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

* 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

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())