Merge pull request #181 from openzim/svg_support

Add minimal support for SVG conversion and probing

Author: benoit74

CHANGELOG.md

@@ -17,6 +17,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Expose new `optimization.get_optimization_method` to get the proper optimization method to call for a given image format
 - Add `optimization.get_optimization_method` to get the proper optimization method to call for a given image format
 - New `creator.Creator.convert_and_check_metadata` to convert metadata to bytes or str for known use cases and check proper type is passed to libzim
+- Add svg2png image conversion function #113
+- Add `conversion.convert_svg2png` image conversion function + support for SVG in `probing.format_for` #113
 
 ## Changed
 

pyproject.toml

@@ -23,6 +23,7 @@ dependencies = [
   # limited and we use only a very small subset of it.
   "regex>=2020.7.14",
   "pymupdf>=1.24.0,<2.0",
+  "CairoSVG>=2.2.0,<3.0",
   # youtube-dl should be updated as frequently as possible
   "yt-dlp"
 ]

src/zimscraperlib/image/conversion.py

@@ -3,9 +3,11 @@
 
 from __future__ import annotations
 
+import io
 import pathlib
 from typing import IO
 
+import cairosvg.svg
 from PIL.Image import open as pilopen
 
 from zimscraperlib.constants import ALPHA_NOT_SUPPORTED
@@ -40,6 +42,39 @@ def convert_image(
         save_image(image, dst, fmt, **params)
 
 
+def convert_svg2png(
+    src: str | pathlib.Path | io.BytesIO,
+    dst: pathlib.Path | IO[bytes],
+    width: int | None = None,
+    height: int | None = None,
+):
+    """Convert a SVG to a PNG
+
+    Output width and height might be specified if resize is needed.
+    PNG background is transparent.
+    """
+    kwargs = {}
+    if isinstance(src, pathlib.Path):
+        src = str(src)
+    if isinstance(src, str):
+        kwargs["url"] = src
+    else:
+        kwargs["bytestring"] = src.getvalue()
+    if width:
+        kwargs["output_width"] = width
+    if height:
+        kwargs["output_height"] = height
+    if isinstance(dst, pathlib.Path):
+        cairosvg.svg2png(write_to=str(dst), **kwargs)
+    else:
+        result = cairosvg.svg2png(**kwargs)
+        if not isinstance(result, bytes):
+            raise Exception(
+                "Unexpected type returned by cairosvg.svg2png"
+            )  # pragma: no cover
+        dst.write(result)
+
+
 def create_favicon(src: pathlib.Path, dst: pathlib.Path) -> None:
     """generate a squared favicon from a source image"""
     if dst.suffix != ".ico":

src/zimscraperlib/image/probing.py

@@ -12,6 +12,8 @@
 import colorthief
 import PIL.Image
 
+from zimscraperlib.filesystem import get_content_mimetype, get_file_mimetype
+
 
 def get_colors(
     src: pathlib.Path, *, use_palette: bool | None = True
@@ -59,8 +61,23 @@ def format_for(
 ) -> str | None:
     """Pillow format of a given filename, either Pillow-detected or from suffix"""
     if not from_suffix:
-        with PIL.Image.open(src) as img:
-            return img.format
+        try:
+            with PIL.Image.open(src) as img:
+                return img.format
+        except PIL.UnidentifiedImageError:
+            # Fallback based on mimetype for SVG which are not supported by PIL
+            if (
+                isinstance(src, pathlib.Path)
+                and get_file_mimetype(src) == "image/svg+xml"
+            ):
+                return "SVG"
+            elif (
+                isinstance(src, io.BytesIO)
+                and get_content_mimetype(src.getvalue()) == "image/svg+xml"
+            ):
+                return "SVG"
+            else:  # pragma: no cover
+                raise
 
     if not isinstance(src, pathlib.Path):
         raise ValueError(
@@ -70,8 +87,11 @@ def format_for(
     from PIL.Image import EXTENSION as PIL_FMT_EXTENSION
     from PIL.Image import init as init_pil
 
-    init_pil()
-    return PIL_FMT_EXTENSION[src.suffix] if src.suffix in PIL_FMT_EXTENSION else None
+    init_pil()  # populate the PIL_FMT_EXTENSION dictionary
+
+    known_extensions = {".svg": "SVG"}
+    known_extensions.update(PIL_FMT_EXTENSION)
+    return known_extensions[src.suffix] if src.suffix in known_extensions else None
 
 
 def is_valid_image(

tests/image/test_image.py

@@ -17,7 +17,11 @@
 from resizeimage.imageexceptions import ImageSizeError
 
 from zimscraperlib.image import presets
-from zimscraperlib.image.conversion import convert_image, create_favicon
+from zimscraperlib.image.conversion import (
+    convert_image,
+    convert_svg2png,
+    create_favicon,
+)
 from zimscraperlib.image.optimization import (
     ensure_matches,
     get_optimization_method,
@@ -64,8 +68,15 @@ def get_src_dst(
     jpg_image: pathlib.Path | None = None,
     gif_image: pathlib.Path | None = None,
     webp_image: pathlib.Path | None = None,
+    svg_image: pathlib.Path | None = None,
 ) -> tuple[pathlib.Path, pathlib.Path]:
-    options = {"png": png_image, "jpg": jpg_image, "webp": webp_image, "gif": gif_image}
+    options = {
+        "png": png_image,
+        "jpg": jpg_image,
+        "webp": webp_image,
+        "gif": gif_image,
+        "svg": svg_image,
+    }
     if fmt not in options:
         raise LookupError(f"Unsupported fmt passed: {fmt}")
     src = options[fmt]
@@ -328,6 +339,42 @@ def test_convert_path_src_io_dst(png_image: pathlib.Path):
     assert dst_image.format == "PNG"
 
 
+def test_convert_svg_io_src_path_dst(svg_image: pathlib.Path, tmp_path: pathlib.Path):
+    src = io.BytesIO(svg_image.read_bytes())
+    dst = tmp_path / "test.png"
+    convert_svg2png(src, dst)
+    dst_image = Image.open(dst)
+    assert dst_image.format == "PNG"
+
+
+def test_convert_svg_io_src_io_dst(svg_image: pathlib.Path):
+    src = io.BytesIO(svg_image.read_bytes())
+    dst = io.BytesIO()
+    convert_svg2png(src, dst)
+    dst_image = Image.open(dst)
+    assert dst_image.format == "PNG"
+
+
+def test_convert_svg_path_src_path_dst(svg_image: pathlib.Path, tmp_path: pathlib.Path):
+    src = svg_image
+    dst = tmp_path / "test.png"
+    convert_svg2png(src, dst, width=96, height=96)
+    dst_image = Image.open(dst)
+    assert dst_image.format == "PNG"
+    assert dst_image.width == 96
+    assert dst_image.height == 96
+
+
+def test_convert_svg_path_src_io_dst(svg_image: pathlib.Path):
+    src = svg_image
+    dst = io.BytesIO()
+    convert_svg2png(src, dst, width=96, height=96)
+    dst_image = Image.open(dst)
+    assert dst_image.format == "PNG"
+    assert dst_image.width == 96
+    assert dst_image.height == 96
+
+
 @pytest.mark.parametrize(
     "fmt,exp_size",
     [("png", 128), ("jpg", 128)],
@@ -576,10 +623,10 @@ def test_ensure_matches(webp_image):
 
 @pytest.mark.parametrize(
     "fmt,expected",
-    [("png", "PNG"), ("jpg", "JPEG"), ("gif", "GIF"), ("webp", "WEBP")],
+    [("png", "PNG"), ("jpg", "JPEG"), ("gif", "GIF"), ("webp", "WEBP"), ("svg", "SVG")],
 )
 def test_format_for_real_images_suffix(
-    png_image, jpg_image, gif_image, webp_image, tmp_path, fmt, expected
+    png_image, jpg_image, gif_image, webp_image, svg_image, tmp_path, fmt, expected
 ):
     src, _ = get_src_dst(
         tmp_path,
@@ -588,16 +635,17 @@ def test_format_for_real_images_suffix(
         jpg_image=jpg_image,
         gif_image=gif_image,
         webp_image=webp_image,
+        svg_image=svg_image,
     )
     assert format_for(src) == expected
 
 
 @pytest.mark.parametrize(
     "fmt,expected",
-    [("png", "PNG"), ("jpg", "JPEG"), ("gif", "GIF"), ("webp", "WEBP")],
+    [("png", "PNG"), ("jpg", "JPEG"), ("gif", "GIF"), ("webp", "WEBP"), ("svg", "SVG")],
 )
 def test_format_for_real_images_content_path(
-    png_image, jpg_image, gif_image, webp_image, tmp_path, fmt, expected
+    png_image, jpg_image, gif_image, webp_image, svg_image, tmp_path, fmt, expected
 ):
     src, _ = get_src_dst(
         tmp_path,
@@ -606,16 +654,17 @@ def test_format_for_real_images_content_path(
         jpg_image=jpg_image,
         gif_image=gif_image,
         webp_image=webp_image,
+        svg_image=svg_image,
     )
     assert format_for(src, from_suffix=False) == expected
 
 
 @pytest.mark.parametrize(
     "fmt,expected",
-    [("png", "PNG"), ("jpg", "JPEG"), ("gif", "GIF"), ("webp", "WEBP")],
+    [("png", "PNG"), ("jpg", "JPEG"), ("gif", "GIF"), ("webp", "WEBP"), ("svg", "SVG")],
 )
 def test_format_for_real_images_content_bytes(
-    png_image, jpg_image, gif_image, webp_image, tmp_path, fmt, expected
+    png_image, jpg_image, gif_image, webp_image, svg_image, tmp_path, fmt, expected
 ):
     src, _ = get_src_dst(
         tmp_path,
@@ -624,6 +673,7 @@ def test_format_for_real_images_content_bytes(
         jpg_image=jpg_image,
         gif_image=gif_image,
         webp_image=webp_image,
+        svg_image=svg_image,
     )
     assert format_for(io.BytesIO(src.read_bytes()), from_suffix=False) == expected
 
@@ -635,6 +685,7 @@ def test_format_for_real_images_content_bytes(
         ("image.jpg", "JPEG"),
         ("image.gif", "GIF"),
         ("image.webp", "WEBP"),
+        ("image.svg", "SVG"),
         ("image.raster", None),
     ],
 )