fix(libreoffice)!: return 500 when a failure is not the client's fault

pkg/modules/libreoffice/api/api.go

@@ -33,12 +33,33 @@ var (
 	// formats option.
 	ErrInvalidPdfFormats = errors.New("invalid PDF formats")
 
-	// ErrUnoException happens when unoconverter returns exit code 5.
+	// ErrUnoException happens when unoconverter returns exit code 5. That code
+	// is the residual bucket of unoconverter's catch-all UNO exception handler:
+	// it covers a malformed page range, a password supplied to a document that
+	// does not need one, a failure to open the document and a failure to write
+	// the output alike. It names the exception class that was caught, not a
+	// cause. See https://github.com/gotenberg/gotenberg/issues/1588.
 	ErrUnoException = errors.New("uno exception")
 
 	// ErrRuntimeException happens when unoconverter returns exit code 6.
+	// unoconverter's own message for it reads "Office probably died", yet a
+	// wrong or missing password also surfaces there. Like [ErrUnoException], it
+	// does not establish who is at fault.
 	ErrRuntimeException = errors.New("runtime exception")
 
+	// ErrIoException happens when unoconverter returns exit code 3. LibreOffice
+	// could not read the source document.
+	ErrIoException = errors.New("io exception")
+
+	// ErrCannotConvertException happens when unoconverter returns exit code 4.
+	// LibreOffice read the document but could not convert it to PDF.
+	ErrCannotConvertException = errors.New("cannot convert exception")
+
+	// ErrIllegalArgumentException happens when unoconverter returns exit code
+	// 8. LibreOffice rejected the source document, usually because its contents
+	// do not match its extension.
+	ErrIllegalArgumentException = errors.New("illegal argument exception")
+
 	// ErrCoreDumped happens randomly; sometimes a conversion will work as
 	// expected, and some other time the same conversion will fail.
 	// See https://github.com/gotenberg/gotenberg/issues/639.
@@ -767,7 +788,10 @@ func conversionRequestAttributes(inputPath string, options Options) []attribute.
 // [gotenberg.ClassifyError].
 func libreofficeErrorType(err error) string {
 	switch {
-	case errors.Is(err, ErrInvalidPdfFormats):
+	case errors.Is(err, ErrInvalidPdfFormats),
+		errors.Is(err, ErrIoException),
+		errors.Is(err, ErrCannotConvertException),
+		errors.Is(err, ErrIllegalArgumentException):
 		return gotenberg.ErrorTypeInvalidInput
 	case errors.Is(err, ErrUnoException), errors.Is(err, ErrRuntimeException):
 		return "libreoffice_exception"

pkg/modules/libreoffice/api/errortype_test.go

@@ -17,6 +17,9 @@ func TestLibreofficeErrorType(t *testing.T) {
 		{"deadline", context.DeadlineExceeded, "timeout"},
 		{"canceled", context.Canceled, "context_cancelled"},
 		{"invalid pdf formats", ErrInvalidPdfFormats, "invalid_input"},
+		{"io exception", ErrIoException, "invalid_input"},
+		{"cannot convert exception", ErrCannotConvertException, "invalid_input"},
+		{"illegal argument exception", ErrIllegalArgumentException, "invalid_input"},
 		{"uno exception", ErrUnoException, "libreoffice_exception"},
 		{"runtime exception", ErrRuntimeException, "libreoffice_exception"},
 		{"queue size exceeded", gotenberg.ErrMaximumQueueSizeExceeded, "libreoffice_unavailable"},

pkg/modules/libreoffice/api/libreoffice.go

@@ -435,9 +435,11 @@ func (p *libreOfficeProcess) pdf(ctx context.Context, logger *slog.Logger, input
 		return nil
 	}
 
-	// LibreOffice's errors are not explicit.
-	// For instance, exit code 5 may be explained by a malformed page range
-	// but also by a not required password.
+	// LibreOffice's errors are not explicit: unoconverter derives its exit code
+	// from the UNO exception class it caught, not from a diagnosis. Exit codes
+	// 5 and 6 are ambiguous in particular, so the route decides the HTTP status
+	// from the request and the document rather than from the code alone.
+	// See https://github.com/gotenberg/gotenberg/issues/1588.
 
 	// We may want to retry in case of a core-dumped event.
 	// See https://github.com/gotenberg/gotenberg/issues/639.
@@ -445,13 +447,17 @@ func (p *libreOfficeProcess) pdf(ctx context.Context, logger *slog.Logger, input
 		return ErrCoreDumped
 	}
 
-	if exitCode == 5 {
-		// Potentially malformed page ranges or password not required.
+	switch exitCode {
+	case 3:
+		return ErrIoException
+	case 4:
+		return ErrCannotConvertException
+	case 5:
 		return ErrUnoException
-	}
-	if exitCode == 6 {
-		// Password potentially required or invalid.
+	case 6:
 		return ErrRuntimeException
+	case 8:
+		return ErrIllegalArgumentException
 	}
 
 	return fmt.Errorf("convert to PDF: %w", err)

pkg/modules/libreoffice/api/protection.go

@@ -0,0 +1,129 @@
+package api
+
+import (
+	"archive/zip"
+	"bytes"
+	"io"
+	"os"
+	"path/filepath"
+	"strings"
+)
+
+// PasswordProtection describes whether a document requires a password to open.
+type PasswordProtection int
+
+const (
+	// PasswordProtectionUnknown means the document's encryption state could not
+	// be determined.
+	PasswordProtectionUnknown PasswordProtection = iota
+
+	// PasswordProtectionNone means the document opens without a password.
+	PasswordProtectionNone
+
+	// PasswordProtectionRequired means the document is encrypted.
+	PasswordProtectionRequired
+)
+
+var (
+	// Compound File Binary magic. An encrypted OOXML document is an
+	// MS-OFFCRYPTO container, which is a compound file. Per MS-CFB 2.2, the
+	// header signature is fixed.
+	ole2Magic = []byte{0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1}
+
+	// Local file header signature. Per APPNOTE.TXT 4.3.7, every ZIP entry
+	// starts with it, so an intact package starts with it too.
+	zipMagic = []byte{0x50, 0x4b, 0x03, 0x04}
+
+	// An unencrypted OOXML document is always a ZIP package, so any of these
+	// extensions over a compound file means the payload is encrypted. Legacy
+	// binary formats (.doc, .xls, .ppt) are compound files either way and are
+	// deliberately absent.
+	ooxmlExtensions = map[string]struct{}{
+		".docx": {}, ".docm": {}, ".dotx": {}, ".dotm": {},
+		".xlsx": {}, ".xlsm": {}, ".xltx": {}, ".xltm": {},
+		".pptx": {}, ".pptm": {}, ".potx": {}, ".potm": {},
+		".ppsx": {}, ".ppsm": {},
+	}
+)
+
+// odfManifestSizeLimit caps how much of an ODF manifest is read. The manifest
+// is a few kilobytes in practice; the cap stops a crafted archive from
+// exhausting memory through its decompressed size.
+const odfManifestSizeLimit = 1 << 20
+
+// DetectPasswordProtection reports whether the document at path is encrypted.
+//
+// Detection is advisory and never fails: an unreadable file, an unknown format
+// or a malformed archive all yield [PasswordProtectionUnknown]. It exists to
+// refine the diagnosis of a conversion that already failed, since LibreOffice's
+// exit codes do not distinguish a missing password from a crash.
+func DetectPasswordProtection(path string) PasswordProtection {
+	f, err := os.Open(path)
+	if err != nil {
+		return PasswordProtectionUnknown
+	}
+	defer func() {
+		_ = f.Close()
+	}()
+
+	magic := make([]byte, 8)
+	n, err := io.ReadFull(f, magic)
+	if err != nil && n < len(zipMagic) {
+		return PasswordProtectionUnknown
+	}
+	magic = magic[:n]
+
+	switch {
+	case bytes.HasPrefix(magic, ole2Magic):
+		if _, ok := ooxmlExtensions[strings.ToLower(filepath.Ext(path))]; ok {
+			return PasswordProtectionRequired
+		}
+		// A legacy binary document is a compound file whether or not it is
+		// encrypted; its encryption lives in a stream this cannot cheaply read.
+		return PasswordProtectionUnknown
+	case bytes.HasPrefix(magic, zipMagic):
+		return detectZipPasswordProtection(f)
+	default:
+		// Flat XML (.fodt), RTF, CSV and everything else carry no encryption.
+		return PasswordProtectionUnknown
+	}
+}
+
+// detectZipPasswordProtection inspects a ZIP package. ODF keeps META-INF/manifest.xml
+// in cleartext even when encrypted, declaring each encrypted entry. An OOXML
+// package has no manifest, and reaching this point already proves it is not an
+// MS-OFFCRYPTO container, so it opens without a password.
+func detectZipPasswordProtection(f *os.File) PasswordProtection {
+	size, err := f.Seek(0, io.SeekEnd)
+	if err != nil {
+		return PasswordProtectionUnknown
+	}
+
+	r, err := zip.NewReader(f, size)
+	if err != nil {
+		return PasswordProtectionUnknown
+	}
+
+	manifest, err := r.Open("META-INF/manifest.xml")
+	if err != nil {
+		// No manifest: an OOXML package, or a ZIP that is not an office
+		// document at all. Neither is encrypted.
+		return PasswordProtectionNone
+	}
+	defer func() {
+		_ = manifest.Close()
+	}()
+
+	content, err := io.ReadAll(io.LimitReader(manifest, odfManifestSizeLimit))
+	if err != nil {
+		return PasswordProtectionUnknown
+	}
+
+	// Per OpenDocument 1.3 part 3, section 4.16, an encrypted entry carries a
+	// <manifest:encryption-data> child.
+	if bytes.Contains(content, []byte("encryption-data")) {
+		return PasswordProtectionRequired
+	}
+
+	return PasswordProtectionNone
+}

pkg/modules/libreoffice/api/protection_test.go

@@ -0,0 +1,193 @@
+package api
+
+import (
+	"archive/zip"
+	"bytes"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+// writeFile writes content to a new file named name inside dir and returns its
+// path.
+func writeFile(t *testing.T, dir, name string, content []byte) string {
+	t.Helper()
+
+	path := filepath.Join(dir, name)
+	err := os.WriteFile(path, content, 0o600)
+	if err != nil {
+		t.Fatalf("write %s: %v", path, err)
+	}
+
+	return path
+}
+
+// writeZip builds a ZIP archive from entries and returns its path.
+func writeZip(t *testing.T, dir, name string, entries map[string]string) string {
+	t.Helper()
+
+	buf := new(bytes.Buffer)
+	w := zip.NewWriter(buf)
+
+	for entryName, content := range entries {
+		f, err := w.Create(entryName)
+		if err != nil {
+			t.Fatalf("create zip entry %s: %v", entryName, err)
+		}
+		_, err = f.Write([]byte(content))
+		if err != nil {
+			t.Fatalf("write zip entry %s: %v", entryName, err)
+		}
+	}
+
+	err := w.Close()
+	if err != nil {
+		t.Fatalf("close zip writer: %v", err)
+	}
+
+	return writeFile(t, dir, name, buf.Bytes())
+}
+
+func TestDetectPasswordProtection(t *testing.T) {
+	dir := t.TempDir()
+
+	ole2 := func(name string) string {
+		return writeFile(t, dir, name, append(ole2Magic, bytes.Repeat([]byte{0x00}, 64)...))
+	}
+
+	for _, tc := range []struct {
+		name string
+		path string
+		want PasswordProtection
+	}{
+		{
+			name: "encrypted OOXML is a compound file",
+			path: ole2("encrypted.docx"),
+			want: PasswordProtectionRequired,
+		},
+		{
+			name: "extension casing is ignored",
+			path: ole2("encrypted.DOCX"),
+			want: PasswordProtectionRequired,
+		},
+		{
+			name: "encrypted spreadsheet",
+			path: ole2("encrypted.xlsx"),
+			want: PasswordProtectionRequired,
+		},
+		{
+			name: "legacy binary document is inconclusive",
+			path: ole2("legacy.doc"),
+			want: PasswordProtectionUnknown,
+		},
+		{
+			name: "plain OOXML package",
+			path: writeZip(t, dir, "plain.docx", map[string]string{
+				"[Content_Types].xml": "<Types/>",
+				"word/document.xml":   "<w:document/>",
+			}),
+			want: PasswordProtectionNone,
+		},
+		{
+			name: "encrypted ODF declares encryption-data in its manifest",
+			path: writeZip(t, dir, "encrypted.odt", map[string]string{
+				"mimetype":              "application/vnd.oasis.opendocument.text",
+				"META-INF/manifest.xml": `<manifest:manifest><manifest:file-entry><manifest:encryption-data manifest:checksum="x"/></manifest:file-entry></manifest:manifest>`,
+				"content.xml":           "<office:document-content/>",
+			}),
+			want: PasswordProtectionRequired,
+		},
+		{
+			name: "plain ODF has a manifest without encryption-data",
+			path: writeZip(t, dir, "plain.odt", map[string]string{
+				"mimetype":              "application/vnd.oasis.opendocument.text",
+				"META-INF/manifest.xml": `<manifest:manifest><manifest:file-entry manifest:full-path="/"/></manifest:manifest>`,
+				"content.xml":           "<office:document-content/>",
+			}),
+			want: PasswordProtectionNone,
+		},
+		{
+			name: "flat XML carries no encryption",
+			path: writeFile(t, dir, "flat.fodt", []byte("<?xml version=\"1.0\"?><office:document/>")),
+			want: PasswordProtectionUnknown,
+		},
+		{
+			name: "plain text",
+			path: writeFile(t, dir, "notes.txt", []byte("hello")),
+			want: PasswordProtectionUnknown,
+		},
+		{
+			name: "file shorter than any magic",
+			path: writeFile(t, dir, "tiny.docx", []byte{0x50}),
+			want: PasswordProtectionUnknown,
+		},
+		{
+			name: "empty file",
+			path: writeFile(t, dir, "empty.docx", nil),
+			want: PasswordProtectionUnknown,
+		},
+		{
+			name: "truncated archive",
+			path: writeFile(t, dir, "truncated.docx", append(zipMagic, bytes.Repeat([]byte{0x00}, 32)...)),
+			want: PasswordProtectionUnknown,
+		},
+		{
+			name: "non-existent path",
+			path: filepath.Join(dir, "does-not-exist.docx"),
+			want: PasswordProtectionUnknown,
+		},
+		{
+			name: "directory",
+			path: dir,
+			want: PasswordProtectionUnknown,
+		},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := DetectPasswordProtection(tc.path); got != tc.want {
+				t.Errorf("DetectPasswordProtection(%s) = %d, want %d", tc.path, got, tc.want)
+			}
+		})
+	}
+}
+
+// TestDetectPasswordProtection_Fixtures anchors detection to the same documents
+// the integration scenarios upload, so a fixture swap cannot silently flip a
+// status code.
+func TestDetectPasswordProtection_Fixtures(t *testing.T) {
+	for _, tc := range []struct {
+		path string
+		want PasswordProtection
+	}{
+		{"../../../../test/integration/testdata/protected_page_1.docx", PasswordProtectionRequired},
+		{"../../../../test/integration/testdata/page_1.docx", PasswordProtectionNone},
+	} {
+		t.Run(filepath.Base(tc.path), func(t *testing.T) {
+			if _, err := os.Stat(tc.path); err != nil {
+				t.Skipf("fixture unavailable: %v", err)
+			}
+			if got := DetectPasswordProtection(tc.path); got != tc.want {
+				t.Errorf("DetectPasswordProtection(%s) = %d, want %d", tc.path, got, tc.want)
+			}
+		})
+	}
+}
+
+// TestDetectPasswordProtection_OversizedManifest verifies that a manifest far
+// larger than the cap still yields a verdict through a bounded read.
+func TestDetectPasswordProtection_OversizedManifest(t *testing.T) {
+	dir := t.TempDir()
+
+	// Well past odfManifestSizeLimit, and highly compressible, so the archive
+	// on disk stays small.
+	filler := strings.Repeat("<manifest:file-entry manifest:full-path=\"pad\"/>", 200_000)
+
+	path := writeZip(t, dir, "oversized.odt", map[string]string{
+		"mimetype":              "application/vnd.oasis.opendocument.text",
+		"META-INF/manifest.xml": "<manifest:manifest>" + filler + "</manifest:manifest>",
+	})
+
+	if got := DetectPasswordProtection(path); got != PasswordProtectionNone {
+		t.Errorf("DetectPasswordProtection(oversized) = %d, want %d", got, PasswordProtectionNone)
+	}
+}