http.Request {
body := &bytes.Buffer{}
writer := multipart.NewWriter(body)
defer func() {
err := writer.Close()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
}()
err := writer.WriteField("foo", "foo")
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
part, err := writer.CreateFormFile("foo.txt", "foo.txt")
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
_, err = part.Write([]byte("foo"))
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
req := httptest.NewRequest(http.MethodPost, URL, body)
req.Header.Set(echo.HeaderContentType, writer.FormDataContentType())
return req
}
recorder = httptest.NewRecorder()
mod.srv.ServeHTTP(recorder, multipartRequest("/forms/foo"))
if recorder.Code != http.StatusOK {
t.Errorf("expected %d status code but got %d", http.StatusOK, recorder.Code)
}
recorder = httptest.NewRecorder()
mod.srv.ServeHTTP(recorder, multipartRequest("/forms/bar"))
if recorder.Code != http.StatusInternalServerError {
t.Errorf("expected %d status code but got %d", http.StatusInternalServerError, recorder.Code)
}
err = mod.srv.Shutdown(context.TODO())
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestAPI_StartupMessage(t *testing.T) {
mod := API{
port: 3000,
}
actual := mod.StartupMessage()
expect := "server listening on port 3000"
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestAPI_Stop(t *testing.T) {
mod := API{
port: 3000,
multipartFormDataRoutes: []MultipartFormDataRoute{
{
Path: "/foo",
Handler: func(_ *Context) error { return nil },
},
},
logger: zap.NewNop(),
}
err := mod.Start()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
err = mod.Stop(context.TODO())
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestAPI_GraceDuration(t *testing.T) {
for i, tc := range []struct {
mod API
expect time.Duration
}{
{
mod: API{
readTimeout: time.Duration(1) * time.Second,
processTimeout: time.Duration(1) * time.Second,
writeTimeout: time.Duration(1) * time.Second,
disableWebhook: true,
},
expect: time.Duration(3) * time.Second,
},
{
mod: API{
readTimeout: time.Duration(1) * time.Second,
processTimeout: time.Duration(1) * time.Second,
writeTimeout: time.Duration(1) * time.Second,
webhookMaxRetry: 5,
webhookRetryMaxWait: time.Duration(5) * time.Second,
},
expect: time.Duration(28) * time.Second,
},
} {
actual := tc.mod.GraceDuration()
if actual != tc.expect {
t.Errorf("test %d: expected '%s' but got '%s'", i, tc.expect, actual)
}
}
}
// Interface guards.
var (
_ gotenberg.Module = (*ProtoModule)(nil)
_ gotenberg.Validator = (*ProtoValidator)(nil)
_ gotenberg.Module = (*ProtoValidator)(nil)
_ MultipartFormDataRouter = (*ProtoMultipartFormDataRouter)(nil)
_ gotenberg.Module = (*ProtoMultipartFormDataRouter)(nil)
_ gotenberg.Validator = (*ProtoMultipartFormDataRouter)(nil)
_ MiddlewareProvider = (*ProtoMiddlewareProvider)(nil)
_ gotenberg.Module = (*ProtoMiddlewareProvider)(nil)
_ gotenberg.Validator = (*ProtoMiddlewareProvider)(nil)
_ HealthChecker = (*ProtoHealthChecker)(nil)
_ gotenberg.Module = (*ProtoHealthChecker)(nil)
_ gotenberg.Validator = (*ProtoHealthChecker)(nil)
_ gotenberg.LoggerProvider = (*ProtoLoggerProvider)(nil)
_ gotenberg.Module = (*ProtoLoggerProvider)(nil)
)

319
pkg/modules/api/context.go Normal file
View File

@@ -0,0 +1,319 @@
package api
import (
"compress/flate"
"context"
"errors"
"fmt"
"io"
"mime/multipart"
"net/http"
"os"
"path/filepath"
"strings"
"time"
"unicode"
"github.com/google/uuid"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/labstack/echo/v4"
"github.com/mholt/archiver/v3"
"go.uber.org/zap"
"golang.org/x/text/runes"
"golang.org/x/text/transform"
"golang.org/x/text/unicode/norm"
)
var (
// ErrContextAlreadyClosed happens when the context has been canceled.
ErrContextAlreadyClosed = errors.New("context already closed")
// ErrOutOfBoundsOutputPath happens when an output path is not within
// context's working directory. It enforces having all the files in the
// same directory.
ErrOutOfBoundsOutputPath = errors.New("output path is not within context's working directory")
)
// Context is the request context for a "multipart/form-data" requests.
type Context struct {
dirPath string
values map[string][]string
files map[string]string
outputPaths []string
cancelled bool
logger *zap.Logger
echoCtx echo.Context
context.Context
}
// newContext returns a Context by parsing a "multipart/form-data" request.
func newContext(echoCtx echo.Context, logger *zap.Logger, timeout time.Duration) (*Context, context.CancelFunc, error) {
processCtx, processCancel := context.WithTimeout(context.Background(), timeout)
ctx := &Context{
outputPaths: make([]string, 0),
cancelled: false,
logger: logger,
echoCtx: echoCtx,
Context: processCtx,
}
// A custom cancel function which removes the context's working directory
// when called.
cancel := func() context.CancelFunc {
return func() {
if ctx.cancelled {
return
}
processCancel()
if ctx.dirPath == "" {
return
}
err := os.RemoveAll(ctx.dirPath)
if err != nil {
ctx.logger.Error(fmt.Sprintf("remove context's working directory: %s", err))
return
}
ctx.logger.Debug(fmt.Sprintf("'%s' removed", ctx.dirPath))
ctx.cancelled = true
}
}()
form, err := echoCtx.MultipartForm()
if err != nil {
if errors.Is(err, http.ErrNotMultipart) {
return nil, cancel, WrapError(
fmt.Errorf("get multipart form: %w", err),
NewSentinelHTTPError(http.StatusUnsupportedMediaType, "Invalid 'Content-Type' header value: want 'multipart/form-data'"),
)
}
if errors.Is(err, http.ErrMissingBoundary) {
return nil, cancel, WrapError(
fmt.Errorf("get multipart form: %w", err),
NewSentinelHTTPError(http.StatusUnsupportedMediaType, "Invalid 'Content-Type' header value: no boundary"),
)
}
if strings.Contains(err.Error(), io.EOF.Error()) {
return nil, cancel, WrapError(
fmt.Errorf("get multipart form: %w", err),
NewSentinelHTTPError(http.StatusBadRequest, "Malformed body: it does not match the 'Content-Type' header boundaries"),
)
}
return nil, cancel, fmt.Errorf("get multipart form: %w", err)
}
dirPath, err := gotenberg.MkdirAll()
if err != nil {
return nil, cancel, fmt.Errorf("create working directory: %w", err)
}
ctx.dirPath = dirPath
ctx.values = form.Value
ctx.files = make(map[string]string)
copyToDisk := func(fh *multipart.FileHeader) error {
// Avoid directory traversal and normalize filename.
// See https://github.com/thecodingmachine/gotenberg/issues/104.
// See https://github.com/thecodingmachine/gotenberg/issues/228.
t := transform.Chain(norm.NFD, runes.Remove(runes.In(unicode.Mn)), norm.NFC)
filename, _, err := transform.String(t, strings.ToLower(filepath.Base(fh.Filename)))
if err != nil {
return fmt.Errorf("transform filename: %w", err)
}
in, err := fh.Open()
if err != nil {
return fmt.Errorf("open multipart file: %w", err)
}
defer func() {
err := in.Close()
if err != nil {
logger.Error(fmt.Sprintf("close file header: %s", err))
}
}()
path := fmt.Sprintf("%s/%s", ctx.dirPath, filename)
out, err := os.Create(path)
if err != nil {
return fmt.Errorf("create local file: %w", err)
}
defer func() {
err := out.Close()
if err != nil {
logger.Error(fmt.Sprintf("close local file: %s", err))
}
}()
_, err = io.Copy(out, in)
if err != nil {
return fmt.Errorf("copy multipart file to local file: %w", err)
}
ctx.files[filename] = path
return nil
}
for _, files := range form.File {
for _, fh := range files {
err = copyToDisk(fh)
if err != nil {
return ctx, cancel, fmt.Errorf("copy to disk: %w", err)
}
}
}
ctx.Log().Debug(fmt.Sprintf("form data values: %+v", ctx.values))
ctx.Log().Debug(fmt.Sprintf("form data files: %+v", ctx.files))
return ctx, cancel, err
}
// Request returns the http.Request.
func (ctx Context) Request() *http.Request {
return ctx.echoCtx.Request()
}
// FormData return a FormData.
func (ctx Context) FormData() *FormData {
return &FormData{
values: ctx.values,
files: ctx.files,
errors: nil,
}
}
// GeneratePath generates a path within the context's working directory. It
// does not create a file.
func (ctx Context) GeneratePath(extension string) string {
return fmt.Sprintf("%s/%s%s", ctx.dirPath, uuid.New(), extension)
}
// AddOutputPaths adds the given paths. Those paths will be used later to build
// the output file.
func (ctx *Context) AddOutputPaths(paths ...string) error {
if ctx.cancelled {
return ErrContextAlreadyClosed
}
for _, path := range paths {
if !strings.HasPrefix(path, ctx.dirPath) {
return ErrOutOfBoundsOutputPath
}
ctx.outputPaths = append(ctx.outputPaths, path)
}
return nil
}
// Log returns the context zap.Logger.
func (ctx Context) Log() *zap.Logger {
return ctx.logger
}
// buildOutputFile builds the output file according to the output paths
// registered in the context. If many output paths, an archive is created.
func (ctx Context) buildOutputFile() (string, error) {
if ctx.cancelled {
return "", ErrContextAlreadyClosed
}
if len(ctx.outputPaths) == 0 {
return "", errors.New("no output path")
}
if len(ctx.outputPaths) == 1 {
ctx.logger.Debug(fmt.Sprintf("only one output file '%s', skip archive creation", ctx.outputPaths[0]))
return ctx.outputPaths[0], nil
}
z := archiver.Zip{
CompressionLevel: flate.DefaultCompression,
MkdirAll: true,
SelectiveCompression: true,
ContinueOnError: false,
OverwriteExisting: false,
ImplicitTopLevelFolder: false,
}
archivePath := ctx.GeneratePath(".zip")
err := z.Archive(ctx.outputPaths, archivePath)
if err != nil {
return "", fmt.Errorf("archive output files: %w", err)
}
ctx.logger.Debug(fmt.Sprintf("archive '%s' created", archivePath))
return archivePath, nil
}
// MockContext is a helper for tests.
//
// ctx := &api.MockContext{Context: &api.Context{}}
type MockContext struct {
*Context
}
// SetDirPath sets the context's working directory path.
//
// ctx := &api.MockContext{Context: &api.Context{}}
// ctx.SetDirPath("/foo")
func (ctx *MockContext) SetDirPath(path string) {
ctx.dirPath = path
}
// SetValues sets the values.
//
// ctx := &api.MockContext{Context: &api.Context{}}
// ctx.SetValues(map[string][]string{
// "url": {
// "foo",
// },
// })
func (ctx *MockContext) SetValues(values map[string][]string) {
ctx.values = values
}
// SetFiles sets the files.
//
// ctx := &api.MockContext{Context: &api.Context{}}
// ctx.SetFiles(map[string]string{
// "foo": "/foo",
// })
func (ctx *MockContext) SetFiles(files map[string]string) {
ctx.files = files
}
// SetCancelled sets if the context is cancelled or not.
//
// ctx := &api.MockContext{Context: &api.Context{}}
// ctx.SetCancelled(true)
func (ctx *MockContext) SetCancelled(cancelled bool) {
ctx.cancelled = cancelled
}
// OutputPaths returns the registered output paths.
// ctx := &api.MockContext{Context: &api.Context{}}
// outputPaths := ctx.OutputPaths()
func (ctx MockContext) OutputPaths() []string {
return ctx.outputPaths
}

View File

@@ -0,0 +1,373 @@
package api
import (
"bytes"
"errors"
"mime/multipart"
"net/http"
"net/http/httptest"
"os"
"reflect"
"strings"
"testing"
"time"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/labstack/echo/v4"
"go.uber.org/zap"
)
func TestNewContext(t *testing.T) {
for i, tc := range []struct {
request *http.Request
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
}{
{
request: httptest.NewRequest(http.MethodPost, "/", nil),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusUnsupportedMediaType,
},
{
request: func() *http.Request {
req := httptest.NewRequest(http.MethodPost, "/", nil)
req.Header.Set(echo.HeaderContentType, echo.MIMEMultipartForm)
return req
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusUnsupportedMediaType,
},
{
request: func() *http.Request {
body := &bytes.Buffer{}
writer := multipart.NewWriter(body)
defer func() {
err := writer.Close()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
}()
err := writer.WriteField("foo", "foo")
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
req := httptest.NewRequest(http.MethodPost, "/", nil)
req.Header.Set(echo.HeaderContentType, writer.FormDataContentType())
return req
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
request: func() *http.Request {
body := &bytes.Buffer{}
writer := multipart.NewWriter(body)
defer func() {
err := writer.Close()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
}()
err := writer.WriteField("foo", "foo")
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
part, err := writer.CreateFormFile("foo.txt", "foo.txt")
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
_, err = part.Write([]byte("foo"))
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
req := httptest.NewRequest(http.MethodPost, "/", body)
req.Header.Set(echo.HeaderContentType, writer.FormDataContentType())
return req
}(),
},
} {
handler := func(c echo.Context) error {
_, cancel, err := newContext(c, zap.NewNop(), time.Duration(10)*time.Second)
defer cancel()
// Context already cancelled.
defer cancel()
if err != nil {
return err
}
return nil
}
recorder := httptest.NewRecorder()
srv := echo.New()
srv.HideBanner = true
srv.HidePort = true
c := srv.NewContext(tc.request, recorder)
err := handler(c)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
}
}
func TestContext_Request(t *testing.T) {
request := httptest.NewRequest(http.MethodPost, "/", nil)
recorder := httptest.NewRecorder()
c := echo.New().NewContext(request, recorder)
ctx := Context{
echoCtx: c,
}
if !reflect.DeepEqual(ctx.Request(), c.Request()) {
t.Errorf("expected %v but got %v", ctx.Request(), c.Request())
}
}
func TestContext_FormData(t *testing.T) {
ctx := Context{
values: map[string][]string{
"foo": {"foo"},
},
files: map[string]string{
"foo.txt": "/foo.txt",
},
}
actual := ctx.FormData()
expect := &FormData{
values: ctx.values,
files: ctx.files,
}
if !reflect.DeepEqual(actual, expect) {
t.Errorf("expected %+v but got %+v", expect, actual)
}
}
func TestContext_GeneratePath(t *testing.T) {
ctx := Context{
dirPath: "/foo",
}
path := ctx.GeneratePath(".pdf")
if !strings.HasPrefix(path, ctx.dirPath) {
t.Errorf("expected '%s' to start with '%s'", path, ctx.dirPath)
}
}
func TestContext_AddOutputPaths(t *testing.T) {
for i, tc := range []struct {
ctx *Context
path string
expectCount int
expectErr bool
}{
{
ctx: &Context{cancelled: true},
expectErr: true,
},
{
ctx: &Context{dirPath: "/foo"},
path: "/bar/foo.txt",
expectErr: true,
},
{
ctx: &Context{dirPath: "/foo"},
path: "/foo/foo.txt",
expectCount: 1,
},
} {
err := tc.ctx.AddOutputPaths(tc.path)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
if len(tc.ctx.outputPaths) != tc.expectCount {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectCount, len(tc.ctx.outputPaths))
}
}
}
func TestContext_Log(t *testing.T) {
expect := zap.NewNop()
ctx := Context{logger: expect}
actual := ctx.Log()
if !reflect.DeepEqual(actual, expect) {
t.Errorf("expected %v but got %v", expect, actual)
}
}
func TestContext_buildOutputFile(t *testing.T) {
for i, tc := range []struct {
ctx *Context
expectErr bool
}{
{
ctx: &Context{cancelled: true},
expectErr: true,
},
{
ctx: &Context{},
expectErr: true,
},
{
ctx: &Context{outputPaths: []string{"foo.txt"}},
},
{
ctx: &Context{outputPaths: []string{"foo.txt", "foo.pdf"}},
expectErr: true,
},
{
ctx: &Context{
outputPaths: []string{
"/tests/test/testdata/api/sample1.txt",
"/tests/test/testdata/api/sample1.txt",
},
},
},
} {
dirPath, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("%d: expected no erro but got: %v", i, err)
}
tc.ctx.dirPath = dirPath
tc.ctx.logger = zap.NewNop()
_, err = tc.ctx.buildOutputFile()
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
err = os.RemoveAll(dirPath)
if err != nil {
t.Fatalf("%d: expected no erro but got: %v", i, err)
}
}
}
func TestMockContext_SetDirPath(t *testing.T) {
mock := &MockContext{&Context{}}
mock.SetDirPath("/foo")
actual := mock.dirPath
expect := "/foo"
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestMockContext_SetValues(t *testing.T) {
mock := &MockContext{&Context{}}
mock.SetValues(map[string][]string{
"foo": {"foo"},
})
actual := mock.values
expect := map[string][]string{
"foo": {"foo"},
}
if !reflect.DeepEqual(actual, expect) {
t.Errorf("expected %+v but got: %+v", expect, actual)
}
}
func TestMockContext_SetFiles(t *testing.T) {
mock := &MockContext{&Context{}}
mock.SetFiles(map[string]string{
"foo": "/foo",
})
actual := mock.files
expect := map[string]string{
"foo": "/foo",
}
if !reflect.DeepEqual(actual, expect) {
t.Errorf("expected %+v but got: %+v", expect, actual)
}
}
func TestMockContext_SetCancelled(t *testing.T) {
mock := &MockContext{&Context{}}
mock.SetCancelled(true)
actual := mock.cancelled
if !actual {
t.Errorf("expected %t but got %t", true, actual)
}
}
func TestMockContext_OutputPaths(t *testing.T) {
mock := MockContext{
&Context{
outputPaths: []string{"/foo"},
},
}
actual := mock.OutputPaths()
expect := []string{"/foo"}
if !reflect.DeepEqual(actual, expect) {
t.Errorf("expected %+v but got: %+v", expect, actual)
}
}

3
pkg/modules/api/doc.go Normal file
View File

@@ -0,0 +1,3 @@
// Package api provides a module which is an HTTP server. Other modules may
// add multipart/form-data routes, middlewares, and health checks.
package api

77
pkg/modules/api/errors.go Normal file
View File

@@ -0,0 +1,77 @@
package api
// Credits: https://www.joeshaw.org/error-handling-in-go-http-applications.
// HTTPError is an interface allowing to retrieve the HTTP details of an error.
type HTTPError interface {
HTTPError() (int, string)
}
// SentinelHTTPError is the HTTP sidekick of an error.
type SentinelHTTPError struct {
status int
message string
}
// NewSentinelHTTPError creates a SentinelHTTPError. The message will be sent
// as the response's body if returned from an handler, so make sure to not leak
// sensible information.
func NewSentinelHTTPError(status int, message string) SentinelHTTPError {
return SentinelHTTPError{
status: status,
message: message,
}
}
// Error returns the message.
func (err SentinelHTTPError) Error() string {
return err.message
}
// HTTPError returns the status and message.
func (err SentinelHTTPError) HTTPError() (int, string) {
return err.status, err.message
}
// sentinelWrappedError contains both the error which will logged and the
// sidekick SentinelHTTPError.
type sentinelWrappedError struct {
error
sentinel SentinelHTTPError
}
func (w sentinelWrappedError) Is(err error) bool {
return w.sentinel == err
}
func (w sentinelWrappedError) HTTPError() (int, string) {
return w.sentinel.HTTPError()
}
// WrapError wraps the given error with a SentinelHTTPError. The wrapped error
// will be displayed in a log, while the SentinelHTTPError will be sent in the
// response.
//
// return api.WrapError(
// // This first error will be logged.
// fmt.Errorf("my action: %w", err),
// // The HTTP error will be sent as a response.
// api.NewSentinelHTTPError(
// http.StatusForbidden,
// "Hey, you did something wrong!"
// ),
// )
func WrapError(err error, sentinel SentinelHTTPError) error {
return sentinelWrappedError{
error: err,
sentinel: sentinel,
}
}
// Interface guards.
var (
_ error = (*SentinelHTTPError)(nil)
_ HTTPError = (*SentinelHTTPError)(nil)
_ error = (*sentinelWrappedError)(nil)
_ HTTPError = (*sentinelWrappedError)(nil)
)

View File

@@ -0,0 +1,108 @@
package api
import (
"errors"
"net/http"
"reflect"
"testing"
)
func TestNewSentinelHTTPError(t *testing.T) {
actual := NewSentinelHTTPError(http.StatusInternalServerError, "foo")
expect := SentinelHTTPError{
status: http.StatusInternalServerError,
message: "foo",
}
if !reflect.DeepEqual(actual, expect) {
t.Errorf("expected %v but got %v", expect, actual)
}
}
func TestSentinelHTTPError_Error(t *testing.T) {
err := SentinelHTTPError{
message: "foo",
}
actual := err.Error()
expect := "foo"
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestSentinelHTTPError_HTTPError(t *testing.T) {
actualStatus, actualMessage := SentinelHTTPError{
status: http.StatusInternalServerError,
message: "foo",
}.HTTPError()
expectStatus := http.StatusInternalServerError
expectMessage := "foo"
if actualStatus != expectStatus {
t.Errorf("expected %d but got %d", expectStatus, actualStatus)
}
if actualMessage != expectMessage {
t.Errorf("expected '%s' but got '%s'", expectMessage, actualMessage)
}
}
func TestSentinelWrappedError_Is(t *testing.T) {
errSentinel := SentinelHTTPError{}
err := sentinelWrappedError{
error: errors.New("foo"),
sentinel: errSentinel,
}
if !err.Is(errSentinel) {
t.Error("expected true")
}
}
func TestSentinelWrappedError_HTTPError(t *testing.T) {
expectStatus, expectMessage := SentinelHTTPError{
status: http.StatusInternalServerError,
message: "foo",
}.HTTPError()
actualStatus, actualMessage := sentinelWrappedError{
error: errors.New("foo"),
sentinel: SentinelHTTPError{
status: http.StatusInternalServerError,
message: "foo",
},
}.HTTPError()
if actualStatus != expectStatus {
t.Errorf("expected %d but got %d", expectStatus, actualStatus)
}
if actualMessage != expectMessage {
t.Errorf("expected '%s' but got '%s'", expectMessage, actualMessage)
}
}
func TestWrapError(t *testing.T) {
errFoo := errors.New("foo")
expect := sentinelWrappedError{
error: errFoo,
sentinel: SentinelHTTPError{
status: http.StatusInternalServerError,
message: "foo",
},
}
actual := WrapError(errFoo, SentinelHTTPError{
status: http.StatusInternalServerError,
message: "foo",
})
if !reflect.DeepEqual(actual, expect) {
t.Errorf("expected %v but got %v", expect, actual)
}
}

439
pkg/modules/api/formdata.go Normal file
View File

@@ -0,0 +1,439 @@
package api
import (
"fmt"
"io/ioutil"
"net/http"
"path/filepath"
"sort"
"strconv"
"strings"
"time"
"go.uber.org/multierr"
)
// FormData is a helper for validating and hydrating values from a
// "multipart/form-data" request.
//
// form := ctx.FormData()
type FormData struct {
values map[string][]string
files map[string]string
errors error
}
// Validate returns nil or an error related to the FormData values, with a
// SentinelHTTPError (status code 400, errors' details as message) wrapped
// inside.
//
// var foo string
//
// err := ctx.FormData().
// MandatoryString("foo", &foo, "bar").
// Validate()
func (form FormData) Validate() error {
if form.errors == nil {
return nil
}
return WrapError(
form.errors,
NewSentinelHTTPError(http.StatusBadRequest, fmt.Sprintf("Invalid form data: %s", form.errors)),
)
}
// String binds a form data value to a string variable.
//
// var foo string
//
// ctx.FormData().String("foo", &foo, "bar")
func (form *FormData) String(key string, target *string, defaultValue string) *FormData {
return form.mustValue(key, target, defaultValue)
}
// MandatoryString binds a form data value to a string variable. It populates
// an error if the value is empty or the "key" does not exist.
//
// var foo string
//
// ctx.FormData().MandatoryString("foo", &foo)
func (form *FormData) MandatoryString(key string, target *string) *FormData {
return form.mustMandatoryValue(key, target)
}
// Bool binds a form data value to a bool variable. It populates an error if
// the value is not bool.
//
// var foo bool
//
// ctx.FormData().Bool("foo", &foo, true)
func (form *FormData) Bool(key string, target *bool, defaultValue bool) *FormData {
return form.mustValue(key, target, defaultValue)
}
// MandatoryBool binds a form data value to a bool variable. It populates an
// error if the value is not bool, is empty, or the "key" does not exist.
//
// var foo bool
//
// ctx.FormData().MandatoryBool("foo", &foo)
func (form *FormData) MandatoryBool(key string, target *bool) *FormData {
return form.mustMandatoryValue(key, target)
}
// Int binds a form data value to an int variable. It populates an error if the
// value is not int.
//
// var foo int
//
// ctx.FormData().Int("foo", &foo, 2)
func (form *FormData) Int(key string, target *int, defaultValue int) *FormData {
return form.mustValue(key, target, defaultValue)
}
// MandatoryInt binds a form data value to an int variable. It populates an
// error if the value is not int, is empty, or the "key" does not exist.
//
// var foo int
//
// ctx.FormData().MandatoryInt("foo", &foo)
func (form *FormData) MandatoryInt(key string, target *int) *FormData {
return form.mustMandatoryValue(key, target)
}
// Float64 binds a form data value to a float64 variable. It populates an error
// if the value is not float64.
//
// var foo float64
//
// ctx.FormData().Float64("foo", &foo, 2.0)
func (form *FormData) Float64(key string, target *float64, defaultValue float64) *FormData {
return form.mustValue(key, target, defaultValue)
}
// MandatoryFloat64 binds a form data value to a float64 variable. It populates
// an error if the is not float64, is empty, or the "key" does not exist.
//
// var foo float64
//
// ctx.FormData().MandatoryFloat64("foo", &foo)
func (form *FormData) MandatoryFloat64(key string, target *float64) *FormData {
return form.mustMandatoryValue(key, target)
}
// Duration binds a form data value to a time.Duration variable. It populates
// an error if the form data value is not time.Duration.
//
// var foo time.Duration
//
// ctx.FormData().Duration("foo", &foo, time.Duration(2) * time.Second)
func (form *FormData) Duration(key string, target *time.Duration, defaultValue time.Duration) *FormData {
return form.mustValue(key, target, defaultValue)
}
// MandatoryDuration binds a form data value to a time.Duration variable. It
// populates an error if the value is not time.Duration, is empty, or the "key"
// does not exist.
//
// var foo time.Duration
//
// ctx.FormData().MandatoryDuration("foo", &foo)
func (form *FormData) MandatoryDuration(key string, target *time.Duration) *FormData {
return form.mustMandatoryValue(key, target)
}
// Custom helps to define a custom binding function for a form data value.
//
// var foo map[string]string
//
// ctx.FormData().Custom("foo", func(value string) error {
// if value == "" {
// foo = "bar"
//
// return nil
// }
//
// err := json.Unmarshal([]byte(value), &foo)
// if err != nil {
// return fmt.Errorf("unmarshal foo: %w", err)
// }
//
// return nil
// })
func (form *FormData) Custom(key string, assign func(value string) error) *FormData {
var value string
form.mustValue(key, &value, "")
err := assign(value)
if err != nil {
form.append(
fmt.Errorf("form value '%s' is invalid (got '%s', resulting to %w)", key, value, err),
)
}
return form
}
// MandatoryCustom helps to define a custom binding function for a form data
// value. It populates an error if the value is empty or the "key" does not
// exist.
//
// var foo map[string]string
//
// ctx.FormData().MandatoryCustom("foo", func(value string) error {
// err := json.Unmarshal([]byte(value), &foo)
// if err != nil {
// return fmt.Errorf("unmarshal foo: %w", err)
// }
//
// return nil
// })
func (form *FormData) MandatoryCustom(key string, assign func(value string) error) *FormData {
var value string
form.mustMandatoryValue(key, &value)
if value == "" {
return form
}
err := assign(value)
if err != nil {
form.append(
fmt.Errorf("form value '%s' is invalid (got '%s', resulting to %w)", key, value, err),
)
}
return form
}
// Path binds the absolute path of a form data file to a string variable.
//
// var path string
//
// ctx.FormData().Path("foo.txt", &path)
func (form *FormData) Path(filename string, target *string) *FormData {
return form.path(filename, target)
}
// MandatoryPath binds the absolute path ofa form data file to a string
// variable. It populates an error if the file does not exist.
//
// var path string
//
// ctx.FormData().MandatoryPath("foo.txt", &path)
func (form *FormData) MandatoryPath(filename string, target *string) *FormData {
return form.mandatoryPath(filename, target)
}
// Content binds the content of a form data file to a string variable.
//
// var content string
//
// ctx.FormData().Content("foo.txt", &content, "bar")
func (form *FormData) Content(filename string, target *string, defaultValue string) *FormData {
var path string
form.path(filename, &path)
if path == "" {
*target = defaultValue
return form
}
return form.readFile(path, filename, target)
}
// MandatoryContent binds the content of a form data file to a string variable.
// It populates an error if the file does not exist.
//
// var content string
//
// ctx.FormData().MandatoryContent("foo.txt", &content)
func (form *FormData) MandatoryContent(filename string, target *string) *FormData {
var path string
form.mandatoryPath(filename, &path)
if path == "" {
return form
}
return form.readFile(path, filename, target)
}
// Paths binds the absolute paths of form data files, according to a list of
// file extensions, to a string slice variable.
//
// var paths []string
//
// ctx.FormData().Paths([]string{".txt"}, &paths)
func (form *FormData) Paths(extensions []string, target *[]string) *FormData {
return form.paths(extensions, target)
}
// MandatoryPaths binds the absolute paths of form data files, according to a
// list of file extensions, to a string slice variable. It populates an error
// if there is no file for given file extensions.
//
// var paths []string
//
// ctx.FormData().MandatoryPaths([]string{".txt"}, &paths)
func (form *FormData) MandatoryPaths(extensions []string, target *[]string) *FormData {
form.paths(extensions, target)
if len(*target) > 0 {
return form
}
form.append(
fmt.Errorf("no form file found for extensions: %v", extensions),
)
return form
}
// paths binds the absolute paths of form data files, according to a list of
// file extensions, to a string slice variable.
func (form *FormData) paths(extensions []string, target *[]string) *FormData {
for filename, path := range form.files {
for _, ext := range extensions {
// See https://github.com/thecodingmachine/gotenberg/issues/228.
if strings.ToLower(filepath.Ext(filename)) == ext {
*target = append(*target, path)
}
}
}
// See https://github.com/thecodingmachine/gotenberg/issues/139.
sort.Strings(*target)
return form
}
// append adds an error to the list of errors.
func (form *FormData) append(err error) {
form.errors = multierr.Append(form.errors, err)
}
// mustValue binds the target interface with a form data value. If the value is
// empty or the "key" does not exist, it binds the default value. Currently,
// only the string, bool, int, float64 and time.Duration types are bindable.
func (form *FormData) mustValue(key string, target interface{}, defaultValue interface{}) *FormData {
val, ok := form.values[key]
if !ok || val[0] == "" {
switch t := (target).(type) {
case *string:
*t = defaultValue.(string)
case *bool:
*t = defaultValue.(bool)
case *int:
*t = defaultValue.(int)
case *float64:
*t = defaultValue.(float64)
case *time.Duration:
*t = defaultValue.(time.Duration)
default:
panic("target type not supported")
}
return form
}
return form.mustAssign(key, val[0], target)
}
// mustMandatoryValue binds the target interface with a form data value. It
// populates an error if the value is empty or the "key" does not exist.
// Currently, only the string, bool, int, float64 and time.Duration types are
// bindable.
func (form *FormData) mustMandatoryValue(key string, target interface{}) *FormData {
val, ok := form.values[key]
if !ok || val[0] == "" {
form.append(
fmt.Errorf("form value '%s' is required", key),
)
return form
}
form.mustAssign(key, val[0], target)
return form
}
// mustAssign parses the string value and tries to convert it to the target
// interface real type. Currently, only the string, bool, int, float64 and
// time.Duration types are bindable.
func (form *FormData) mustAssign(key, value string, target interface{}) *FormData {
var err error
switch t := (target).(type) {
case *string:
*t = value
case *bool:
*t, err = strconv.ParseBool(value)
case *int:
*t, err = strconv.Atoi(value)
case *float64:
*t, err = strconv.ParseFloat(value, 64)
case *time.Duration:
*t, err = time.ParseDuration(value)
default:
panic("target type not supported")
}
if err != nil {
form.append(
fmt.Errorf("form value '%s' is invalid (got '%s', resulting to %w)", key, value, err),
)
}
return form
}
// path binds the absolute path of a form data file to a string variable.
func (form *FormData) path(filename string, target *string) *FormData {
for name, path := range form.files {
if name == filename {
*target = path
return form
}
}
return form
}
// mandatoryPath binds the absolute path of a form data file to a string
// variable. It populates an error if the file does not exist.
func (form *FormData) mandatoryPath(filename string, target *string) *FormData {
form.path(filename, target)
if *target != "" {
return form
}
form.append(
fmt.Errorf("form file '%s' is required", filename),
)
return form
}
// readFile binds the content of a file to a string variable. It populates an
// error if it fails to read the file content.
func (form *FormData) readFile(path, filename string, target *string) *FormData {
b, err := ioutil.ReadFile(path)
if err != nil {
form.append(
fmt.Errorf("form file '%s' is invalid (%w)", filename, err),
)
return form
}
*target = string(b)
return form
}

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,576 @@
package api
import (
"bufio"
"bytes"
"context"
"encoding/json"
"errors"
"fmt"
"net/http"
"os"
"path/filepath"
"regexp"
"strconv"
"strings"
"time"
"github.com/google/uuid"
"github.com/hashicorp/go-retryablehttp"
"github.com/labstack/echo/v4"
"go.uber.org/zap"
)
// httpErrorHandler is the centralized HTTP error handler. It parses the error,
// returns either a response as "text/plain; charset=UTF-8" or, if a webhook
// client exists in the echo.Context, sends a request to the webhook error URL
// with a JSON body containing the trace, the status and the error message.
func httpErrorHandler(traceHeader string) echo.HTTPErrorHandler {
return func(err error, c echo.Context) {
parseError := func(err error) (int, string) {
echoErr, ok := err.(*echo.HTTPError)
if ok {
return echoErr.Code, http.StatusText(echoErr.Code)
}
if errors.Is(err, context.DeadlineExceeded) {
return http.StatusServiceUnavailable, http.StatusText(http.StatusServiceUnavailable)
}
var httpErr HTTPError
if errors.As(err, &httpErr) {
return httpErr.HTTPError()
}
// Default 500 status code.
return http.StatusInternalServerError, http.StatusText(http.StatusInternalServerError)
}
status, message := parseError(err)
logger := c.Get("logger").(*zap.Logger)
clientOrNil := c.Get("webhookClient")
// No webhook client, meaning we can send the error as a response.
if clientOrNil == nil {
c.Response().Header().Add(echo.HeaderContentType, echo.MIMETextPlainCharsetUTF8)
err = c.String(status, message)
if err != nil {
logger.Error(fmt.Sprintf("send error response: %s", err.Error()))
}
return
}
// We have to send the error to the webhook.
client := clientOrNil.(*webhookClient)
body := struct {
Status int `json:"status"`
Message string `json:"message"`
}{
Status: status,
Message: message,
}
b, err := json.Marshal(body)
if err != nil {
logger.Error(fmt.Sprintf("marshal JSON: %s", err.Error()))
return
}
headers := map[string]string{
echo.HeaderContentType: echo.MIMEApplicationJSONCharsetUTF8,
traceHeader: c.Get("trace").(string),
}
err = client.send(bytes.NewReader(b), headers, true)
if err != nil {
logger.Error(fmt.Sprintf("send error response to webhook: %s", err.Error()))
}
}
}
// latencyMiddleware sets the start time in the echo.Context under "startTime".
// Its value will be used later to calculate a request latency.
//
// startTime := c.Get("startTime").(time.Time)
func latencyMiddleware() echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
// First piece for calculating the latency.
startTime := time.Now()
c.Set("startTime", startTime)
// Call the next middleware in the chain.
return next(c)
}
}
}
// rootPathMiddleware sets the root path in the echo.Context under "rootPath".
// Its value may be used to skip a middleware execution based on a request
// URI.
//
// rootPath := c.Get("rootPath").(string)
// healthURI := fmt.Sprintf("%shealth", rootPath)
//
// // Skip the middleware if health check URI.
// if c.Request().RequestURI == healthURI {
// // Call the next middleware in the chain.
// return next(c)
// }
func rootPathMiddleware(rootPath string) echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
c.Set("rootPath", rootPath)
// Call the next middleware in the chain.
return next(c)
}
}
}
// traceMiddleware sets the request identifier in the echo.Context under
// "trace". Its value is either retrieved from the trace header or generated if
// the header is not present / its value is empty.
//
// trace := c.Get("trace").(string)
func traceMiddleware(header string) echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
// Get or create the request identifier.
trace := c.Request().Header.Get(header)
if trace == "" {
trace = uuid.New().String()
}
c.Set("trace", trace)
c.Response().Header().Add(header, trace)
// Call the next middleware in the chain.
return next(c)
}
}
}
// loggerMiddleware sets the logger in the echo.Context under "logger" and logs
// a request result (but does not log a webhook call result, which is the job
// of the webhookClient).
//
// logger := c.Get("logger").(*zap.Logger)
func loggerMiddleware(logger *zap.Logger, skipHealthRouteLogging bool) echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
startTime := c.Get("startTime").(time.Time)
trace := c.Get("trace").(string)
// Create the request logger and add it to our locals.
reqLogger := logger.With(zap.String("trace", trace))
c.Set("logger", reqLogger)
// Call the next middleware in the chain.
err := next(c)
if err != nil {
c.Error(err)
}
if skipHealthRouteLogging {
rootPath := c.Get("rootPath").(string)
healthURI := fmt.Sprintf("%shealth", rootPath)
if c.Request().RequestURI == healthURI {
return nil
}
}
// Last piece for calculating the latency.
finishTime := time.Now()
// Now, let's log!
fields := make([]zap.Field, 12)
fields[0] = zap.String("remote_ip", c.RealIP())
fields[1] = zap.String("host", c.Request().Host)
fields[2] = zap.String("uri", c.Request().RequestURI)
fields[3] = zap.String("method", c.Request().Method)
fields[4] = zap.String("path", func() string {
path := c.Request().URL.Path
if path == "" {
path = "/"
}
return path
}())
fields[5] = zap.String("referer", c.Request().Referer())
fields[6] = zap.String("user_agent", c.Request().UserAgent())
fields[7] = zap.Int("status", c.Response().Status)
fields[8] = zap.Int64("latency", int64(finishTime.Sub(startTime)))
fields[9] = zap.String("latency_human", finishTime.Sub(startTime).String())
fields[10] = zap.Int64("bytes_in", c.Request().ContentLength)
fields[11] = zap.Int64("bytes_out", c.Response().Size)
if err != nil {
reqLogger.Error(err.Error(), fields...)
} else {
reqLogger.Info("request handled", fields...)
}
return nil
}
}
}
type contextMiddlewareConfig struct {
traceHeader string
timeout struct {
process time.Duration
write time.Duration
}
webhook struct {
allowList *regexp.Regexp
denyList *regexp.Regexp
errorAllowList *regexp.Regexp
errorDenyList *regexp.Regexp
maxRetry int
retryMinWait time.Duration
retryMaxWait time.Duration
disable bool
}
}
// contextMiddleware handles the result of a "multipart/form-data" request. If
// a webhook URL is present in the headers, exit early and process the result
// in a goroutine.
func contextMiddleware(cfg contextMiddlewareConfig) echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
webhookURL := c.Request().Header.Get("Gotenberg-Webhook-Url")
logger := c.Get("logger").(*zap.Logger).With(zap.Bool("webhook", webhookURL != ""))
// We create a context with a timeout so that underlying processes are
// able to stop early and handle correctly a timeout scenario.
ctx, cancel, err := newContext(c, logger, cfg.timeout.process)
if err != nil {
cancel()
return fmt.Errorf("create request context: %w", err)
}
c.Set("context", ctx)
// Helper function for retrieving/creating the output filename.
outputFilename := func(outputPath string) string {
filename := c.Request().Header.Get("Gotenberg-Output-Filename")
if filename == "" {
return filepath.Base(outputPath)
}
return fmt.Sprintf("%s%s", filename, filepath.Ext(outputPath))
}
if webhookURL == "" {
defer cancel()
// No webhook URL, call the next middleware in the chain.
err := next(c)
if err != nil {
return err
}
// No error, let's build the output file.
outputPath, err := ctx.buildOutputFile()
if err != nil {
return fmt.Errorf("build output file: %w", err)
}
// Send the output file.
err = c.Attachment(outputPath, outputFilename(outputPath))
if err != nil {
return fmt.Errorf("send response: %w", err)
}
return nil
}
// Ok, we got a webhook URL.
if cfg.webhook.disable {
// The client requested the webhook feature, but it has been
// disabled. Let's tell the client about that.
cancel()
return WrapError(
errors.New("webhook feature requested but it is disabled"),
NewSentinelHTTPError(http.StatusForbidden, "Invalid 'Gotenberg-Webhook-Url' header: feature is disabled"),
)
}
// Do we have a webhook error URL in case of... error?
webhookErrorURL := c.Request().Header.Get("Gotenberg-Webhook-Error-Url")
if webhookErrorURL == "" {
cancel()
return WrapError(
errors.New("empty webhook error URL"),
NewSentinelHTTPError(http.StatusBadRequest, "Invalid 'Gotenberg-Webhook-Error-Url' header: empty value or header not provided"),
)
}
// Let's check if the webhook URLs are acceptable according to our
// allowed/denied lists.
filter := func(URL, header string, allowList, denyList *regexp.Regexp) error {
if !allowList.MatchString(URL) {
return WrapError(
fmt.Errorf("'%s' does not match the expression from the allowed list", URL),
NewSentinelHTTPError(
http.StatusForbidden,
fmt.Sprintf("Invalid '%s' header value: '%s' does not match the authorized URLs", header, URL),
),
)
}
if denyList.String() != "" && denyList.MatchString(URL) {
return WrapError(
fmt.Errorf("'%s' matches the expression from the denied list", URL),
NewSentinelHTTPError(
http.StatusForbidden,
fmt.Sprintf("Invalid '%s' header value: '%s' does not match the authorized URLs", header, URL),
),
)
}
return nil
}
err = filter(webhookURL, "Gotenberg-Webhook-Url", cfg.webhook.allowList, cfg.webhook.denyList)
if err != nil {
cancel()
return fmt.Errorf("filter webhook URL: %w", err)
}
err = filter(webhookErrorURL, "Gotenberg-Webhook-Error-Url", cfg.webhook.errorAllowList, cfg.webhook.errorDenyList)
if err != nil {
cancel()
return fmt.Errorf("filter webhook error URL: %w", err)
}
// Let's check the HTTP methods for calling the webhook URLs.
methodFromHeader := func(header string) (string, error) {
method := c.Request().Header.Get(header)
if method == "" {
return http.MethodPost, nil
}
method = strings.ToUpper(method)
switch method {
case http.MethodPost:
return method, nil
case http.MethodPatch:
return method, nil
case http.MethodPut:
return method, nil
}
return "", WrapError(
fmt.Errorf("webhook method '%s' is not '%s', '%s' or '%s'", method, http.MethodPost, http.MethodPatch, http.MethodPut),
NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("Invalid '%s' header value: expected '%s', '%s' or '%s', but got '%s'", header, http.MethodPost, http.MethodPatch, http.MethodPut, method),
),
)
}
webhookMethod, err := methodFromHeader("Gotenberg-Webhook-Method")
if err != nil {
cancel()
return fmt.Errorf("get method to use for webhook: %w", err)
}
webhookErrorMethod, err := methodFromHeader("Gotenberg-Webhook-Error-Method")
if err != nil {
cancel()
return fmt.Errorf("get method to use for webhook error: %w", err)
}
// What about extra HTTP headers?
var extraHTTPHeaders map[string]string
extraHTTPHeadersJSON := c.Request().Header.Get("Gotenberg-Webhook-Extra-Http-Headers")
if extraHTTPHeadersJSON != "" {
err = json.Unmarshal([]byte(extraHTTPHeadersJSON), &extraHTTPHeaders)
if err != nil {
cancel()
return WrapError(
fmt.Errorf("unmarshal webhook extra HTTP headers: %w", err),
NewSentinelHTTPError(http.StatusBadRequest, fmt.Sprintf("Invalid 'Gotenberg-Webhook-Extra-Http-Headers' header value: %s", err.Error())),
)
}
}
client := &webhookClient{
url: webhookURL,
method: webhookMethod,
errorURL: webhookErrorURL,
errorMethod: webhookErrorMethod,
extraHTTPHeaders: extraHTTPHeaders,
startTime: c.Get("startTime").(time.Time),
client: &retryablehttp.Client{
HTTPClient: &http.Client{
Timeout: cfg.timeout.write,
},
RetryMax: cfg.webhook.maxRetry,
RetryWaitMin: cfg.webhook.retryMinWait,
RetryWaitMax: cfg.webhook.retryMaxWait,
Logger: leveledLogger{
logger: logger,
},
CheckRetry: retryablehttp.DefaultRetryPolicy,
Backoff: retryablehttp.DefaultBackoff,
},
logger: logger,
}
c.Set("webhookClient", client)
// As a webhook URL has been given, we handle the request in a
// goroutine and return immediately.
go func() {
defer cancel()
// Call the next middleware in the chain.
err := next(c)
if err != nil {
// The process failed for whatever reason. Let's send the
// details to the webhook.
ctx.Log().Error(err.Error())
c.Error(err)
return
}
// No error, let's get build the output file.
outputPath, err := ctx.buildOutputFile()
if err != nil {
ctx.Log().Error(fmt.Sprintf("build output file: %s", err))
c.Error(err)
return
}
outputFile, err := os.Open(outputPath)
if err != nil {
ctx.Log().Error(fmt.Sprintf("open output file: %s", err))
c.Error(err)
return
}
defer func() {
err := outputFile.Close()
if err != nil {
ctx.Log().Error(fmt.Sprintf("close output file: %s", err))
}
}()
fileHeader := make([]byte, 512)
_, err = outputFile.Read(fileHeader)
if err != nil {
ctx.Log().Error(fmt.Sprintf("read header of output file: %s", err))
c.Error(err)
return
}
fileStat, err := outputFile.Stat()
if err != nil {
ctx.Log().Error(fmt.Sprintf("get stat from output file: %s", err))
c.Error(err)
return
}
_, err = outputFile.Seek(0, 0)
if err != nil {
ctx.Log().Error(fmt.Sprintf("reset output file reader: %s", err))
c.Error(err)
return
}
headers := map[string]string{
echo.HeaderContentDisposition: fmt.Sprintf("attachement; filename=%q", outputFilename(outputPath)),
echo.HeaderContentType: http.DetectContentType(fileHeader),
echo.HeaderContentLength: strconv.FormatInt(fileStat.Size(), 10),
cfg.traceHeader: c.Get("trace").(string),
}
// Send the output file to the webhook.
err = client.send(bufio.NewReader(outputFile), headers, false)
if err != nil {
ctx.Log().Error(fmt.Sprintf("send output file to webhook: %s", err))
c.Error(err)
}
}()
return c.NoContent(http.StatusNoContent)
}
}
}
// timeoutMiddleware manages hard timeout scenarios, i.e., when a route handler
// fails to timeout as expected.
func timeoutMiddleware(hardTimeout time.Duration) echo.MiddlewareFunc {
return func(next echo.HandlerFunc) echo.HandlerFunc {
return func(c echo.Context) error {
logger := c.Get("logger").(*zap.Logger)
// Define a hard timeout if the route handler fails to timeout as
// expected.
hardTimeoutCtx, hardTimeoutCancel := context.WithTimeout(
context.Background(),
hardTimeout,
)
defer hardTimeoutCancel()
errChan := make(chan error, 1)
go func() {
// In case of hard timeout, a panic may occur.
// This deferred function allows us to recover from such scenarios.
defer func() {
if r := recover(); r != nil {
logger.Debug(fmt.Sprintf("recovering from a panic (possible cause being a hard timeout): %s", r))
}
}()
// Call the next middleware in the chain.
errChan <- next(c)
}()
select {
case err := <-errChan:
return err
case <-hardTimeoutCtx.Done():
logger.Debug("hard timeout as the route handler did not timeout as expected")
return fmt.Errorf("hard timeout: %w", hardTimeoutCtx.Err())
}
}
}
}

File diff suppressed because it is too large Load Diff

139
pkg/modules/api/webhook.go Normal file
View File

@@ -0,0 +1,139 @@
package api
import (
"fmt"
"io"
"strconv"
"time"
"github.com/hashicorp/go-retryablehttp"
"github.com/labstack/echo/v4"
"go.uber.org/zap"
)
// webhookClient gathers all the data required to send a request to a webhook.
type webhookClient struct {
url string
method string
errorURL string
errorMethod string
extraHTTPHeaders map[string]string
startTime time.Time
client *retryablehttp.Client
logger *zap.Logger
}
// send call the webhook either to send the success response or the error response.
func (webhook webhookClient) send(body io.Reader, headers map[string]string, erroed bool) error {
URL := webhook.url
if erroed {
URL = webhook.errorURL
}
method := webhook.method
if erroed {
method = webhook.errorMethod
}
req, err := retryablehttp.NewRequest(method, URL, body)
if err != nil {
return fmt.Errorf("create '%s' request to '%s': %w", method, URL, err)
}
req.Header.Set("User-Agent", "Gotenberg")
// Extra HTTP headers are the custom headers from the user.
for key, value := range webhook.extraHTTPHeaders {
req.Header.Set(key, value)
}
// Middleware caller's headers > extra HTTP headers from the user.
contentLength, ok := headers[echo.HeaderContentLength]
if ok {
// Golang "http" package should automatically calculate the size of the
// body. But, when using a buffered file reader, it does not work.
// Worse, the "Content-Length" header is also removed. Therefore, in
// order to keep this valuable information, we have to trust the caller
// by reading the value of the "Content-Length" entry and set it as the
// content length of the request. It's kinda sub-optimal, but hey, at
// least it works.
bodySize, err := strconv.ParseInt(contentLength, 10, 64)
if err != nil {
return fmt.Errorf("parse content length entry: %w", err)
}
req.ContentLength = bodySize
}
for key, value := range headers {
req.Header.Set(key, value)
}
resp, err := webhook.client.Do(req)
if err != nil {
return fmt.Errorf("send '%s' request to '%s': %w", method, URL, err)
}
defer func() {
err := resp.Body.Close()
if err != nil {
webhook.logger.Error(fmt.Sprintf("close response body from '%s': %s", URL, err))
}
}()
// Last piece for calculating the latency.
finishTime := time.Now()
// Now let's log!
fields := make([]zap.Field, 5)
fields[0] = zap.String("webhook_url", URL)
fields[1] = zap.String("method", method)
fields[2] = zap.Int64("latency", int64(finishTime.Sub(webhook.startTime)))
fields[3] = zap.String("latency_human", finishTime.Sub(webhook.startTime).String())
fields[4] = zap.Int64("bytes_out", req.ContentLength)
if erroed {
webhook.logger.Warn("request to webhook with error details handled", fields...)
return nil
}
webhook.logger.Info("request to webhook handled", fields...)
return nil
}
// leveledLogger is wrapper around a zap.Logger which is used by the
// retryablehttp.Client.
type leveledLogger struct {
logger *zap.Logger
}
// Error logs a message at error level using the wrapped zap.Logger.
func (leveled leveledLogger) Error(msg string, keysAndValues ...interface{}) {
leveled.logger.Error(fmt.Sprintf("%s: %+v", msg, keysAndValues))
}
// Warn logs a message at warning level using the wrapped zap.Logger.
func (leveled leveledLogger) Warn(msg string, keysAndValues ...interface{}) {
leveled.logger.Warn(fmt.Sprintf("%s: %+v", msg, keysAndValues))
}
// Info logs a message at info level using the wrapped zap.Logger.
func (leveled leveledLogger) Info(msg string, keysAndValues ...interface{}) {
leveled.logger.Info(fmt.Sprintf("%s: %+v", msg, keysAndValues))
}
// Debug logs a message at debug level using the wrapped zap.Logger.
func (leveled leveledLogger) Debug(msg string, keysAndValues ...interface{}) {
leveled.logger.Debug(fmt.Sprintf("%s: %+v", msg, keysAndValues))
}
// Interface guards.
var (
_ retryablehttp.LeveledLogger = (*leveledLogger)(nil)
)

View File

@@ -0,0 +1,23 @@
package api
import (
"testing"
"go.uber.org/zap"
)
func TestLeveledLogger_Error(t *testing.T) {
leveledLogger{logger: zap.NewNop()}.Error("foo")
}
func TestLeveledLogger_Warn(t *testing.T) {
leveledLogger{logger: zap.NewNop()}.Warn("foo")
}
func TestLeveledLogger_Info(t *testing.T) {
leveledLogger{logger: zap.NewNop()}.Info("foo")
}
func TestLeveledLogger_Debug(t *testing.T) {
leveledLogger{logger: zap.NewNop()}.Debug("foo")
}

View File

@@ -0,0 +1,477 @@
package chromium
import (
"context"
"errors"
"fmt"
"io/ioutil"
"os"
"regexp"
"strings"
"time"
"github.com/chromedp/cdproto/network"
"github.com/chromedp/cdproto/page"
"github.com/chromedp/chromedp"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
flag "github.com/spf13/pflag"
"go.uber.org/zap"
)
func init() {
gotenberg.MustRegisterModule(Chromium{})
}
var (
// ErrURLNotAuthorized happens if a URL is not acceptable according to the
// allowed/denied lists.
ErrURLNotAuthorized = errors.New("URL not authorized")
// ErrInvalidPrinterSettings happens if the Options have one or more
// aberrant values.
ErrInvalidPrinterSettings = errors.New("invalid printer settings")
// ErrPageRangesSyntaxError happens if the Options have an invalid page
// ranges.
ErrPageRangesSyntaxError = errors.New("page ranges syntax error")
// ErrRpccMessageTooLarge happens when the messages received by
// ChromeDevTools are larger than 100 MB.
ErrRpccMessageTooLarge = errors.New("rpcc message too large")
)
// Chromium is a module which provides both an API and routes for converting
// HTML document to PDF.
type Chromium struct {
binPath string
engine gotenberg.PDFEngine
userAgent string
incognito bool
ignoreCertificateErrors bool
allowList *regexp.Regexp
denyList *regexp.Regexp
disableRoutes bool
}
// Options are the available options for converting HTML document to PDF.
type Options struct {
// WaitDelay is the duration to wait when loading an HTML document before
// converting it to PDF.
// Optional.
WaitDelay time.Duration
// WaitWindowStatus is the window.status value to wait for before
// converting an HTML document to PDF.
// Optional.
WaitWindowStatus string
// ExtraHTTPHeaders are the HTTP headers to send by Chromium while loading
// the HTML document.
// Optional.
ExtraHTTPHeaders map[string]string
// Landscape sets the paper orientation.
// Optional.
Landscape bool
// PrintBackground prints the background graphics.
// Optional.
PrintBackground bool
// Scale is the scale of the page rendering.
// Optional.
Scale float64
// PaperWidth is the paper width, in inches.
// Optional.
PaperWidth float64
// PaperHeight is the paper height, in inches.
// Optional.
PaperHeight float64
// MarginTop is the top margin, in inches.
// Optional.
MarginTop float64
// MarginBottom is the bottom margin, in inches.
// Optional.
MarginBottom float64
// MarginLeft is the left margin, in inches.
// Optional.
MarginLeft float64
// MarginRight is the right margin, in inches.
// Optional.
MarginRight float64
// Page ranges to print, e.g., '1-5, 8, 11-13'. Empty means all pages.
// Optional.
PageRanges string
// HeaderTemplate is the HTML template of the header. It should be valid
// HTML markup with following classes used to inject printing values into
// them:
// - date: formatted print date
// - title: document title
// - url: document location
// - pageNumber: current page number
// - totalPages: total pages in the document
// For example, <span class=title></span> would generate span containing
// the title.
// Optional.
HeaderTemplate string
// FooterTemplate is the HTML template of the footer. It should use the
// same format as the HeaderTemplate.
// Optional.
FooterTemplate string
// PreferCSSPageSize defines whether to prefer page size as defined by CSS.
// If false, the content will be scaled to fit the paper size.
// Optional.
PreferCSSPageSize bool
}
// DefaultOptions returns the default values for Options.
func DefaultOptions() Options {
return Options{
WaitDelay: 0,
WaitWindowStatus: "",
ExtraHTTPHeaders: nil,
Landscape: false,
PrintBackground: false,
Scale: 1.0,
PaperWidth: 8.5,
PaperHeight: 11,
MarginTop: 0.39,
MarginBottom: 0.39,
MarginLeft: 0.39,
MarginRight: 0.39,
PageRanges: "",
HeaderTemplate: "<html><head></head><body></body></html>",
FooterTemplate: "<html><head></head><body></body></html>",
PreferCSSPageSize: false,
}
}
// API helps to interact with Chromium for converting HTML documents to PDF.
type API interface {
PDF(ctx context.Context, logger *zap.Logger, URL, outputPath string, options Options) error
}
// Provider is a module interface which exposes a method for creating an API
// for other modules.
//
// func (m *YourModule) Provision(ctx *gotenberg.Context) error {
// provider, _ := ctx.Module(new(chromium.Provider))
// chromium, _ := provider.(chromium.Provider).Chromium()
// }
type Provider interface {
Chromium() (API, error)
}
// Descriptor returns a Chromium's module descriptor.
func (mod Chromium) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "chromium",
FlagSet: func() *flag.FlagSet {
fs := flag.NewFlagSet("chromium", flag.ExitOnError)
fs.String("chromium-user-agent", "", "Override the default User-Agent header")
fs.Bool("chromium-incognito", false, "Start Chromium with incognito mode")
fs.Bool("chromium-ignore-certificate-errors", false, "Ignore the certificate errors")
fs.String("chromium-allow-list", "", "Set the allowed URLs for Chromium using a regular expression")
fs.String("chromium-deny-list", "", "Set the denied URLs for Chromium using a regular expression")
fs.Bool("chromium-disable-routes", false, "Disable the routes")
return fs
}(),
New: func() gotenberg.Module { return new(Chromium) },
}
}
// Provision sets the module properties.
func (mod *Chromium) Provision(ctx *gotenberg.Context) error {
flags := ctx.ParsedFlags()
mod.ignoreCertificateErrors = flags.MustBool("chromium-ignore-certificate-errors")
mod.allowList = flags.MustRegexp("chromium-allow-list")
mod.denyList = flags.MustRegexp("chromium-deny-list")
mod.disableRoutes = flags.MustBool("chromium-disable-routes")
binPath, ok := os.LookupEnv("CHROMIUM_BIN_PATH")
if !ok {
return errors.New("CHROMIUM_BIN_PATH environment variable is not set")
}
mod.binPath = binPath
provider, err := ctx.Module(new(gotenberg.PDFEngineProvider))
if err != nil {
return fmt.Errorf("get PDF engine provider: %w", err)
}
engine, err := provider.(gotenberg.PDFEngineProvider).PDFEngine()
if err != nil {
return fmt.Errorf("get PDF engine: %w", err)
}
mod.engine = engine
return nil
}
// Validate validates the module properties.
func (mod Chromium) Validate() error {
_, err := os.Stat(mod.binPath)
if os.IsNotExist(err) {
return fmt.Errorf("chromium binary path does not exist: %w", err)
}
return nil
}
// Chromium returns an API for interacting with Chromium for converting HTML
// documents to PDF.
func (mod Chromium) Chromium() (API, error) {
return mod, nil
}
// Routes returns the API routes.
func (mod Chromium) Routes() ([]api.MultipartFormDataRoute, error) {
if mod.disableRoutes {
return nil, nil
}
return []api.MultipartFormDataRoute{
convertURLRoute(mod, mod.engine),
convertHTMLRoute(mod, mod.engine),
convertMarkdownRoute(mod, mod.engine),
}, nil
}
// PDF converts a URL to PDF. It creates a dedicated Chromium instance.
// Substantial calls to this method may increase CPU and memory usage
// drastically. In such a scenario, the given context may also be done before
// the end of the conversion.
func (mod Chromium) PDF(ctx context.Context, logger *zap.Logger, URL, outputPath string, options Options) error {
userProfileDirPath := gotenberg.NewDirPath()
args := append(chromedp.DefaultExecAllocatorOptions[:],
chromedp.ExecPath(mod.binPath),
chromedp.NoSandbox,
// See:
// https://github.com/puppeteer/puppeteer/issues/661
// https://github.com/puppeteer/puppeteer/issues/2410
chromedp.Flag("font-render-hinting", "none"),
chromedp.UserDataDir(userProfileDirPath),
)
if mod.userAgent != "" {
args = append(args, chromedp.UserAgent(mod.userAgent))
}
if mod.incognito {
args = append(args, chromedp.Flag("incognito", mod.incognito))
}
if mod.ignoreCertificateErrors {
args = append(args, chromedp.IgnoreCertErrors)
}
allocatorCtx, cancel := chromedp.NewExecAllocator(ctx, args...)
defer cancel()
taskCtx, cancel := chromedp.NewContext(allocatorCtx)
defer cancel()
if !mod.allowList.MatchString(URL) {
return fmt.Errorf("'%s' does not match the expression from the allowed list: %w", URL, ErrURLNotAuthorized)
}
if mod.denyList.String() != "" && mod.denyList.MatchString(URL) {
return fmt.Errorf("'%s' matches the expression from the denied list: %w", URL, ErrURLNotAuthorized)
}
printToPDF := func(URL string, options Options, result *[]byte) chromedp.Tasks {
return chromedp.Tasks{
network.Enable(),
chromedp.ActionFunc(func(ctx context.Context) error {
if len(options.ExtraHTTPHeaders) == 0 {
logger.Debug("no extra HTTP headers")
return nil
}
logger.Debug(fmt.Sprintf("extra HTTP headers: %+v", options.ExtraHTTPHeaders))
headers := make(network.Headers, len(options.ExtraHTTPHeaders))
for key, value := range options.ExtraHTTPHeaders {
headers[key] = value
}
err := network.SetExtraHTTPHeaders(headers).Do(ctx)
if err == nil {
return nil
}
return fmt.Errorf("set extra HTTP headers: %w", err)
}),
chromedp.ActionFunc(func(ctx context.Context) error {
logger.Debug(fmt.Sprintf("navigate to '%s'", URL))
_, _, _, err := page.Navigate(URL).Do(ctx)
if err != nil {
return fmt.Errorf("navigate to '%s': %w", URL, err)
}
err = runBatch(
ctx,
waitForEventDomContentEventFired(ctx, logger),
waitForEventLoadEventFired(ctx, logger),
waitForEventNetworkIdle(ctx, logger),
waitForEventLoadingFinished(ctx, logger),
)
if err == nil {
return nil
}
return fmt.Errorf("wait for events: %w", err)
}),
chromedp.ActionFunc(func(ctx context.Context) error {
if options.WaitDelay > 0 {
// We wait for a given amount of time so that JavaScript
// scripts have a chance to finish before printing the page
// to PDF.
logger.Debug(fmt.Sprintf("wait '%s' before print", options.WaitDelay))
select {
case <-ctx.Done():
return fmt.Errorf("wait delay: %w", ctx.Err())
case <-time.After(options.WaitDelay):
return nil
}
}
return nil
}),
chromedp.ActionFunc(func(ctx context.Context) error {
if options.WaitWindowStatus == "" {
return nil
}
// We wait until the evaluation of
// "window.status === options.WaitWindowStatus" is true or
// until the context is done.
logger.Debug(fmt.Sprintf("wait for window.status === '%s' before print", options.WaitWindowStatus))
ticker := time.NewTicker(time.Duration(100) * time.Millisecond)
for {
select {
case <-ctx.Done():
ticker.Stop()
return fmt.Errorf("wait for window.status === '%s': %w", options.WaitWindowStatus, ctx.Err())
case <-ticker.C:
var ok bool
evaluate := chromedp.Evaluate(fmt.Sprintf("window.status === '%s'", options.WaitWindowStatus), &ok)
err := evaluate.Do(ctx)
if err != nil {
return fmt.Errorf("evaluate: %w", err)
}
if ok {
ticker.Stop()
return nil
}
continue
}
}
}),
chromedp.ActionFunc(func(ctx context.Context) error {
printToPDF := page.PrintToPDF().
WithLandscape(options.Landscape).
WithPrintBackground(options.PrintBackground).
WithScale(options.Scale).
WithPaperWidth(options.PaperWidth).
WithPaperHeight(options.PaperHeight).
WithMarginTop(options.MarginTop).
WithMarginBottom(options.MarginBottom).
WithMarginLeft(options.MarginLeft).
WithMarginRight(options.MarginRight).
WithIgnoreInvalidPageRanges(false).
WithPageRanges(options.PageRanges).
WithDisplayHeaderFooter(true).
WithHeaderTemplate(options.HeaderTemplate).
WithFooterTemplate(options.FooterTemplate).
WithPreferCSSPageSize(options.PreferCSSPageSize)
logger.Debug(fmt.Sprintf("print to PDF with: %+v", printToPDF))
data, _, err := printToPDF.Do(ctx)
if err != nil {
return fmt.Errorf("print to PDF: %w", err)
}
*result = data
return nil
}),
}
}
var buffer []byte
err := chromedp.Run(taskCtx, printToPDF(URL, options, &buffer))
// Always remove the user profile directory created by Chromium.
go func() {
logger.Debug(fmt.Sprintf("remove user profile directory '%s'", userProfileDirPath))
err := os.RemoveAll(userProfileDirPath)
if err != nil {
logger.Error(fmt.Sprintf("remove user profile directory: %s", err))
}
}()
if err != nil {
errMessage := err.Error()
if strings.Contains(errMessage, "Show invalid printer settings error (-32000)") {
return ErrInvalidPrinterSettings
}
if strings.Contains(errMessage, "Page range syntax error") {
return ErrPageRangesSyntaxError
}
if strings.Contains(errMessage, "rpcc: message too large") {
return ErrRpccMessageTooLarge
}
return fmt.Errorf("chromium PDF: %w", err)
}
err = ioutil.WriteFile(outputPath, buffer, 0600)
if err != nil {
return fmt.Errorf("write result to output path: %w", err)
}
return nil
}
// Interface guards.
var (
_ gotenberg.Module = (*Chromium)(nil)
_ gotenberg.Provisioner = (*Chromium)(nil)
_ gotenberg.Validator = (*Chromium)(nil)
_ api.MultipartFormDataRouter = (*Chromium)(nil)
_ API = (*Chromium)(nil)
_ Provider = (*Chromium)(nil)
)

View File

@@ -0,0 +1,366 @@
package chromium
import (
"context"
"errors"
"io/ioutil"
"os"
"reflect"
"regexp"
"testing"
"time"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
type ProtoModule struct {
descriptor func() gotenberg.ModuleDescriptor
}
func (mod ProtoModule) Descriptor() gotenberg.ModuleDescriptor {
return mod.descriptor()
}
type ProtoAPI struct {
pdf func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error
}
func (mod ProtoAPI) PDF(ctx context.Context, logger *zap.Logger, URL, outputPath string, options Options) error {
return mod.pdf(ctx, logger, URL, outputPath, options)
}
type ProtoPDFEngineProvider struct {
ProtoModule
pdfEngine func() (gotenberg.PDFEngine, error)
}
func (mod ProtoPDFEngineProvider) PDFEngine() (gotenberg.PDFEngine, error) {
return mod.pdfEngine()
}
type ProtoPDFEngine struct {
merge func(_ context.Context, _ *zap.Logger, _ []string, _ string) error
convert func(_ context.Context, _ *zap.Logger, _, _, _ string) error
}
func (mod ProtoPDFEngine) Merge(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
return mod.merge(ctx, logger, inputPaths, outputPath)
}
func (mod ProtoPDFEngine) Convert(ctx context.Context, logger *zap.Logger, format, inputPath, outputPath string) error {
return mod.convert(ctx, logger, format, inputPath, outputPath)
}
func TestDefaultOptions(t *testing.T) {
actual := DefaultOptions()
notExpect := Options{}
if reflect.DeepEqual(actual, notExpect) {
t.Errorf("expected %v and got identical %v", actual, notExpect)
}
}
func TestChromium_Descriptor(t *testing.T) {
descriptor := Chromium{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(Chromium))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestChromium_Provision(t *testing.T) {
for i, tc := range []struct {
ctx *gotenberg.Context
expectErr bool
}{
{
ctx: func() *gotenberg.Context {
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(Chromium).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoPDFEngineProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "bar", New: func() gotenberg.Module { return mod }}
}
mod.pdfEngine = func() (gotenberg.PDFEngine, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(Chromium).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoPDFEngineProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "bar", New: func() gotenberg.Module { return mod }}
}
mod.pdfEngine = func() (gotenberg.PDFEngine, error) {
return struct{ ProtoPDFEngine }{}, nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(Chromium).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
},
} {
mod := new(Chromium)
err := mod.Provision(tc.ctx)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestChromium_Validate(t *testing.T) {
for i, tc := range []struct {
binPath string
expectErr bool
}{
{
expectErr: true,
},
{
binPath: "/foo",
expectErr: true,
},
{
binPath: os.Getenv("CHROMIUM_BIN_PATH"),
},
} {
mod := new(Chromium)
mod.binPath = tc.binPath
err := mod.Validate()
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestChromium_Chromium(t *testing.T) {
mod := new(Chromium)
_, err := mod.Chromium()
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestChromium_Routes(t *testing.T) {
for i, tc := range []struct {
expectRoutes int
disableRoutes bool
}{
{
expectRoutes: 3,
},
{
disableRoutes: true,
},
} {
mod := new(Chromium)
mod.disableRoutes = tc.disableRoutes
routes, err := mod.Routes()
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
if tc.expectRoutes != len(routes) {
t.Errorf("test %d: expected %d routes but got %d", i, tc.expectRoutes, len(routes))
}
}
}
func TestChromium_PDF(t *testing.T) {
for i, tc := range []struct {
timeout time.Duration
cancel context.CancelFunc
URL string
options Options
userAgent string
incognito bool
ignoreCertificateErrors bool
allowList *regexp.Regexp
denyList *regexp.Regexp
expectErr bool
}{
{
URL: "https://google.com",
allowList: regexp.MustCompile("https://google.fr"),
expectErr: true,
},
{
URL: "https://google.com",
denyList: regexp.MustCompile("https://google.com"),
expectErr: true,
},
{
URL: "",
options: Options{
ExtraHTTPHeaders: map[string]string{
"foo": "bar",
},
},
expectErr: true,
},
{
URL: "https://google.com",
options: Options{
WaitDelay: time.Duration(1) * time.Nanosecond,
},
},
{
timeout: time.Duration(3) * time.Second,
URL: "file:///tests/test/testdata/chromium/html/sample2/index.html",
options: Options{
WaitWindowStatus: "foo",
},
expectErr: true,
},
{
timeout: time.Duration(3) * time.Second,
URL: "file:///tests/test/testdata/chromium/html/sample2/index.html",
options: Options{
WaitWindowStatus: "ready",
},
},
{
URL: "https://google.com",
options: Options{
MarginBottom: 100,
},
expectErr: true,
},
{
URL: "https://google.com",
options: Options{
PageRanges: "foo",
},
expectErr: true,
},
{
URL: "https://google.com",
userAgent: "foo",
incognito: true,
ignoreCertificateErrors: true,
},
{
URL: "file:///tests/test/testdata/chromium/html/sample1/index.html",
},
{
URL: "https://google.com",
options: Options{
HeaderTemplate: func() string {
b, err := ioutil.ReadFile("/tests/test/testdata/chromium/url/sample2/header.html")
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return string(b)
}(),
FooterTemplate: func() string {
b, err := ioutil.ReadFile("/tests/test/testdata/chromium/url/sample2/footer.html")
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return string(b)
}(),
},
},
} {
func() {
mod := new(Chromium)
mod.binPath = os.Getenv("CHROMIUM_BIN_PATH")
mod.userAgent = tc.userAgent
mod.incognito = tc.incognito
mod.ignoreCertificateErrors = tc.ignoreCertificateErrors
if tc.allowList == nil {
tc.allowList = regexp.MustCompile("")
}
if tc.denyList == nil {
tc.denyList = regexp.MustCompile("")
}
mod.allowList = tc.allowList
mod.denyList = tc.denyList
outputDir, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
defer func() {
err := os.RemoveAll(outputDir)
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
}()
if tc.timeout == 0 {
err = mod.PDF(context.Background(), zap.NewNop(), tc.URL, outputDir+"/foo.pdf", tc.options)
} else {
ctx, cancel := context.WithTimeout(context.Background(), tc.timeout)
defer cancel()
err = mod.PDF(ctx, zap.NewNop(), tc.URL, outputDir+"/foo.pdf", tc.options)
}
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}()
}
}
// Interface guards.
var (
_ gotenberg.Module = (*ProtoModule)(nil)
_ API = (*ProtoAPI)(nil)
_ gotenberg.PDFEngineProvider = (*ProtoPDFEngineProvider)(nil)
_ gotenberg.Module = (*ProtoPDFEngineProvider)(nil)
_ gotenberg.PDFEngine = (*ProtoPDFEngine)(nil)
)

View File

@@ -0,0 +1,4 @@
// Package chromium provides a module which adds routes for converting HTML
// documents to PDF. Other modules may also retrieve the API provided by this
// module.
package chromium

View File

@@ -0,0 +1,122 @@
package chromium
import (
"context"
"fmt"
"github.com/chromedp/cdproto/network"
"github.com/chromedp/cdproto/page"
"github.com/chromedp/chromedp"
"go.uber.org/zap"
"golang.org/x/sync/errgroup"
)
// waitForEventDomContentEventFired waits until the event DomContentEventFired
// is fired or the context timeout.
func waitForEventDomContentEventFired(ctx context.Context, logger *zap.Logger) func() error {
return func() error {
ch := make(chan struct{})
cctx, cancel := context.WithCancel(ctx)
chromedp.ListenTarget(cctx, func(ev interface{}) {
switch ev.(type) {
case *page.EventDomContentEventFired:
cancel()
close(ch)
}
})
select {
case <-ch:
logger.Debug("event DomContentEventFired fired")
return nil
case <-ctx.Done():
return fmt.Errorf("wait for event DomContentEventFired: %w", ctx.Err())
}
}
}
// waitForEventLoadEventFired waits until the event LoadEventFired is fired or
// the context timeout.
func waitForEventLoadEventFired(ctx context.Context, logger *zap.Logger) func() error {
return func() error {
ch := make(chan struct{})
cctx, cancel := context.WithCancel(ctx)
chromedp.ListenTarget(cctx, func(ev interface{}) {
switch ev.(type) {
case *page.EventLoadEventFired:
cancel()
close(ch)
}
})
select {
case <-ch:
logger.Debug("event LoadEventFired fired")
return nil
case <-ctx.Done():
return fmt.Errorf("wait for event LoadEventFired: %w", ctx.Err())
}
}
}
// waitForEventNetworkIdle waits until the event networkIdle is fired or the
// context timeout.
func waitForEventNetworkIdle(ctx context.Context, logger *zap.Logger) func() error {
return func() error {
ch := make(chan struct{})
cctx, cancel := context.WithCancel(ctx)
chromedp.ListenTarget(cctx, func(ev interface{}) {
switch e := ev.(type) {
case *page.EventLifecycleEvent:
if e.Name == "networkIdle" {
cancel()
close(ch)
}
}
})
select {
case <-ch:
logger.Debug("event networkIdle fired")
return nil
case <-ctx.Done():
return fmt.Errorf("wait for event networkIdle: %w", ctx.Err())
}
}
}
// waitForEventLoadingFinished waits until the event LoadingFinished is fired
// or the context timeout.
func waitForEventLoadingFinished(ctx context.Context, logger *zap.Logger) func() error {
return func() error {
ch := make(chan struct{})
cctx, cancel := context.WithCancel(ctx)
chromedp.ListenTarget(cctx, func(ev interface{}) {
switch ev.(type) {
case *network.EventLoadingFinished:
cancel()
close(ch)
}
})
select {
case <-ch:
logger.Debug("event LoadingFinished fired")
return nil
case <-ctx.Done():
return fmt.Errorf("wait for event LoadingFinished: %w", ctx.Err())
}
}
}
// runBatch runs all functions simultaneously and waits until all of them are
// completed or an error is encountered.
func runBatch(ctx context.Context, fn ...func() error) error {
eg, _ := errgroup.WithContext(ctx)
for _, f := range fn {
eg.Go(f)
}
return eg.Wait()
}

View File

@@ -0,0 +1,340 @@
package chromium
import (
"bytes"
"encoding/json"
"errors"
"fmt"
"html/template"
"io/ioutil"
"net/http"
"os"
"path/filepath"
"time"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
"github.com/microcosm-cc/bluemonday"
"github.com/russross/blackfriday/v2"
"go.uber.org/multierr"
)
// FormDataChromiumPDFOptions creates Options form the form data. Fallback to
// default value if the considered key is not present.
func FormDataChromiumPDFOptions(ctx *api.Context) (*api.FormData, Options) {
defaultOptions := DefaultOptions()
var (
waitDelay time.Duration
waitWindowStatus string
extraHTTPHeaders map[string]string
landscape, printBackground bool
scale, paperWidth, paperHeight float64
marginTop, marginBottom, marginLeft, marginRight float64
pageRanges string
headerTemplate, footerTemplate string
preferCSSPageSize bool
)
form := ctx.FormData().
Duration("waitDelay", &waitDelay, defaultOptions.WaitDelay).
String("waitWindowStatus", &waitWindowStatus, defaultOptions.WaitWindowStatus).
Custom("extraHttpHeaders", func(value string) error {
if value == "" {
extraHTTPHeaders = defaultOptions.ExtraHTTPHeaders
return nil
}
err := json.Unmarshal([]byte(value), &extraHTTPHeaders)
if err != nil {
return fmt.Errorf("unmarshal extra HTTP headers: %w", err)
}
return nil
}).
Bool("landscape", &landscape, defaultOptions.Landscape).
Bool("printBackground", &printBackground, defaultOptions.PrintBackground).
Float64("scale", &scale, defaultOptions.Scale).
Float64("paperWidth", &paperWidth, defaultOptions.PaperWidth).
Float64("paperHeight", &paperHeight, defaultOptions.PaperHeight).
Float64("marginTop", &marginTop, defaultOptions.MarginTop).
Float64("marginBottom", &marginBottom, defaultOptions.MarginBottom).
Float64("marginLeft", &marginLeft, defaultOptions.MarginLeft).
Float64("marginRight", &marginRight, defaultOptions.MarginRight).
String("nativePageRanges", &pageRanges, defaultOptions.PageRanges).
Content("header.html", &headerTemplate, defaultOptions.HeaderTemplate).
Content("footer.html", &footerTemplate, defaultOptions.FooterTemplate).
Bool("preferCssPageSize", &preferCSSPageSize, defaultOptions.PreferCSSPageSize)
options := Options{
WaitDelay: waitDelay,
WaitWindowStatus: waitWindowStatus,
ExtraHTTPHeaders: extraHTTPHeaders,
Landscape: landscape,
PrintBackground: printBackground,
Scale: scale,
PaperWidth: paperWidth,
PaperHeight: paperHeight,
MarginTop: marginTop,
MarginBottom: marginBottom,
MarginLeft: marginLeft,
MarginRight: marginRight,
PageRanges: pageRanges,
HeaderTemplate: headerTemplate,
FooterTemplate: footerTemplate,
PreferCSSPageSize: preferCSSPageSize,
}
return form, options
}
// convertURLRoute returns an api.MultipartFormDataRoute route which can
// convert a URL to PDF.
func convertURLRoute(chromium API, engine gotenberg.PDFEngine) api.MultipartFormDataRoute {
return api.MultipartFormDataRoute{
Path: "/chromium/convert/url",
Handler: func(ctx *api.Context) error {
form, options := FormDataChromiumPDFOptions(ctx)
var (
URL string
PDFformat string
)
err := form.
MandatoryString("url", &URL).
String("pdfFormat", &PDFformat, "").
Validate()
if err != nil {
return fmt.Errorf("validate form data: %w", err)
}
err = convertURL(ctx, chromium, engine, URL, PDFformat, options)
if err != nil {
return fmt.Errorf("convert URL to PDF: %w", err)
}
return nil
},
}
}
// convertHTMLRoute returns an api.MultipartFormDataRoute route which can
// convert an HTML file to PDF.
func convertHTMLRoute(chromium API, engine gotenberg.PDFEngine) api.MultipartFormDataRoute {
return api.MultipartFormDataRoute{
Path: "/chromium/convert/html",
Handler: func(ctx *api.Context) error {
form, options := FormDataChromiumPDFOptions(ctx)
var (
inputPath string
PDFformat string
)
err := form.
MandatoryPath("index.html", &inputPath).
String("pdfFormat", &PDFformat, "").
Validate()
if err != nil {
return fmt.Errorf("validate form data: %w", err)
}
URL := fmt.Sprintf("file://%s", inputPath)
err = convertURL(ctx, chromium, engine, URL, PDFformat, options)
if err != nil {
return fmt.Errorf("convert HTML to PDF: %w", err)
}
return nil
},
}
}
// convertMarkdownRoute returns an api.MultipartFormDataRoute route which can
// convert markdown files to PDF.
func convertMarkdownRoute(chromium API, engine gotenberg.PDFEngine) api.MultipartFormDataRoute {
return api.MultipartFormDataRoute{
Path: "/chromium/convert/markdown",
Handler: func(ctx *api.Context) error {
form, options := FormDataChromiumPDFOptions(ctx)
var (
inputPath string
markdownPaths []string
PDFformat string
)
err := form.
MandatoryPath("index.html", &inputPath).
MandatoryPaths([]string{".md"}, &markdownPaths).
String("pdfFormat", &PDFformat, "").
Validate()
if err != nil {
return fmt.Errorf("validate form data: %w", err)
}
// We have to convert each markdown file referenced in the HTML
// file to... HTML. Thanks to the "html/template" package, we are
// able to provide the "toHTML" function which the user may call
// directly inside the HTML file.
var markdownFilesNotFoundErr error
tmpl, err := template.
New(filepath.Base(inputPath)).
Funcs(template.FuncMap{
"toHTML": func(filename string) (template.HTML, error) {
var path string
for _, markdownPath := range markdownPaths {
markdownFilename := filepath.Base(markdownPath)
if filename == markdownFilename {
path = markdownPath
break
}
}
if path == "" {
markdownFilesNotFoundErr = multierr.Append(
markdownFilesNotFoundErr,
fmt.Errorf("'%s'", filename),
)
return "", nil
}
b, err := ioutil.ReadFile(path)
if err != nil {
return "", fmt.Errorf("read markdown file '%s': %w", filename, err)
}
unsafe := blackfriday.Run(b)
sanitized := bluemonday.UGCPolicy().SanitizeBytes(unsafe)
// #nosec
return template.HTML(sanitized), nil
},
}).ParseFiles(inputPath)
if err != nil {
return fmt.Errorf("parse template file: %w", err)
}
var buffer bytes.Buffer
err = tmpl.Execute(&buffer, &struct{}{})
if err != nil {
return fmt.Errorf("execute template: %w", err)
}
if markdownFilesNotFoundErr != nil {
return api.WrapError(
fmt.Errorf("markdown files not found: %w", markdownFilesNotFoundErr),
api.NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("Markdown file(s) not found: %s", markdownFilesNotFoundErr),
),
)
}
inputPath = ctx.GeneratePath(".html")
err = os.WriteFile(inputPath, buffer.Bytes(), 0600)
if err != nil {
return fmt.Errorf("write template result: %w", err)
}
URL := fmt.Sprintf("file://%s", inputPath)
err = convertURL(ctx, chromium, engine, URL, PDFformat, options)
if err != nil {
return fmt.Errorf("convert markdown to PDF: %w", err)
}
return nil
},
}
}
// convertURL is a stub which is called by the other methods of this file.
func convertURL(ctx *api.Context, chromium API, engine gotenberg.PDFEngine, URL, PDFformat string, options Options) error {
outputPath := ctx.GeneratePath(".pdf")
err := chromium.PDF(ctx, ctx.Log(), URL, outputPath, options)
if err != nil {
if errors.Is(err, ErrURLNotAuthorized) {
return api.WrapError(
fmt.Errorf("convert to PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusForbidden,
fmt.Sprintf("'%s' does not match the authorized URLs", URL),
),
)
}
if errors.Is(err, ErrInvalidPrinterSettings) {
return api.WrapError(
fmt.Errorf("convert to PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusBadRequest,
"Chromium does not handle the provided settings; please check for aberrant form values",
),
)
}
if errors.Is(err, ErrPageRangesSyntaxError) {
return api.WrapError(
fmt.Errorf("convert to PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("Chromium does not handle the page ranges '%s' (nativePageRanges)", options.PageRanges),
),
)
}
return fmt.Errorf("convert to PDF: %w", err)
}
// So far so good, the URL has been converted to PDF.
// Now, let's check if the client want to convert this result PDF
// to a specific PDF format.
if PDFformat != "" {
convertInputPath := outputPath
convertOutputPath := ctx.GeneratePath(".pdf")
err = engine.Convert(ctx, ctx.Log(), PDFformat, convertInputPath, convertOutputPath)
if err != nil {
if errors.Is(err, gotenberg.ErrPDFFormatNotAvailable) {
return api.WrapError(
fmt.Errorf("convert PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("At least one PDF engine does not handle the PDF format '%s' (pdfFormat), while other have failed to convert for other reasons", PDFformat),
),
)
}
return fmt.Errorf("convert PDF: %w", err)
}
// Important: the output path is now the converted file.
outputPath = convertOutputPath
}
err = ctx.AddOutputPaths(outputPath)
if err != nil {
return fmt.Errorf("add output path: %w", err)
}
return nil
}

View File

@@ -0,0 +1,676 @@
package chromium
import (
"context"
"errors"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"net/http"
"os"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
"go.uber.org/zap"
)
func TestFormDataChromiumPDFOptions(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
options Options
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
options: DefaultOptions(),
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetValues(map[string][]string{
"extraHttpHeaders": {
"foo",
},
})
return ctx
}(),
options: DefaultOptions(),
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetValues(map[string][]string{
"extraHttpHeaders": {
`{"foo":"bar"}`,
},
})
return ctx
}(),
options: func() Options {
options := DefaultOptions()
options.ExtraHTTPHeaders = map[string]string{
"foo": "bar",
}
return options
}(),
},
} {
_, actual := FormDataChromiumPDFOptions(tc.ctx.Context)
if !reflect.DeepEqual(actual, tc.options) {
t.Errorf("test %d: expected %v but got: %v", i, tc.options, actual)
}
}
}
func TestConvertURLHandler(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
api API
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
expectOutputPathsCount int
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetValues(map[string][]string{
"url": {
"",
},
})
return ctx
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetValues(map[string][]string{
"url": {
"foo",
},
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return errors.New("foo")
}
return chromiumAPI
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetValues(map[string][]string{
"url": {
"foo",
},
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
expectOutputPathsCount: 1,
},
} {
err := convertURLRoute(tc.api, nil).Handler(tc.ctx.Context)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr api.HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
if tc.expectOutputPathsCount != len(tc.ctx.OutputPaths()) {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectOutputPathsCount, len(tc.ctx.OutputPaths()))
}
}
}
func TestConvertHTMLHandler(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
api API
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
expectOutputPathsCount int
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.html": "/foo/foo.html",
})
return ctx
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"index.html": "/foo/foo.html",
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return errors.New("foo")
}
return chromiumAPI
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"index.html": "/foo/foo.html",
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
expectOutputPathsCount: 1,
},
} {
err := convertHTMLRoute(tc.api, nil).Handler(tc.ctx.Context)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr api.HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
if tc.expectOutputPathsCount != len(tc.ctx.OutputPaths()) {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectOutputPathsCount, len(tc.ctx.OutputPaths()))
}
}
}
func TestConvertMarkdownHandler(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
api API
outputDir string
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
expectOutputPathsCount int
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.html": "/foo/foo.html",
})
return ctx
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"index.html": "/foo/foo.html",
})
return ctx
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"index.html": "/foo/foo.html",
"markdown.md": "/foo/markdown.md",
})
return ctx
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"index.html": "/tests/test/testdata/chromium/markdown/sample2/index.html",
"markdown1.md": "/foo/markdown1.md",
})
return ctx
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"index.html": "/tests/test/testdata/chromium/markdown/sample1/index.html",
"markdown1.md": "/foo/markdown1.md",
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return errors.New("foo")
}
return chromiumAPI
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"index.html": "/tests/test/testdata/chromium/markdown/sample1/index.html",
"markdown1.md": "/tests/test/testdata/chromium/markdown/sample1/markdown1.md",
"markdown2.md": "/tests/test/testdata/chromium/markdown/sample1/markdown2.md",
"markdown3.md": "/tests/test/testdata/chromium/markdown/sample1/markdown3.md",
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return errors.New("foo")
}
return chromiumAPI
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetDirPath("/tmp/foo")
ctx.SetFiles(map[string]string{
"index.html": "/tests/test/testdata/chromium/markdown/sample1/index.html",
"markdown1.md": "/tests/test/testdata/chromium/markdown/sample1/markdown1.md",
"markdown2.md": "/tests/test/testdata/chromium/markdown/sample1/markdown2.md",
"markdown3.md": "/tests/test/testdata/chromium/markdown/sample1/markdown3.md",
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return errors.New("foo")
}
return chromiumAPI
}(),
outputDir: "/tmp/foo",
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetDirPath("/tmp/foo")
ctx.SetFiles(map[string]string{
"index.html": "/tests/test/testdata/chromium/markdown/sample1/index.html",
"markdown1.md": "/tests/test/testdata/chromium/markdown/sample1/markdown1.md",
"markdown2.md": "/tests/test/testdata/chromium/markdown/sample1/markdown2.md",
"markdown3.md": "/tests/test/testdata/chromium/markdown/sample1/markdown3.md",
})
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
outputDir: "/tmp/foo",
expectOutputPathsCount: 1,
},
} {
func() {
if tc.outputDir != "" {
err := os.MkdirAll(tc.outputDir, 0755)
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
defer func() {
err := os.RemoveAll(tc.outputDir)
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
}()
}
err := convertMarkdownRoute(tc.api, nil).Handler(tc.ctx.Context)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr api.HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
if tc.expectOutputPathsCount != len(tc.ctx.OutputPaths()) {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectOutputPathsCount, len(tc.ctx.OutputPaths()))
}
}()
}
}
func TestConvertURL(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
api API
engine gotenberg.PDFEngine
PDFformat string
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
expectOutputPathsCount int
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return ErrURLNotAuthorized
}
return chromiumAPI
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusForbidden,
},
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return ErrInvalidPrinterSettings
}
return chromiumAPI
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return ErrPageRangesSyntaxError
}
return chromiumAPI
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return errors.New("foo")
}
return chromiumAPI
}(),
expectErr: true,
},
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return gotenberg.ErrPDFFormatNotAvailable
},
}
}(),
PDFformat: "foo",
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
}
}(),
PDFformat: "foo",
expectErr: true,
},
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
PDFformat: "foo",
expectOutputPathsCount: 1,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetCancelled(true)
return ctx
}(),
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
expectErr: true,
},
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() API {
chromiumAPI := struct{ ProtoAPI }{}
chromiumAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ Options) error {
return nil
}
return chromiumAPI
}(),
expectOutputPathsCount: 1,
},
} {
err := convertURL(tc.ctx.Context, tc.api, tc.engine, "", tc.PDFformat, DefaultOptions())
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr api.HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
if tc.expectOutputPathsCount != len(tc.ctx.OutputPaths()) {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectOutputPathsCount, len(tc.ctx.OutputPaths()))
}
}
}

3
pkg/modules/gc/doc.go Normal file
View File

@@ -0,0 +1,3 @@
// Package gc provides a module for removing files and directories that have
// expired.
package gc

235
pkg/modules/gc/gc.go Normal file
View File

@@ -0,0 +1,235 @@
package gc
import (
"context"
"errors"
"fmt"
"os"
"os/signal"
"path/filepath"
"strings"
"sync"
"time"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
func init() {
gotenberg.MustRegisterModule(GarbageCollector{})
}
// GarbageCollector is a module for removing files and directories that have
// expired. It allows us to make sure that the application does not leak files
// or directories when running.
type GarbageCollector struct {
rootPath string
graceDuration time.Duration
excludeSubstr []string
ticker *time.Ticker
done chan bool
logger *zap.Logger
}
// GarbageCollectorGraceDurationModifier is a module interface which allows to
// update the expiration time of files and directories parsed by the garbage
// collector. For instance, if the grace duration is 30s, the garbage collector
// will remove paths that have a modification time older than 30s. If there are
// many GarbageCollectorGraceDurationModifier, only the longest grace duration
// is selected.
type GarbageCollectorGraceDurationModifier interface {
GraceDuration() time.Duration
}
// GarbageCollectorExcludeSubstrModifier is a module interface which adds the
// given substrings to the exclude list of the garbage collector. If a path
// contains one of those substrings, the garbage collector ignores it.
type GarbageCollectorExcludeSubstrModifier interface {
ExcludeSubstr() []string
}
// Descriptor returns a GarbageCollector's module descriptor.
func (gc GarbageCollector) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "gc",
New: func() gotenberg.Module { return new(GarbageCollector) },
}
}
// Provision sets the module properties.
func (gc *GarbageCollector) Provision(ctx *gotenberg.Context) error {
gc.rootPath = gotenberg.TmpPath()
graceDurationModifiers, err := ctx.Modules(new(GarbageCollectorGraceDurationModifier))
if err != nil {
return fmt.Errorf("get grace duration modifiers: %w", err)
}
for _, graceDurationModifier := range graceDurationModifiers {
modifier := graceDurationModifier.(GarbageCollectorGraceDurationModifier)
if gc.graceDuration < modifier.GraceDuration() {
gc.graceDuration = modifier.GraceDuration()
}
}
excludeSubstrModifiers, err := ctx.Modules(new(GarbageCollectorExcludeSubstrModifier))
if err != nil {
return fmt.Errorf("get exclude substr modifiers: %w", err)
}
gc.excludeSubstr = strings.Split(os.Getenv("GC_EXCLUDE_SUBSTR"), ",")
for _, excludeSubstrModifier := range excludeSubstrModifiers {
modifier := excludeSubstrModifier.(GarbageCollectorExcludeSubstrModifier)
gc.excludeSubstr = append(gc.excludeSubstr, modifier.ExcludeSubstr()...)
}
loggerProvider, err := ctx.Module(new(gotenberg.LoggerProvider))
if err != nil {
return fmt.Errorf("get logger provider: %w", err)
}
logger, err := loggerProvider.(gotenberg.LoggerProvider).Logger(gc)
if err != nil {
return fmt.Errorf("get logger: %w", err)
}
gc.logger = logger
return nil
}
// Start starts the garbage collector.
func (gc *GarbageCollector) Start() error {
gc.ticker = time.NewTicker(gc.graceDuration + time.Duration(1)*time.Second)
gc.done = make(chan bool, 1)
go func() {
for {
func() {
gcMu.RLock()
defer gcMu.RUnlock()
select {
case <-gc.done:
return
case <-gc.ticker.C:
gc.collect(false)
}
}()
}
}()
return nil
}
// collect parses the root path of the garbage collector and removes files or
// directories that have expired. It ignores the expiration date if the "force"
// argument is set to true.
func (gc GarbageCollector) collect(force bool) {
expirationTime := time.Now().Add(-gc.graceDuration)
// To make sure that the next Walk method stays on
// the root level of the considered path, we have to
// return a filepath.SkipDir error if the current path
// is a directory.
skipDirOrNil := func(info os.FileInfo) error {
if info.IsDir() {
return filepath.SkipDir
}
return nil
}
removePath := func(path string) {
err := os.RemoveAll(path)
if err != nil {
gc.logger.Error(fmt.Sprintf("remove '%s': %s", path, err))
}
gc.logger.Debug(fmt.Sprintf("'%s' removed", path))
}
err := filepath.Walk(gc.rootPath, func(path string, info os.FileInfo, pathErr error) error {
if pathErr != nil {
// For whatever reasons, the Walk method failed
// to process the current path.
return pathErr
}
if path == gc.rootPath {
return nil
}
for _, substr := range gc.excludeSubstr {
if strings.Contains(info.Name(), substr) {
return skipDirOrNil(info)
}
}
if force {
removePath(path)
return skipDirOrNil(info)
}
if info.ModTime().Before(expirationTime) {
removePath(path)
}
return skipDirOrNil(info)
})
if err != nil {
gc.logger.Error(err.Error())
}
}
// StartupMessage returns an empty string.
func (gc GarbageCollector) StartupMessage() string {
return ""
}
// Stop stops the garbage collector.
func (gc *GarbageCollector) Stop(ctx context.Context) error {
_, ok := ctx.Deadline()
if !ok {
return errors.New("no context dead line")
}
// Block until the context is done so that other module may gracefully stop
// before we do a shutdown cleanup. We skip this step if we receive a
// SIGINT in the meantime.
gc.logger.Debug("wait for the end of grace duration")
quit := make(chan os.Signal, 1)
signal.Notify(quit, os.Interrupt)
select {
case <-quit:
return nil
case <-ctx.Done():
break
}
gc.ticker.Stop()
gc.done <- true
gc.logger.Debug("shutdown cleanup...")
gc.collect(true)
return nil
}
var gcMu sync.RWMutex
// Interface guards.
var (
_ gotenberg.Module = (*GarbageCollector)(nil)
_ gotenberg.Provisioner = (*GarbageCollector)(nil)
_ gotenberg.App = (*GarbageCollector)(nil)
)

468
pkg/modules/gc/gc_test.go Normal file
View File

@@ -0,0 +1,468 @@
package gc
import (
"context"
"errors"
"os"
"reflect"
"strings"
"testing"
"time"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
type ProtoModule struct {
descriptor func() gotenberg.ModuleDescriptor
}
func (mod ProtoModule) Descriptor() gotenberg.ModuleDescriptor {
return mod.descriptor()
}
type ProtoValidator struct {
ProtoModule
validate func() error
}
func (mod ProtoValidator) Validate() error {
return mod.validate()
}
type ProtoGarbageCollectorGraceDurationModifier struct {
ProtoValidator
graceDuration func() time.Duration
}
func (mod ProtoGarbageCollectorGraceDurationModifier) GraceDuration() time.Duration {
return mod.graceDuration()
}
type ProtoGarbageCollectorExcludeSubstrModifier struct {
ProtoValidator
excludeSubstr func() []string
}
func (mod ProtoGarbageCollectorExcludeSubstrModifier) ExcludeSubstr() []string {
return mod.excludeSubstr()
}
type ProtoLoggerProvider struct {
ProtoModule
logger func(mod gotenberg.Module) (*zap.Logger, error)
}
func (factory ProtoLoggerProvider) Logger(mod gotenberg.Module) (*zap.Logger, error) {
return factory.logger(mod)
}
func TestGarbageCollector_Descriptor(t *testing.T) {
descriptor := GarbageCollector{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(GarbageCollector))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestGarbageCollector_Provision(t *testing.T) {
for i, tc := range []struct {
ctx *gotenberg.Context
expectGraceDuration time.Duration
expectExcludeSubstr []string
expectErr bool
}{
{
ctx: func() *gotenberg.Context {
mod := struct {
ProtoGarbageCollectorGraceDurationModifier
}{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error { return errors.New("foo") }
return gotenberg.NewContext(gotenberg.ParsedFlags{}, []gotenberg.ModuleDescriptor{
mod.Descriptor(),
})
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct {
ProtoGarbageCollectorExcludeSubstrModifier
}{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.validate = func() error { return errors.New("foo") }
return gotenberg.NewContext(gotenberg.ParsedFlags{}, []gotenberg.ModuleDescriptor{
mod.Descriptor(),
})
}(),
expectErr: true,
},
{
ctx: gotenberg.NewContext(gotenberg.ParsedFlags{}, make([]gotenberg.ModuleDescriptor, 0)),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct {
ProtoLoggerProvider
}{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.logger = func(mod gotenberg.Module) (*zap.Logger, error) { return nil, errors.New("foo") }
return gotenberg.NewContext(gotenberg.ParsedFlags{}, []gotenberg.ModuleDescriptor{
mod.Descriptor(),
})
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct {
ProtoLoggerProvider
}{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.logger = func(mod gotenberg.Module) (*zap.Logger, error) { return zap.NewNop(), nil }
return gotenberg.NewContext(gotenberg.ParsedFlags{}, []gotenberg.ModuleDescriptor{
mod.Descriptor(),
})
}(),
},
{
ctx: func() *gotenberg.Context {
mod1 := struct {
ProtoGarbageCollectorGraceDurationModifier
}{}
mod1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod1 }}
}
mod1.graceDuration = func() time.Duration { return time.Duration(10) * time.Second }
mod1.validate = func() error { return nil }
mod2 := struct {
ProtoGarbageCollectorGraceDurationModifier
}{}
mod2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "bar", New: func() gotenberg.Module { return mod2 }}
}
mod2.graceDuration = func() time.Duration { return time.Duration(20) * time.Second }
mod2.validate = func() error { return nil }
mod3 := struct {
ProtoLoggerProvider
}{}
mod3.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "baz", New: func() gotenberg.Module { return mod3 }}
}
mod3.logger = func(mod gotenberg.Module) (*zap.Logger, error) { return zap.NewNop(), nil }
return gotenberg.NewContext(gotenberg.ParsedFlags{}, []gotenberg.ModuleDescriptor{
mod1.Descriptor(),
mod2.Descriptor(),
mod3.Descriptor(),
})
}(),
expectGraceDuration: time.Duration(20) * time.Second,
},
{
ctx: func() *gotenberg.Context {
mod1 := struct {
ProtoGarbageCollectorExcludeSubstrModifier
}{}
mod1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod1 }}
}
mod1.excludeSubstr = func() []string { return []string{"foo"} }
mod1.validate = func() error { return nil }
mod2 := struct {
ProtoGarbageCollectorExcludeSubstrModifier
}{}
mod2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "bar", New: func() gotenberg.Module { return mod2 }}
}
mod2.excludeSubstr = func() []string { return []string{"bar"} }
mod2.validate = func() error { return nil }
mod3 := struct {
ProtoLoggerProvider
}{}
mod3.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "baz", New: func() gotenberg.Module { return mod3 }}
}
mod3.logger = func(mod gotenberg.Module) (*zap.Logger, error) { return zap.NewNop(), nil }
return gotenberg.NewContext(gotenberg.ParsedFlags{}, []gotenberg.ModuleDescriptor{
mod1.Descriptor(),
mod2.Descriptor(),
mod3.Descriptor(),
})
}(),
expectExcludeSubstr: func() []string {
expect := strings.Split(os.Getenv("GC_EXCLUDE_SUBSTR"), ",")
return append(expect, "foo", "bar")
}(),
},
} {
mod := new(GarbageCollector)
err := mod.Provision(tc.ctx)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
if tc.expectGraceDuration != 0 && tc.expectGraceDuration != mod.graceDuration {
t.Errorf("test %d: expected grace duration of '%s' but got '%s'", i, tc.expectGraceDuration, mod.graceDuration)
}
if tc.expectExcludeSubstr != nil && !reflect.DeepEqual(tc.expectExcludeSubstr, mod.excludeSubstr) {
t.Errorf("test %d: expected exclude substr '%s' but got '%s'", i, tc.expectExcludeSubstr, mod.excludeSubstr)
}
}
}
func TestGarbageCollector_Start(t *testing.T) {
mod := new(GarbageCollector)
mod.logger = zap.NewNop()
path, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
mod.rootPath = path
err = mod.Start()
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
time.Sleep(time.Duration(2) * time.Second)
mod.ticker.Stop()
mod.done <- true
}
func TestGarbageCollector_collect(t *testing.T) {
for i, tc := range []struct {
gc *GarbageCollector
expectNotExists []string
expectExists []string
force bool
}{
{
gc: func() *GarbageCollector {
mod := new(GarbageCollector)
mod.logger = zap.NewNop()
path, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
mod.rootPath = path
err = os.WriteFile(path+"/foo", []byte{1}, 0755)
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
mod.excludeSubstr = []string{
"foo",
}
return mod
}(),
expectExists: []string{
"/foo",
},
},
{
gc: func() *GarbageCollector {
mod := new(GarbageCollector)
mod.logger = zap.NewNop()
path, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
mod.rootPath = path
err = os.WriteFile(path+"/foo", []byte{1}, 0755)
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
err = os.MkdirAll(path+"/bar", 0755)
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return mod
}(),
expectNotExists: []string{
"/foo",
"/bar",
},
force: true,
},
{
gc: func() *GarbageCollector {
mod := new(GarbageCollector)
mod.logger = zap.NewNop()
mod.graceDuration = time.Duration(10) * time.Second
path, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
mod.rootPath = path
err = os.WriteFile(path+"/foo", []byte{1}, 0755)
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
newTime := time.Now().Add(-time.Duration(20) * time.Second)
err = os.Chtimes(path+"/foo", newTime, newTime)
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
err = os.WriteFile(path+"/bar", []byte{1}, 0755)
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
newTime = time.Now().Add(time.Duration(10) * time.Second)
err = os.Chtimes(path+"/bar", newTime, newTime)
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return mod
}(),
expectNotExists: []string{
"/foo",
},
expectExists: []string{
"/bar",
},
},
} {
tc.gc.collect(tc.force)
for _, name := range tc.expectNotExists {
path := tc.gc.rootPath + name
_, err := os.Stat(path)
if !os.IsNotExist(err) {
t.Errorf("test %d: expected '%s' not to exist but got: %v", i, path, err)
}
}
for _, name := range tc.expectExists {
path := tc.gc.rootPath + name
_, err := os.Stat(path)
if os.IsNotExist(err) {
t.Errorf("test %d: expected '%s' to exist but got: %v", i, path, err)
}
}
err := os.RemoveAll(tc.gc.rootPath)
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestGarbageCollector_StartupMessage(t *testing.T) {
actual := new(GarbageCollector).StartupMessage()
expect := ""
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestGarbageCollector_Stop(t *testing.T) {
for i, tc := range []struct {
timeout time.Duration
expectErr bool
}{
{
expectErr: true,
},
{
timeout: time.Duration(1) * time.Nanosecond,
},
} {
func() {
mod := new(GarbageCollector)
mod.logger = zap.NewNop()
path, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
mod.rootPath = path
err = mod.Start()
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
if tc.timeout == 0 {
err = mod.Stop(context.TODO())
} else {
ctx, cancel := context.WithTimeout(context.Background(), tc.timeout)
defer cancel()
err = mod.Stop(ctx)
}
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}()
}
}
// Interface guards.
var (
_ gotenberg.Module = (*ProtoModule)(nil)
_ gotenberg.Validator = (*ProtoValidator)(nil)
_ GarbageCollectorGraceDurationModifier = (*ProtoGarbageCollectorGraceDurationModifier)(nil)
_ gotenberg.Module = (*ProtoGarbageCollectorGraceDurationModifier)(nil)
_ gotenberg.Validator = (*ProtoGarbageCollectorGraceDurationModifier)(nil)
_ GarbageCollectorExcludeSubstrModifier = (*ProtoGarbageCollectorExcludeSubstrModifier)(nil)
_ gotenberg.Module = (*ProtoGarbageCollectorExcludeSubstrModifier)(nil)
_ gotenberg.Validator = (*ProtoGarbageCollectorExcludeSubstrModifier)(nil)
_ gotenberg.LoggerProvider = (*ProtoLoggerProvider)(nil)
_ gotenberg.Module = (*ProtoLoggerProvider)(nil)
)

View File

@@ -0,0 +1,3 @@
// Package libreoffice provides a module which adds a route for converting
// document to PDF with LibreOffice.
package libreoffice

View File

@@ -0,0 +1,86 @@
package libreoffice
import (
"fmt"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
"github.com/gotenberg/gotenberg/v7/pkg/modules/libreoffice/unoconv"
flag "github.com/spf13/pflag"
)
func init() {
gotenberg.MustRegisterModule(LibreOffice{})
}
// LibreOffice is a module which provides a route for converting documents to
// PDF with LibreOffice.
type LibreOffice struct {
unoconv unoconv.API
engine gotenberg.PDFEngine
disableRoutes bool
}
// Descriptor returns a LibreOffice's module descriptor.
func (LibreOffice) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "libreoffice",
FlagSet: func() *flag.FlagSet {
fs := flag.NewFlagSet("libreoffice", flag.ExitOnError)
fs.Bool("libreoffice-disable-routes", false, "Disable the routes")
return fs
}(),
New: func() gotenberg.Module { return new(LibreOffice) },
}
}
// Provision sets the module properties.
func (mod *LibreOffice) Provision(ctx *gotenberg.Context) error {
flags := ctx.ParsedFlags()
mod.disableRoutes = flags.MustBool("libreoffice-disable-routes")
provider, err := ctx.Module(new(unoconv.Provider))
if err != nil {
return fmt.Errorf("get unoconv provider: %w", err)
}
uno, err := provider.(unoconv.Provider).Unoconv()
if err != nil {
return fmt.Errorf("get unoconv API: %w", err)
}
mod.unoconv = uno
provider, err = ctx.Module(new(gotenberg.PDFEngineProvider))
if err != nil {
return fmt.Errorf("get PDF engine provider: %w", err)
}
engine, err := provider.(gotenberg.PDFEngineProvider).PDFEngine()
if err != nil {
return fmt.Errorf("get PDF engine: %w", err)
}
mod.engine = engine
return nil
}
// Routes returns the API routes.
func (mod LibreOffice) Routes() ([]api.MultipartFormDataRoute, error) {
if mod.disableRoutes {
return nil, nil
}
return []api.MultipartFormDataRoute{
convertRoute(mod.unoconv, mod.engine),
}, nil
}
// Interface guards.
var (
_ gotenberg.Module = (*LibreOffice)(nil)
_ gotenberg.Provisioner = (*LibreOffice)(nil)
_ api.MultipartFormDataRouter = (*LibreOffice)(nil)
)

View File

@@ -0,0 +1,250 @@
package libreoffice
import (
"context"
"errors"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/libreoffice/unoconv"
"go.uber.org/zap"
)
type ProtoModule struct {
descriptor func() gotenberg.ModuleDescriptor
}
func (mod ProtoModule) Descriptor() gotenberg.ModuleDescriptor {
return mod.descriptor()
}
type ProtoUnoconvProvider struct {
ProtoModule
unoconv func() (unoconv.API, error)
}
func (mod ProtoUnoconvProvider) Unoconv() (unoconv.API, error) {
return mod.unoconv()
}
type ProtoUnoconvAPI struct {
pdf func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error
extensions func() []string
}
func (mod ProtoUnoconvAPI) PDF(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options unoconv.Options) error {
return mod.pdf(ctx, logger, inputPath, outputPath, options)
}
func (mod ProtoUnoconvAPI) Extensions() []string {
return mod.extensions()
}
type ProtoPDFEngineProvider struct {
ProtoModule
pdfEngine func() (gotenberg.PDFEngine, error)
}
func (mod ProtoPDFEngineProvider) PDFEngine() (gotenberg.PDFEngine, error) {
return mod.pdfEngine()
}
type ProtoPDFEngine struct {
merge func(_ context.Context, _ *zap.Logger, _ []string, _ string) error
convert func(_ context.Context, _ *zap.Logger, _, _, _ string) error
}
func (mod ProtoPDFEngine) Merge(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
return mod.merge(ctx, logger, inputPaths, outputPath)
}
func (mod ProtoPDFEngine) Convert(ctx context.Context, logger *zap.Logger, format, inputPath, outputPath string) error {
return mod.convert(ctx, logger, format, inputPath, outputPath)
}
func TestLibreOffice_Descriptor(t *testing.T) {
descriptor := LibreOffice{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(LibreOffice))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestLibreOffice_Provision(t *testing.T) {
for i, tc := range []struct {
ctx *gotenberg.Context
expectErr bool
}{
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoModule }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(LibreOffice).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoUnoconvProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.unoconv = func() (unoconv.API, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(LibreOffice).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoUnoconvProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.unoconv = func() (unoconv.API, error) {
return struct{ ProtoUnoconvAPI }{}, nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(LibreOffice).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod1 := struct{ ProtoUnoconvProvider }{}
mod1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod1 }}
}
mod1.unoconv = func() (unoconv.API, error) {
return struct{ ProtoUnoconvAPI }{}, nil
}
mod2 := struct{ ProtoPDFEngineProvider }{}
mod2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "bar", New: func() gotenberg.Module { return mod2 }}
}
mod2.pdfEngine = func() (gotenberg.PDFEngine, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(LibreOffice).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod1.Descriptor(),
mod2.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod1 := struct{ ProtoUnoconvProvider }{}
mod1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod1 }}
}
mod1.unoconv = func() (unoconv.API, error) {
return struct{ ProtoUnoconvAPI }{}, nil
}
mod2 := struct{ ProtoPDFEngineProvider }{}
mod2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "bar", New: func() gotenberg.Module { return mod2 }}
}
mod2.pdfEngine = func() (gotenberg.PDFEngine, error) {
return struct{ ProtoPDFEngine }{}, nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(LibreOffice).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
mod1.Descriptor(),
mod2.Descriptor(),
},
)
}(),
},
} {
mod := new(LibreOffice)
err := mod.Provision(tc.ctx)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestLibreOffice_Routes(t *testing.T) {
for i, tc := range []struct {
expectRoutes int
disableRoutes bool
}{
{
expectRoutes: 1,
},
{
disableRoutes: true,
},
} {
mod := new(LibreOffice)
mod.disableRoutes = tc.disableRoutes
routes, err := mod.Routes()
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
if tc.expectRoutes != len(routes) {
t.Errorf("test %d: expected %d routes but got %d", i, tc.expectRoutes, len(routes))
}
}
}
// Interface guards.
var (
_ gotenberg.Module = (*ProtoModule)(nil)
_ unoconv.Provider = (*ProtoUnoconvProvider)(nil)
_ gotenberg.Module = (*ProtoUnoconvProvider)(nil)
_ unoconv.API = (*ProtoUnoconvAPI)(nil)
_ gotenberg.PDFEngineProvider = (*ProtoPDFEngineProvider)(nil)
_ gotenberg.Module = (*ProtoPDFEngineProvider)(nil)
_ gotenberg.PDFEngine = (*ProtoPDFEngine)(nil)
)

View File

@@ -0,0 +1,3 @@
// Package pdfengine provides a module which abstracts the CLI tool unoconv and
// implements the gotenberg.PDFEngine interface.
package pdfengine

View File

@@ -0,0 +1,76 @@
package pdfengine
import (
"context"
"fmt"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/libreoffice/unoconv"
"go.uber.org/zap"
)
func init() {
gotenberg.MustRegisterModule(UnoconvPDFEngine{})
}
// UnoconvPDFEngine abstracts the CLI tool unoconv and implements the
// gotenberg.PDFEngine interface.
type UnoconvPDFEngine struct {
unoconv unoconv.API
}
// Descriptor returns a UnoconvPDFEngine's module descriptor.
func (engine UnoconvPDFEngine) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "unoconv-pdfengine",
New: func() gotenberg.Module { return new(UnoconvPDFEngine) },
}
}
// Provision sets the module properties.
func (engine *UnoconvPDFEngine) Provision(ctx *gotenberg.Context) error {
provider, err := ctx.Module(new(unoconv.Provider))
if err != nil {
return fmt.Errorf("get unoconv provider: %w", err)
}
uno, err := provider.(unoconv.Provider).Unoconv()
if err != nil {
return fmt.Errorf("get unoconv API: %w", err)
}
engine.unoconv = uno
return nil
}
// Merge is not available for this PDF engine.
func (engine UnoconvPDFEngine) Merge(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return fmt.Errorf("merge PDFs with unoconv: %w", gotenberg.ErrPDFEngineMethodNotAvailable)
}
// Convert converts the given PDF to a specific PDF format. Currently, only the
// PDF/A-1 format is available. If another PDF format is requested, it returns
// a gotenberg.ErrPDFFormatNotAvailable error.
func (engine UnoconvPDFEngine) Convert(ctx context.Context, logger *zap.Logger, format, inputPath, outputPath string) error {
if format != gotenberg.FormatPDFA1a {
return fmt.Errorf("convert PDF to '%s' with unoconv: %w", format, gotenberg.ErrPDFFormatNotAvailable)
}
err := engine.unoconv.PDF(ctx, logger, inputPath, outputPath, unoconv.Options{
PDFArchive: true,
})
if err == nil {
return nil
}
return fmt.Errorf("convert PDF to '%s' with unoconv: %w", format, err)
}
// Interface guards.
var (
_ gotenberg.Module = (*UnoconvPDFEngine)(nil)
_ gotenberg.Provisioner = (*UnoconvPDFEngine)(nil)
_ gotenberg.PDFEngine = (*UnoconvPDFEngine)(nil)
)

View File

@@ -0,0 +1,197 @@
package pdfengine
import (
"context"
"errors"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/libreoffice/unoconv"
flag "github.com/spf13/pflag"
"go.uber.org/zap"
)
type ProtoModule struct {
descriptor func() gotenberg.ModuleDescriptor
}
func (mod ProtoModule) Descriptor() gotenberg.ModuleDescriptor {
return mod.descriptor()
}
type ProtoUnoconvProvider struct {
ProtoModule
unoconv func() (unoconv.API, error)
}
func (mod ProtoUnoconvProvider) Unoconv() (unoconv.API, error) {
return mod.unoconv()
}
type ProtoUnoconvAPI struct {
pdf func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options unoconv.Options) error
}
func (mod ProtoUnoconvAPI) PDF(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options unoconv.Options) error {
return mod.pdf(ctx, logger, inputPath, outputPath, options)
}
func (mod ProtoUnoconvAPI) Extensions() []string {
return nil
}
func TestUnoconvPDFEngine_Descriptor(t *testing.T) {
descriptor := UnoconvPDFEngine{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(UnoconvPDFEngine))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestUnoconvPDFEngine_Provision(t *testing.T) {
for i, tc := range []struct {
ctx *gotenberg.Context
expectErr bool
}{
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoModule }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: flag.NewFlagSet("foo", flag.ExitOnError),
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoUnoconvProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.unoconv = func() (unoconv.API, error) {
return nil, errors.New("foo")
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: flag.NewFlagSet("foo", flag.ExitOnError),
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
mod := struct{ ProtoUnoconvProvider }{}
mod.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return mod }}
}
mod.unoconv = func() (unoconv.API, error) {
return struct{ ProtoUnoconvAPI }{}, nil
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: flag.NewFlagSet("foo", flag.ExitOnError),
},
[]gotenberg.ModuleDescriptor{
mod.Descriptor(),
},
)
}(),
},
} {
mod := new(UnoconvPDFEngine)
err := mod.Provision(tc.ctx)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestUnoconvPDFEngine_Merge(t *testing.T) {
mod := new(UnoconvPDFEngine)
err := mod.Merge(context.TODO(), zap.NewNop(), nil, "")
if !errors.Is(err, gotenberg.ErrPDFEngineMethodNotAvailable) {
t.Errorf("expected error %v, but got: %v", gotenberg.ErrPDFEngineMethodNotAvailable, err)
}
}
func TestUnoconvPDFEngine_Convert(t *testing.T) {
for i, tc := range []struct {
api unoconv.API
format string
expectErr bool
}{
{
format: "",
expectErr: true,
},
{
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, __ string, _ unoconv.Options) error {
return errors.New("foo")
}
return unoconvAPI
}(),
format: gotenberg.FormatPDFA1a,
expectErr: true,
},
{
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, __ string, _ unoconv.Options) error {
return nil
}
return unoconvAPI
}(),
format: gotenberg.FormatPDFA1a,
},
} {
mod := new(UnoconvPDFEngine)
mod.unoconv = tc.api
err := mod.Convert(context.TODO(), zap.NewNop(), tc.format, "", "")
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
// Interface guards.
var (
_ gotenberg.Module = (*ProtoModule)(nil)
_ unoconv.Provider = (*ProtoUnoconvProvider)(nil)
_ gotenberg.Module = (*ProtoUnoconvProvider)(nil)
_ unoconv.API = (*ProtoUnoconvAPI)(nil)
)

View File

@@ -0,0 +1,177 @@
package libreoffice
import (
"errors"
"fmt"
"net/http"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
"github.com/gotenberg/gotenberg/v7/pkg/modules/libreoffice/unoconv"
)
// convertRoute returns an api.MultipartFormDataRoute which can convert
// LibreOffice documents to PDF.
func convertRoute(uno unoconv.API, engine gotenberg.PDFEngine) api.MultipartFormDataRoute {
return api.MultipartFormDataRoute{
Path: "/libreoffice/convert",
Handler: func(ctx *api.Context) error {
// Let's get the data from the form and validate them.
var (
inputPaths []string
landscape bool
nativePageRanges string
nativePDFA1aFormat bool
PDFformat string
merge bool
)
err := ctx.FormData().
MandatoryPaths(uno.Extensions(), &inputPaths).
Bool("landscape", &landscape, false).
String("nativePageRanges", &nativePageRanges, "").
Bool("nativePdfA1aFormat", &nativePDFA1aFormat, false).
String("pdfFormat", &PDFformat, "").
Bool("merge", &merge, false).
Validate()
if err != nil {
return fmt.Errorf("validate form data: %w", err)
}
if nativePDFA1aFormat && PDFformat != "" {
return api.WrapError(
errors.New("got both 'pdfFormat' and 'nativePdfA1aFormat' form values"),
api.NewSentinelHTTPError(http.StatusBadRequest, "Both 'pdfFormat' and 'nativePdfA1aFormat' form values are provided"),
)
}
// Alright, let's convert each document to PDF.
outputPaths := make([]string, len(inputPaths))
for i, inputPath := range inputPaths {
outputPaths[i] = ctx.GeneratePath(".pdf")
options := unoconv.Options{
Landscape: landscape,
PageRanges: nativePageRanges,
PDFArchive: nativePDFA1aFormat,
}
err = uno.PDF(ctx, ctx.Log(), inputPath, outputPaths[i], options)
if err != nil {
if errors.Is(err, unoconv.ErrMalformedPageRanges) {
return api.WrapError(
fmt.Errorf("convert to PDF: %w", err),
api.NewSentinelHTTPError(http.StatusBadRequest, fmt.Sprintf("Malformed page ranges '%s' (nativePageRanges)", options.PageRanges)),
)
}
return fmt.Errorf("convert to PDF: %w", err)
}
}
// So far so good, let's check if we have to merge the PDFs. Quick
// win: if there is only one PDF, skip this step.
if len(outputPaths) > 1 && merge {
outputPath := ctx.GeneratePath(".pdf")
err = engine.Merge(ctx, ctx.Log(), outputPaths, outputPath)
if err != nil {
return fmt.Errorf("merge PDFs: %w", err)
}
// Now, let's check if the client want to convert this result
// PDF to a specific PDF format.
// Note: nativePdfA1aFormat has not been specified if we reach
// this part of the code. Indeed, the handler returns early on
// an error if both nativePdfA1aFormat and pdfFormat are
// present.
if PDFformat != "" {
convertInputPath := outputPath
convertOutputPath := ctx.GeneratePath(".pdf")
err = engine.Convert(ctx, ctx.Log(), PDFformat, convertInputPath, convertOutputPath)
if err != nil {
if errors.Is(err, gotenberg.ErrPDFFormatNotAvailable) {
return api.WrapError(
fmt.Errorf("convert PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("At least one PDF engine does not handle the PDF format '%s' (pdfFormat), while other have failed to convert for other reasons", PDFformat),
),
)
}
return fmt.Errorf("convert PDF: %w", err)
}
// Important: the output path is now the converted file.
outputPath = convertOutputPath
}
// Last but not least, add the output path to the context so that
// the API is able to send it as a response to the client.
err = ctx.AddOutputPaths(outputPath)
if err != nil {
return fmt.Errorf("add output path: %w", err)
}
return nil
}
// Ok, we don't have to merge the PDFs. Let's check if the client
// want to convert each PDF to a specific PDF format.
// Note: nativePdfA1aFormat has not been specified if we reach this
// part of the code. Indeed, the handler returns early on an error
// if both nativePdfA1aFormat and pdfFormat are present.
if PDFformat != "" {
convertOutputPaths := make([]string, len(outputPaths))
for i, outputPath := range outputPaths {
convertInputPath := outputPath
convertOutputPaths[i] = ctx.GeneratePath(".pdf")
err = engine.Convert(ctx, ctx.Log(), PDFformat, convertInputPath, convertOutputPaths[i])
if err != nil {
if errors.Is(err, gotenberg.ErrPDFFormatNotAvailable) {
return api.WrapError(
fmt.Errorf("convert PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("At least one PDF engine does not handle the PDF format '%s' (pdfFormat), while other have failed to convert for other reasons", PDFformat),
),
)
}
return fmt.Errorf("convert PDF: %w", err)
}
}
// Important: the output paths are now the converted files.
outputPaths = convertOutputPaths
}
// Last but not least, add the output paths to the context so that
// the API is able to send them as a response to the client.
err = ctx.AddOutputPaths(outputPaths...)
if err != nil {
return fmt.Errorf("add output paths: %w", err)
}
return nil
},
}
}

View File

@@ -0,0 +1,541 @@
package libreoffice
import (
"context"
"errors"
"net/http"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
"github.com/gotenberg/gotenberg/v7/pkg/modules/libreoffice/unoconv"
"go.uber.org/zap"
)
func TestConvertHandler(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
api unoconv.API
engine gotenberg.PDFEngine
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
expectOutputPathsCount int
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.extensions = func() []string {
return []string{
".foo",
}
}
return unoconvAPI
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
})
ctx.SetValues(map[string][]string{
"nativePdfA1aFormat": {
"true",
},
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return unoconv.ErrMalformedPageRanges
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return errors.New("foo")
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return errors.New("foo")
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return gotenberg.ErrPDFFormatNotAvailable
},
}
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetCancelled(true)
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectOutputPathsCount: 1,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return gotenberg.ErrPDFFormatNotAvailable
},
}
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetCancelled(true)
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.docx": "/foo/foo.docx",
"bar.docx": "/foo/bar.docx",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
api: func() unoconv.API {
unoconvAPI := struct{ ProtoUnoconvAPI }{}
unoconvAPI.pdf = func(_ context.Context, _ *zap.Logger, _, _ string, _ unoconv.Options) error {
return nil
}
unoconvAPI.extensions = func() []string {
return []string{
".docx",
}
}
return unoconvAPI
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectOutputPathsCount: 2,
},
} {
err := convertRoute(tc.api, tc.engine).Handler(tc.ctx.Context)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr api.HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
if tc.expectOutputPathsCount != len(tc.ctx.OutputPaths()) {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectOutputPathsCount, len(tc.ctx.OutputPaths()))
}
}
}

View File

@@ -0,0 +1,2 @@
// Package unoconv provides a module which abstracts the CLI tool unoconv.
package unoconv

View File

@@ -0,0 +1,287 @@
package unoconv
import (
"context"
"errors"
"fmt"
"net"
"os"
"strconv"
"strings"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
func init() {
gotenberg.MustRegisterModule(Unoconv{})
}
// ErrMalformedPageRanges happens if the page ranges option cannot be
// interpreted by LibreOffice.
var ErrMalformedPageRanges = errors.New("page ranges are malformed")
// Unoconv is a module which provides an API to interact with unoconv.
type Unoconv struct {
binPath string
}
// Options gathers available options when converting a document to PDF.
type Options struct {
// Landscape allows to change the orientation of the resulting PDF.
// Optional.
Landscape bool
// PageRanges allows to select the pages to convert.
// TODO: should prefer a method form PDFEngine.
// Optional.
PageRanges string
// PDFArchive allows to convert the resulting PDF to PDF/A-1a.
// In a module, prefer the Convert method from the gotenberg.PDFEngine
// interface.
// Optional.
PDFArchive bool
}
// API is an abstraction on top of unoconv.
//
// See https://github.com/unoconv/unoconv.
type API interface {
PDF(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options Options) error
Extensions() []string
}
// Provider is a module interface which exposes a method for creating an API
// for other modules.
//
// func (m *YourModule) Provision(ctx *gotenberg.Context) error {
// provider, _ := ctx.Module(new(unoconv.Provider))
// uno, _ := provider.(unoconv.Provider).Unoconv()
// }
type Provider interface {
Unoconv() (API, error)
}
// Descriptor returns a Unoconv's module descriptor.
func (Unoconv) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "unoconv",
New: func() gotenberg.Module { return new(Unoconv) },
}
}
// Provision sets the module properties. It returns an error if the environment
// variable UNOCONV_BIN_PATH is not set.
func (mod *Unoconv) Provision(_ *gotenberg.Context) error {
binPath, ok := os.LookupEnv("UNOCONV_BIN_PATH")
if !ok {
return errors.New("UNOCONV_BIN_PATH environment variable is not set")
}
mod.binPath = binPath
return nil
}
// Validate validates the module properties.
func (mod Unoconv) Validate() error {
_, err := os.Stat(mod.binPath)
if os.IsNotExist(err) {
return fmt.Errorf("unoconv binary path does not exist: %w", err)
}
return nil
}
// Unoconv returns an API for interacting with unoconv.
func (mod Unoconv) Unoconv() (API, error) {
return mod, nil
}
// PDF converts a document to PDF. It creates a dedicated LibreOffice instance
// thanks to a custom user profile directory and a free port. Substantial calls
// to this method may increase CPU and memory usage drastically. In such a
// scenario, the given context may also be done before the end of the
// conversion.
func (mod Unoconv) PDF(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options Options) error {
port, err := func() (int, error) {
listener, err := net.Listen("tcp", "127.0.0.1:0")
if err != nil {
return 0, fmt.Errorf("listen on the local network address: %w", err)
}
defer func() {
err := listener.Close()
if err != nil {
logger.Error(fmt.Sprintf("close listener: %s", err.Error()))
}
}()
addr := listener.Addr().String()
_, portStr, err := net.SplitHostPort(addr)
if err != nil {
return 0, fmt.Errorf("get free port from host: %w", err)
}
return strconv.Atoi(portStr)
}()
if err != nil {
return fmt.Errorf("get free port: %w", err)
}
userProfileDirPath := gotenberg.NewDirPath()
args := []string{
"--user-profile",
fmt.Sprintf("//%s", userProfileDirPath),
"--port",
fmt.Sprintf("%d", port),
"--format",
"pdf",
}
if options.Landscape {
args = append(args, "--printer", "PaperOrientation=landscape")
}
if options.PageRanges != "" {
args = append(args, "--export", fmt.Sprintf("PageRange=%s", options.PageRanges))
}
if options.PDFArchive {
args = append(args, "--export", "SelectPdfVersion=1")
}
args = append(args, "--output", outputPath, inputPath)
cmd, err := gotenberg.CommandContext(ctx, logger, mod.binPath, args...)
if err != nil {
return fmt.Errorf("create unoconv command: %w", err)
}
logger.Debug(fmt.Sprintf("print to PDF with: %+v", options))
err = cmd.Exec()
// Always remove the user profile directory created by LibreOffice.
// See https://github.com/thecodingmachine/gotenberg/issues/192.
go func() {
logger.Debug(fmt.Sprintf("remove user profile directory '%s'", userProfileDirPath))
err := os.RemoveAll(userProfileDirPath)
if err != nil {
logger.Error(fmt.Sprintf("remove user profile directory: %s", err))
}
}()
if err == nil {
return nil
}
// Unoconv/LibreOffice errors are not explicit.
// That's why we have to make an educated guess according to the exit code
// and given inputs.
if strings.Contains(err.Error(), "exit status 5") && options.PageRanges != "" {
return ErrMalformedPageRanges
}
// Possible errors:
// 1. Unoconv/LibreOffice failed for some reason.
// 2. Context done.
//
// On the second scenario, LibreOffice might not had time to remove some of
// its temporary files, as it has been killed without warning. The garbage
// collector will delete them for us (if the module is loaded).
return fmt.Errorf("unoconv PDF: %w", err)
}
// Extensions returns the file extensions available with unoconv.
func (mod Unoconv) Extensions() []string {
return []string{
".bib",
".doc",
".xml",
".docx",
".fodt",
".html",
".ltx",
".txt",
".odt",
".ott",
".pdb",
".pdf",
".psw",
".rtf",
".sdw",
".stw",
".sxw",
".uot",
".vor",
".wps",
".epub",
".png",
".bmp",
".emf",
".eps",
".fodg",
".gif",
".jpg",
".met",
".odd",
".otg",
".pbm",
".pct",
".pgm",
".ppm",
".ras",
".std",
".svg",
".svm",
".swf",
".sxd",
".sxw",
".tiff",
".xhtml",
".xpm",
".fodp",
".potm",
".pot",
".pptx",
".pps",
".ppt",
".pwp",
".sda",
".sdd",
".sti",
".sxi",
".uop",
".wmf",
".csv",
".dbf",
".dif",
".fods",
".ods",
".ots",
".pxl",
".sdc",
".slk",
".stc",
".sxc",
".uos",
".xls",
".xlt",
".xlsx",
}
}
// Interface guards.
var (
_ gotenberg.Module = (*Unoconv)(nil)
_ gotenberg.Provisioner = (*Unoconv)(nil)
_ gotenberg.Validator = (*Unoconv)(nil)
_ API = (*Unoconv)(nil)
_ Provider = (*Unoconv)(nil)
)

View File

@@ -0,0 +1,154 @@
package unoconv
import (
"context"
"os"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
func TestUnoconv_Descriptor(t *testing.T) {
descriptor := Unoconv{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(Unoconv))
if actual != expect {
t.Errorf("expected '%'s' but got '%s'", expect, actual)
}
}
func TestUnoconv_Provision(t *testing.T) {
mod := new(Unoconv)
ctx := gotenberg.NewContext(gotenberg.ParsedFlags{}, nil)
err := mod.Provision(ctx)
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestUnoconv_Validate(t *testing.T) {
for i, tc := range []struct {
binPath string
expectErr bool
}{
{
expectErr: true,
},
{
binPath: "/foo",
expectErr: true,
},
{
binPath: os.Getenv("UNOCONV_BIN_PATH"),
},
} {
mod := new(Unoconv)
mod.binPath = tc.binPath
err := mod.Validate()
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestUnoconv_Unoconv(t *testing.T) {
mod := new(Unoconv)
_, err := mod.Unoconv()
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestUnoconv_PDF(t *testing.T) {
for i, tc := range []struct {
ctx context.Context
inputPath string
options Options
expectErr bool
}{
{
expectErr: true,
},
{
ctx: context.Background(),
inputPath: "/tests/test/testdata/libreoffice/sample1.docx",
options: Options{
Landscape: true,
PageRanges: "1-2",
PDFArchive: true,
},
},
{
ctx: context.Background(),
inputPath: "/tests/test/testdata/libreoffice/sample1.docx",
options: Options{
PageRanges: "foo",
},
expectErr: true,
},
{
ctx: func() context.Context {
ctx, cancel := context.WithCancel(context.TODO())
defer cancel()
return ctx
}(),
inputPath: "/tests/test/testdata/libreoffice/sample1.docx",
expectErr: true,
},
} {
func() {
mod := new(Unoconv)
err := mod.Provision(nil)
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
outputDir, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
defer func() {
err := os.RemoveAll(outputDir)
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
}()
err = mod.PDF(tc.ctx, zap.NewNop(), tc.inputPath, outputDir+"/foo.pdf", tc.options)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}()
}
}
func TestUnoconv_Extensions(t *testing.T) {
mod := new(Unoconv)
extensions := mod.Extensions()
actual := len(extensions)
expect := 73
if actual != expect {
t.Errorf("expected %d extentions but got %d", expect, actual)
}
}

View File

@@ -0,0 +1,3 @@
// Package logging provides a module which creates a zap.Logger for other
// modules.
package logging

View File

@@ -0,0 +1,169 @@
package logging
import (
"fmt"
"os"
"time"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
flag "github.com/spf13/pflag"
"go.uber.org/multierr"
"go.uber.org/zap"
"go.uber.org/zap/zapcore"
"golang.org/x/term"
)
func init() {
gotenberg.MustRegisterModule(Logging{})
}
const (
errorLoggingLevel = "error"
warnLoggingLevel = "warn"
infoLoggingLevel = "info"
debugLoggingLevel = "debug"
)
const (
autoLoggingFormat = "auto"
jsonLoggingFormat = "json"
textLoggingFormat = "text"
)
// Logging is a module which implements the gotenberg.LoggerProvider interface.
type Logging struct {
level string
format string
}
// Descriptor returns a Logging's module descriptor.
func (Logging) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "logging",
FlagSet: func() *flag.FlagSet {
fs := flag.NewFlagSet("logging", flag.ExitOnError)
fs.String("log-level", infoLoggingLevel, fmt.Sprintf("Set the log level - %s, %s, %s, or %s", errorLoggingLevel, warnLoggingLevel, infoLoggingLevel, debugLoggingLevel))
fs.String("log-format", autoLoggingFormat, fmt.Sprintf("Set log format - %s, %s, or %s", autoLoggingFormat, jsonLoggingFormat, textLoggingFormat))
return fs
}(),
New: func() gotenberg.Module { return new(Logging) },
}
}
// Provision sets the log level and format.
func (log *Logging) Provision(ctx *gotenberg.Context) error {
flags := ctx.ParsedFlags()
log.level = flags.MustString("log-level")
log.format = flags.MustString("log-format")
return nil
}
// Validate validates the log level and format.
func (log Logging) Validate() error {
var err error
switch log.level {
case errorLoggingLevel, warnLoggingLevel, infoLoggingLevel, debugLoggingLevel:
break
default:
err = multierr.Append(
err,
fmt.Errorf("log level must be either %s, %s, %s or %s", errorLoggingLevel, warnLoggingLevel, infoLoggingLevel, debugLoggingLevel),
)
}
switch log.format {
case autoLoggingFormat, jsonLoggingFormat, textLoggingFormat:
break
default:
err = multierr.Append(
err,
fmt.Errorf("log format must be either %s, %s or %s", autoLoggingFormat, jsonLoggingFormat, textLoggingFormat),
)
}
return err
}
// Logger returns a zap.Logger.
func (log Logging) Logger(mod gotenberg.Module) (*zap.Logger, error) {
if logger == nil {
lvl, err := newLogLevel(log.level)
if err != nil {
return nil, fmt.Errorf("get log level: %w", err)
}
encoder, err := newLogEncoder(log.format)
if err != nil {
return nil, fmt.Errorf("get log encoder: %w", err)
}
core := zapcore.NewCore(encoder, os.Stderr, lvl)
logger = zap.New(core)
// nolint
defer logger.Sync()
}
return logger.Named(mod.Descriptor().ID), nil
}
func newLogLevel(level string) (zapcore.Level, error) {
switch level {
case errorLoggingLevel:
return zap.ErrorLevel, nil
case warnLoggingLevel:
return zap.WarnLevel, nil
case infoLoggingLevel:
return zap.InfoLevel, nil
case debugLoggingLevel:
return zap.DebugLevel, nil
default:
return -2, fmt.Errorf("%s is not a recognized log level", level)
}
}
func newLogEncoder(format string) (zapcore.Encoder, error) {
isTerminal := term.IsTerminal(int(os.Stdout.Fd()))
encCfg := zap.NewProductionEncoderConfig()
if isTerminal {
// If interactive terminal, make output more human-readable by default.
// Credits: https://github.com/caddyserver/caddy/blob/v2.1.1/logging.go#L671.
encCfg.EncodeTime = func(ts time.Time, encoder zapcore.PrimitiveArrayEncoder) {
encoder.AppendString(ts.UTC().Format("2006/01/02 15:04:05.000"))
}
if format == textLoggingFormat || format == autoLoggingFormat {
encCfg.EncodeLevel = zapcore.CapitalColorLevelEncoder
}
}
if format == autoLoggingFormat && isTerminal {
format = textLoggingFormat
} else if format == autoLoggingFormat {
format = jsonLoggingFormat
}
switch format {
case textLoggingFormat:
return zapcore.NewConsoleEncoder(encCfg), nil
case jsonLoggingFormat:
return zapcore.NewJSONEncoder(encCfg), nil
default:
return nil, fmt.Errorf("%s is not a recognized log format", format)
}
}
var logger *zap.Logger = nil
// Interface guards.
var (
_ gotenberg.Module = (*Logging)(nil)
_ gotenberg.Provisioner = (*Logging)(nil)
_ gotenberg.Validator = (*Logging)(nil)
_ gotenberg.LoggerProvider = (*Logging)(nil)
)

View File

@@ -0,0 +1,199 @@
package logging
import (
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap/zapcore"
)
type ProtoModule struct {
descriptor func() gotenberg.ModuleDescriptor
}
func (mod ProtoModule) Descriptor() gotenberg.ModuleDescriptor {
return mod.descriptor()
}
func TestLogging_Descriptor(t *testing.T) {
descriptor := Logging{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(Logging))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestLogging_Provision(t *testing.T) {
logging := new(Logging)
fs := logging.Descriptor().FlagSet
err := fs.Parse([]string{""})
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
ctx := gotenberg.NewContext(gotenberg.ParsedFlags{FlagSet: fs}, nil)
err = logging.Provision(ctx)
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestLogging_Validate(t *testing.T) {
for i, tc := range []struct {
level, format string
expectErr bool
}{
{
level: "foo",
expectErr: true,
},
{
level: debugLoggingLevel,
format: "foo",
expectErr: true,
},
{
level: debugLoggingLevel,
format: autoLoggingFormat,
},
} {
mod := new(Logging)
mod.level = tc.level
mod.format = tc.format
err := mod.Validate()
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestLogging_Logger(t *testing.T) {
for i, tc := range []struct {
level, format string
expectErr bool
}{
{
level: "foo",
expectErr: true,
},
{
level: debugLoggingLevel,
format: "foo",
expectErr: true,
},
{
level: debugLoggingLevel,
format: autoLoggingFormat,
},
} {
mod := new(Logging)
mod.level = tc.level
mod.format = tc.format
_, err := mod.Logger(ProtoModule{
descriptor: func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: nil}
},
})
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestNewLogLevel(t *testing.T) {
for i, tc := range []struct {
level string
expectZapLevel zapcore.Level
expectErr bool
}{
{
level: errorLoggingLevel,
expectZapLevel: zapcore.ErrorLevel,
},
{
level: warnLoggingLevel,
expectZapLevel: zapcore.WarnLevel,
},
{
level: infoLoggingLevel,
expectZapLevel: zapcore.InfoLevel,
},
{
level: debugLoggingLevel,
expectZapLevel: zapcore.DebugLevel,
},
{
level: "foo",
expectZapLevel: -2,
expectErr: true,
},
} {
actual, err := newLogLevel(tc.level)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
if tc.expectZapLevel != actual {
t.Errorf("test %d: expected %d level but got %d", i, tc.expectZapLevel, actual)
}
}
}
func TestNewLogEncoder(t *testing.T) {
for i, tc := range []struct {
format string
expectErr bool
}{
{
format: autoLoggingFormat,
},
{
format: textLoggingFormat,
},
{
format: jsonLoggingFormat,
},
{
format: "foo",
expectErr: true,
},
} {
_, err := newLogEncoder(tc.format)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
// Interface guards.
var (
_ gotenberg.Module = (*ProtoModule)(nil)
)

View File

@@ -0,0 +1,4 @@
// Package pdfcpu provides a module which wraps the
// https://github.com/pdfcpu/pdfcpu library and implements the
// gotenberg.PDFEngine interface.
package pdfcpu

View File

@@ -0,0 +1,62 @@
package pdfcpu
import (
"context"
"fmt"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
pdfcpuAPI "github.com/pdfcpu/pdfcpu/pkg/api"
pdfcpuLog "github.com/pdfcpu/pdfcpu/pkg/log"
pdfcpuConfig "github.com/pdfcpu/pdfcpu/pkg/pdfcpu"
"go.uber.org/zap"
)
func init() {
gotenberg.MustRegisterModule(PDFcpu{})
}
// PDFcpu is a module which wraps the https://github.com/pdfcpu/pdfcpu library
// and implements the gotenberg.PDFEngine interface.
type PDFcpu struct {
conf *pdfcpuConfig.Configuration
}
// Descriptor returns a PDFcpu's module descriptor.
func (engine PDFcpu) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "pdfcpu",
New: func() gotenberg.Module { return new(PDFcpu) },
}
}
// Provision sets the engine properties.
func (engine *PDFcpu) Provision(_ *gotenberg.Context) error {
pdfcpuConfig.ConfigPath = "disable"
pdfcpuLog.DisableLoggers()
engine.conf = pdfcpuConfig.NewDefaultConfiguration()
return nil
}
// Merge merges the given PDFs into a unique PDF.
func (engine PDFcpu) Merge(_ context.Context, _ *zap.Logger, inputPaths []string, outputPath string) error {
err := pdfcpuAPI.MergeCreateFile(inputPaths, outputPath, engine.conf)
if err == nil {
return nil
}
return fmt.Errorf("merge PDFs with PDFcpu: %w", err)
}
// Convert is not available for this PDF engine.
func (engine PDFcpu) Convert(_ context.Context, _ *zap.Logger, format, _, _ string) error {
return fmt.Errorf("convert PDF to '%s' with PDFcpu: %w", format, gotenberg.ErrPDFEngineMethodNotAvailable)
}
// Interface guards.
var (
_ gotenberg.Module = (*PDFcpu)(nil)
_ gotenberg.Provisioner = (*PDFcpu)(nil)
_ gotenberg.PDFEngine = (*PDFcpu)(nil)
)

View File

@@ -0,0 +1,98 @@
package pdfcpu
import (
"context"
"errors"
"os"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
func TestPDFcpu_Descriptor(t *testing.T) {
descriptor := PDFcpu{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(PDFcpu))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestPDFcpu_Provision(t *testing.T) {
mod := new(PDFcpu)
ctx := gotenberg.NewContext(gotenberg.ParsedFlags{}, nil)
err := mod.Provision(ctx)
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestPDFcpu_Merge(t *testing.T) {
for i, tc := range []struct {
inputPaths []string
expectErr bool
}{
{
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
},
},
{
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
"/tests/test/testdata/pdfengines/sample2.pdf",
},
},
{
inputPaths: []string{
"foo",
},
expectErr: true,
},
} {
func() {
mod := new(PDFcpu)
err := mod.Provision(nil)
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
outputDir, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
defer func() {
err := os.RemoveAll(outputDir)
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
}()
err = mod.Merge(nil, nil, tc.inputPaths, outputDir+"/foo.pdf")
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}()
}
}
func TestPDFcpu_Convert(t *testing.T) {
mod := new(PDFcpu)
err := mod.Convert(context.TODO(), zap.NewNop(), "", "", "")
if !errors.Is(err, gotenberg.ErrPDFEngineMethodNotAvailable) {
t.Errorf("expected error %v, but got: %v", gotenberg.ErrPDFEngineMethodNotAvailable, err)
}
}

View File

@@ -0,0 +1,3 @@
// Package pdfengines provides a module which gathers modules that implements
// the gotenberg.PDFEngine interface.
package pdfengines

View File

@@ -0,0 +1,81 @@
package pdfengines
import (
"context"
"fmt"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/multierr"
"go.uber.org/zap"
)
// multiPDFEngines implements the gotenberg.PDFEngine interface and gathers one
// or more gotenberg.PDFEngine. It provides a sort of fallback mechanism: if an
// engine's method returns an error, it calls the same method from another
// engine.
type multiPDFEngines struct {
engines []gotenberg.PDFEngine
}
// newMultiPDFEngines returns a multiPDFEngines. Arguments' order determines the
// order of the engines called.
func newMultiPDFEngines(engines ...gotenberg.PDFEngine) *multiPDFEngines {
return &multiPDFEngines{
engines: engines,
}
}
// Merge tries to merge the given PDFs into a unique PDF thanks to its
// children. If the context is done, it stops and returns an error.
func (multi multiPDFEngines) Merge(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
var err error
errChan := make(chan error, 1)
for _, engine := range multi.engines {
go func(engine gotenberg.PDFEngine) {
errChan <- engine.Merge(ctx, logger, inputPaths, outputPath)
}(engine)
select {
case mergeErr := <-errChan:
errored := multierr.AppendInto(&err, mergeErr)
if !errored {
return nil
}
case <-ctx.Done():
return ctx.Err()
}
}
return fmt.Errorf("merge PDFs with multi PDF engines: %w", err)
}
// Convert converts the given PDF to a specific PDF format. thanks to its
// children. If the context is done, it stops and returns an error.
func (multi multiPDFEngines) Convert(ctx context.Context, logger *zap.Logger, format, inputPath, outputPath string) error {
var err error
errChan := make(chan error, 1)
for _, engine := range multi.engines {
go func(engine gotenberg.PDFEngine) {
errChan <- engine.Convert(ctx, logger, format, inputPath, outputPath)
}(engine)
select {
case mergeErr := <-errChan:
errored := multierr.AppendInto(&err, mergeErr)
if !errored {
return nil
}
case <-ctx.Done():
return ctx.Err()
}
}
return fmt.Errorf("convert PDF to '%s' with multi PDF engines: %w", format, err)
}
// Interface guards.
var (
_ gotenberg.PDFEngine = (*multiPDFEngines)(nil)
)

View File

@@ -0,0 +1,215 @@
package pdfengines
import (
"context"
"errors"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
func TestNewMultiPDFEngines(t *testing.T) {
engine1 := &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
engine2 := &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return errors.New("foo")
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
}
multi := newMultiPDFEngines(engine1, engine2)
if len(multi.engines) != 2 {
t.Fatalf("expected %d engines but got %d", 2, len(multi.engines))
}
if !reflect.DeepEqual(engine1, multi.engines[0]) {
t.Errorf("expected %v, but got: %v", engine1, multi.engines[0])
}
if !reflect.DeepEqual(engine2, multi.engines[1]) {
t.Errorf("expected %v, but got: %v", engine2, multi.engines[1])
}
}
func TestMultiPDFEngines_Merge(t *testing.T) {
for i, tc := range []struct {
ctx context.Context
engines []gotenberg.PDFEngine
expectErr bool
}{
{
ctx: context.TODO(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
},
}
}(),
},
{
ctx: context.TODO(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return errors.New("foo")
},
},
ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
},
}
}(),
},
{
ctx: context.TODO(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return errors.New("foo")
},
},
ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return errors.New("bar")
},
},
}
}(),
expectErr: true,
},
{
ctx: func() context.Context {
ctx, cancel := context.WithCancel(context.TODO())
defer cancel()
return ctx
}(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
},
}
}(),
expectErr: true,
},
} {
multi := newMultiPDFEngines(tc.engines...)
err := multi.Merge(tc.ctx, nil, nil, "")
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestMultiPDFEngines_Convert(t *testing.T) {
for i, tc := range []struct {
ctx context.Context
engines []gotenberg.PDFEngine
expectErr bool
}{
{
ctx: context.TODO(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
},
}
}(),
},
{
ctx: context.TODO(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
},
ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
},
}
}(),
},
{
ctx: context.TODO(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
},
ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("bar")
},
},
}
}(),
expectErr: true,
},
{
ctx: func() context.Context {
ctx, cancel := context.WithCancel(context.TODO())
defer cancel()
return ctx
}(),
engines: func() []gotenberg.PDFEngine {
return []gotenberg.PDFEngine{
ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
},
}
}(),
expectErr: true,
},
} {
multi := newMultiPDFEngines(tc.engines...)
err := multi.Convert(tc.ctx, nil, "", "", "")
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}

View File

@@ -0,0 +1,160 @@
package pdfengines
import (
"errors"
"fmt"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
flag "github.com/spf13/pflag"
)
func init() {
gotenberg.MustRegisterModule(PDFEngines{})
}
// PDFEngines is a module which gathers available gotenberg.PDFEngine modules.
// The available gotenberg.PDFEngine modules can be either all
// gotenberg.PDFEngine modules or the modules selected by the user thanks to
// the "engines" flag.
//
// PDFEngines wraps the gotenberg.PDFEngine modules in an internal struct which
// also implements gotenberg.PDFEngine. This struct provides a sort of fallback
// mechanism: if an engine's method returns an error, it calls the same method
// from another engine.
//
// This module implements the gotenberg.PDFEngineProvider interface.
type PDFEngines struct {
names []string
engines []gotenberg.PDFEngine
disableRoutes bool
}
// Descriptor returns a PDFEngines' module descriptor.
func (PDFEngines) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "pdfengines",
FlagSet: func() *flag.FlagSet {
fs := flag.NewFlagSet("pdfengines", flag.ExitOnError)
fs.StringSlice("pdfengines-engines", make([]string, 0), "Set the PDF engines - all by default")
fs.Bool("pdfengines-disable-routes", false, "Disable the routes")
return fs
}(),
New: func() gotenberg.Module { return new(PDFEngines) },
}
}
// Provision gets either all gotenberg.PDFEngine modules or the modules
// selected by the user thanks to the "engines" flag.
func (mod *PDFEngines) Provision(ctx *gotenberg.Context) error {
flags := ctx.ParsedFlags()
names := flags.MustStringSlice("pdfengines-engines")
mod.disableRoutes = flags.MustBool("pdfengines-disable-routes")
engines, err := ctx.Modules(new(gotenberg.PDFEngine))
if err != nil {
return fmt.Errorf("get PDF engines: %w", err)
}
mod.engines = make([]gotenberg.PDFEngine, len(engines))
for i, engine := range engines {
mod.engines[i] = engine.(gotenberg.PDFEngine)
}
if len(names) > 0 {
// Selection from user.
mod.names = names
return nil
}
// No selection from user, use all PDF engines available.
mod.names = make([]string, len(mod.engines))
for i, engine := range mod.engines {
mod.names[i] = engine.(gotenberg.Module).Descriptor().ID
}
return nil
}
// Validate validates there is at least one gotenberg.PDFEngine module
// available. It also validates that selected gotenberg.PDFEngine modules
// actually exist.
func (mod PDFEngines) Validate() error {
if len(mod.engines) == 0 {
return errors.New("no PDF engine")
}
availableEngines := make([]string, len(mod.engines))
for i, engine := range mod.engines {
availableEngines[i] = engine.(gotenberg.Module).Descriptor().ID
}
nonExistingEngines := make([]string, 0)
for _, name := range mod.names {
engineExists := false
for _, engine := range mod.engines {
if name == engine.(gotenberg.Module).Descriptor().ID {
engineExists = true
break
}
}
if !engineExists {
nonExistingEngines = append(nonExistingEngines, name)
}
}
if len(nonExistingEngines) == 0 {
return nil
}
return fmt.Errorf("non-existing PDF engine(s): %s - available PDF engine(s): %s", nonExistingEngines, availableEngines)
}
// PDFEngine returns a gotenberg.PDFEngine.
func (mod PDFEngines) PDFEngine() (gotenberg.PDFEngine, error) {
engines := make([]gotenberg.PDFEngine, len(mod.engines))
i := 0
for _, engine := range mod.engines {
engines[i] = engine
i++
}
return newMultiPDFEngines(engines...), nil
}
// Routes returns the API routes.
func (mod PDFEngines) Routes() ([]api.MultipartFormDataRoute, error) {
if mod.disableRoutes {
return nil, nil
}
engine, err := mod.PDFEngine()
if err != nil {
// Should not happen, unless our provider implementation
// changes in the future.
return nil, fmt.Errorf("get pdf engine: %w", err)
}
return []api.MultipartFormDataRoute{
mergeRoute(engine),
convertRoute(engine),
}, nil
}
// Interface guards.
var (
_ gotenberg.Module = (*PDFEngines)(nil)
_ gotenberg.Provisioner = (*PDFEngines)(nil)
_ gotenberg.Validator = (*PDFEngines)(nil)
_ gotenberg.PDFEngineProvider = (*PDFEngines)(nil)
_ api.MultipartFormDataRouter = (*PDFEngines)(nil)
)

View File

@@ -0,0 +1,304 @@
package pdfengines
import (
"context"
"errors"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
type ProtoModule struct {
descriptor func() gotenberg.ModuleDescriptor
}
func (mod ProtoModule) Descriptor() gotenberg.ModuleDescriptor {
return mod.descriptor()
}
type ProtoValidator struct {
ProtoModule
validate func() error
}
func (mod ProtoValidator) Validate() error {
return mod.validate()
}
type ProtoPDFEngine struct {
ProtoValidator
merge func(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error
convert func(ctx context.Context, logger *zap.Logger, format, inputPath, outputPath string) error
}
func (mod ProtoPDFEngine) Merge(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
return mod.merge(ctx, logger, inputPaths, outputPath)
}
func (mod ProtoPDFEngine) Convert(ctx context.Context, logger *zap.Logger, format, inputPath, outputPath string) error {
return mod.convert(ctx, logger, format, inputPath, outputPath)
}
func TestPDFEngine_Descriptor(t *testing.T) {
descriptor := PDFEngines{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(PDFEngines))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestPDFEngine_Provision(t *testing.T) {
for i, tc := range []struct {
ctx *gotenberg.Context
expectNames []string
expectEnginesCount int
expectErr bool
}{
{
ctx: func() *gotenberg.Context {
engine := struct{ ProtoPDFEngine }{}
engine.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return engine }}
}
engine.validate = func() error { return errors.New("foo") }
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(PDFEngines).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
engine.Descriptor(),
},
)
}(),
expectErr: true,
},
{
ctx: func() *gotenberg.Context {
engine := struct{ ProtoPDFEngine }{}
engine.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return engine }}
}
engine.validate = func() error { return nil }
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: new(PDFEngines).Descriptor().FlagSet,
},
[]gotenberg.ModuleDescriptor{
engine.Descriptor(),
},
)
}(),
expectNames: []string{"foo"},
expectEnginesCount: 1,
},
{
ctx: func() *gotenberg.Context {
engine1 := struct{ ProtoPDFEngine }{}
engine1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "a", New: func() gotenberg.Module { return engine1 }}
}
engine1.validate = func() error { return nil }
engine2 := struct{ ProtoPDFEngine }{}
engine2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "b", New: func() gotenberg.Module { return engine2 }}
}
engine2.validate = func() error { return nil }
fs := new(PDFEngines).Descriptor().FlagSet
err := fs.Parse([]string{"--pdfengines-engines=b", "--pdfengines-engines=a"})
if err != nil {
t.Fatalf("expected no error but got: %v", err)
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: fs,
},
[]gotenberg.ModuleDescriptor{
engine1.Descriptor(),
engine2.Descriptor(),
},
)
}(),
expectNames: []string{"b", "a"},
expectEnginesCount: 2,
},
{
ctx: func() *gotenberg.Context {
engine1 := struct{ ProtoPDFEngine }{}
engine1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "a", New: func() gotenberg.Module { return engine1 }}
}
engine1.validate = func() error { return nil }
engine2 := struct{ ProtoPDFEngine }{}
engine2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "b", New: func() gotenberg.Module { return engine2 }}
}
engine2.validate = func() error { return nil }
fs := new(PDFEngines).Descriptor().FlagSet
err := fs.Parse([]string{"--pdfengines-engines=b"})
if err != nil {
t.Fatalf("expected error but got: %v", err)
}
return gotenberg.NewContext(
gotenberg.ParsedFlags{
FlagSet: fs,
},
[]gotenberg.ModuleDescriptor{
engine1.Descriptor(),
engine2.Descriptor(),
},
)
}(),
expectNames: []string{"b"},
expectEnginesCount: 2,
},
} {
mod := new(PDFEngines)
err := mod.Provision(tc.ctx)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
if len(tc.expectNames) != len(mod.names) {
t.Errorf("test %d: expected %d names but got %d", i, len(tc.expectNames), len(mod.names))
}
if tc.expectEnginesCount != len(mod.engines) {
t.Errorf("test %d: expected %d engines but got %d", i, tc.expectEnginesCount, len(mod.engines))
}
for index, name := range mod.names {
if name != tc.expectNames[index] {
t.Errorf("test %d: expected name at index %d to be %s, but got: %s", i, index, name, tc.expectNames[index])
}
}
}
}
func TestPDFEngine_Validate(t *testing.T) {
for i, tc := range []struct {
names []string
engines []gotenberg.PDFEngine
expectErr bool
}{
{
expectErr: true,
},
{
names: []string{"foo"},
engines: func() []gotenberg.PDFEngine {
engine := struct{ ProtoPDFEngine }{}
engine.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return engine }}
}
return []gotenberg.PDFEngine{
engine,
}
}(),
},
{
names: []string{"foo", "bar", "baz"},
engines: func() []gotenberg.PDFEngine {
engine1 := struct{ ProtoPDFEngine }{}
engine1.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "foo", New: func() gotenberg.Module { return engine1 }}
}
engine2 := struct{ ProtoPDFEngine }{}
engine2.descriptor = func() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{ID: "baz", New: func() gotenberg.Module { return engine2 }}
}
return []gotenberg.PDFEngine{
engine1,
engine2,
}
}(),
expectErr: true,
},
} {
mod := new(PDFEngines)
mod.names = tc.names
mod.engines = tc.engines
err := mod.Validate()
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestPDFEngine_PDFEngine(t *testing.T) {
mod := new(PDFEngines)
mod.engines = []gotenberg.PDFEngine{
struct{ ProtoPDFEngine }{},
}
_, err := mod.PDFEngine()
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestPDFEngine_Routes(t *testing.T) {
for i, tc := range []struct {
expectRoutes int
disableRoutes bool
}{
{
expectRoutes: 2,
},
{
disableRoutes: true,
},
} {
mod := new(PDFEngines)
mod.engines = []gotenberg.PDFEngine{
struct{ ProtoPDFEngine }{},
}
mod.disableRoutes = tc.disableRoutes
routes, err := mod.Routes()
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
if tc.expectRoutes != len(routes) {
t.Errorf("test %d: expected %d routes but got %d", i, tc.expectRoutes, len(routes))
}
}
}
// Interface guards.
var (
_ gotenberg.Module = (*ProtoModule)(nil)
_ gotenberg.Validator = (*ProtoValidator)(nil)
_ gotenberg.Module = (*ProtoValidator)(nil)
_ gotenberg.PDFEngine = (*ProtoPDFEngine)(nil)
_ gotenberg.Module = (*ProtoPDFEngine)(nil)
_ gotenberg.Validator = (*ProtoPDFEngine)(nil)
)

View File

@@ -0,0 +1,138 @@
package pdfengines
import (
"errors"
"fmt"
"net/http"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
)
// mergeRoute returns an api.MultipartFormDataRoute which can merge PDFs.
func mergeRoute(engine gotenberg.PDFEngine) api.MultipartFormDataRoute {
return api.MultipartFormDataRoute{
Path: "/pdfengines/merge",
Handler: func(ctx *api.Context) error {
// Let's get the data from the form and validate them.
var (
inputPaths []string
PDFformat string
)
err := ctx.FormData().
MandatoryPaths([]string{".pdf"}, &inputPaths).
String("pdfFormat", &PDFformat, "").
Validate()
if err != nil {
return fmt.Errorf("validate form data: %w", err)
}
// Alright, let's merge the PDFs.
outputPath := ctx.GeneratePath(".pdf")
err = engine.Merge(ctx, ctx.Log(), inputPaths, outputPath)
if err != nil {
return fmt.Errorf("merge PDFs: %w", err)
}
// So far so good, the PDFs are merged into one unique PDF.
// Now, let's check if the client want to convert this result PDF
// to a specific PDF format.
if PDFformat != "" {
convertInputPath := outputPath
convertOutputPath := ctx.GeneratePath(".pdf")
err = engine.Convert(ctx, ctx.Log(), PDFformat, convertInputPath, convertOutputPath)
if err != nil {
if errors.Is(err, gotenberg.ErrPDFFormatNotAvailable) {
return api.WrapError(
fmt.Errorf("convert PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("At least one PDF engine does not handle the PDF format '%s' (pdfFormat), while other have failed to convert for other reasons", PDFformat),
),
)
}
return fmt.Errorf("convert PDF: %w", err)
}
// Important: the output path is now the converted file.
outputPath = convertOutputPath
}
// Last but not least, add the output path to the context so that
// the API is able to send it as a response to the client.
err = ctx.AddOutputPaths(outputPath)
if err != nil {
return fmt.Errorf("add output path: %w", err)
}
return nil
},
}
}
// convertRoute returns an api.MultipartFormDataRoute which can convert a PDF
// to a specific PDF format.
func convertRoute(engine gotenberg.PDFEngine) api.MultipartFormDataRoute {
return api.MultipartFormDataRoute{
Path: "/pdfengines/convert",
Handler: func(ctx *api.Context) error {
// Let's get the data from the form and validate them.
var (
inputPaths []string
PDFformat string
)
err := ctx.FormData().
MandatoryPaths([]string{".pdf"}, &inputPaths).
MandatoryString("pdfFormat", &PDFformat).
Validate()
if err != nil {
return fmt.Errorf("validate form data: %w", err)
}
// Alright, let's merge the PDFs.
outputPaths := make([]string, len(inputPaths))
for i, inputPath := range inputPaths {
outputPaths[i] = ctx.GeneratePath(".pdf")
err = engine.Convert(ctx, ctx.Log(), PDFformat, inputPath, outputPaths[i])
if err != nil {
if errors.Is(err, gotenberg.ErrPDFFormatNotAvailable) {
return api.WrapError(
fmt.Errorf("convert PDF: %w", err),
api.NewSentinelHTTPError(
http.StatusBadRequest,
fmt.Sprintf("At least one PDF engine does not handle the PDF format '%s' (pdfFormat), while other have failed to convert for other reasons", PDFformat),
),
)
}
return fmt.Errorf("convert PDF: %w", err)
}
}
// Last but not least, add the output paths to the context so that
// the API is able to send them as a response to the client.
err = ctx.AddOutputPaths(outputPaths...)
if err != nil {
return fmt.Errorf("add output paths: %w", err)
}
return nil
},
}
}

View File

@@ -0,0 +1,392 @@
package pdfengines
import (
"context"
"errors"
"net/http"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"github.com/gotenberg/gotenberg/v7/pkg/modules/api"
"go.uber.org/zap"
)
func TestMergeHandler(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
engine gotenberg.PDFEngine
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
expectOutputPathsCount int
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return errors.New("foo")
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return gotenberg.ErrPDFFormatNotAvailable
},
}
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetCancelled(true)
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectOutputPathsCount: 1,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
merge: func(_ context.Context, _ *zap.Logger, _ []string, _ string) error {
return nil
},
}
}(),
expectOutputPathsCount: 1,
},
} {
err := mergeRoute(tc.engine).Handler(tc.ctx.Context)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr api.HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
if tc.expectOutputPathsCount != len(tc.ctx.OutputPaths()) {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectOutputPathsCount, len(tc.ctx.OutputPaths()))
}
}
}
func TestConvertHandler(t *testing.T) {
for i, tc := range []struct {
ctx *api.MockContext
engine gotenberg.PDFEngine
expectErr bool
expectHTTPErr bool
expectHTTPStatus int
expectOutputPathsCount int
}{
{
ctx: &api.MockContext{Context: &api.Context{}},
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return gotenberg.ErrPDFFormatNotAvailable
},
}
}(),
expectErr: true,
expectHTTPErr: true,
expectHTTPStatus: http.StatusBadRequest,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return errors.New("foo")
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetCancelled(true)
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectErr: true,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectOutputPathsCount: 1,
},
{
ctx: func() *api.MockContext {
ctx := &api.MockContext{Context: &api.Context{}}
ctx.SetFiles(map[string]string{
"foo.pdf": "/foo/foo.pdf",
"bar.pdf": "/foo/bar.pdf",
})
ctx.SetValues(map[string][]string{
"pdfFormat": {
"foo",
},
})
return ctx
}(),
engine: func() gotenberg.PDFEngine {
return &ProtoPDFEngine{
convert: func(_ context.Context, _ *zap.Logger, _, _, _ string) error {
return nil
},
}
}(),
expectOutputPathsCount: 2,
},
} {
err := convertRoute(tc.engine).Handler(tc.ctx.Context)
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
var httpErr api.HTTPError
isHTTPErr := errors.As(err, &httpErr)
if tc.expectHTTPErr && !isHTTPErr {
t.Errorf("test %d: expected HTTP error but got: %v", i, err)
}
if !tc.expectHTTPErr && isHTTPErr {
t.Errorf("test %d: expected no HTTP error but got one: %v", i, httpErr)
}
if err != nil && tc.expectHTTPErr && isHTTPErr {
status, _ := httpErr.HTTPError()
if status != tc.expectHTTPStatus {
t.Errorf("test %d: expected %d HTTP status code but got %d", i, tc.expectHTTPStatus, status)
}
}
if tc.expectOutputPathsCount != len(tc.ctx.OutputPaths()) {
t.Errorf("test %d: expected %d output paths but got %d", i, tc.expectOutputPathsCount, len(tc.ctx.OutputPaths()))
}
}
}

3
pkg/modules/pdftk/doc.go Normal file
View File

@@ -0,0 +1,3 @@
// Package pdftk provides a module which abstracts the CLI tool PDFtk and
// implements the gotenberg.PDFEngine interface.
package pdftk

View File

@@ -0,0 +1,84 @@
package pdftk
import (
"context"
"errors"
"fmt"
"os"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
func init() {
gotenberg.MustRegisterModule(PDFtk{})
}
// PDFtk abstracts the CLI tool PDFtk and implements the gotenberg.PDFEngine
// interface.
type PDFtk struct {
binPath string
}
// Descriptor returns a PDFtk's module descriptor.
func (engine PDFtk) Descriptor() gotenberg.ModuleDescriptor {
return gotenberg.ModuleDescriptor{
ID: "pdftk",
New: func() gotenberg.Module { return new(PDFtk) },
}
}
// Provision sets the modules properties. It returns an error if the
// environment variable PDFTK_BIN_PATH is not set.
func (engine *PDFtk) Provision(_ *gotenberg.Context) error {
binPath, ok := os.LookupEnv("PDFTK_BIN_PATH")
if !ok {
return errors.New("PDFTK_BIN_PATH environment variable is not set")
}
engine.binPath = binPath
return nil
}
// Validate validates the module properties.
func (engine PDFtk) Validate() error {
_, err := os.Stat(engine.binPath)
if os.IsNotExist(err) {
return fmt.Errorf("PDFtk binary path does not exist: %w", err)
}
return nil
}
// Merge merges the given PDFs into a unique PDF.
func (engine PDFtk) Merge(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
var args []string
args = append(args, inputPaths...)
args = append(args, "cat", "output", outputPath)
cmd, err := gotenberg.CommandContext(ctx, logger, engine.binPath, args...)
if err != nil {
return fmt.Errorf("create command: %w", err)
}
err = cmd.Exec()
if err == nil {
return nil
}
return fmt.Errorf("merge PDFs with PDFtk: %w", err)
}
// Convert is not available for this PDF engine.
func (engine PDFtk) Convert(_ context.Context, _ *zap.Logger, format, _, _ string) error {
return fmt.Errorf("convert PDF to '%s' with PDFtk: %w", format, gotenberg.ErrPDFEngineMethodNotAvailable)
}
// Interface guards.
var (
_ gotenberg.Module = (*PDFtk)(nil)
_ gotenberg.Provisioner = (*PDFtk)(nil)
_ gotenberg.Validator = (*PDFtk)(nil)
_ gotenberg.PDFEngine = (*PDFtk)(nil)
)

View File

@@ -0,0 +1,136 @@
package pdftk
import (
"context"
"errors"
"os"
"reflect"
"testing"
"github.com/gotenberg/gotenberg/v7/pkg/gotenberg"
"go.uber.org/zap"
)
func TestPDFtk_Descriptor(t *testing.T) {
descriptor := PDFtk{}.Descriptor()
actual := reflect.TypeOf(descriptor.New())
expect := reflect.TypeOf(new(PDFtk))
if actual != expect {
t.Errorf("expected '%s' but got '%s'", expect, actual)
}
}
func TestPDFtk_Provision(t *testing.T) {
mod := new(PDFtk)
ctx := gotenberg.NewContext(gotenberg.ParsedFlags{}, nil)
err := mod.Provision(ctx)
if err != nil {
t.Errorf("expected no error but got: %v", err)
}
}
func TestPDFtk_Validate(t *testing.T) {
for i, tc := range []struct {
binPath string
expectErr bool
}{
{
expectErr: true,
},
{
binPath: "/foo",
expectErr: true,
},
{
binPath: os.Getenv("PDFTK_BIN_PATH"),
},
} {
mod := new(PDFtk)
mod.binPath = tc.binPath
err := mod.Validate()
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}
}
func TestPDFtk_Merge(t *testing.T) {
for i, tc := range []struct {
ctx context.Context
inputPaths []string
expectErr bool
}{
{
ctx: context.TODO(),
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
},
},
{
ctx: context.TODO(),
inputPaths: []string{
"/tests/test/testdata/pdfengines/sample1.pdf",
"/tests/test/testdata/pdfengines/sample2.pdf",
},
},
{
ctx: nil,
expectErr: true,
},
{
ctx: context.TODO(),
inputPaths: []string{
"foo",
},
expectErr: true,
},
} {
func() {
mod := new(PDFtk)
err := mod.Provision(nil)
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
outputDir, err := gotenberg.MkdirAll()
if err != nil {
t.Fatalf("test %d: expected error but got: %v", i, err)
}
defer func() {
err := os.RemoveAll(outputDir)
if err != nil {
t.Fatalf("test %d: expected no error but got: %v", i, err)
}
}()
err = mod.Merge(tc.ctx, zap.NewNop(), tc.inputPaths, outputDir+"/foo.pdf")
if tc.expectErr && err == nil {
t.Errorf("test %d: expected error but got: %v", i, err)
}
if !tc.expectErr && err != nil {
t.Errorf("test %d: expected no error but got: %v", i, err)
}
}()
}
}
func TestPDFtk_Convert(t *testing.T) {
mod := new(PDFtk)
err := mod.Convert(context.TODO(), zap.NewNop(), "", "", "")
if !errors.Is(err, gotenberg.ErrPDFEngineMethodNotAvailable) {
t.Errorf("expected error %v, but got: %v", gotenberg.ErrPDFEngineMethodNotAvailable, err)
}
}