feat: add ePub thumbnail support (port #387) (#539)

* feat: add ePub thumbnail support

Co-Authored-By: Jorge Rui Da Silva Barrios <29062316+jorgerui@users.noreply.github.com>

* tests: compare epub cover against png snapshot

Co-Authored-By: yed <yedpodtrzitko@users.noreply.github.com>

* test: optimize epub test file

---------

Co-authored-by: Jorge Rui Da Silva Barrios <29062316+jorgerui@users.noreply.github.com>
Co-authored-by: yed <yedpodtrzitko@users.noreply.github.com>

Author: 3 people

tagstudio/src/core/media_types.py

@@ -23,6 +23,7 @@ class MediaType(str, Enum):
     DATABASE: str = "database"
     DISK_IMAGE: str = "disk_image"
     DOCUMENT: str = "document"
+    EBOOK: str = "ebook"
     FONT: str = "font"
     IMAGE_ANIMATED: str = "image_animated"
     IMAGE_RAW: str = "image_raw"
@@ -160,6 +161,25 @@ class MediaCategories:
         ".wpd",
         ".wps",
     }
+    _EBOOK_SET: set[str] = {
+        ".epub",
+        # ".azw",
+        # ".azw3",
+        # ".cb7",
+        # ".cba",
+        # ".cbr",
+        # ".cbt",
+        # ".cbz",
+        # ".djvu",
+        # ".fb2",
+        # ".ibook",
+        # ".inf",
+        # ".kfx",
+        # ".lit",
+        # ".mobi",
+        # ".pdb"
+        # ".prc",
+    }
     _FONT_SET: set[str] = {
         ".fon",
         ".otf",
@@ -347,6 +367,11 @@ class MediaCategories:
         extensions=_DOCUMENT_SET,
         is_iana=False,
     )
+    EBOOK_TYPES: MediaCategory = MediaCategory(
+        media_type=MediaType.EBOOK,
+        extensions=_EBOOK_SET,
+        is_iana=False,
+    )
     FONT_TYPES: MediaCategory = MediaCategory(
         media_type=MediaType.FONT,
         extensions=_FONT_SET,
@@ -448,6 +473,7 @@ class MediaCategories:
         DATABASE_TYPES,
         DISK_IMAGE_TYPES,
         DOCUMENT_TYPES,
+        EBOOK_TYPES,
         FONT_TYPES,
         IMAGE_ANIMATED_TYPES,
         IMAGE_RAW_TYPES,

tagstudio/src/qt/widgets/thumb_renderer.py

@@ -5,6 +5,7 @@
 
 import math
 import struct
+import zipfile
 from copy import deepcopy
 from io import BytesIO
 from pathlib import Path
@@ -616,6 +617,29 @@ def _source_engine(self, filepath: Path) -> Image.Image:
                 logger.error("Couldn't render thumbnail", filepath=filepath, error=e)
         return im
 
+    def _epub_cover(self, filepath: Path) -> Image.Image:
+        """Extracts and returns the first image found in the ePub file at the given filepath.
+
+        Args:
+            filepath (Path): The path to the ePub file.
+
+        Returns:
+            Image: The first image found in the ePub file, or None by default.
+        """
+        im: Image.Image = None
+        try:
+            with zipfile.ZipFile(filepath, "r") as zip_file:
+                for file_name in zip_file.namelist():
+                    if file_name.lower().endswith(
+                        (".png", ".jpg", ".jpeg", ".gif", ".bmp", ".svg")
+                    ):
+                        image_data = zip_file.read(file_name)
+                        im = Image.open(BytesIO(image_data))
+        except Exception as e:
+            logger.error("Couldn't render thumbnail", filepath=filepath, error=e)
+
+        return im
+
     def _font_short_thumb(self, filepath: Path, size: int) -> Image.Image:
         """Render a small font preview ("Aa") thumbnail from a font file.
 
@@ -1045,6 +1069,11 @@ def render(
                         image = self._audio_waveform_thumb(_filepath, ext, adj_size, pixel_ratio)
                         if image is not None:
                             image = self._apply_overlay_color(image, UiColor.GREEN)
+                # Ebooks =======================================================
+                elif MediaCategories.is_ext_in_category(
+                    ext, MediaCategories.EBOOK_TYPES, mime_fallback=True
+                ):
+                    image = self._epub_cover(_filepath)
                 # Blender ======================================================
                 elif MediaCategories.is_ext_in_category(
                     ext, MediaCategories.BLENDER_TYPES, mime_fallback=True
@@ -1060,7 +1089,6 @@ def render(
                     ext, MediaCategories.SOURCE_ENGINE_TYPES, mime_fallback=True
                 ):
                     image = self._source_engine(_filepath)
-
                 # No Rendered Thumbnail ========================================
                 if not _filepath.exists():
                     raise FileNotFoundError

tagstudio/tests/fixtures/sample.epub

@@ -10,6 +10,18 @@
 from syrupy.extensions.image import PNGImageSnapshotExtension
 
 
+def test_epub_preview(cwd, snapshot):
+    file_path: Path = cwd / "fixtures" / "sample.epub"
+    tr = ThumbRenderer()
+    img: Image.Image = tr._epub_cover(file_path)
+
+    img_bytes = io.BytesIO()
+    img.save(img_bytes, format="PNG")
+    img_bytes.seek(0)
+
+    assert img_bytes.read() == snapshot(extension_class=PNGImageSnapshotExtension)
+
+
 def test_pdf_preview(cwd, snapshot):
     file_path: Path = cwd / "fixtures" / "sample.pdf"
     renderer = ThumbRenderer()
@@ -18,6 +30,7 @@ def test_pdf_preview(cwd, snapshot):
     img_bytes = io.BytesIO()
     img.save(img_bytes, format="PNG")
     img_bytes.seek(0)
+
     assert img_bytes.read() == snapshot(extension_class=PNGImageSnapshotExtension)
 
 
@@ -29,4 +42,5 @@ def test_svg_preview(cwd, snapshot):
     img_bytes = io.BytesIO()
     img.save(img_bytes, format="PNG")
     img_bytes.seek(0)
+
     assert img_bytes.read() == snapshot(extension_class=PNGImageSnapshotExtension)