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

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)
}
}