This commit is contained in:
Julien Neuhart
2018-12-10 16:09:19 +01:00
committed by GitHub
parent c1f6100382
commit 495203c112
473 changed files with 4631 additions and 174161 deletions

View File

@@ -0,0 +1,52 @@
/*
Package printer contains structs which convert
a specific file type to PDF:
// converting HTML to PDF.
p := &printer.HTML{
Context: context.Background(),
PaperWidth: 8.27,
PaperHeight: 11.7,
Landscape: false,
}
p.WithLocalURL("index.html")
if err := p.Print("result.pdf"); err != nil {
return err
}
// converting Markdown to PDF:
// it assumes here that our template "index.html"
// will call toHTML method to convert
// markdown files to HTML.
p := &printer.Markdown{
Context: context.Background(),
TemplatePath: "index.html",
PaperWidth: 8.27,
PaperHeight: 11.7,
Landscape: false,
}
if err := p.Print("result.pdf"); err != nil {
return err
}
// converting Office documents to PDF:
// it converts each files independently and
// then merge them.
//
// Also, as unoconv cannot perform
// concurrent conversions, a lock is applied.
p := &printer.Office{
Context: ctx,
FilePaths: []string{"document.docx", "presentation.pptx"}
}
if err := p.Print("result.pdf"); err != nil {
return err
}
It is also able to merge a list of PDF files:
if err := printer.Merge([]string{"foo.pdf", "bar.pdf"}, "result.pdf"); err != nil {
return err
}
*/
package printer

View File

@@ -0,0 +1,188 @@
package printer
import (
"context"
"errors"
"fmt"
"io/ioutil"
"strings"
"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/runtime"
"github.com/mafredri/cdp/protocol/target"
"github.com/mafredri/cdp/rpcc"
)
// HTML facilitates HTML to PDF conversion.
type HTML struct {
Context context.Context
URL string
HeaderHTML string
FooterHTML string
PaperWidth float64
PaperHeight float64
MarginTop float64
MarginBottom float64
MarginLeft float64
MarginRight float64
Landscape bool
}
const defaultHeaderFooterHTML string = "<html><head></head><body></body></html>"
// Print converts HTML to PDF.
// Credits: https://medium.com/compass-true-north/go-service-to-convert-web-pages-to-pdf-using-headless-chrome-5fd9ffbae1af
func (html *HTML) Print(destination string) error {
// use the DevTools HTTP/JSON API to manage targets (e.g. pages, webworkers).
devt := devtool.New("http://127.0.0.1:9222")
pt, err := devt.Create(html.Context)
if err != nil {
return fmt.Errorf("creating DevTools target: %v", err)
}
// open a new RPC connection to the Chrome Debugging Protocol target.
conn, err := rpcc.DialContext(html.Context, pt.WebSocketDebuggerURL)
if err != nil {
return fmt.Errorf("creating RPC connection: %v", err)
}
defer conn.Close()
// create new browser context.
baseBrowser := cdp.NewClient(conn)
newContextTarget, err := baseBrowser.Target.CreateBrowserContext(html.Context)
if err != nil {
return fmt.Errorf("creating new browser context: %v", err)
}
// create a new blank target with the new browser context.
newTargetArgs := target.NewCreateTargetArgs("about:blank").
SetBrowserContextID(newContextTarget.BrowserContextID)
newTarget, err := baseBrowser.Target.CreateTarget(html.Context, newTargetArgs)
if err != nil {
return fmt.Errorf("creating new blank target: %v", err)
}
// connect the client to the new target.
newTargetWsURL := fmt.Sprintf("ws://127.0.0.1:9222/devtools/page/%s", newTarget.TargetID)
newContextConn, err := rpcc.DialContext(html.Context, newTargetWsURL)
if err != nil {
return fmt.Errorf("connecting client to blank target: %v", err)
}
defer newContextConn.Close()
// close the target when done.
closeTargetArgs := target.NewCloseTargetArgs(newTarget.TargetID)
defer baseBrowser.Target.CloseTarget(html.Context, closeTargetArgs)
c := cdp.NewClient(newContextConn)
// enable the runtime.
if err := c.Runtime.Enable(html.Context); err != nil {
return fmt.Errorf("enabling runtime: %v", err)
}
// enable the network.
if err := c.Network.Enable(html.Context, network.NewEnableArgs()); err != nil {
return fmt.Errorf("enabling network: %v", err)
}
// enable events on the page domain.
if err := c.Page.Enable(html.Context); err != nil {
return fmt.Errorf("enabling events on page domain: %v", err)
}
// create a client to listen for the load event to be fired.
loadEventFiredClient, err := c.Page.LoadEventFired(html.Context)
if err != nil {
return fmt.Errorf("creating client listening for load event: %v", err)
}
defer loadEventFiredClient.Close()
// tell the page to navigate to the URL.
navArgs := page.NewNavigateArgs(html.URL)
_, err = c.Page.Navigate(html.Context, navArgs)
if err != nil {
return fmt.Errorf("%s: navigating to URL: %v", html.URL, err)
}
// wait for the page to finish loading.
_, err = loadEventFiredClient.Recv()
if err != nil {
return fmt.Errorf("waiting for page loading: %v", err)
}
// inject a script to make sure web fonts are loaded.
script := `new Promise((resolve, reject) => {
document.fonts.ready.then(function () {
resolve('fonts loaded');
});
setTimeout(resolve.bind(resolve, 'timeout'), %.0f);
});`
scriptArg := runtime.NewEvaluateArgs(script).SetAwaitPromise(true)
returnObj, _ := c.Runtime.Evaluate(html.Context, scriptArg)
loadFontsResult := string(returnObj.Result.Value)
if strings.Contains(loadFontsResult, "timeout") {
return errors.New("timed out loading fonts")
}
// if no header or footer, use the default template
// for avoiding displaying default Chrome templates.
if html.HeaderHTML == "" {
html.HeaderHTML = defaultHeaderFooterHTML
}
if html.FooterHTML == "" {
html.FooterHTML = defaultHeaderFooterHTML
}
print, err := c.Page.PrintToPDF(
html.Context,
page.NewPrintToPDFArgs().
SetPaperWidth(html.PaperWidth).
SetPaperHeight(html.PaperHeight).
SetMarginTop(html.MarginTop).
SetMarginBottom(html.MarginBottom).
SetMarginLeft(html.MarginLeft).
SetMarginRight(html.MarginRight).
SetLandscape(html.Landscape).
SetDisplayHeaderFooter(true).
SetHeaderTemplate(html.HeaderHTML).
SetFooterTemplate(html.FooterHTML).
SetPrintBackground(true),
)
if err != nil {
return fmt.Errorf("printing page to PDF: %v", err)
}
return writeBytesToFile(destination, print.Data)
}
// WithLocalURL sets a local URL from a file path.
func (html *HTML) WithLocalURL(fpath string) {
html.URL = fmt.Sprintf("file://%s", fpath)
}
// WithHeaderFile sets header content from a file.
func (html *HTML) WithHeaderFile(fpath string) error {
if fpath == "" {
return nil
}
contentHTML, err := fileContentToString(fpath)
if err != nil {
return err
}
html.HeaderHTML = contentHTML
return nil
}
// WithFooterFile sets footer content from a file.
func (html *HTML) WithFooterFile(fpath string) error {
if fpath == "" {
return nil
}
contentHTML, err := fileContentToString(fpath)
if err != nil {
return err
}
html.FooterHTML = contentHTML
return nil
}
func fileContentToString(fpath string) (string, error) {
b, err := ioutil.ReadFile(fpath)
if err != nil {
return "", fmt.Errorf("%s: reading file: %v", fpath, err)
}
return string(b), nil
}
// Compile-time checks to ensure type implements desired interfaces.
var (
_ = Printer(new(HTML))
)

View File

@@ -0,0 +1,39 @@
package printer
import (
"context"
"fmt"
"os"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/thecodingmachine/gotenberg/test"
)
func TestHTML(t *testing.T) {
dirPath := test.HTMLTestDirPath(t)
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
html := &HTML{
Context: ctx,
PaperWidth: 8.27,
PaperHeight: 11.7,
MarginTop: 1,
MarginBottom: 1,
MarginLeft: 1,
MarginRight: 1,
}
html.WithLocalURL(fmt.Sprintf("%s/%s", dirPath, "index.html"))
err := html.WithHeaderFile(fmt.Sprintf("%s/%s", dirPath, "header.html"))
require.Nil(t, err)
err = html.WithFooterFile(fmt.Sprintf("%s/%s", dirPath, "footer.html"))
require.Nil(t, err)
dst := fmt.Sprintf("%s/%s", dirPath, "foo.pdf")
err = html.Print(dst)
require.Nil(t, err)
require.FileExists(t, dst)
err = os.RemoveAll(dirPath)
assert.Nil(t, err)
}

View File

@@ -0,0 +1,111 @@
package printer
import (
"bytes"
"context"
"fmt"
"html/template"
"io/ioutil"
"path/filepath"
"github.com/microcosm-cc/bluemonday"
"github.com/russross/blackfriday/v2"
"github.com/thecodingmachine/gotenberg/internal/pkg/rand"
)
// Markdown facilitates Markdown to PDF conversion.
type Markdown struct {
Context context.Context
TemplatePath string
HeaderHTML string
FooterHTML string
PaperWidth float64
PaperHeight float64
MarginTop float64
MarginBottom float64
MarginLeft float64
MarginRight float64
Landscape bool
html *HTML
}
type templateData struct {
DirPath string
}
// Print converts markdown to PDF.
func (md *Markdown) Print(destination string) error {
if md.html == nil {
md.html = &HTML{Context: md.Context}
}
if md.HeaderHTML != "" {
md.html.HeaderHTML = md.HeaderHTML
}
if md.FooterHTML != "" {
md.html.FooterHTML = md.FooterHTML
}
md.html.PaperWidth = md.PaperWidth
md.html.PaperHeight = md.PaperHeight
md.html.MarginTop = md.MarginTop
md.html.MarginBottom = md.MarginBottom
md.html.MarginLeft = md.MarginLeft
md.html.MarginRight = md.MarginRight
md.html.Landscape = md.Landscape
tmpl, err := template.
New(filepath.Base(md.TemplatePath)).
Funcs(template.FuncMap{"toHTML": toHTML}).
ParseFiles(md.TemplatePath)
if err != nil {
return fmt.Errorf("%s: parsing template: %v", md.TemplatePath, err)
}
dirPath := filepath.Dir(md.TemplatePath)
data := &templateData{DirPath: dirPath}
var buffer bytes.Buffer
if err := tmpl.Execute(&buffer, data); err != nil {
return fmt.Errorf("%s: executing template: %v", md.TemplatePath, err)
}
baseFilename, err := rand.Get()
if err != nil {
return err
}
dst := fmt.Sprintf("%s/%s.html", dirPath, baseFilename)
if err := writeBytesToFile(dst, buffer.Bytes()); err != nil {
return err
}
md.html.WithLocalURL(dst)
return md.html.Print(destination)
}
func toHTML(dirPath, filename string) (template.HTML, error) {
fpath := fmt.Sprintf("%s/%s", dirPath, filename)
b, err := ioutil.ReadFile(fpath)
if err != nil {
return "", fmt.Errorf("%s: reading file: %v", fpath, err)
}
unsafe := blackfriday.Run(b)
contentHTML := bluemonday.UGCPolicy().SanitizeBytes(unsafe)
return template.HTML(contentHTML), nil
}
// WithHeaderFile sets header content from a file.
func (md *Markdown) WithHeaderFile(fpath string) error {
if md.html == nil {
md.html = &HTML{Context: md.Context}
}
return md.html.WithHeaderFile(fpath)
}
// WithFooterFile sets footer content from a file.
func (md *Markdown) WithFooterFile(fpath string) error {
if md.html == nil {
md.html = &HTML{Context: md.Context}
}
return md.html.WithFooterFile(fpath)
}
// Compile-time checks to ensure type implements desired interfaces.
var (
_ = Printer(new(Markdown))
)

View File

@@ -0,0 +1,39 @@
package printer
import (
"context"
"fmt"
"os"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/thecodingmachine/gotenberg/test"
)
func TestMarkdown(t *testing.T) {
dirPath := test.MarkdownTestDirPath(t)
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
markdown := &Markdown{
Context: ctx,
TemplatePath: fmt.Sprintf("%s/%s", dirPath, "index.html"),
PaperWidth: 8.27,
PaperHeight: 11.7,
MarginTop: 1,
MarginBottom: 1,
MarginLeft: 1,
MarginRight: 1,
}
err := markdown.WithHeaderFile(fmt.Sprintf("%s/%s", dirPath, "header.html"))
require.Nil(t, err)
err = markdown.WithFooterFile(fmt.Sprintf("%s/%s", dirPath, "footer.html"))
require.Nil(t, err)
dst := fmt.Sprintf("%s/%s", dirPath, "foo.pdf")
err = markdown.Print(dst)
require.Nil(t, err)
require.FileExists(t, dst)
err = os.RemoveAll(dirPath)
assert.Nil(t, err)
}

View File

@@ -0,0 +1,62 @@
package printer
import (
"context"
"errors"
"fmt"
"os"
"os/exec"
"path/filepath"
"sync"
"github.com/thecodingmachine/gotenberg/internal/pkg/rand"
)
var mu sync.Mutex
// Office facilitates Office documents to PDF conversion.
type Office struct {
Context context.Context
FilePaths []string
}
// Print converts Office documents to PDF.
func (o *Office) Print(destination string) error {
mu.Lock()
defer mu.Unlock()
fpaths := make([]string, len(o.FilePaths))
dirPath := filepath.Dir(destination)
for i, fpath := range o.FilePaths {
baseFilename, err := rand.Get()
if err != nil {
return err
}
tmpDest := fmt.Sprintf("%s/%s.pdf", dirPath, baseFilename)
cmd := exec.CommandContext(
o.Context,
"unoconv",
"--format",
"pdf",
"--output",
tmpDest,
fpath,
)
_, err = cmd.Output()
if o.Context.Err() == context.DeadlineExceeded {
return errors.New("unoconv: command timed out")
}
if err != nil {
return fmt.Errorf("unoconv: non-zero exit code: %v", err)
}
fpaths[i] = tmpDest
}
if len(fpaths) == 1 {
return os.Rename(fpaths[0], destination)
}
return Merge(fpaths, destination)
}
// Compile-time checks to ensure type implements desired interfaces.
var (
_ = Printer(new(Office))
)

View File

@@ -0,0 +1,31 @@
package printer
import (
"context"
"fmt"
"os"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/thecodingmachine/gotenberg/test"
)
func TestOffice(t *testing.T) {
dirPath := test.OfficeTestDirPath(t)
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
office := &Office{
Context: ctx,
FilePaths: []string{
fmt.Sprintf("%s/%s", dirPath, "document.docx"),
},
}
dst := fmt.Sprintf("%s/%s", dirPath, "foo.pdf")
err := office.Print(dst)
require.Nil(t, err)
require.FileExists(t, dst)
err = os.RemoveAll(dirPath)
assert.Nil(t, err)
}

View File

@@ -0,0 +1,29 @@
package printer
import (
"fmt"
"io/ioutil"
pdfcpuAPI "github.com/hhrutter/pdfcpu/pkg/api"
pdfcpuConfig "github.com/hhrutter/pdfcpu/pkg/pdfcpu"
)
// Printer is a type that can create a PDF file from a source.
// The source is defined in the underlying implementation.
type Printer interface {
Print(destination string) error
}
// Merge merges PDF files.
func Merge(fpaths []string, destination string) error {
cmd := pdfcpuAPI.MergeCommand(fpaths, destination, pdfcpuConfig.NewDefaultConfiguration())
_, err := pdfcpuAPI.Merge(cmd)
return err
}
func writeBytesToFile(dst string, b []byte) error {
if err := ioutil.WriteFile(dst, b, 0644); err != nil {
return fmt.Errorf("%s: writting file: %v", dst, err)
}
return nil
}

View File

@@ -0,0 +1,27 @@
package printer
import (
"fmt"
"os"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/thecodingmachine/gotenberg/test"
)
func TestMerge(t *testing.T) {
dirPath := test.PDFTestDirPath(t)
dst := fmt.Sprintf("%s/%s", dirPath, "foo.pdf")
err := Merge(
[]string{
fmt.Sprintf("%s/%s", dirPath, "gotenberg.pdf"),
fmt.Sprintf("%s/%s", dirPath, "gotenberg.pdf"),
},
dst,
)
require.Nil(t, err)
require.FileExists(t, dst)
err = os.RemoveAll(dirPath)
assert.Nil(t, err)
}