feat: add option to convert progressive JPEG to baseline

sacad/__init__.py

@@ -42,6 +42,7 @@ async def search_and_download(
     amazon_tlds: Sequence[str] = (),
     source_classes: Optional[Sequence[Any]] = None,
     preserve_format: bool = False,
+    convert_progressive_jpeg: bool = False,
 ) -> bool:
     """Search and download a cover, return True if success, False instead."""
     logger = logging.getLogger("Main")
@@ -100,6 +101,7 @@ async def search_and_download(
                 size_tolerance_prct,
                 out_filepath,
                 preserve_format=preserve_format,
+                convert_progressive_jpeg=convert_progressive_jpeg,
             )
         except Exception as e:
             logger.warning(f"Download of {result} failed: {e.__class__.__qualname__} {e}")
@@ -153,6 +155,12 @@ def setup_common_args(arg_parser: argparse.ArgumentParser) -> None:
         default=False,
         help="Preserve source image format if possible. Target format will still be prefered when sorting results.",
     )
+    arg_parser.add_argument(
+        "--convert-progressive-jpeg",
+        action="store_true",
+        default=False,
+        help="Convert progressive JPEG to baseline if needed. May result in bigger files and loss of quality.",
+    )
 
 
 def cl_main() -> None:
@@ -222,6 +230,7 @@ def cl_main() -> None:
         amazon_tlds=args.amazon_tlds,
         source_classes=args.cover_sources,
         preserve_format=args.preserve_format,
+        convert_progressive_jpeg=args.convert_progressive_jpeg,
     )
     future = asyncio.ensure_future(coroutine)
     asyncio.get_event_loop().run_until_complete(future)

sacad/cover.py

@@ -13,6 +13,7 @@
 import pickle
 import shutil
 import urllib.parse
+from typing import Dict
 
 import appdirs
 import bitarray
@@ -166,13 +167,22 @@ def __str__(self):
             s += f" [x{len(self.urls)}]"
         return s
 
-    async def get(self, target_format, target_size, size_tolerance_prct, out_filepath, *, preserve_format=False):
+    async def get(
+        self,
+        target_format: CoverImageFormat,
+        target_size: int,
+        size_tolerance_prct: float,
+        out_filepath: str,
+        *,
+        preserve_format: bool = False,
+        convert_progressive_jpeg: bool = False,
+    ) -> None:
         """Download cover and process it."""
         images_data = []
         for i, url in enumerate(self.urls):
             # download
             logging.getLogger("Cover").info(f"Downloading cover {url!r} (part {i + 1}/{len(self.urls)})...")
-            headers = {}
+            headers: Dict[str, str] = {}
             self.source.updateHttpHeaders(headers)
 
             async def pre_cache_callback(img_data):
@@ -193,7 +203,11 @@ async def pre_cache_callback(img_data):
             abs(max(self.size) - target_size) > target_size * size_tolerance_prct / 100
         )
         need_join = len(images_data) > 1
-        if (need_format_change and (not preserve_format)) or need_join or need_size_change:
+        need_post_process = (need_format_change and (not preserve_format)) or need_join or need_size_change
+        need_post_process = need_post_process or (
+            __class__.isProgressiveJpegData(images_data[0]) and convert_progressive_jpeg
+        )
+        if need_post_process:
             # post process
             image_data = self.postProcess(
                 images_data, target_format if need_format_change else None, target_size if need_size_change else None
@@ -213,6 +227,16 @@ async def pre_cache_callback(img_data):
         with open(out_filepath, "wb") as file:
             file.write(image_data)
 
+    @staticmethod
+    def isProgressiveJpegData(data: bytes) -> bool:
+        """Return True if data is from a progressive JPEG."""
+        in_bytes = io.BytesIO(data)
+        try:
+            img = PIL.Image.open(in_bytes)
+            return bool(img.info["progressive"])
+        except Exception:
+            return False
+
     def postProcess(self, images_data, new_format, new_size):
         """
         Convert image binary data.

sacad/recurse.py

@@ -288,7 +288,6 @@ def ichunk(iterable, n):
 def get_covers(work, args):
     """Get missing covers."""
     with contextlib.ExitStack() as cm:
-
         if args.cover_pattern == EMBEDDED_ALBUM_ART_SYMBOL:
             tmp_prefix = f"{os.path.splitext(os.path.basename(inspect.getfile(inspect.currentframe())))[0]}_"
             tmp_dir = cm.enter_context(tempfile.TemporaryDirectory(prefix=tmp_prefix))
@@ -361,6 +360,7 @@ def post_download(future):
                     amazon_tlds=args.amazon_tlds,
                     source_classes=args.cover_sources,
                     preserve_format=args.preserve_format,
+                    convert_progressive_jpeg=args.convert_progressive_jpeg,
                 )
                 future = asyncio.ensure_future(coroutine)
                 futures[future] = cur_work