feat(pdfengines): add support for flattening annotations (#1105)

* initial changes

* Add tests

* Fix edge case when we need to regenerate appearances

* Fix comments

* Add missing comment

* Add missing comment

* Add missing comment

* Add flatten option to the merge route

* Add flatten option to the libreoffice convert route

* Add flatten option to the chromium convert route

* Revert "Add flatten option to the chromium convert route"

This reverts commit cdab8b4e6b.

* Ignore lint false positives

* Add missing tests

* Add flatten route tests

* Replace input instead of creating a new file

* create copy before flatten in tests

---------

Co-authored-by: Peter Chakalov <peter.chakalov@abraxa.com>
This commit is contained in:
Peter Chakalov
2025-01-28 15:38:27 +02:00
committed by GitHub
parent a9b44ee39e
commit b418f1eb05
23 changed files with 634 additions and 4 deletions

View File

@@ -101,6 +101,27 @@ func (engine *QPdf) Merge(ctx context.Context, logger *zap.Logger, inputPaths []
return fmt.Errorf("merge PDFs with QPDF: %w", err)
}
// Flatten merges annotation appearances with page content, deleting the original annotations.
func (engine *QPdf) Flatten(ctx context.Context, logger *zap.Logger, inputPath string) error {
var args []string
args = append(args, "--generate-appearances")
args = append(args, "--flatten-annotations=all")
args = append(args, "--replace-input")
args = append(args, inputPath)
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("flatten PDFs with QPDF: %w", err)
}
// Convert is not available in this implementation.
func (engine *QPdf) Convert(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return fmt.Errorf("convert PDF to '%+v' with QPDF: %w", formats, gotenberg.ErrPdfEngineMethodNotSupported)

View File

@@ -3,6 +3,8 @@ package qpdf
import (
"context"
"errors"
"fmt"
"io"
"os"
"reflect"
"testing"
@@ -232,6 +234,101 @@ func TestQPdf_Split(t *testing.T) {
}
}
func TestQPdf_Flatten(t *testing.T) {
for _, tc := range []struct {
scenario string
ctx context.Context
inputPath string
createCopy bool
expectError bool
}{
{
scenario: "invalid context",
ctx: nil,
expectError: true,
},
{
scenario: "invalid input path",
ctx: context.TODO(),
inputPath: "foo.pdf",
expectError: true,
},
{
scenario: "success",
ctx: context.TODO(),
inputPath: "/tests/test/testdata/pdfengines/sample3.pdf",
createCopy: true,
expectError: false,
},
} {
t.Run(tc.scenario, func(t *testing.T) {
engine := new(QPdf)
err := engine.Provision(nil)
if err != nil {
t.Fatalf("expected error but got: %v", err)
}
var destinationPath string
if tc.createCopy {
fs := gotenberg.NewFileSystem(new(gotenberg.OsMkdirAll))
outputDir, err := fs.MkdirAll()
if err != nil {
t.Fatalf("expected error no but got: %v", err)
}
defer func() {
err = os.RemoveAll(fs.WorkingDirPath())
if err != nil {
t.Fatalf("expected no error while cleaning up but got: %v", err)
}
}()
destinationPath = fmt.Sprintf("%s/copy_temp.pdf", outputDir)
source, err := os.Open(tc.inputPath)
if err != nil {
t.Fatalf("open source file: %v", err)
}
defer func(source *os.File) {
err := source.Close()
if err != nil {
t.Fatalf("close file: %v", err)
}
}(source)
destination, err := os.Create(destinationPath)
if err != nil {
t.Fatalf("create destination file: %v", err)
}
defer func(destination *os.File) {
err := destination.Close()
if err != nil {
t.Fatalf("close file: %v", err)
}
}(destination)
_, err = io.Copy(destination, source)
if err != nil {
t.Fatalf("copy source into destination: %v", err)
}
} else {
destinationPath = tc.inputPath
}
err = engine.Flatten(tc.ctx, zap.NewNop(), destinationPath)
if !tc.expectError && err != nil {
t.Fatalf("expected no error but got: %v", err)
}
if tc.expectError && err == nil {
t.Fatal("expected error but got none")
}
})
}
}
func TestQPdf_Convert(t *testing.T) {
engine := new(QPdf)
err := engine.Convert(context.TODO(), zap.NewNop(), gotenberg.PdfFormats{}, "", "")