feat(pdfengines): optimize PDF images to reduce file size (#359)

pkg/modules/pdfengines/multi.go

@@ -18,6 +18,7 @@ type multiPdfEngines struct {
 	splitEngines          []gotenberg.PdfEngine
 	flattenEngines        []gotenberg.PdfEngine
 	convertEngines        []gotenberg.PdfEngine
+	optimizeImagesEngines []gotenberg.PdfEngine
 	readMetadataEngines   []gotenberg.PdfEngine
 	writeMetadataEngines  []gotenberg.PdfEngine
 	passwordEngines       []gotenberg.PdfEngine
@@ -36,6 +37,7 @@ func newMultiPdfEngines(
 	splitEngines,
 	flattenEngines,
 	convertEngines,
+	optimizeImagesEngines,
 	readMetadataEngines,
 	writeMetadataEngines,
 	passwordEngines,
@@ -53,6 +55,7 @@ func newMultiPdfEngines(
 		splitEngines:          splitEngines,
 		flattenEngines:        flattenEngines,
 		convertEngines:        convertEngines,
+		optimizeImagesEngines: optimizeImagesEngines,
 		readMetadataEngines:   readMetadataEngines,
 		writeMetadataEngines:  writeMetadataEngines,
 		passwordEngines:       passwordEngines,
@@ -189,6 +192,17 @@ func (multi *multiPdfEngines) Flatten(ctx context.Context, logger *slog.Logger,
 	)
 }
 
+// OptimizeImages re-encodes the images of a PDF using the first available
+// engine that supports image optimization.
+func (multi *multiPdfEngines) OptimizeImages(ctx context.Context, logger *slog.Logger, imageQuality int, inputPath string) error {
+	return runWithFallbackVoid(ctx, "pdfengines.OptimizeImages", multi.optimizeImagesEngines,
+		func(ctx context.Context, engine gotenberg.PdfEngine) error {
+			return engine.OptimizeImages(ctx, logger, imageQuality, inputPath)
+		},
+		func(err error) error { return fmt.Errorf("optimize PDF images with multi PDF engines: %w", err) },
+	)
+}
+
 // Convert transforms the given PDF to a specific PDF format using the first
 // available engine that supports PDF conversion.
 func (multi *multiPdfEngines) Convert(ctx context.Context, logger *slog.Logger, formats gotenberg.PdfFormats, inputPath, outputPath string) error {

pkg/modules/pdfengines/pdfengines.go

@@ -32,6 +32,7 @@ type PdfEngines struct {
 	splitNames          []string
 	flattenNames        []string
 	convertNames        []string
+	optimizeImagesNames []string
 	readMetadataNames   []string
 	writeMetadataNames  []string
 	encryptNames        []string
@@ -57,6 +58,7 @@ func (mod *PdfEngines) Descriptor() gotenberg.ModuleDescriptor {
 			fs.StringSlice("pdfengines-split-engines", []string{"pdfcpu", "qpdf", "pdftk"}, "Set the PDF engines and their order for the split feature - empty means all")
 			fs.StringSlice("pdfengines-flatten-engines", []string{"qpdf"}, "Set the PDF engines and their order for the flatten feature - empty means all")
 			fs.StringSlice("pdfengines-convert-engines", []string{"libreoffice-pdfengine"}, "Set the PDF engines and their order for the convert feature - empty means all")
+			fs.StringSlice("pdfengines-optimize-images-engines", []string{"pdfcpu"}, "Set the PDF engines and their order for the image optimization feature - empty means all")
 			fs.StringSlice("pdfengines-read-metadata-engines", []string{"exiftool"}, "Set the PDF engines and their order for the read metadata feature - empty means all")
 			fs.StringSlice("pdfengines-write-metadata-engines", []string{"exiftool"}, "Set the PDF engines and their order for the write metadata feature - empty means all")
 			fs.StringSlice("pdfengines-encrypt-engines", []string{"qpdf", "pdftk", "pdfcpu"}, "Set the PDF engines and their order for the password protection feature - empty means all")
@@ -91,6 +93,7 @@ func (mod *PdfEngines) Provision(ctx *gotenberg.Context) error {
 	splitNames := flags.MustStringSlice("pdfengines-split-engines")
 	flattenNames := flags.MustStringSlice("pdfengines-flatten-engines")
 	convertNames := flags.MustStringSlice("pdfengines-convert-engines")
+	optimizeImagesNames := flags.MustStringSlice("pdfengines-optimize-images-engines")
 	readMetadataNames := flags.MustStringSlice("pdfengines-read-metadata-engines")
 	writeMetadataNames := flags.MustStringSlice("pdfengines-write-metadata-engines")
 	encryptNames := flags.MustStringSlice("pdfengines-encrypt-engines")
@@ -148,6 +151,11 @@ func (mod *PdfEngines) Provision(ctx *gotenberg.Context) error {
 		mod.convertNames = convertNames
 	}
 
+	mod.optimizeImagesNames = defaultNames
+	if len(optimizeImagesNames) > 0 {
+		mod.optimizeImagesNames = optimizeImagesNames
+	}
+
 	mod.readMetadataNames = defaultNames
 	if len(readMetadataNames) > 0 {
 		mod.readMetadataNames = readMetadataNames
@@ -247,6 +255,7 @@ func (mod *PdfEngines) Validate() error {
 	findNonExistingEngines(mod.mergeNames)
 	findNonExistingEngines(mod.splitNames)
 	findNonExistingEngines(mod.flattenNames)
+	findNonExistingEngines(mod.optimizeImagesNames)
 	findNonExistingEngines(mod.convertNames)
 	findNonExistingEngines(mod.readMetadataNames)
 	findNonExistingEngines(mod.writeMetadataNames)
@@ -275,6 +284,7 @@ func (mod *PdfEngines) SystemMessages() []string {
 		fmt.Sprintf("split engines - %s", strings.Join(mod.splitNames, " ")),
 		fmt.Sprintf("flatten engines - %s", strings.Join(mod.flattenNames, " ")),
 		fmt.Sprintf("convert engines - %s", strings.Join(mod.convertNames, " ")),
+		fmt.Sprintf("optimize images engines - %s", strings.Join(mod.optimizeImagesNames, " ")),
 		fmt.Sprintf("read metadata engines - %s", strings.Join(mod.readMetadataNames, " ")),
 		fmt.Sprintf("write metadata engines - %s", strings.Join(mod.writeMetadataNames, " ")),
 		fmt.Sprintf("encrypt engines - %s", strings.Join(mod.encryptNames, " ")),
@@ -310,6 +320,7 @@ func (mod *PdfEngines) PdfEngine() (gotenberg.PdfEngine, error) {
 		engines(mod.splitNames),
 		engines(mod.flattenNames),
 		engines(mod.convertNames),
+		engines(mod.optimizeImagesNames),
 		engines(mod.readMetadataNames),
 		engines(mod.writeMetadataNames),
 		engines(mod.encryptNames),
@@ -341,6 +352,7 @@ func (mod *PdfEngines) Routes() ([]api.Route, error) {
 		mergeRoute(engine),
 		splitRoute(engine),
 		flattenRoute(engine),
+		optimizeRoute(engine),
 		convertRoute(engine),
 		readMetadataRoute(engine),
 		writeMetadataRoute(engine),

pkg/modules/pdfengines/routes.go

@@ -344,6 +344,65 @@ func FlattenStub(ctx *api.Context, engine gotenberg.PdfEngine, inputPaths []stri
 	return nil
 }
 
+// defaultImageQuality is the JPEG quality applied by the image optimization
+// feature when the imageQuality form field is not set.
+const defaultImageQuality = 80
+
+// FormDataPdfOptimize extracts the image-optimization options from the form
+// data: whether to optimize the images, and the JPEG quality (1 to 100) to
+// apply to each re-encoded image.
+func FormDataPdfOptimize(form *api.FormData) (bool, int) {
+	var (
+		optimizeImages bool
+		imageQuality   int
+	)
+
+	form.
+		Bool("optimizeImages", &optimizeImages, false).
+		Custom("imageQuality", func(value string) error {
+			if value == "" {
+				imageQuality = defaultImageQuality
+				return nil
+			}
+
+			intValue, err := strconv.Atoi(value)
+			if err != nil {
+				return err
+			}
+
+			if intValue < 1 {
+				return errors.New("value is inferior to 1")
+			}
+
+			if intValue > 100 {
+				return errors.New("value is superior to 100")
+			}
+
+			imageQuality = intValue
+			return nil
+		})
+
+	return optimizeImages, imageQuality
+}
+
+// OptimizeStub re-encodes the images of each given PDF to shrink the file when
+// optimizeImages is set, leaving text, vectors and structure untouched. It does
+// nothing when optimizeImages is false.
+func OptimizeStub(ctx *api.Context, engine gotenberg.PdfEngine, optimizeImages bool, imageQuality int, inputPaths []string) error {
+	if !optimizeImages {
+		return nil
+	}
+
+	for _, inputPath := range inputPaths {
+		err := engine.OptimizeImages(ctx, ctx.Log(), imageQuality, inputPath)
+		if err != nil {
+			return fmt.Errorf("optimize images of '%s': %w", inputPath, err)
+		}
+	}
+
+	return nil
+}
+
 // ConvertStub transforms a given PDF to the specified formats defined in
 // [gotenberg.PdfFormats]. If no format, it does nothing and returns the input
 // paths.
@@ -931,6 +990,7 @@ func mergeRoute(engine gotenberg.PdfEngine) api.Route {
 			angle, rotatePages := FormDataPdfRotate(form, false)
 			embedsMetadata := FormDataPdfEmbedsMetadata(form)
 			facturX, facturxXmlPath := FormDataPdfFacturX(form)
+			optimizeImages, imageQuality := FormDataPdfOptimize(form)
 
 			var inputPaths []string
 			var flatten bool
@@ -998,6 +1058,11 @@ func mergeRoute(engine gotenberg.PdfEngine) api.Route {
 				}
 			}
 
+			err = OptimizeStub(ctx, engine, optimizeImages, imageQuality, outputPaths)
+			if err != nil {
+				return fmt.Errorf("optimize PDF images: %w", err)
+			}
+
 			pdfFormats = FacturXPdfFormats(ctx, engine, facturX, pdfFormats, false, outputPaths)
 
 			outputPaths, err = ConvertStub(ctx, engine, pdfFormats, outputPaths)
@@ -1108,6 +1173,7 @@ func splitRoute(engine gotenberg.PdfEngine) api.Route {
 			angle, rotatePages := FormDataPdfRotate(form, false)
 			embedsMetadata := FormDataPdfEmbedsMetadata(form)
 			facturX, facturxXmlPath := FormDataPdfFacturX(form)
+			optimizeImages, imageQuality := FormDataPdfOptimize(form)
 
 			var inputPaths []string
 			var flatten bool
@@ -1170,6 +1236,11 @@ func splitRoute(engine gotenberg.PdfEngine) api.Route {
 				}
 			}
 
+			err = OptimizeStub(ctx, engine, optimizeImages, imageQuality, outputPaths)
+			if err != nil {
+				return fmt.Errorf("optimize PDF images: %w", err)
+			}
+
 			pdfFormats = FacturXPdfFormats(ctx, engine, facturX, pdfFormats, false, outputPaths)
 
 			convertOutputPaths, err := ConvertStub(ctx, engine, pdfFormats, outputPaths)
@@ -1262,6 +1333,42 @@ func flattenRoute(engine gotenberg.PdfEngine) api.Route {
 
 // convertRoute returns an [api.Route] which can convert PDFs to a specific ODF
 // format.
+func optimizeRoute(engine gotenberg.PdfEngine) api.Route {
+	return api.Route{
+		Method:      http.MethodPost,
+		Path:        "/forms/pdfengines/optimize",
+		IsMultipart: true,
+		Handler: func(c echo.Context) error {
+			ctx := c.Get("context").(*api.Context)
+
+			form := ctx.FormData()
+			// This route optimizes unconditionally, so the optimizeImages toggle
+			// is ignored; only the image quality is read.
+			_, imageQuality := FormDataPdfOptimize(form)
+
+			var inputPaths []string
+			err := form.
+				MandatoryPaths([]string{".pdf"}, &inputPaths).
+				Validate()
+			if err != nil {
+				return fmt.Errorf("validate form data: %w", err)
+			}
+
+			err = OptimizeStub(ctx, engine, true, imageQuality, inputPaths)
+			if err != nil {
+				return fmt.Errorf("optimize PDF images: %w", err)
+			}
+
+			err = ctx.AddOutputPaths(inputPaths...)
+			if err != nil {
+				return fmt.Errorf("add output paths: %w", err)
+			}
+
+			return nil
+		},
+	}
+}
+
 func convertRoute(engine gotenberg.PdfEngine) api.Route {
 	return api.Route{
 		Method:      http.MethodPost,