Add Canon CR2 and CR3 support for v0.8

app/main.py

@@ -31,6 +31,7 @@ FORMAT_GROUPS: dict[str, tuple[str, ...]] = {
     "tiff": (".tif", ".tiff"),
     "heif": (".heic", ".heif", ".hif"),
     "dng": (".dng",),
+    "canon": (".cr2", ".cr3"),
 }
 FORMAT_LABELS = {
     "jpg": "JPG",
@@ -38,6 +39,7 @@ FORMAT_LABELS = {
     "tiff": "TIFF",
     "heif": "HEIF",
     "dng": "DNG",
+    "canon": "Canon RAW",
 }
 
 
@@ -83,7 +85,7 @@ APP_SUBTITLE = (
     or f"Local folder · in-place {ENABLED_FORMAT_LABEL} metadata"
 )
 APP_URL = os.getenv("APP_URL", "http://localhost:8080")
-APP_VERSION = "0.7.1"
+APP_VERSION = "0.8.0"
 MAX_UPLOAD_MB = int(os.getenv("MAX_UPLOAD_MB", "150"))
 MAX_UPLOAD_BYTES = MAX_UPLOAD_MB * 1024 * 1024
 PREVIEW_MAX_EDGE = int(os.getenv("PREVIEW_MAX_EDGE", "2400"))
@@ -93,8 +95,10 @@ Image.MAX_IMAGE_PIXELS = PREVIEW_MAX_PIXELS
 TIFF_EXTENSIONS = {".tif", ".tiff"}
 HEIF_EXTENSIONS = {".heic", ".heif", ".hif"}
 DNG_EXTENSIONS = {".dng"}
+CANON_EXTENSIONS = {".cr2", ".cr3"}
+IPTC_EXTENSIONS = {".jpg", ".jpeg", ".tif", ".tiff", ".dng", ".cr2"}
 SERVER_PREVIEW_EXTENSIONS = (
-    TIFF_EXTENSIONS | HEIF_EXTENSIONS | DNG_EXTENSIONS
+    TIFF_EXTENSIONS | HEIF_EXTENSIONS | DNG_EXTENSIONS | CANON_EXTENSIONS
 ) & SUPPORTED_EXTENSIONS
 MEDIA_TYPES = {
     ".jpg": "image/jpeg",
@@ -106,6 +110,8 @@ MEDIA_TYPES = {
     ".heif": "image/heif",
     ".hif": "image/heif",
     ".dng": "image/x-adobe-dng",
+    ".cr2": "image/x-canon-cr2",
+    ".cr3": "image/x-canon-cr3",
 }
 LOG_LEVEL = os.getenv("LOG_LEVEL", "INFO").upper()
 
@@ -465,6 +471,10 @@ def _format_name(suffix: str) -> str:
         return "HEIF"
     if suffix in DNG_EXTENSIONS:
         return "DNG"
+    if suffix == ".cr2":
+        return "CR2"
+    if suffix == ".cr3":
+        return "CR3"
     return suffix.lstrip(".").upper() or "image"
 
 
@@ -492,6 +502,20 @@ def _validate_image_signature(path: Path, suffix: str) -> None:
         # DNG is TIFF-based. ExifTool performs the deeper DNG structure
         # validation when reading, extracting a preview, or writing metadata.
         valid = signature.startswith((b"II*\x00", b"MM\x00*"))
+    elif suffix == ".cr2":
+        valid = (
+            signature.startswith((b"II*\x00", b"MM\x00*"))
+            and signature[8:12] == b"CR\x02\x00"
+        )
+    elif suffix == ".cr3":
+        valid = (
+            len(signature) >= 16
+            and signature[4:8] == b"ftyp"
+            and any(
+                signature[offset:offset + 4] == b"crx "
+                for offset in range(8, len(signature) - 3, 4)
+            )
+        )
 
     if not valid:
         raise HTTPException(
@@ -652,7 +676,7 @@ def _ensure_heif_decoder() -> None:
 
 
 def _get_rawpy() -> Any:
-    """Import rawpy once, on the first DNG that needs a rendered fallback."""
+    """Import rawpy once, on the first RAW file that needs a rendered fallback."""
     global _rawpy_module
     if _rawpy_module is not None:
         return _rawpy_module
@@ -661,12 +685,14 @@ def _get_rawpy() -> Any:
             import rawpy
 
             _rawpy_module = rawpy
-            logger.info("DNG LibRaw preview decoder loaded.")
+            logger.info("LibRaw preview decoder loaded.")
     return _rawpy_module
 
 
-def _embedded_dng_preview(path: Path) -> Image.Image | None:
-    """Return the first usable DNG preview, preferring the largest common tags."""
+def _embedded_raw_preview(path: Path, format_name: str) -> Image.Image | None:
+    """Extract the largest usable JPEG from the common RAW preview tags."""
+    best: Image.Image | None = None
+    best_area = 0
     for tag in ("JpgFromRaw", "PreviewImage", "ThumbnailImage"):
         try:
             completed = subprocess.run(
@@ -676,28 +702,38 @@ def _embedded_dng_preview(path: Path) -> Image.Image | None:
                 check=False,
             )
         except subprocess.TimeoutExpired:
-            logger.warning("Timed out extracting DNG %s data.", tag)
+            logger.warning("Timed out extracting %s %s data.", format_name, tag)
             continue
         if completed.returncode != 0 or not completed.stdout:
             continue
         try:
             with Image.open(io.BytesIO(completed.stdout)) as source:
                 source.load()
-                return source.copy()
+                candidate = source.copy()
+            area = candidate.width * candidate.height
+            if area > best_area:
+                if best is not None:
+                    best.close()
+                best = candidate
+                best_area = area
+            else:
+                candidate.close()
         except (UnidentifiedImageError, OSError, ValueError):
-            logger.debug("DNG %s data was not a usable image preview.", tag)
-    return None
+            logger.debug("%s %s data was not a usable image preview.", format_name, tag)
+    return best
 
 
-def _render_dng(path: Path) -> Image.Image:
-    """Render a half-size RGB fallback when a DNG has no embedded JPEG preview."""
+def _render_raw(path: Path, format_name: str) -> Image.Image:
+    """Render a half-size RGB fallback when RAW has no embedded JPEG preview."""
     rawpy = _get_rawpy()
     try:
         with rawpy.imread(str(path)) as raw:
             width = int(raw.sizes.width)
             height = int(raw.sizes.height)
             if width <= 0 or height <= 0 or width * height > PREVIEW_MAX_PIXELS:
-                raise ValueError("DNG dimensions exceed the configured preview pixel limit.")
+                raise ValueError(
+                    f"{format_name} dimensions exceed the configured preview pixel limit."
+                )
             rgb = raw.postprocess(
                 use_camera_wb=True,
                 use_auto_wb=False,
@@ -706,13 +742,13 @@ def _render_dng(path: Path) -> Image.Image:
                 output_bps=8,
             )
     except rawpy.LibRawError as exc:
-        raise ValueError("LibRaw could not decode this DNG.") from exc
+        raise ValueError(f"LibRaw could not decode this {format_name} file.") from exc
     return Image.fromarray(rgb)
 
 
 @app.post("/api/preview")
 async def create_preview(file: Annotated[UploadFile, File(...)]) -> Response:
-    """Generate a temporary browser-friendly preview for TIFF, HEIF, and DNG images."""
+    """Generate a browser-friendly preview for TIFF, HEIF, DNG, CR2, and CR3."""
     filename = file.filename or "photo.tiff"
     suffix = Path(filename).suffix.lower()
     if suffix not in SUPPORTED_EXTENSIONS:
@@ -723,17 +759,18 @@ async def create_preview(file: Annotated[UploadFile, File(...)]) -> Response:
     if suffix not in SERVER_PREVIEW_EXTENSIONS:
         raise HTTPException(
             status_code=415,
-            detail="Server-generated previews are only required for TIFF, HEIF, and DNG files.",
+            detail="Server-generated previews are only required for TIFF, HEIF, DNG, CR2, and CR3 files.",
         )
 
     with tempfile.TemporaryDirectory(prefix="photo-date-editor-preview-") as temp_dir:
         temp_path = Path(temp_dir) / f"source{suffix}"
         await _save_upload(file, temp_path, suffix)
         try:
-            if suffix in DNG_EXTENSIONS:
-                preview = _embedded_dng_preview(temp_path)
+            if suffix in DNG_EXTENSIONS | CANON_EXTENSIONS:
+                format_name = _format_name(suffix)
+                preview = _embedded_raw_preview(temp_path, format_name)
                 if preview is None:
-                    preview = _render_dng(temp_path)
+                    preview = _render_raw(temp_path, format_name)
                 preview = ImageOps.exif_transpose(preview)
             else:
                 if suffix in HEIF_EXTENSIONS:
@@ -881,7 +918,7 @@ async def process_photo(
         temp_path = Path(temp_dir) / f"working{suffix}"
         await _save_upload(file, temp_path, suffix)
 
-        supports_iptc = suffix in {".jpg", ".jpeg", ".tif", ".tiff", ".dng"}
+        supports_iptc = suffix in IPTC_EXTENSIONS
         command = [
             "exiftool",
             "-overwrite_original",