fix(pdfengines): better workflow when applying PDF/A or PDF/UA compliance

This commit is contained in:
Julien Neuhart
2026-03-18 15:24:31 +01:00
parent 7fb4c89832
commit 21e300fcec
11 changed files with 252 additions and 112 deletions

View File

@@ -781,24 +781,36 @@ func convertUrl(ctx *api.Context, chromium Api, engine gotenberg.PdfEngine, url
return fmt.Errorf("convert to PDF: %w", err)
}
err = pdfengines.ValidatePdfFormatsCompat(pdfFormats, userPassword, embedPaths)
if err != nil {
return err
}
outputPaths, err := pdfengines.SplitPdfStub(ctx, engine, mode, []string{outputPath})
if err != nil {
return fmt.Errorf("split PDF: %w", err)
}
err = pdfengines.WatermarkStub(ctx, engine, watermark, outputPaths)
if err != nil {
return fmt.Errorf("watermark PDFs: %w", err)
}
err = pdfengines.StampStub(ctx, engine, stamp, outputPaths)
if err != nil {
return fmt.Errorf("stamp PDFs: %w", err)
}
convertOutputPaths, err := pdfengines.ConvertStub(ctx, engine, pdfFormats, outputPaths)
if err != nil {
return fmt.Errorf("convert PDF(s): %w", err)
}
err = pdfengines.WatermarkStub(ctx, engine, watermark, convertOutputPaths)
// Metadata, embeds are written after Convert, as LibreOffice
// strips them during PDF/A conversion.
err = pdfengines.WriteMetadataStub(ctx, engine, metadata, convertOutputPaths)
if err != nil {
return fmt.Errorf("watermark PDFs: %w", err)
}
err = pdfengines.StampStub(ctx, engine, stamp, convertOutputPaths)
if err != nil {
return fmt.Errorf("stamp PDFs: %w", err)
return fmt.Errorf("write metadata: %w", err)
}
err = pdfengines.EmbedFilesStub(ctx, engine, embedPaths, convertOutputPaths)
@@ -806,11 +818,6 @@ func convertUrl(ctx *api.Context, chromium Api, engine gotenberg.PdfEngine, url
return fmt.Errorf("embed files into PDFs: %w", err)
}
err = pdfengines.WriteMetadataStub(ctx, engine, metadata, convertOutputPaths)
if err != nil {
return fmt.Errorf("write metadata: %w", err)
}
err = pdfengines.EncryptPdfStub(ctx, engine, userPassword, ownerPassword, convertOutputPaths)
if err != nil {
return fmt.Errorf("encrypt PDFs: %w", err)

View File

@@ -195,6 +195,14 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
stamp.Expression = stampFiles[0]
}
err = pdfengines.ValidatePdfFormatsCompat(pdfFormats, userPassword, embedPaths)
if err != nil {
return err
}
hasPostProcessing := watermark.Source != "" || stamp.Source != "" ||
len(embedPaths) > 0 || len(metadata) > 0 || flatten
outputPaths := make([]string, len(inputPaths))
for i, inputPath := range inputPaths {
outputPaths[i] = ctx.GeneratePath(".pdf")
@@ -230,9 +238,10 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
NativeTiledWatermarkText: nativeTiledWatermarkText,
}
if nativePdfFormats && splitMode == zeroValuedSplitMode {
if nativePdfFormats && splitMode == zeroValuedSplitMode && !hasPostProcessing {
// Only natively apply given PDF formats if we're not
// splitting the PDF later.
// splitting the PDF later and no post-processing features
// are enabled (as they would degrade compliance).
options.PdfFormats = pdfFormats
}
@@ -298,7 +307,27 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
}
}
if !nativePdfFormats || (nativePdfFormats && splitMode != zeroValuedSplitMode) {
err = pdfengines.WatermarkStub(ctx, engine, watermark, outputPaths)
if err != nil {
return fmt.Errorf("watermark PDFs: %w", err)
}
err = pdfengines.StampStub(ctx, engine, stamp, outputPaths)
if err != nil {
return fmt.Errorf("stamp PDFs: %w", err)
}
if flatten {
err = pdfengines.FlattenStub(ctx, engine, outputPaths)
if err != nil {
return fmt.Errorf("flatten PDFs: %w", err)
}
}
needsConvertStub := !nativePdfFormats ||
(nativePdfFormats && splitMode != zeroValuedSplitMode) ||
(nativePdfFormats && hasPostProcessing)
if needsConvertStub {
convertOutputPaths, err := pdfengines.ConvertStub(ctx, engine, pdfFormats, outputPaths)
if err != nil {
return fmt.Errorf("convert PDFs: %w", err)
@@ -318,31 +347,16 @@ func convertRoute(libreOffice libreofficeapi.Uno, engine gotenberg.PdfEngine) ap
}
}
err = pdfengines.WatermarkStub(ctx, engine, watermark, outputPaths)
if err != nil {
return fmt.Errorf("watermark PDFs: %w", err)
}
err = pdfengines.StampStub(ctx, engine, stamp, outputPaths)
if err != nil {
return fmt.Errorf("stamp PDFs: %w", err)
}
err = pdfengines.EmbedFilesStub(ctx, engine, embedPaths, outputPaths)
if err != nil {
return fmt.Errorf("embed files into PDFs: %w", err)
}
// Metadata, embeds are written after Convert, as LibreOffice
// strips them during PDF/A conversion.
err = pdfengines.WriteMetadataStub(ctx, engine, metadata, outputPaths)
if err != nil {
return fmt.Errorf("write metadata: %w", err)
}
if flatten {
err = pdfengines.FlattenStub(ctx, engine, outputPaths)
if err != nil {
return fmt.Errorf("flatten PDFs: %w", err)
}
err = pdfengines.EmbedFilesStub(ctx, engine, embedPaths, outputPaths)
if err != nil {
return fmt.Errorf("embed files into PDFs: %w", err)
}
err = pdfengines.EncryptPdfStub(ctx, engine, userPassword, ownerPassword, outputPaths)

View File

@@ -166,6 +166,41 @@ func FormDataPdfBookmarks(form *api.FormData, mandatory bool) any {
return bookmarks
}
// ValidatePdfFormatsCompat checks for incompatible combinations of PDF formats
// with other features and returns an appropriate error if found.
func ValidatePdfFormatsCompat(pdfFormats gotenberg.PdfFormats, userPassword string, embedPaths []string) error {
zeroValued := gotenberg.PdfFormats{}
if pdfFormats == zeroValued {
return nil
}
// PDF/A forbids encryption per the standard.
if pdfFormats.PdfA != "" && userPassword != "" {
return api.WrapError(
errors.New("PDF/A format is incompatible with encryption"),
api.NewSentinelHttpError(
http.StatusBadRequest,
"Invalid form data: PDF/A format is incompatible with encryption",
),
)
}
// Only PDF/A-3 variants allow embedded file attachments.
if pdfFormats.PdfA != "" && len(embedPaths) > 0 {
if pdfFormats.PdfA != gotenberg.PdfA3a && pdfFormats.PdfA != gotenberg.PdfA3b && pdfFormats.PdfA != gotenberg.PdfA3u {
return api.WrapError(
fmt.Errorf("PDF format '%s' does not support embedded files", pdfFormats.PdfA),
api.NewSentinelHttpError(
http.StatusBadRequest,
fmt.Sprintf("Invalid form data: PDF format '%s' does not support embedded files; only PDF/A-3 variants allow attachments", pdfFormats.PdfA),
),
)
}
}
return nil
}
// MergeStub merges given PDFs. If only one input PDF, it does nothing and
// returns the corresponding input path.
func MergeStub(ctx *api.Context, engine gotenberg.PdfEngine, inputPaths []string) (string, error) {
@@ -546,16 +581,18 @@ func mergeRoute(engine gotenberg.PdfEngine) api.Route {
stamp.Expression = stampFiles[0]
}
err = ValidatePdfFormatsCompat(pdfFormats, userPassword, embedPaths)
if err != nil {
return err
}
outputPath := ctx.GeneratePath(".pdf")
err = engine.Merge(ctx, ctx.Log(), inputPaths, outputPath)
if err != nil {
return fmt.Errorf("merge PDFs: %w", err)
}
outputPaths, err := ConvertStub(ctx, engine, pdfFormats, []string{outputPath})
if err != nil {
return fmt.Errorf("convert PDF: %w", err)
}
outputPaths := []string{outputPath}
err = WatermarkStub(ctx, engine, watermark, outputPaths)
if err != nil {
@@ -567,11 +604,21 @@ func mergeRoute(engine gotenberg.PdfEngine) api.Route {
return fmt.Errorf("stamp PDFs: %w", err)
}
err = EmbedFilesStub(ctx, engine, embedPaths, outputPaths)
if err != nil {
return fmt.Errorf("embed files into PDFs: %w", err)
if flatten {
err = FlattenStub(ctx, engine, outputPaths)
if err != nil {
return fmt.Errorf("flatten PDFs: %w", err)
}
}
outputPaths, err = ConvertStub(ctx, engine, pdfFormats, outputPaths)
if err != nil {
return fmt.Errorf("convert PDF: %w", err)
}
// Bookmarks, metadata, and embeds are written after Convert,
// as LibreOffice strips them during PDF/A conversion.
var finalBookmarks []gotenberg.Bookmark
if b, ok := bookmarks.([]gotenberg.Bookmark); ok {
finalBookmarks = b
@@ -620,11 +667,9 @@ func mergeRoute(engine gotenberg.PdfEngine) api.Route {
return fmt.Errorf("write metadata: %w", err)
}
if flatten {
err = FlattenStub(ctx, engine, outputPaths)
if err != nil {
return fmt.Errorf("flatten PDFs: %w", err)
}
err = EmbedFilesStub(ctx, engine, embedPaths, outputPaths)
if err != nil {
return fmt.Errorf("embed files into PDFs: %w", err)
}
err = EncryptPdfStub(ctx, engine, userPassword, ownerPassword, outputPaths)
@@ -679,41 +724,48 @@ func splitRoute(engine gotenberg.PdfEngine) api.Route {
stamp.Expression = stampFiles[0]
}
err = ValidatePdfFormatsCompat(pdfFormats, userPassword, embedPaths)
if err != nil {
return err
}
outputPaths, err := SplitPdfStub(ctx, engine, mode, inputPaths)
if err != nil {
return fmt.Errorf("split PDFs: %w", err)
}
err = WatermarkStub(ctx, engine, watermark, outputPaths)
if err != nil {
return fmt.Errorf("watermark PDFs: %w", err)
}
err = StampStub(ctx, engine, stamp, outputPaths)
if err != nil {
return fmt.Errorf("stamp PDFs: %w", err)
}
if flatten {
err = FlattenStub(ctx, engine, outputPaths)
if err != nil {
return fmt.Errorf("flatten PDFs: %w", err)
}
}
convertOutputPaths, err := ConvertStub(ctx, engine, pdfFormats, outputPaths)
if err != nil {
return fmt.Errorf("convert PDFs: %w", err)
}
err = WatermarkStub(ctx, engine, watermark, convertOutputPaths)
if err != nil {
return fmt.Errorf("watermark PDFs: %w", err)
}
err = StampStub(ctx, engine, stamp, convertOutputPaths)
if err != nil {
return fmt.Errorf("stamp PDFs: %w", err)
}
err = EmbedFilesStub(ctx, engine, embedPaths, convertOutputPaths)
if err != nil {
return fmt.Errorf("embed files into PDFs: %w", err)
}
// Metadata, embeds are written after Convert, as LibreOffice
// strips them during PDF/A conversion.
err = WriteMetadataStub(ctx, engine, metadata, convertOutputPaths)
if err != nil {
return fmt.Errorf("write metadata: %w", err)
}
if flatten {
err = FlattenStub(ctx, engine, convertOutputPaths)
if err != nil {
return fmt.Errorf("flatten PDFs: %w", err)
}
err = EmbedFilesStub(ctx, engine, embedPaths, convertOutputPaths)
if err != nil {
return fmt.Errorf("embed files into PDFs: %w", err)
}
err = EncryptPdfStub(ctx, engine, userPassword, ownerPassword, convertOutputPaths)