mirror of
https://github.com/gotenberg/gotenberg.git
synced 2026-08-17 20:52:14 +01:00
process load balancing: broken but in progress
This commit is contained in:
265
internal/pkg/print/chrome.go
Normal file
265
internal/pkg/print/chrome.go
Normal file
@@ -0,0 +1,265 @@
|
||||
package print
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"io/ioutil"
|
||||
"time"
|
||||
|
||||
"github.com/mafredri/cdp"
|
||||
"github.com/mafredri/cdp/devtool"
|
||||
"github.com/mafredri/cdp/protocol/network"
|
||||
"github.com/mafredri/cdp/protocol/page"
|
||||
"github.com/mafredri/cdp/protocol/target"
|
||||
"github.com/mafredri/cdp/rpcc"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/process"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xerrgroup"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xerror"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xlog"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xtime"
|
||||
)
|
||||
|
||||
type chromePrint struct {
|
||||
logger xlog.Logger
|
||||
url string
|
||||
opts ChromePrintOptions
|
||||
}
|
||||
|
||||
// ChromePrintOptions helps customizing the
|
||||
// Google Chrome Print result.
|
||||
type ChromePrintOptions struct {
|
||||
WaitDelay float64
|
||||
HeaderHTML string
|
||||
FooterHTML string
|
||||
PaperWidth float64
|
||||
PaperHeight float64
|
||||
MarginTop float64
|
||||
MarginBottom float64
|
||||
MarginLeft float64
|
||||
MarginRight float64
|
||||
Landscape bool
|
||||
}
|
||||
|
||||
// DefaultChromePrintOptions returns the default
|
||||
// Google Chrome Print options.
|
||||
func DefaultChromePrintOptions() ChromePrintOptions {
|
||||
const defaultHeaderFooterHTML string = "<html><head></head><body></body></html>"
|
||||
return ChromePrintOptions{
|
||||
WaitDelay: 0.0,
|
||||
HeaderHTML: defaultHeaderFooterHTML,
|
||||
FooterHTML: defaultHeaderFooterHTML,
|
||||
PaperWidth: 8.27,
|
||||
PaperHeight: 11.7,
|
||||
MarginTop: 1.0,
|
||||
MarginBottom: 1.0,
|
||||
MarginLeft: 1.0,
|
||||
MarginRight: 1.0,
|
||||
Landscape: false,
|
||||
}
|
||||
}
|
||||
|
||||
func (p chromePrint) Print(ctx context.Context, dest string, proc process.Process) error {
|
||||
const op string = "print.chromePrint.Print"
|
||||
resolver := func() error {
|
||||
devtEndpoint := fmt.Sprintf("http://%s:%d", proc.Host(), proc.Port())
|
||||
devt, err := devtool.New(devtEndpoint).Version(ctx)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
// connect to WebSocket URL (page) that speaks the Chrome DevTools Protocol.
|
||||
devtConn, err := rpcc.DialContext(ctx, devt.WebSocketDebuggerURL)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer devtConn.Close() // nolint: errcheck
|
||||
// create a new CDP Client that uses conn.
|
||||
devtClient := cdp.NewClient(devtConn)
|
||||
newContextTarget, err := devtClient.Target.CreateBrowserContext(ctx)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
/*
|
||||
close the browser context when done.
|
||||
we're not using the "default" context
|
||||
as it may timeout before actually closing
|
||||
the browser context.
|
||||
see: https://github.com/mafredri/cdp/issues/101#issuecomment-524533670
|
||||
*/
|
||||
disposeBrowserContextArgs := target.NewDisposeBrowserContextArgs(newContextTarget.BrowserContextID)
|
||||
defer devtClient.Target.DisposeBrowserContext(context.Background(), disposeBrowserContextArgs) // nolint: errcheck
|
||||
// create a new blank target with the new browser context.
|
||||
createTargetArgs := target.
|
||||
NewCreateTargetArgs("about:blank").
|
||||
SetBrowserContextID(newContextTarget.BrowserContextID)
|
||||
newTarget, err := devtClient.Target.CreateTarget(ctx, createTargetArgs)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
// connect the client to the new target.
|
||||
newTargetWsURL := fmt.Sprintf("ws://%s:%d/devtools/page/%s", proc.Host(), proc.Port(), newTarget.TargetID)
|
||||
newContextConn, err := rpcc.DialContext(ctx, newTargetWsURL)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer newContextConn.Close() // nolint: errcheck
|
||||
// create a new CDP Client that uses newContextConn.
|
||||
targetClient := cdp.NewClient(newContextConn)
|
||||
/*
|
||||
close the target when done.
|
||||
we're not using the "default" context
|
||||
as it may timeout before actually closing
|
||||
the target.
|
||||
see: https://github.com/mafredri/cdp/issues/101#issuecomment-524533670
|
||||
*/
|
||||
closeTargetArgs := target.NewCloseTargetArgs(newTarget.TargetID)
|
||||
defer targetClient.Target.CloseTarget(context.Background(), closeTargetArgs) // nolint: errcheck
|
||||
// enable all events.
|
||||
if err := p.enableEvents(ctx, targetClient); err != nil {
|
||||
return err
|
||||
}
|
||||
// listen for all events.
|
||||
if err := p.listenEvents(ctx, targetClient); err != nil {
|
||||
return err
|
||||
}
|
||||
// apply a wait delay (if any).
|
||||
if p.opts.WaitDelay > 0.0 {
|
||||
// wait for a given amount of time (useful for javascript delay).
|
||||
p.logger.DebugfOp(op, "applying a wait delay of '%.2fs'...", p.opts.WaitDelay)
|
||||
time.Sleep(xtime.Duration(p.opts.WaitDelay))
|
||||
} else {
|
||||
p.logger.DebugOp(op, "no wait delay to apply, moving on...")
|
||||
}
|
||||
// print the page to PDF.
|
||||
print, err := targetClient.Page.PrintToPDF(
|
||||
ctx,
|
||||
page.NewPrintToPDFArgs().
|
||||
SetPaperWidth(p.opts.PaperWidth).
|
||||
SetPaperHeight(p.opts.PaperHeight).
|
||||
SetMarginTop(p.opts.MarginTop).
|
||||
SetMarginBottom(p.opts.MarginBottom).
|
||||
SetMarginLeft(p.opts.MarginLeft).
|
||||
SetMarginRight(p.opts.MarginRight).
|
||||
SetLandscape(p.opts.Landscape).
|
||||
SetDisplayHeaderFooter(true).
|
||||
SetHeaderTemplate(p.opts.HeaderHTML).
|
||||
SetFooterTemplate(p.opts.FooterHTML).
|
||||
SetPrintBackground(true),
|
||||
)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
if err := ioutil.WriteFile(dest, print.Data, 0644); err != nil {
|
||||
return err
|
||||
}
|
||||
return nil
|
||||
}
|
||||
if err := resolver(); err != nil {
|
||||
return xerror.New(op, err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
func (p chromePrint) enableEvents(ctx context.Context, client *cdp.Client) error {
|
||||
const op string = "print.chromePrint.enableEvents"
|
||||
// enable all the domain events that we're interested in.
|
||||
if err := xerrgroup.Run(
|
||||
func() error { return client.DOM.Enable(ctx) },
|
||||
func() error { return client.Network.Enable(ctx, network.NewEnableArgs()) },
|
||||
func() error { return client.Page.Enable(ctx) },
|
||||
func() error {
|
||||
return client.Page.SetLifecycleEventsEnabled(ctx, page.NewSetLifecycleEventsEnabledArgs(true))
|
||||
},
|
||||
func() error { return client.Runtime.Enable(ctx) },
|
||||
); err != nil {
|
||||
return xerror.New(op, err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
func (p chromePrint) listenEvents(ctx context.Context, client *cdp.Client) error {
|
||||
const op string = "print.chromePrint.listenEvents"
|
||||
resolver := func() error {
|
||||
// make sure Page events are enabled.
|
||||
if err := client.Page.Enable(ctx); err != nil {
|
||||
return err
|
||||
}
|
||||
// make sure Network events are enabled.
|
||||
if err := client.Network.Enable(ctx, nil); err != nil {
|
||||
return err
|
||||
}
|
||||
// create all clients for events.
|
||||
domContentEventFired, err := client.Page.DOMContentEventFired(ctx)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer domContentEventFired.Close() // nolint: errcheck
|
||||
loadEventFired, err := client.Page.LoadEventFired(ctx)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer loadEventFired.Close() // nolint: errcheck
|
||||
lifecycleEvent, err := client.Page.LifecycleEvent(ctx)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer lifecycleEvent.Close() // nolint: errcheck
|
||||
loadingFinished, err := client.Network.LoadingFinished(ctx)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer loadingFinished.Close() // nolint: errcheck
|
||||
if _, err := client.Page.Navigate(ctx, page.NewNavigateArgs(p.url)); err != nil {
|
||||
return err
|
||||
}
|
||||
// wait for all events.
|
||||
return xerrgroup.Run(
|
||||
func() error {
|
||||
_, err := domContentEventFired.Recv()
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
p.logger.DebugOp(op, "event 'domContentEventFired' received")
|
||||
return nil
|
||||
},
|
||||
func() error {
|
||||
_, err := loadEventFired.Recv()
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
p.logger.DebugOp(op, "event 'loadEventFired' received")
|
||||
return nil
|
||||
},
|
||||
func() error {
|
||||
const networkIdleEventName string = "networkIdle"
|
||||
for {
|
||||
ev, err := lifecycleEvent.Recv()
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
p.logger.DebugfOp(op, "event '%s' received", ev.Name)
|
||||
if ev.Name == networkIdleEventName {
|
||||
break
|
||||
}
|
||||
}
|
||||
return nil
|
||||
},
|
||||
func() error {
|
||||
_, err := loadingFinished.Recv()
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
p.logger.DebugOp(op, "event 'loadingFinished' received")
|
||||
return nil
|
||||
},
|
||||
)
|
||||
}
|
||||
if err := resolver(); err != nil {
|
||||
return xerror.New(op, err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// Compile-time checks to ensure type implements desired interfaces.
|
||||
var (
|
||||
_ = Print(new(chromePrint))
|
||||
)
|
||||
1
internal/pkg/print/doc.go
Normal file
1
internal/pkg/print/doc.go
Normal file
@@ -0,0 +1 @@
|
||||
package print
|
||||
18
internal/pkg/print/html.go
Normal file
18
internal/pkg/print/html.go
Normal file
@@ -0,0 +1,18 @@
|
||||
package print
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xlog"
|
||||
)
|
||||
|
||||
// NewHTMLPrint returns a Print for
|
||||
// converting an HTML file to PDF.
|
||||
func NewHTMLPrint(logger xlog.Logger, fpath string, opts ChromePrintOptions) Print {
|
||||
URL := fmt.Sprintf("file://%s", fpath)
|
||||
return chromePrint{
|
||||
logger: logger,
|
||||
url: URL,
|
||||
opts: opts,
|
||||
}
|
||||
}
|
||||
70
internal/pkg/print/markdown.go
Normal file
70
internal/pkg/print/markdown.go
Normal file
@@ -0,0 +1,70 @@
|
||||
package print
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"fmt"
|
||||
"html/template"
|
||||
"io/ioutil"
|
||||
"path/filepath"
|
||||
|
||||
"github.com/microcosm-cc/bluemonday"
|
||||
"github.com/russross/blackfriday/v2"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xerror"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xlog"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xrand"
|
||||
)
|
||||
|
||||
// NewMarkdownPrint returns a Print for
|
||||
// converting a Markdown file to PDF.
|
||||
func NewMarkdownPrint(logger xlog.Logger, fpath string, opts ChromePrintOptions) (Print, error) {
|
||||
const op string = "print.NewMarkdownPrint"
|
||||
resolver := func() (string, error) {
|
||||
tmpl, err := template.
|
||||
New(filepath.Base(fpath)).
|
||||
Funcs(template.FuncMap{"toHTML": markdownToHTML}).
|
||||
ParseFiles(fpath)
|
||||
if err != nil {
|
||||
return "", err
|
||||
}
|
||||
dirPath := filepath.Dir(fpath)
|
||||
data := &templateData{DirPath: dirPath}
|
||||
logger.DebugOp(op, "converting Markdown files to HTML...")
|
||||
var buffer bytes.Buffer
|
||||
if err := tmpl.Execute(&buffer, data); err != nil {
|
||||
return "", err
|
||||
}
|
||||
baseFilename := xrand.Get()
|
||||
dst := fmt.Sprintf("%s/%s.html", dirPath, baseFilename)
|
||||
logger.DebugOp(op, "writing the HTML from previous conversion(s) into new file...")
|
||||
if err := ioutil.WriteFile(dst, buffer.Bytes(), 0644); err != nil {
|
||||
return "", err
|
||||
}
|
||||
return fmt.Sprintf("file://%s", dst), nil
|
||||
}
|
||||
URL, err := resolver()
|
||||
if err != nil {
|
||||
return chromePrint{}, xerror.New(op, err)
|
||||
}
|
||||
return chromePrint{
|
||||
logger: logger,
|
||||
url: URL,
|
||||
opts: opts,
|
||||
}, nil
|
||||
}
|
||||
|
||||
type templateData struct {
|
||||
DirPath string
|
||||
}
|
||||
|
||||
func markdownToHTML(dirPath, filename string) (template.HTML, error) {
|
||||
const op string = "print.markdownToHTML"
|
||||
fpath := fmt.Sprintf("%s/%s", dirPath, filename)
|
||||
b, err := ioutil.ReadFile(fpath)
|
||||
if err != nil {
|
||||
return "", xerror.New(op, err)
|
||||
}
|
||||
unsafe := blackfriday.Run(b)
|
||||
content := bluemonday.UGCPolicy().SanitizeBytes(unsafe)
|
||||
/* #nosec */
|
||||
return template.HTML(content), nil
|
||||
}
|
||||
44
internal/pkg/print/merge.go
Normal file
44
internal/pkg/print/merge.go
Normal file
@@ -0,0 +1,44 @@
|
||||
package print
|
||||
|
||||
import (
|
||||
"context"
|
||||
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/process"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xerror"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xexec"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xlog"
|
||||
)
|
||||
|
||||
type mergePrint struct {
|
||||
logger xlog.Logger
|
||||
fpaths []string
|
||||
}
|
||||
|
||||
// NewMergePrint returns a Print for
|
||||
// merging PDF files.
|
||||
func NewMergePrint(logger xlog.Logger, fpaths []string) Print {
|
||||
return mergePrint{
|
||||
logger: logger,
|
||||
fpaths: fpaths,
|
||||
}
|
||||
}
|
||||
|
||||
func (p mergePrint) Print(ctx context.Context, dest string, proc process.Process) error {
|
||||
const op string = "print.mergePrint.Print"
|
||||
p.logger.DebugfOp(op, "merging '%v'...", p.fpaths)
|
||||
resolver := func() error {
|
||||
var args []string
|
||||
args = append(args, p.fpaths...)
|
||||
args = append(args, "cat", "output", dest)
|
||||
cmd, err := xexec.CommandContext(ctx, p.logger, "pdftk", args...)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
xexec.LogBeforeExecute(p.logger, cmd)
|
||||
return cmd.Run()
|
||||
}
|
||||
if err := resolver(); err != nil {
|
||||
return xerror.New(op, err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
109
internal/pkg/print/office.go
Normal file
109
internal/pkg/print/office.go
Normal file
@@ -0,0 +1,109 @@
|
||||
package print
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"os"
|
||||
"path/filepath"
|
||||
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/process"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xerror"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xexec"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xlog"
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xrand"
|
||||
)
|
||||
|
||||
type officePrint struct {
|
||||
logger xlog.Logger
|
||||
fpaths []string
|
||||
opts OfficePrintOptions
|
||||
}
|
||||
|
||||
// OfficePrintOptions helps customizing the
|
||||
// LibreOffice Print result.
|
||||
type OfficePrintOptions struct {
|
||||
Landscape bool
|
||||
}
|
||||
|
||||
// DefaultOfficePrinterOptions returns the default
|
||||
// LibreOffice Print options.
|
||||
func DefaultOfficePrinterOptions() OfficePrintOptions {
|
||||
return OfficePrintOptions{
|
||||
Landscape: false,
|
||||
}
|
||||
}
|
||||
|
||||
// NewOfficePrint returns a Print for
|
||||
// converting Office documents to PDF.
|
||||
func NewOfficePrint(logger xlog.Logger, fpaths []string, opts OfficePrintOptions) Print {
|
||||
return officePrint{
|
||||
logger: logger,
|
||||
fpaths: fpaths,
|
||||
opts: opts,
|
||||
}
|
||||
}
|
||||
|
||||
func (p officePrint) Print(ctx context.Context, dest string, proc process.Process) error {
|
||||
const op string = "print.officePrint.Print"
|
||||
resolver := func() error {
|
||||
fpaths := make([]string, len(p.fpaths))
|
||||
dirPath := filepath.Dir(dest)
|
||||
for i, fpath := range p.fpaths {
|
||||
baseFilename := xrand.Get()
|
||||
tmpDest := fmt.Sprintf("%s/%d%s.pdf", dirPath, i, baseFilename)
|
||||
p.logger.DebugfOp(op, "converting '%s' to PDF...", fpath)
|
||||
if err := unoconv(ctx, p.logger, fpath, tmpDest, p.opts); err != nil {
|
||||
return err
|
||||
}
|
||||
p.logger.DebugfOp(op, "'%s.pdf' created", baseFilename)
|
||||
fpaths[i] = tmpDest
|
||||
}
|
||||
if len(fpaths) == 1 {
|
||||
p.logger.DebugOp(op, "only one PDF created, nothing to merge")
|
||||
if err := os.Rename(fpaths[0], dest); err != nil {
|
||||
return err
|
||||
}
|
||||
return nil
|
||||
}
|
||||
merger := NewMergePrint(p.logger, fpaths)
|
||||
return merger.Print(ctx, dest, nil)
|
||||
}
|
||||
if err := resolver(); err != nil {
|
||||
return xerror.New(op, err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
func unoconv(ctx context.Context, logger xlog.Logger, fpath, dest string, opts OfficePrintOptions) error {
|
||||
const op string = "print.unoconv"
|
||||
resolver := func() error {
|
||||
args := []string{
|
||||
"--format",
|
||||
"pdf",
|
||||
}
|
||||
if opts.Landscape {
|
||||
args = append(args, "--printer", "PaperOrientation=landscape")
|
||||
}
|
||||
args = append(args, "--output", dest, fpath)
|
||||
cmd, err := xexec.CommandContext(
|
||||
ctx,
|
||||
logger,
|
||||
"unoconv",
|
||||
args...,
|
||||
)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
xexec.LogBeforeExecute(logger, cmd)
|
||||
return cmd.Run()
|
||||
}
|
||||
if err := resolver(); err != nil {
|
||||
return xerror.New(op, err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// Compile-time checks to ensure type implements desired interfaces.
|
||||
var (
|
||||
_ = Print(new(officePrint))
|
||||
)
|
||||
13
internal/pkg/print/print.go
Normal file
13
internal/pkg/print/print.go
Normal file
@@ -0,0 +1,13 @@
|
||||
package print
|
||||
|
||||
import (
|
||||
"context"
|
||||
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/process"
|
||||
)
|
||||
|
||||
// Print is a type that can create a PDF file from a source.
|
||||
// The source is defined in the underlying implementation.
|
||||
type Print interface {
|
||||
Print(ctx context.Context, dest string, proc process.Process) error
|
||||
}
|
||||
15
internal/pkg/print/url.go
Normal file
15
internal/pkg/print/url.go
Normal file
@@ -0,0 +1,15 @@
|
||||
package print
|
||||
|
||||
import (
|
||||
"github.com/thecodingmachine/gotenberg/internal/pkg/xlog"
|
||||
)
|
||||
|
||||
// NewURLPrint returns a Print for
|
||||
// converting a URL to PDF.
|
||||
func NewURLPrint(logger xlog.Logger, url string, opts ChromePrintOptions) Print {
|
||||
return chromePrint{
|
||||
logger: logger,
|
||||
url: url,
|
||||
opts: opts,
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user