feat(pdfengines): add PDF/UA (#714)

pkg/modules/libreoffice/api/api.go

@@ -47,17 +47,17 @@ type Options struct {
 	Landscape bool
 
 	// PageRanges allows to select the pages to convert.
-	// TODO: should prefer a method form PDFEngine.
+	// TODO: should prefer a method form PdfEngine.
 	// Optional.
 	PageRanges string
 
-	// PdfFormat allows to convert the resulting PDF to PDF/A-1a, PDF/A-2b, or
-	// PDF/A-3b.
+	// PdfFormats allows to convert the resulting PDF to PDF/A-1a, PDF/A-2b,
+	// PDF/A-3b and PDF/UA.
 	// Optional.
-	PdfFormat string
+	PdfFormats gotenberg.PdfFormats
 }
 
-// Uno is an abstraction on top of Api.
+// Uno is an abstraction on top of the Universal Network Objects API.
 type Uno interface {
 	Pdf(ctx context.Context, logger *zap.Logger, inputPath, outputPath string, options Options) error
 	Extensions() []string
@@ -265,13 +265,13 @@ func (a *Api) Metrics() ([]gotenberg.Metric, error) {
 func (a *Api) Checks() ([]health.CheckerOption, error) {
 	return []health.CheckerOption{
 		health.WithCheck(health.Check{
-			Name: "api",
+			Name: "libreoffice",
 			Check: func(_ context.Context) error {
 				if a.supervisor.Healthy() {
 					return nil
 				}
 
-				return errors.New("LibreOffice unhealthy")
+				return errors.New("LibreOffice is unhealthy")
 			},
 		}),
 	}, nil

pkg/modules/libreoffice/api/libreoffice.go

@@ -270,18 +270,26 @@ func (p *libreOfficeProcess) pdf(ctx context.Context, logger *zap.Logger, inputP
 		args = append(args, "--export", fmt.Sprintf("PageRange=%s", options.PageRanges))
 	}
 
-	switch options.PdfFormat {
+	switch options.PdfFormats.PdfA {
 	case "":
-	case gotenberg.FormatPDFA1a:
+	case gotenberg.PdfA1a:
 		args = append(args, "--export", "SelectPdfVersion=1")
-	case gotenberg.FormatPDFA2b:
+	case gotenberg.PdfA2b:
 		args = append(args, "--export", "SelectPdfVersion=2")
-	case gotenberg.FormatPDFA3b:
+	case gotenberg.PdfA3b:
 		args = append(args, "--export", "SelectPdfVersion=3")
 	default:
 		return ErrInvalidPdfFormat
 	}
 
+	if options.PdfFormats.PdfUa {
+		args = append(
+			args,
+			"--export", "EnableTextAccessForAccessibilityTools=true",
+			"--export", "UseTaggedPDF=true",
+		)
+	}
+
 	args = append(args, "--output", outputPath, inputPath)
 
 	cmd, err := gotenberg.CommandContext(ctx, logger, p.arguments.unoBinPath, args...)

pkg/modules/libreoffice/api/libreoffice_test.go

@@ -243,7 +243,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
 				return p
 			}(),
 			fs:            gotenberg.NewFileSystem(),
-			options:       Options{PdfFormat: "foo"},
+			options:       Options{PdfFormats: gotenberg.PdfFormats{PdfA: "foo"}},
 			cancelledCtx:  false,
 			start:         false,
 			expectError:   true,
@@ -417,7 +417,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
 
 				return fs
 			}(),
-			options:      Options{PdfFormat: gotenberg.FormatPDFA1a},
+			options:      Options{PdfFormats: gotenberg.PdfFormats{PdfA: gotenberg.PdfA1a}},
 			cancelledCtx: false,
 			start:        true,
 			expectError:  false,
@@ -446,7 +446,7 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
 
 				return fs
 			}(),
-			options:      Options{PdfFormat: gotenberg.FormatPDFA2b},
+			options:      Options{PdfFormats: gotenberg.PdfFormats{PdfA: gotenberg.PdfA2b}},
 			cancelledCtx: false,
 			start:        true,
 			expectError:  false,
@@ -475,7 +475,36 @@ func TestLibreOfficeProcess_pdf(t *testing.T) {
 
 				return fs
 			}(),
-			options:      Options{PdfFormat: gotenberg.FormatPDFA3b},
+			options:      Options{PdfFormats: gotenberg.PdfFormats{PdfA: gotenberg.PdfA3b}},
+			cancelledCtx: false,
+			start:        true,
+			expectError:  false,
+		},
+		{
+			scenario: "success (PDF/UA)",
+			libreOffice: newLibreOfficeProcess(
+				libreOfficeArguments{
+					binPath:      os.Getenv("LIBREOFFICE_BIN_PATH"),
+					unoBinPath:   os.Getenv("UNOCONV_BIN_PATH"),
+					startTimeout: 5 * time.Second,
+				},
+			),
+			fs: func() *gotenberg.FileSystem {
+				fs := gotenberg.NewFileSystem()
+
+				err := os.MkdirAll(fs.WorkingDirPath(), 0o755)
+				if err != nil {
+					t.Fatalf(fmt.Sprintf("expected no error but got: %v", err))
+				}
+
+				err = os.WriteFile(fmt.Sprintf("%s/document.txt", fs.WorkingDirPath()), []byte("Landscape"), 0o755)
+				if err != nil {
+					t.Fatalf("expected no error but got: %v", err)
+				}
+
+				return fs
+			}(),
+			options:      Options{PdfFormats: gotenberg.PdfFormats{PdfUa: true}},
 			cancelledCtx: false,
 			start:        true,
 			expectError:  false,