adalal/gotenberg
1
0
Fork 0
mirror of https://github.com/gotenberg/gotenberg.git synced 2026-08-17 12:42:16 +01:00
Code Issues Packages Projects Releases Wiki Activity

feat(ExifTool): add capability to overwrite metadata of the PDF generated (#776)

Browse Source 
* add new metadata functions to pdf engine interface

* add exiftool module with relevant test cases

* use exiftool to overwrite metadata in libreoffice

* use exiftool to overwrite metadata in chromium

* fix linter issues

* fix more linter issues

* more test cases for better coverage

* remove utils

* minor changes

* remove metadata from pdfoptions

* correct indentation

* read/write metadata one file at a time.
This commit is contained in:
Piyush Srivastava
2024-02-16 18:46:50 +00:00
committed by Julien Neuhart
parent 31e7582216
commit 71911cb1a7
24 changed files with 1433 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
pkg/modules/libreoffice/pdfengine/pdfengine.go
View File

@@ -71,6 +71,16 @@ func (engine *LibreOfficePdfEngine) Convert(ctx context.Context, logger *zap.Log
return fmt.Errorf("convert PDF to '%+v' with LibreOffice: %w", formats, err)
}
// ReadMetadata is not available in this implementation.
func (engine *LibreOfficePdfEngine) ReadMetadata(ctx context.Context, logger *zap.Logger, inputPaths string, metadata map[string]interface{}) error {
return fmt.Errorf("read PDF metadata with LibreOffice: %w", gotenberg.ErrPdfEngineMethodNotSupported)
}
// WriteMetadata is not available in this implementation.
func (engine *LibreOfficePdfEngine) WriteMetadata(ctx context.Context, logger *zap.Logger, inputPaths string, newMetadata map[string]interface{}) error {
return fmt.Errorf("write PDF metadata with LibreOffice: %w", gotenberg.ErrPdfEngineMethodNotSupported)
}
// Interface guards.
var (
_ gotenberg.Module = (*LibreOfficePdfEngine)(nil)

18
pkg/modules/libreoffice/pdfengine/pdfengine_test.go
View File

@@ -166,3 +166,21 @@ func TestLibreOfficePdfEngine_Convert(t *testing.T) {
})
}
}
func TestLibreOfficePdfEngine_ReadMetadata(t *testing.T) {
engine := new(LibreOfficePdfEngine)
err := engine.ReadMetadata(context.Background(), zap.NewNop(), "", nil)
if !errors.Is(err, gotenberg.ErrPdfEngineMethodNotSupported) {
t.Errorf("expected error %v, but got: %v", gotenberg.ErrPdfEngineMethodNotSupported, err)
}
}
func TestLibreOfficePdfEngine_WriteMetadata(t *testing.T) {
engine := new(LibreOfficePdfEngine)
err := engine.WriteMetadata(context.Background(), zap.NewNop(), "", nil)
if !errors.Is(err, gotenberg.ErrPdfEngineMethodNotSupported) {
t.Errorf("expected error %v, but got: %v", gotenberg.ErrPdfEngineMethodNotSupported, err)
}
}

34
pkg/modules/libreoffice/routes.go
View File

@@ -1,6 +1,7 @@
package libreoffice
import (
"encoding/json"
"errors"
"fmt"
"net/http"
@@ -27,22 +28,33 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
inputPaths []string
landscape bool
nativePageRanges string
exportFormFields bool
pdfa string
pdfua bool
nativePdfFormats bool
merge bool
exportFormFields bool
metadata map[string]interface{}
)
err := ctx.FormData().
MandatoryPaths(libreOffice.Extensions(), &inputPaths).
Bool("landscape", &landscape, false).
String("nativePageRanges", &nativePageRanges, "").
Bool("exportFormFields", &exportFormFields, true).
String("pdfa", &pdfa, "").
Bool("pdfua", &pdfua, false).
Bool("nativePdfFormats", &nativePdfFormats, true).
Bool("merge", &merge, false).
Bool("exportFormFields", &exportFormFields, true).
Custom("metadata", func(value string) error {
metadata = map[string]interface{}{}
if len(value) > 0 {
err := json.Unmarshal([]byte(value), &metadata)
if err != nil {
return err
}
}
return nil
}).
Validate()
if err != nil {
return fmt.Errorf("validate form data: %w", err)
@@ -116,6 +128,14 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
outputPath = convertOutputPath
}
// Writes and potentially overrides metadata entries, if any.
if len(metadata) > 0 {
err = engine.WriteMetadata(ctx, ctx.Log(), outputPath, metadata)
if err != nil {
return fmt.Errorf("write metadata failure: %w", err)
}
}
// 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)
@@ -160,6 +180,16 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
}
}
// Writes and potentially overrides metadata entries, if any.
if len(metadata) > 0 {
for _, outputPath := range outputPaths {
err = engine.WriteMetadata(ctx, ctx.Log(), outputPath, metadata)
if err != nil {
return fmt.Errorf("write metadata: %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...)

309
pkg/modules/libreoffice/routes_test.go
View File

@@ -546,6 +546,315 @@ func TestConvertRoute(t *testing.T) {
expectHttpError: false,
expectOutputPathsCount: 1,
},
{
scenario: "success with metadata (single file)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
})
ctx.SetValues(map[string][]string{
"metadata": {
"{\"Creator\": \"foo\", \"Producer\": \"bar\" }",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
ConvertMock: func(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return nil
},
WriteMetadataMock: func(ctx context.Context, logger *zap.Logger, inputPath string, newMetadata map[string]interface{}) error {
return nil
},
},
expectError: false,
expectHttpError: false,
expectOutputPathsCount: 1,
},
{
scenario: "success with metadata (many files)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"metadata": {
"{\"Creator\": \"foo\", \"Producer\": \"bar\" }",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
ConvertMock: func(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return nil
},
WriteMetadataMock: func(ctx context.Context, logger *zap.Logger, inputPath string, newMetadata map[string]interface{}) error {
return nil
},
},
expectError: false,
expectHttpError: false,
expectOutputPathsCount: 2,
},
{
scenario: "error with metadata (single file)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
})
ctx.SetValues(map[string][]string{
"metadata": {
// invalid json
"{\"Creator\"",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
expectError: true,
expectHttpError: true,
expectHttpStatus: http.StatusBadRequest,
expectOutputPathsCount: 0,
},
{
scenario: "error with metadata (many files)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"metadata": {
// invalid json
"{\"Creator\"",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
expectError: true,
expectHttpError: true,
expectHttpStatus: http.StatusBadRequest,
expectOutputPathsCount: 0,
},
{
scenario: "error with metadata write failure (single file)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
})
ctx.SetValues(map[string][]string{
"metadata": {
"{\"Creator\": \"foo\", \"Producer\": \"bar\" }",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
ConvertMock: func(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return nil
},
WriteMetadataMock: func(ctx context.Context, logger *zap.Logger, inputPath string, newMetadata map[string]interface{}) error {
return errors.New("foo")
},
},
expectError: true,
expectHttpError: false,
expectOutputPathsCount: 0,
},
{
scenario: "error with metadata write failure (many files)",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"metadata": {
"{\"Creator\": \"foo\", \"Producer\": \"bar\" }",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
ConvertMock: func(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return nil
},
WriteMetadataMock: func(ctx context.Context, logger *zap.Logger, inputPath string, newMetadata map[string]interface{}) error {
return errors.New("foo")
},
},
expectError: true,
expectHttpError: false,
expectOutputPathsCount: 0,
},
{
scenario: "success merge with metadata",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"metadata": {
"{\"Creator\": \"foo\", \"Producer\": \"bar\" }",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
MergeMock: func(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
return nil
},
ConvertMock: func(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return nil
},
WriteMetadataMock: func(ctx context.Context, logger *zap.Logger, inputPath string, newMetadata map[string]interface{}) error {
return nil
},
},
expectError: false,
expectHttpError: false,
expectOutputPathsCount: 1,
},
{
scenario: "error merge with metadata",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"metadata": {
// invalid json
"{\"Creator",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
expectError: true,
expectHttpError: true,
expectHttpStatus: http.StatusBadRequest,
expectOutputPathsCount: 0,
},
{
scenario: "error merge with metadata write failure",
ctx: func() *api.ContextMock {
ctx := &api.ContextMock{Context: new(api.Context)}
ctx.SetFiles(map[string]string{
"document.docx": "/document.docx",
"document2.docx": "/document2.docx",
})
ctx.SetValues(map[string][]string{
"merge": {
"true",
},
"metadata": {
"{\"Creator\": \"foo\", \"Producer\": \"bar\" }",
},
})
return ctx
}(),
libreOffice: &libreofficeapi.ApiMock{
PdfMock: func(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options libreofficeapi.Options) error {
return nil
},
ExtensionsMock: func() []string {
return []string{".docx"}
},
},
engine: &gotenberg.PdfEngineMock{
MergeMock: func(ctx context.Context, logger *zap.Logger, inputPaths []string, outputPath string) error {
return nil
},
ConvertMock: func(ctx context.Context, logger *zap.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {
return nil
},
WriteMetadataMock: func(ctx context.Context, logger *zap.Logger, inputPath string, newMetadata map[string]interface{}) error {
return errors.New("foo")
},
},
expectError: true,
expectHttpError: false,
expectOutputPathsCount: 0,
},
} {
t.Run(tc.scenario, func(t *testing.T) {
tc.ctx.SetLogger(zap.NewNop())