fix: support vw/vh and percents in SVG dimensions

.gitignore

@@ -161,7 +161,7 @@ cython_debug/
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/
+.idea/
 
 ### Python Patch ###
 # Poetry local configuration file - https://python-poetry.org/docs/configuration/#local-configuration

app/svg_utils.py

@@ -8,6 +8,10 @@
 from pathlib import Path
 from uuid import uuid4
 
+from defusedxml import ElementTree as ET
+
+SPECIAL_UNITS = ("vw", "vh", "%")  # Special units that require viewBox context
+
 IMAGE_PNG = "image/png"
 IMAGE_SVG = "image/svg+xml"
 
@@ -64,11 +68,11 @@ def replace_img_base64(match: re.Match[str]) -> str:
 
 # Checks that base64 encoded content is a svg image and replaces it with the png screenshot made by chromium
 def replace_svg_with_png(svg_content: str) -> tuple[str, str | bytes]:
-    width, height = extract_svg_dimensions_as_px(svg_content)
+    width, height, updated_svg_content = extract_svg_dimensions_as_px(svg_content)
     if not width or not height:
         return IMAGE_SVG, svg_content
 
-    svg_filepath, png_filepath = prepare_temp_files(svg_content)
+    svg_filepath, png_filepath = prepare_temp_files(updated_svg_content)
     if not svg_filepath or not png_filepath:
         return IMAGE_SVG, svg_content
 
@@ -82,21 +86,98 @@ def replace_svg_with_png(svg_content: str) -> tuple[str, str | bytes]:
     return IMAGE_PNG, png_content
 
 
-# Extract the width and height from the SVG tag (and convert it to px)
-def extract_svg_dimensions_as_px(svg_content: str) -> tuple[int | None, int | None]:
-    width_match = re.search(r'<svg[^>]+?width="(?P<width>[\d.]+)(?P<unit>\w+)?', svg_content)
-    height_match = re.search(r'<svg[^>]+?height="(?P<height>[\d.]+)(?P<unit>\w+)?', svg_content)
+def extract_svg_dimensions_as_px(svg_content: str) -> tuple[int | None, int | None, str]:
+    """
+    Extract width and height from the SVG tag and convert them to px.
+    If units are vw/vh/% and viewBox exists, compute their pixel equivalents.
+    Returns updated SVG content with replaced width/height if necessary.
+    """
+    width, width_unit = parse_svg_dimension(svg_content, "width")
+    height, height_unit = parse_svg_dimension(svg_content, "height")
+    vb_width, vb_height = parse_viewbox(svg_content)
+
+    width_px = calculate_dimension(width, width_unit, vb_width)
+    height_px = calculate_dimension(height, height_unit, vb_height)
+
+    if vb_width is not None and vb_height is not None:
+        if width_px is None:
+            width_px = math.ceil(vb_width)
+        if height_px is None:
+            height_px = math.ceil(vb_height)
+        svg_content = replace_svg_size_attributes(svg_content, width_px, height_px)
+
+    if width_px is None or height_px is None:
+        return None, None, svg_content
+
+    return width_px, height_px, svg_content
+
+
+def parse_svg_dimension(svg_content: str, dimension: str) -> tuple[str | None, str | None]:
+    match = re.search(
+        rf'<svg[^>]*?\b{dimension}\s*=\s*["\'](?P<value>[\d.]+)(?P<unit>\w+|%)?["\']',
+        svg_content,
+        flags=re.IGNORECASE,
+    )
+    if match:
+        return match.group("value"), match.group("unit")
+    return None, None
+
+
+def parse_viewbox(svg_content: str) -> tuple[float | None, float | None]:
+    match = re.search(
+        r'<svg[^>]*?\bviewBox\s*=\s*["\']'
+        r"[\d.\-]+\s+[\d.\-]+\s+"
+        r"(?P<vb_width>[\d.\-]+)\s+"
+        r"(?P<vb_height>[\d.\-]+)"
+        r'["\']',
+        svg_content,
+        flags=re.IGNORECASE,
+    )
+    if match:
+        return float(match.group("vb_width")), float(match.group("vb_height"))
+    return None, None
+
+
+def calculate_dimension(value: str | None, unit: str | None, vb_dimension: float | None) -> int | None:
+    if value is None:
+        return None
 
-    width = width_match.group("width") if width_match else None
-    height = height_match.group("height") if height_match else None
+    if unit in SPECIAL_UNITS:
+        if vb_dimension is None:
+            raise ValueError(f"{unit} units require a viewBox to be defined")
+        return calculate_special_unit(value, unit, vb_dimension)
 
-    if not width or not height:
-        logging.error(f"Cannot find SVG dimensions. Width: {width}, Height: {height}")
+    return convert_to_px(value, unit)
+
+
+def replace_svg_size_attributes(svg_content: str, width_px: int, height_px: int) -> str:
+    try:
+        root = ET.fromstring(svg_content)
+    except ET.ParseError as e:
+        raise ValueError("Invalid SVG content") from e
+
+    # Set or replace width and height attributes
+    root.set("width", f"{width_px}px")
+    root.set("height", f"{height_px}px")
 
-    width_unit = width_match.group("unit") if width_match else None
-    height_unit = height_match.group("unit") if height_match else None
+    # Convert XML tree back to a string
+    svg_with_attributes = ET.tostring(root, encoding="unicode")
 
-    return convert_to_px(width, width_unit), convert_to_px(height, height_unit)
+    return svg_with_attributes
+
+
+# Calculates the pixel value for vw, vh, or % units based on viewBox dimensions
+def calculate_special_unit(value: str, unit: str | None, viewbox_dimension: float) -> int:
+    val = float(value)
+
+    if unit in SPECIAL_UNITS:
+        return math.ceil((val / 100) * viewbox_dimension)
+
+    fallback = convert_to_px(value, unit)
+    if fallback is None:
+        raise ValueError(f"Cannot convert unit '{unit}' to px")
+
+    return fallback
 
 
 # Save the SVG content to a temporary file and return the file paths for the SVG and PNG.
@@ -185,6 +266,10 @@ def convert_to_px(value: str | None, unit: str | None) -> int | None:
         if value is None:
             raise ValueError()
         value_f64 = float(value)
+
+        if unit in SPECIAL_UNITS:
+            return None
+
         return math.ceil(value_f64 * get_px_conversion_ratio(unit))
     except ValueError:
         logging.error(f"Invalid value for conversion: {value}")

app/weasyprint_service_application.py

@@ -3,7 +3,8 @@
 
 from app import weasyprint_controller  # type: ignore
 
-if __name__ == "__main__":
+
+def main() -> None:
     parser = argparse.ArgumentParser(description="Weasyprint service")
     parser.add_argument("--port", default=9080, type=int, required=False, help="Service port")
     args = parser.parse_args()
@@ -13,3 +14,7 @@
     logging.getLogger().setLevel(logging.WARN)
 
     weasyprint_controller.start_server(args.port)
+
+
+if __name__ == "__main__":
+    main()

pyproject.toml

@@ -11,7 +11,8 @@ requires-python=">=3.13"
 dependencies = [
     "flask (>=3.1.0,<4.0.0)",
     "gevent (>=24.11.1,<25.0.0)",
-    "weasyprint (>=64.0,<65)"
+    "weasyprint (>=64.0,<65)",
+    "defusedxml (>=0.7.1,<0.8.0)"
 ]
 
 [tool.poetry]
@@ -25,6 +26,7 @@ python = "^3.13"
 pre-commit = "4.1.0"
 ruff = "0.11.0"
 mypy = "1.15.0"
+types-defusedxml = "^0.7.0.20240218"
 
 [tool.poetry.group.test.dependencies]
 tox = "^4.18.0"
@@ -73,3 +75,5 @@ line-ending = "lf"
 explicit_package_bases = true
 disallow_untyped_defs = true
 exclude = "tests/.*"
+install_types = true
+non_interactive = true

tests/test_svg_utils.py

@@ -110,15 +110,104 @@ def test_replace_svg_with_png():
 
 @setup_env_variables
 def test_extract_svg_dimensions_as_px():
-    # Return width and height with valid input
-    svg_content = r'<svg height="200px" width="100px">'
-    width, height = extract_svg_dimensions_as_px(svg_content)
+    # Valid width and height with absolute units
+    svg_content = r'<svg height="200px" width="100px"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
     assert width == 100 and height == 200
+    assert 'width="100px"' in updated_svg
+    assert 'height="200px"' in updated_svg
 
-    # Return None values for incorrectly formatted svg tag attributes
-    svg_content = r'<svg height=200px" width "100px'
-    width, height = extract_svg_dimensions_as_px(svg_content)
+    # Invalid width and height attribute formatting → None
+    svg_content = r'<svg height=200px" width "100px></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width is None and height is None
+    assert updated_svg == svg_content
+
+    # Valid viewBox fallback when width and height are missing
+    svg_content = r'<svg viewBox="0 0 300 150"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width == 300 and height == 150
+    assert 'width="300px"' in updated_svg
+    assert 'height="150px"' in updated_svg
+
+    # viewBox with incorrect format → None
+    svg_content = r'<svg viewBox="0 0 300"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width is None and height is None
+    assert updated_svg == svg_content
+
+    # Missing everything → None, None
+    svg_content = r"<svg></svg>"
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width is None and height is None
+    assert updated_svg == svg_content
+
+    # Partial attributes: width only, no height, but with valid viewBox
+    svg_content = r'<svg width="100px" viewBox="0 0 400 200"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width == 100 and height == 200
+    assert 'width="100px"' in updated_svg
+    assert 'height="200px"' in updated_svg
+
+    # Partial attributes: height only, no width, but with valid viewBox
+    svg_content = r'<svg height="50px" viewBox="0 0 400 200"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width == 400 and height == 50
+    assert 'width="400px"' in updated_svg
+    assert 'height="50px"' in updated_svg
+
+    # Non-numeric width and height → None
+    svg_content = r'<svg width="abc" height="xyz"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
     assert width is None and height is None
+    assert updated_svg == svg_content
+
+
+@setup_env_variables
+def test_extract_svg_dimensions_as_px_relative():
+    # width/height in vw/vh without viewBox → raise ValueError
+    svg_content = r'<svg width="100vw" height="100vh"></svg>'
+    try:
+        extract_svg_dimensions_as_px(svg_content)
+        raise AssertionError("Expected ValueError due to missing viewBox for vw/vh units")
+    except ValueError as e:
+        assert "vw units require a viewBox to be defined" in str(e)
+
+    # width/height in % without viewBox → raise ValueError
+    svg_content = r'<svg width="100%" height="100%"></svg>'
+    try:
+        extract_svg_dimensions_as_px(svg_content)
+        raise AssertionError("Expected ValueError due to missing viewBox for vw/vh units")
+    except ValueError as e:
+        assert "% units require a viewBox to be defined" in str(e)
+
+    # width/height in vw/vh with valid viewBox
+    svg_content = r'<svg width="100vw" height="100vh" viewBox="0 0 800 600"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width == 800 and height == 600
+    assert 'width="800px"' in updated_svg
+    assert 'height="600px"' in updated_svg
+
+    # width/height in vw/vh with valid viewBox
+    svg_content = r'<svg width="50vw" height="50vh" viewBox="0 0 800 600"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width == 400 and height == 300
+    assert 'width="400px"' in updated_svg
+    assert 'height="300px"' in updated_svg
+
+    # 100% / 100% with viewBox = 800x600
+    svg_content = r'<svg width="100%" height="100%" viewBox="0 0 800 600"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width == 800 and height == 600
+    assert 'width="800px"' in updated_svg
+    assert 'height="600px"' in updated_svg
+
+    # 50% / 25% with viewBox = 800x600
+    svg_content = r'<svg width="50%" height="25%" viewBox="0 0 800 600"></svg>'
+    width, height, updated_svg = extract_svg_dimensions_as_px(svg_content)
+    assert width == 400 and height == 150
+    assert 'width="400px"' in updated_svg
+    assert 'height="150px"' in updated_svg
 
 
 @setup_env_variables
@@ -189,6 +278,9 @@ def test_convert_to_px():
     assert convert_to_px("1", "mm") == 378
     assert convert_to_px(None, "px") is None
     assert convert_to_px("abc", "px") is None
+    assert convert_to_px("100", "vh") is None
+    assert convert_to_px("100", "vw") is None
+    assert convert_to_px("100", "%") is None
 
 
 @setup_env_variables

tests/test_weasyprint_service_application.py

@@ -0,0 +1,17 @@
+import logging
+import sys
+
+from app import weasyprint_service_application
+
+
+def test_main_runs(monkeypatch):
+    monkeypatch.setattr(sys, "argv", ["weasyprint_service_application.py", "--port", "9999"])
+
+    logger = logging.getLogger("test")
+
+    def fake_start_server(port):
+        logger.info(f"Fake server started on port {port}")
+
+    monkeypatch.setattr(weasyprint_service_application.weasyprint_controller, "start_server", fake_start_server)
+
+    weasyprint_service_application.main()