refactor: split ThumbRenderer class

pyproject.toml

@@ -88,7 +88,7 @@ filterwarnings = [
 ignore = [
     ".venv/**",
     "src/tagstudio/core/library/json/",
-    "src/tagstudio/qt/previews/vendored/pydub/",
+    "src/tagstudio/renderers/vendored/pydub/",
 ]
 include = ["src/tagstudio", "tests"]
 # Reference for the settings here: https://github.com/microsoft/pyright/blob/main/docs/configuration.md
@@ -117,7 +117,7 @@ ignore = ["D100", "D101", "D102", "D103", "D104", "D105", "D106", "D107"]
 
 [tool.ruff.lint.per-file-ignores]
 "tests/**" = ["D", "E402"]
-"src/tagstudio/qt/previews/vendored/**" = ["B", "E", "N", "UP", "SIM115"]
+"src/tagstudio/renderers/vendored/**" = ["B", "E", "N", "UP", "SIM115"]
 
 [tool.ruff.lint.pydocstyle]
 convention = "google"

src/tagstudio/qt/file_renderer.py

@@ -0,0 +1,7 @@
+from PySide6.QtCore import QObject
+
+
+class FileRenderer(QObject): ...
+
+
+"""A Qt-specific entry point for rendering file previews and thumbnails."""

src/tagstudio/qt/helpers/file_tester.py

@@ -6,7 +6,7 @@ from pathlib import Path
 
 import ffmpeg
 
-from tagstudio.qt.previews.vendored.probe import probe
+from tagstudio.renderers.vendored.probe import probe
 
 
 def is_readable_video(filepath: Path | str):
@@ -23,11 +23,7 @@ def is_readable_video(filepath: Path | str):
             return False
         for stream in result["streams"]:
             # DRM check
-            if stream.get("codec_tag_string") in [
-                "drma",
-                "drms",
-                "drmi",
-            ]:
+            if stream.get("codec_tag_string") in ["drma", "drms", "drmi"]:
                 return False
     except ffmpeg.Error:
         return False

src/tagstudio/renderers/apple.py

@@ -0,0 +1,56 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import zipfile
+from io import BytesIO
+from pathlib import Path
+
+import structlog
+from PIL import Image
+
+logger = structlog.get_logger(__name__)
+
+
+def apple_embedded_thumb(filepath: Path) -> Image.Image | None:
+    """Extract and render an apple embedded thumbnail (iWork, Apple Creative Studio).
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    thumb_files: list[str] = [
+        "preview.jpg",
+        "QuickLook/Preview.heic",
+        "QuickLook/Thumbnail.jpg",
+        "QuickLook/Thumbnail.heic",
+        "QuickLook/Thumbnail.webp",
+        "QuickLook/Icon.webp",
+    ]
+    im: Image.Image | None = None
+
+    def get_image(path: str) -> Image.Image | None:
+        thumb_im: Image.Image | None = None
+        # Read the specific file into memory
+        file_data = zip_file.read(path)
+        thumb_im = Image.open(BytesIO(file_data))
+        return thumb_im
+
+    try:
+        with zipfile.ZipFile(filepath, "r") as zip_file:
+            thumb: Image.Image | None = None
+
+            # Check if the file exists in the zip
+            for thumb_file in thumb_files:
+                if thumb_file in zip_file.namelist():
+                    thumb = get_image(thumb_file)
+                    break
+            else:
+                logger.error("Couldn't render thumbnail", filepath=filepath)
+
+            if thumb:
+                im = Image.new("RGB", thumb.size, color="#1e1e1e")
+                im.paste(thumb)
+    except zipfile.BadZipFile as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=e)
+
+    return im

src/tagstudio/renderers/archive.py

@@ -0,0 +1,118 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import tarfile
+import zipfile
+from io import BytesIO
+from pathlib import Path
+from typing import Literal
+
+import py7zr
+import py7zr.io
+import rarfile
+import structlog
+from PIL import Image
+
+from tagstudio.core.media_types import MediaCategories
+from tagstudio.core.utils.types import unwrap
+from tagstudio.renderers.raster_image import load_raster_image
+
+logger = structlog.get_logger(__name__)
+
+type Archive = zipfile.ZipFile | rarfile.RarFile | SevenZipFile | TarFile
+
+
+class SevenZipFile(py7zr.SevenZipFile):
+    """Wrapper around py7zr.SevenZipFile to mimic zipfile.ZipFile's API."""
+
+    def __init__(self, filepath: Path, mode: Literal["r"]) -> None:
+        super().__init__(filepath, mode)
+
+    def read(self, name: str) -> bytes:
+        # SevenZipFile must be reset after every extraction
+        # See https://py7zr.readthedocs.io/en/stable/api.html#py7zr.SevenZipFile.extract
+        self.reset()
+        factory = py7zr.io.BytesIOFactory(limit=10485760)  # 10 MiB
+        self.extract(targets=[name], factory=factory)
+        return factory.get(name).read()
+
+
+class TarFile:
+    """Wrapper around tarfile.TarFile to mimic zipfile.ZipFile's API."""
+
+    def __init__(self, filepath: Path, mode: Literal["r"]) -> None:
+        self.tar: tarfile.TarFile
+        self.filepath = filepath
+        self.mode: Literal["r"] = mode
+
+    def namelist(self) -> list[str]:
+        return self.tar.getnames()
+
+    def read(self, name: str) -> bytes:
+        return unwrap(self.tar.extractfile(name)).read()
+
+    def __enter__(self) -> "TarFile":
+        self.tar = tarfile.open(name=self.filepath, mode=self.mode).__enter__()
+        return self
+
+    def __exit__(self, *args) -> None:  # pyright: ignore[reportUnknownParameterType, reportMissingParameterType]
+        self.tar.__exit__(*args)
+
+
+def archive_thumb(filepath: Path, ext: str) -> Image.Image | None:
+    """Extract the first image found in the archive.
+
+    Args:
+        filepath (Path): The path to the archive.
+        ext (str): The file extension.
+
+    Returns:
+        Image: The first image found in the archive.
+    """
+    im: Image.Image | None = None
+    try:
+        with open_archive(filepath, ext) as archive:
+            im = first_image(archive)
+    except Exception as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+
+    return im
+
+
+def open_archive(filepath: Path, ext: str) -> Archive:
+    """Open an archive with its corresponding archiver.
+
+    Args:
+        filepath (Path): The path to the archive.
+        ext (str): The file extension.
+
+    Returns:
+        Archive: The opened archive.
+    """
+    archiver: type[Archive] = zipfile.ZipFile
+    if ext in {".7z", ".cb7", ".s7z"}:
+        archiver = SevenZipFile
+    elif ext in {".cbr", ".rar"}:
+        archiver = rarfile.RarFile
+    elif ext in {".cbt", ".tar", ".tgz"}:
+        archiver = TarFile
+    return archiver(filepath, "r")
+
+
+def first_image(archive: Archive) -> Image.Image | None:
+    """Find and extract the first renderable image in the archive.
+
+    Args:
+        archive (Archive): The current archive.
+
+    Returns:
+        Image: The first renderable image in the archive.
+    """
+    for file_name in archive.namelist():  # pyright: ignore[reportUnknownVariableType]
+        ext = Path(file_name).suffix
+        if MediaCategories.IMAGE_RASTER_TYPES.contains(ext):
+            image_data = archive.read(file_name)  # pyright: ignore[reportUnknownVariableType]
+            return load_raster_image(BytesIO(image_data))
+
+    return None

src/tagstudio/renderers/audio.py

@@ -0,0 +1,149 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import math
+from io import BytesIO
+from pathlib import Path
+from warnings import catch_warnings
+
+import numpy as np
+import structlog
+from mutagen import flac, id3, mp4
+from mutagen._util import MutagenError
+from PIL import Image, ImageDraw
+
+from tagstudio.renderers.vendored.pydub.audio_segment import (
+    _AudioSegment as AudioSegment,  # pyright: ignore[reportPrivateUsage]
+)
+
+logger = structlog.get_logger(__name__)
+
+
+def audio_album_thumb(filepath: Path, ext: str) -> Image.Image | None:
+    """Return an album cover thumb from an audio file if a cover is present.
+
+    Args:
+        filepath (Path): The path of the file.
+        ext (str): The file extension (with leading ".").
+    """
+    image: Image.Image | None = None
+    try:
+        if not filepath.is_file():
+            raise FileNotFoundError
+
+        artwork = None
+        if ext in [".mp3"]:
+            id3_tags: id3.ID3 = id3.ID3(filepath)
+            id3_covers: list = id3_tags.getall("APIC")  # pyright: ignore[reportUnknownVariableType]
+            if id3_covers:
+                artwork = Image.open(BytesIO(id3_covers[0].data))
+        elif ext in [".flac"]:
+            flac_tags: flac.FLAC = flac.FLAC(filepath)
+            flac_covers: list = flac_tags.pictures  # pyright: ignore[reportUnknownVariableType]
+            if flac_covers:
+                artwork = Image.open(BytesIO(flac_covers[0].data))
+        elif ext in [".mp4", ".m4a", ".aac"]:
+            mp4_tags: mp4.MP4 = mp4.MP4(filepath)
+            mp4_covers: list | None = mp4_tags.get("covr")  # pyright: ignore[reportUnknownVariableType]
+            if mp4_covers:
+                artwork = Image.open(BytesIO(mp4_covers[0]))
+        if artwork:
+            image = artwork
+    except (
+        FileNotFoundError,
+        id3.ID3NoHeaderError,  # pyright: ignore[reportPrivateImportUsage]
+        mp4.MP4MetadataError,
+        mp4.MP4StreamInfoError,
+        MutagenError,
+    ) as e:
+        logger.error("Couldn't read album artwork", path=filepath, error=type(e).__name__)
+    return image
+
+
+def audio_waveform_thumb(
+    filepath: Path, ext: str, size: int, pixel_ratio: float
+) -> Image.Image | None:
+    """Render a waveform image from an audio file.
+
+    Args:
+        filepath (Path): The path of the file.
+        ext (str): The file extension (with leading ".").
+        size (tuple[int,int]): The size of the thumbnail.
+        pixel_ratio (float): The screen pixel ratio.
+    """
+    # BASE_SCALE used for drawing on a larger image and resampling down
+    # to provide an antialiased effect.
+    base_scale: int = 2
+    samples_per_bar: int = 3
+    size_scaled: int = size * base_scale
+    allow_small_min: bool = False
+    im: Image.Image | None = None
+
+    try:
+        bar_count: int = min(math.floor((size // pixel_ratio) / 5), 64)
+        audio = AudioSegment.from_file(filepath, ext[1:])  # pyright: ignore[reportUnknownVariableType]
+        data = np.frombuffer(buffer=audio._data, dtype=np.int16)
+        data_indices = np.linspace(1, len(data), num=bar_count * samples_per_bar)
+        bar_margin: float = ((size_scaled / (bar_count * 3)) * base_scale) / 2
+        line_width: float = ((size_scaled - bar_margin) / (bar_count * 3)) * base_scale
+        bar_height: float = (size_scaled) - (size_scaled // bar_margin)
+
+        count: int = 0
+        maximum_item: int = 0
+        max_array: list[int] = []
+        highest_line: int = 0
+
+        for i in range(-1, len(data_indices)):
+            d = data[math.ceil(data_indices[i]) - 1]
+            if count < samples_per_bar:
+                count = count + 1
+                with catch_warnings(record=True):
+                    if abs(d) > maximum_item:
+                        maximum_item = int(abs(d))
+            else:
+                max_array.append(maximum_item)
+
+                if maximum_item > highest_line:
+                    highest_line = maximum_item
+
+                maximum_item = 0
+                count = 1
+
+        line_ratio = max(highest_line / bar_height, 1)
+
+        im = Image.new("RGB", (size_scaled, size_scaled), color="#000000")
+        draw = ImageDraw.Draw(im)
+
+        current_x = bar_margin
+        for item in max_array:
+            item_height = item / line_ratio
+
+            # If small minimums are not allowed, raise all values
+            # smaller than the line width to the same value.
+            if not allow_small_min:
+                item_height = max(item_height, line_width)
+
+            current_y = (bar_height - item_height + (size_scaled // bar_margin)) // 2
+
+            draw.rounded_rectangle(
+                (
+                    current_x,
+                    current_y,
+                    (current_x + line_width),
+                    (current_y + item_height),
+                ),
+                radius=100 * base_scale,
+                fill=("#FF0000"),
+                outline=("#FFFF00"),
+                width=max(math.ceil(line_width / 6), base_scale),
+            )
+
+            current_x = current_x + line_width + bar_margin
+
+        im.resize((size, size), Image.Resampling.BILINEAR)
+
+    except Exception as e:
+        logger.error("Couldn't render waveform", path=filepath.name, error=type(e).__name__)
+
+    return im

src/tagstudio/renderers/blender.py

@@ -0,0 +1,43 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+from pathlib import Path
+
+import structlog
+from PIL import Image
+from PySide6.QtCore import Qt
+from PySide6.QtGui import QGuiApplication
+
+from tagstudio.renderers.vendored.blender_renderer import (
+    blend_thumb,  # pyright: ignore[reportUnknownVariableType]
+)
+
+logger = structlog.get_logger(__name__)
+
+
+def blender(filepath: Path) -> Image.Image | None:
+    """Get an emended thumbnail from a Blender file, if a thumbnail is present.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    bg_color: str = (
+        "#1e1e1e"
+        if QGuiApplication.styleHints().colorScheme() is Qt.ColorScheme.Dark
+        else "#FFFFFF"
+    )
+    im: Image.Image | None = None
+    try:
+        if (blend_image := blend_thumb(str(filepath))) is not None:
+            bg = Image.new("RGB", blend_image.size, color=bg_color)
+            bg.paste(blend_image, mask=blend_image.getchannel(3))
+            im = bg
+        else:
+            logger.info(
+                f"[ThumbRenderer][BLENDER][INFO] {filepath.name} "
+                "Doesn't have an embedded thumbnail."
+            )
+    except Exception as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im

src/tagstudio/renderers/clip_studio.py

@@ -0,0 +1,80 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import base64
+import sqlite3
+import struct
+import xml.etree.ElementTree as ET
+from io import BytesIO
+from pathlib import Path
+
+import structlog
+from PIL import Image
+
+logger = structlog.get_logger(__name__)
+
+
+def clip_thumb(filepath: Path) -> Image.Image | None:
+    """Extract the thumbnail from the SQLite database embedded in a .clip file.
+
+    Args:
+        filepath (Path): The path of the .clip file.
+
+    Returns:
+        Image: The embedded thumbnail, if extractable.
+    """
+    im: Image.Image | None = None
+    try:
+        with open(filepath, "rb") as f:
+            blob = f.read()
+            sqlite_index = blob.find(b"SQLite format 3")
+            if sqlite_index == -1:
+                return im
+
+        with sqlite3.connect(":memory:") as conn:
+            conn.deserialize(blob[sqlite_index:])
+            thumbnail = conn.execute("SELECT ImageData FROM CanvasPreview").fetchone()
+            if thumbnail:
+                im = Image.open(BytesIO(thumbnail[0]))
+        conn.close()
+    except Exception as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+
+    return im
+
+
+def pdn_thumb(filepath: Path) -> Image.Image | None:
+    """Extract the base64-encoded thumbnail from a .pdn file header.
+
+    Args:
+        filepath (Path): The path of the .pdn file.
+
+    Returns:
+        Image: the decoded PNG thumbnail or None by default.
+    """
+    im: Image.Image | None = None
+    with open(filepath, "rb") as f:
+        try:
+            # First 4 bytes are the magic number
+            if f.read(4) != b"PDN3":
+                return im
+
+            # Header length is a little-endian 24-bit int
+            header_size = struct.unpack("<i", f.read(3) + b"\x00")[0]
+            thumb_element = ET.fromstring(f.read(header_size)).find("./*thumb")
+            if thumb_element is None:
+                return im
+
+            encoded_png = thumb_element.get("png")
+            if encoded_png:
+                decoded_png = base64.b64decode(encoded_png)
+                im = Image.open(BytesIO(decoded_png))
+                if im.mode == "RGBA":
+                    new_bg = Image.new("RGB", im.size, color="#1e1e1e")
+                    new_bg.paste(im, mask=im.getchannel(3))
+                    im = new_bg
+        except Exception as e:
+            logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+
+    return im

src/tagstudio/renderers/ebook.py

@@ -0,0 +1,73 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import xml.etree.ElementTree as ET
+from io import BytesIO
+from pathlib import Path
+from xml.etree.ElementTree import Element
+
+import structlog
+from PIL import Image
+
+from tagstudio.core.media_types import MediaCategories
+from tagstudio.core.utils.types import unwrap
+from tagstudio.renderers.archive import Archive, first_image, open_archive
+from tagstudio.renderers.raster_image import load_raster_image
+
+logger = structlog.get_logger(__name__)
+
+
+def epub_cover(filepath: Path, ext: str) -> Image.Image | None:
+    """Extracts the cover specified by ComicInfo.xml or first image found in the ePub file.
+
+    Args:
+        filepath (Path): The path to the ePub file.
+        ext (str): The file extension.
+
+    Returns:
+        Image: The cover specified in ComicInfo.xml,
+        the first image found in the ePub file, or None by default.
+    """
+    im: Image.Image | None = None
+    try:
+        with open_archive(filepath, ext) as archive:
+            if "ComicInfo.xml" in archive.namelist():
+                comic_info = ET.fromstring(archive.read("ComicInfo.xml"))
+                im = cover_from_comic_info(archive, comic_info, "FrontCover")
+                if not im:
+                    im = cover_from_comic_info(archive, comic_info, "InnerCover")
+
+            if not im:
+                im = first_image(archive)
+    except Exception as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+
+    return im
+
+
+def cover_from_comic_info(
+    archive: Archive, comic_info: Element, cover_type: str
+) -> Image.Image | None:
+    """Extract the cover specified in ComicInfo.xml.
+
+    Args:
+        archive (Archive): The current ePub file.
+        comic_info (Element): The parsed ComicInfo.xml.
+        cover_type (str): The type of cover to load.
+
+    Returns:
+        Image: The cover specified in ComicInfo.xml.
+    """
+    im: Image.Image | None = None
+
+    cover = comic_info.find(f"./*Page[@Type='{cover_type}']")
+    if cover is not None:
+        pages = [f for f in archive.namelist() if f != "ComicInfo.xml"]  # pyright: ignore[reportUnknownVariableType]
+        page_name = pages[int(unwrap(cover.get("Image")))]  # pyright: ignore[reportUnknownVariableType]
+        ext = Path(page_name).suffix
+        if MediaCategories.IMAGE_RASTER_TYPES.contains(ext):
+            image_data = archive.read(page_name)  # pyright: ignore[reportUnknownVariableType]
+            im = load_raster_image(BytesIO(image_data))
+
+    return im

src/tagstudio/renderers/font.py

@@ -0,0 +1,108 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import math
+from pathlib import Path
+from typing import cast
+
+import numpy as np
+import structlog
+from PIL import Image, ImageDraw, ImageFont
+
+from tagstudio.core.constants import FONT_SAMPLE_SIZES, FONT_SAMPLE_TEXT
+from tagstudio.qt.helpers.color_overlay import auto_theme_overlay
+from tagstudio.qt.helpers.text_wrapper import wrap_full_text
+
+logger = structlog.get_logger(__name__)
+
+
+def font_short_thumb(filepath: Path, size: int) -> Image.Image | None:
+    """Render a small font preview ("Aa") thumbnail from a font file.
+
+    Args:
+        filepath (Path): The path of the file.
+        size (tuple[int,int]): The size of the thumbnail.
+    """
+    im: Image.Image | None = None
+    try:
+        bg = Image.new("RGB", (size, size), color="#000000")
+        raw = Image.new("RGB", (size * 3, size * 3), color="#000000")
+        draw = ImageDraw.Draw(raw)
+        font = ImageFont.truetype(filepath, size=size)
+        # NOTE: While a stroke effect is desired, the text
+        # method only allows for outer strokes, which looks
+        # a bit weird when rendering fonts.
+        draw.text(
+            (size // 8, size // 8),
+            "Aa",
+            font=font,
+            fill="#FF0000",
+            # stroke_width=math.ceil(size / 96),
+            # stroke_fill="#FFFF00",
+        )
+        # NOTE: Change to getchannel(1) if using an outline.
+        data = np.asarray(raw.getchannel(0))
+
+        m, n = data.shape[:2]
+        col: np.ndarray = cast(np.ndarray, data.any(0))
+        row: np.ndarray = cast(np.ndarray, data.any(1))
+        cropped_data = np.asarray(raw)[
+            row.argmax() : m - row[::-1].argmax(),
+            col.argmax() : n - col[::-1].argmax(),
+        ]
+        cropped_im: Image.Image = Image.fromarray(cropped_data, "RGB")
+
+        margin: int = math.ceil(size // 16)
+
+        orig_x, orig_y = cropped_im.size
+        new_x, new_y = (size, size)
+        if orig_x > orig_y:
+            new_x = size
+            new_y = math.ceil(size * (orig_y / orig_x))
+        elif orig_y > orig_x:
+            new_y = size
+            new_x = math.ceil(size * (orig_x / orig_y))
+
+        cropped_im = cropped_im.resize(
+            size=(new_x - (margin * 2), new_y - (margin * 2)),
+            resample=Image.Resampling.BILINEAR,
+        )
+        bg.paste(
+            cropped_im,
+            box=(margin, margin + ((size - new_y) // 2)),
+        )
+        im = bg
+    except OSError as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im
+
+
+def font_long_thumb(filepath: Path, size: int) -> Image.Image | None:
+    """Render a large font preview ("Alphabet") thumbnail from a font file.
+
+    Args:
+        filepath (Path): The path of the file.
+        size (tuple[int,int]): The size of the thumbnail.
+    """
+    # Scale the sample font sizes to the preview image
+    # resolution,assuming the sizes are tuned for 256px.
+    im: Image.Image | None = None
+    try:
+        scaled_sizes: list[int] = [math.floor(x * (size / 256)) for x in FONT_SAMPLE_SIZES]
+        bg = Image.new("RGBA", (size, size), color="#00000000")
+        draw = ImageDraw.Draw(bg)
+        lines_of_padding = 2
+        y_offset = 0.0
+
+        for font_size in scaled_sizes:
+            font = ImageFont.truetype(filepath, size=font_size)
+            text_wrapped: str = wrap_full_text(FONT_SAMPLE_TEXT, font=font, width=size, draw=draw)
+            draw.multiline_text((0, y_offset), text_wrapped, font=font)
+            y_offset += (len(text_wrapped.split("\n")) + lines_of_padding) * draw.textbbox(
+                (0, 0), "A", font=font
+            )[-1]
+        im = auto_theme_overlay(bg, use_alpha=False)
+    except OSError as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im

src/tagstudio/renderers/krita.py

@@ -0,0 +1,35 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import zipfile
+from io import BytesIO
+from pathlib import Path
+
+import structlog
+from PIL import Image
+
+logger = structlog.get_logger(__name__)
+
+
+def krita_thumb(filepath: Path) -> Image.Image | None:
+    """Extract and render a thumbnail for a Krita file.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    file_path_within_zip = "preview.png"
+    im: Image.Image | None = None
+    with zipfile.ZipFile(filepath, "r") as zip_file:
+        # Check if the file exists in the zip
+        if file_path_within_zip in zip_file.namelist():
+            # Read the specific file into memory
+            file_data = zip_file.read(file_path_within_zip)
+            thumb_im = Image.open(BytesIO(file_data))
+            if thumb_im:
+                im = Image.new("RGB", thumb_im.size, color="#1e1e1e")
+                im.paste(thumb_im)
+        else:
+            logger.error("Couldn't render thumbnail", filepath=filepath)
+
+    return im

src/tagstudio/renderers/medibang_paint.py

@@ -0,0 +1,58 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import os
+import struct
+import xml.etree.ElementTree as ET
+import zlib
+from pathlib import Path
+
+import structlog
+from PIL import Image
+
+from tagstudio.core.utils.types import unwrap
+
+logger = structlog.get_logger(__name__)
+
+
+def mdp_thumb(filepath: Path) -> Image.Image | None:
+    """Extract the thumbnail from a .mdp file.
+
+    Args:
+        filepath (Path): The path of the .mdp file.
+
+    Returns:
+        Image: The embedded thumbnail.
+    """
+    im: Image.Image | None = None
+    try:
+        with open(filepath, "rb") as f:
+            magic = struct.unpack("<7sx", f.read(8))[0]
+            if magic != b"mdipack":
+                return im
+
+            bin_header = struct.unpack("<LLL", f.read(12))
+            xml_header = ET.fromstring(f.read(bin_header[1]))
+            mdibin_count = len(xml_header.findall("./*Layer")) + 1
+            for _ in range(mdibin_count):
+                pac_header = struct.unpack("<3sxLLLL48s64s", f.read(132))
+                if not pac_header[6].startswith(b"thumb"):
+                    f.seek(pac_header[3], os.SEEK_CUR)
+                    continue
+
+                thumb_element = unwrap(xml_header.find("Thumb"))
+                dimensions = (
+                    int(unwrap(thumb_element.get("width"))),
+                    int(unwrap(thumb_element.get("height"))),
+                )
+                thumb_blob = f.read(pac_header[3])
+                if pac_header[2] == 1:
+                    thumb_blob = zlib.decompress(thumb_blob, bufsize=pac_header[4])
+
+                im = Image.frombytes("RGBA", dimensions, thumb_blob, "raw", "BGRA")
+                break
+    except Exception as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+
+    return im

src/tagstudio/renderers/microsoft_office.py

@@ -0,0 +1,38 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import zipfile
+from io import BytesIO
+from pathlib import Path
+
+import structlog
+from PIL import Image
+
+logger = structlog.get_logger(__name__)
+
+
+def powerpoint_thumb(filepath: Path) -> Image.Image | None:
+    """Extract and render a thumbnail for a Microsoft PowerPoint file.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    file_path_within_zip = "docProps/thumbnail.jpeg"
+    im: Image.Image | None = None
+    try:
+        with zipfile.ZipFile(filepath, "r") as zip_file:
+            # Check if the file exists in the zip
+            if file_path_within_zip in zip_file.namelist():
+                # Read the specific file into memory
+                file_data = zip_file.read(file_path_within_zip)
+                thumb_im = Image.open(BytesIO(file_data))
+                if thumb_im:
+                    im = Image.new("RGB", thumb_im.size, color="#1e1e1e")
+                    im.paste(thumb_im)
+            else:
+                logger.error("Couldn't render thumbnail", filepath=filepath)
+    except zipfile.BadZipFile as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=e)
+
+    return im

src/tagstudio/renderers/open_document.py

@@ -0,0 +1,35 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import zipfile
+from io import BytesIO
+from pathlib import Path
+
+import structlog
+from PIL import Image
+
+logger = structlog.get_logger(__name__)
+
+
+def open_doc_thumb(filepath: Path) -> Image.Image | None:
+    """Extract and render a thumbnail for an OpenDocument file.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    file_path_within_zip = "Thumbnails/thumbnail.png"
+    im: Image.Image | None = None
+    with zipfile.ZipFile(filepath, "r") as zip_file:
+        # Check if the file exists in the zip
+        if file_path_within_zip in zip_file.namelist():
+            # Read the specific file into memory
+            file_data = zip_file.read(file_path_within_zip)
+            thumb_im = Image.open(BytesIO(file_data))
+            if thumb_im:
+                im = Image.new("RGB", thumb_im.size, color="#1e1e1e")
+                im.paste(thumb_im)
+        else:
+            logger.error("Couldn't render thumbnail", filepath=filepath)
+
+    return im

src/tagstudio/renderers/pdf.py

@@ -0,0 +1,65 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+from io import BytesIO
+from pathlib import Path
+
+import structlog
+from PIL import (
+    Image,
+)
+from PySide6.QtCore import QBuffer, QFile, QFileDevice, QIODeviceBase, QSizeF
+from PySide6.QtGui import QImage
+from PySide6.QtPdf import QPdfDocument, QPdfDocumentRenderOptions
+
+from tagstudio.qt.helpers.image_effects import replace_transparent_pixels
+
+logger = structlog.get_logger(__name__)
+
+
+def pdf_thumb(filepath: Path, size: int, ext: str) -> Image.Image | None:
+    """Render a thumbnail for a PDF or Adobe Illustrator file.
+
+    filepath (Path): The path of the file.
+        size (int): The size of the icon.
+        ext (str): The file extension.
+    """
+    im: Image.Image | None = None
+
+    file: QFile = QFile(filepath)
+    success: bool = file.open(QIODeviceBase.OpenModeFlag.ReadOnly, QFileDevice.Permission.ReadUser)
+    if not success:
+        logger.error("Couldn't render thumbnail", filepath=filepath)
+        return im
+    document: QPdfDocument = QPdfDocument()
+    document.load(file)
+    file.close()
+    # Transform page_size in points to pixels with proper aspect ratio
+    page_size: QSizeF = document.pagePointSize(0)
+    ratio_hw: float = page_size.height() / page_size.width()
+    if ratio_hw >= 1:
+        page_size *= size / page_size.height()
+    else:
+        page_size *= size / page_size.width()
+    # Enlarge image for anti-aliasing
+    scale_factor = 2.5 if ext in {".pdf"} else 1
+    page_size *= scale_factor
+    # Render image with no anti-aliasing for speed
+    render_options: QPdfDocumentRenderOptions = QPdfDocumentRenderOptions()
+    render_options.setRenderFlags(
+        QPdfDocumentRenderOptions.RenderFlag.TextAliased
+        | QPdfDocumentRenderOptions.RenderFlag.ImageAliased
+        | QPdfDocumentRenderOptions.RenderFlag.PathAliased
+    )
+    # Convert QImage to PIL Image
+    q_image: QImage = document.render(0, page_size.toSize(), render_options)
+    buffer: QBuffer = QBuffer()
+    buffer.open(QBuffer.OpenModeFlag.ReadWrite)
+    try:
+        q_image.save(buffer, "PNG")  # pyright: ignore
+        im = Image.open(BytesIO(buffer.buffer().data()))
+    finally:
+        buffer.close()
+    # Replace transparent pixels with white (otherwise Background defaults to transparent)
+    return replace_transparent_pixels(im)

src/tagstudio/renderers/raster_image.py

@@ -0,0 +1,121 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import os
+from io import BytesIO
+from pathlib import Path
+
+import cv2
+import numpy as np
+import rawpy
+import structlog
+from PIL import Image, ImageOps, UnidentifiedImageError
+from PIL.Image import DecompressionBombError
+from pillow_heif import register_heif_opener  # pyright: ignore[reportUnknownVariableType]
+from rawpy import (
+    LibRawFileUnsupportedError,  # pyright: ignore[reportPrivateImportUsage]
+    LibRawIOError,  # pyright: ignore[reportPrivateImportUsage]
+)
+
+from tagstudio.core.utils.types import unwrap
+
+register_heif_opener()
+os.environ["OPENCV_IO_ENABLE_OPENEXR"] = "1"
+
+logger = structlog.get_logger(__name__)
+
+
+def image_thumb(filepath: Path) -> Image.Image | None:
+    """Render a thumbnail for a standard image type.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    im: Image.Image | None = None
+    try:
+        with filepath.open("rb") as file:
+            im = load_raster_image(BytesIO(file.read()))
+    except (
+        FileNotFoundError,
+        UnidentifiedImageError,
+        DecompressionBombError,
+        NotImplementedError,
+    ) as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im
+
+
+def image_exr_thumb(filepath: Path) -> Image.Image | None:
+    """Render a thumbnail for a EXR image type.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    im: Image.Image | None = None
+    try:
+        # Load the EXR data to an array and rotate the color space from BGRA -> RGBA
+        raw_array = cv2.imread(str(filepath), cv2.IMREAD_UNCHANGED)
+        assert raw_array is not None
+        raw_array[..., :3] = raw_array[..., 2::-1]
+
+        # Correct the gamma of the raw array
+        gamma = 2.2
+        array_gamma = np.power(np.clip(raw_array, 0, 1), 1 / gamma)
+        array = (array_gamma * 255).astype(np.uint8)
+
+        im = Image.fromarray(array, mode="RGBA")
+
+        # Paste solid background
+        if im.mode == "RGBA":
+            new_bg = Image.new("RGB", im.size, color="#1e1e1e")
+            new_bg.paste(im, mask=im.getchannel(3))
+            im = new_bg
+
+    except Exception as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im
+
+
+def image_raw_thumb(filepath: Path) -> Image.Image | None:
+    """Render a thumbnail for a RAW image type.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    im: Image.Image | None = None
+    try:
+        with rawpy.imread(str(filepath)) as raw:
+            rgb = raw.postprocess(use_camera_wb=True)
+            im = Image.frombytes(
+                "RGB",
+                (rgb.shape[1], rgb.shape[0]),
+                rgb,
+                decoder_name="raw",
+            )
+    except (
+        DecompressionBombError,
+        LibRawIOError,
+        LibRawFileUnsupportedError,
+    ) as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im
+
+
+def load_raster_image(image_data: BytesIO) -> Image.Image:
+    """Load a raster image and add a background if it's transparent.
+
+    Args:
+        image_data (BytesIO): The binary image data.
+
+    Returns:
+        Image.Image: The loaded raster image, with a background if needed.
+    """
+    im: Image.Image = Image.open(image_data)
+    if im.mode != "RGB" and im.mode != "RGBA":
+        im = im.convert(mode="RGBA")
+    if im.mode == "RGBA":
+        new_bg = Image.new("RGB", im.size, color="#1e1e1e")
+        new_bg.paste(im, mask=im.getchannel(3))
+        im = new_bg
+    return unwrap(ImageOps.exif_transpose(im))

src/tagstudio/renderers/source_engine.py

@@ -0,0 +1,30 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+from pathlib import Path
+
+import srctools
+import structlog
+from PIL import Image
+
+logger = structlog.get_logger(__name__)
+
+
+def vtf_thumb(filepath: Path) -> Image.Image | None:
+    """Extract and render a thumbnail for VTF (Valve Texture Format) images.
+
+    Uses the srctools library for reading VTF files.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    im: Image.Image | None = None
+    try:
+        with open(filepath, "rb") as f:
+            vtf = srctools.VTF.read(f)
+            im = vtf.get(frame=0).to_PIL()
+
+    except (ValueError, FileNotFoundError) as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im

src/tagstudio/renderers/text.py

@@ -0,0 +1,55 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+from pathlib import Path
+
+import cv2
+import structlog
+from PIL import Image, ImageDraw, UnidentifiedImageError
+from PIL.Image import DecompressionBombError
+from PySide6.QtCore import Qt
+from PySide6.QtGui import QGuiApplication
+
+from tagstudio.core.utils.encoding import detect_char_encoding
+
+logger = structlog.get_logger(__name__)
+
+
+def text_thumb(filepath: Path) -> Image.Image | None:
+    """Render a thumbnail for a plaintext file.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    im: Image.Image | None = None
+
+    bg_color: str = (
+        "#1e1e1e"
+        if QGuiApplication.styleHints().colorScheme() is Qt.ColorScheme.Dark
+        else "#FFFFFF"
+    )
+    fg_color: str = (
+        "#FFFFFF"
+        if QGuiApplication.styleHints().colorScheme() is Qt.ColorScheme.Dark
+        else "#111111"
+    )
+
+    try:
+        encoding = detect_char_encoding(filepath)
+        with open(filepath, encoding=encoding) as text_file:
+            text = text_file.read(256)
+        bg = Image.new("RGB", (256, 256), color=bg_color)
+        draw = ImageDraw.Draw(bg)
+        draw.text((16, 16), text, fill=fg_color)
+        im = bg
+    except (
+        UnidentifiedImageError,
+        cv2.error,
+        DecompressionBombError,
+        UnicodeDecodeError,
+        OSError,
+        FileNotFoundError,
+    ) as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im

src/tagstudio/renderers/vector_image.py

@@ -0,0 +1,54 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+from io import BytesIO
+from pathlib import Path
+
+import structlog
+from PIL import (
+    Image,
+    UnidentifiedImageError,
+)
+from PySide6.QtCore import QBuffer, Qt
+from PySide6.QtGui import QImage, QPainter
+from PySide6.QtSvg import QSvgRenderer
+
+logger = structlog.get_logger(__name__)
+
+
+def image_vector_thumb(filepath: Path, size: int) -> Image.Image:
+    """Render a thumbnail for a vector image, such as SVG.
+
+    Args:
+        filepath (Path): The path of the file.
+        size (tuple[int,int]): The size of the thumbnail.
+    """
+    im: Image.Image | None = None
+    # Create an image to draw the svg to and a painter to do the drawing
+    q_image: QImage = QImage(size, size, QImage.Format.Format_ARGB32)
+    q_image.fill("#1e1e1e")
+
+    # Create an svg renderer, then render to the painter
+    svg: QSvgRenderer = QSvgRenderer(str(filepath))
+
+    if not svg.isValid():
+        raise UnidentifiedImageError
+
+    painter: QPainter = QPainter(q_image)
+    svg.setAspectRatioMode(Qt.AspectRatioMode.KeepAspectRatio)
+    svg.render(painter)
+    painter.end()
+
+    # Write the image to a buffer as png
+    buffer: QBuffer = QBuffer()
+    buffer.open(QBuffer.OpenModeFlag.ReadWrite)
+    q_image.save(buffer, "PNG")  # pyright: ignore[reportCallIssue, reportArgumentType]
+
+    # Load the image from the buffer
+    im = Image.new("RGB", (size, size), color="#1e1e1e")
+    im.paste(Image.open(BytesIO(buffer.data().data())))
+    im = im.convert(mode="RGB")
+
+    buffer.close()
+    return im

src/tagstudio/renderers/vendored/pydub/audio_segment.py

@@ -43,7 +43,7 @@ from pydub.utils import (
 )
 
 from tagstudio.core.utils.silent_subprocess import silent_popen
-from tagstudio.qt.previews.vendored.pydub.utils import _mediainfo_json
+from tagstudio.renderers.vendored.pydub.utils import _mediainfo_json
 
 basestring = str
 xrange = range
@@ -256,7 +256,7 @@ class _AudioSegment:
             self.sample_width = 4
             self.frame_width = self.channels * self.sample_width
 
-        super(_AudioSegment, self).__init__(*args, **kwargs)
+        super().__init__(*args, **kwargs)
 
     @property
     def raw_data(self):

src/tagstudio/renderers/video.py

@@ -0,0 +1,55 @@
+# SPDX-FileCopyrightText: (c) TagStudio Contributors
+# SPDX-License-Identifier: GPL-3.0-only
+
+
+import math
+from pathlib import Path
+
+import cv2
+import structlog
+from cv2.typing import MatLike
+from PIL import Image, UnidentifiedImageError
+from PIL.Image import DecompressionBombError
+
+from tagstudio.qt.helpers.file_tester import is_readable_video
+
+logger = structlog.get_logger(__name__)
+
+
+def video_thumb(filepath: Path) -> Image.Image | None:
+    """Render a thumbnail for a video file.
+
+    Args:
+        filepath (Path): The path of the file.
+    """
+    im: Image.Image | None = None
+    frame: MatLike | None = None
+    try:
+        if is_readable_video(filepath):
+            video = cv2.VideoCapture(str(filepath), cv2.CAP_FFMPEG)
+            # TODO: Move this check to is_readable_video()
+            if video.get(cv2.CAP_PROP_FRAME_COUNT) <= 0:
+                raise cv2.error("File is invalid or has 0 frames")
+            video.set(
+                cv2.CAP_PROP_POS_FRAMES,
+                (video.get(cv2.CAP_PROP_FRAME_COUNT) // 2),
+            )
+            # NOTE: Depending on the video format, compression, and
+            # frame count, seeking halfway does not work and the thumb
+            # must be pulled from the earliest available frame.
+            max_frame_seek: int = 10
+            for i in range(
+                0,
+                min(max_frame_seek, math.floor(video.get(cv2.CAP_PROP_FRAME_COUNT))),
+            ):
+                success, frame = video.read()
+                if not success:
+                    video.set(cv2.CAP_PROP_POS_FRAMES, i)
+                else:
+                    break
+            if frame is not None:
+                frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+                im = Image.fromarray(frame)
+    except (UnidentifiedImageError, cv2.error, DecompressionBombError, OSError) as e:
+        logger.error("Couldn't render thumbnail", filepath=filepath, error=type(e).__name__)
+    return im