fix: select correct unicode subset from font provider CSS

docs/changelog.md

@@ -1,3 +1,7 @@
+## Unreleased
+
+- **NEW** and **FIX**: Add `subset` parameter to `load_google_font()` and `load_bunny_font()` (defaults to `"latin"`). Multi-subset fonts (e.g., Chakra Petch) now return the correct character set instead of whichever subset appears first in the provider CSS.
+
 ## 1.3.0 (stable)
 
 - **NEW** and **FIX**: `set_default_font()` function now fetches and adds all font variants (weight, style) ([issue #23](https://github.com/y-sunflower/pyfonts/issues/23), [PR #39](https://github.com/y-sunflower/pyfonts/pull/39))

pyfonts/bunny.py

@@ -12,6 +12,7 @@ def load_bunny_font(
     weight: Optional[Union[int, str]] = None,
     italic: Optional[bool] = None,
     allowed_formats: List[str] = ["woff", "ttf", "otf"],
+    subset: str = "latin",
     use_cache: bool = True,
     danger_not_verify_ssl: bool = False,
 ) -> FontProperties:
@@ -30,6 +31,7 @@ def load_bunny_font(
         allowed_formats: List of acceptable font file formats. Defaults to ["woff", "ttf", "otf"].
             Note that for `woff2` fonts to work, you must have [brotli](https://github.com/google/brotli)
             installed.
+        subset: Unicode subset to select (e.g., "latin", "thai"). Defaults to "latin".
         use_cache: Whether or not to cache fonts (to make pyfonts faster). Default to `True`.
         danger_not_verify_ssl: Whether or not to to skip SSL certificate on
             `ssl.SSLCertVerificationError`. If `True`, it's a **security risk** (such as data breaches or
@@ -57,6 +59,7 @@ def load_bunny_font(
         italic=italic,
         allowed_formats=allowed_formats,
         use_cache=use_cache,
+        subset=subset,
     )
 
     font = load_font(
@@ -70,6 +73,7 @@ def load_bunny_font(
         endpoint=_BUNNY_ENDPOINT,
         family=family,
         allowed_formats=allowed_formats,
+        subset=subset,
         use_cache=use_cache,
         danger_not_verify_ssl=danger_not_verify_ssl,
     )

pyfonts/cache.py

@@ -12,13 +12,16 @@
 _MEMORY_CACHE: dict = {}
 
 
-def _cache_key(family: str, weight, italic, allowed_formats: list[str]) -> str:
+def _cache_key(
+    family: str, weight, italic, allowed_formats: list[str], subset: str
+) -> str:
     key_str: str = json.dumps(
         {
             "family": family,
             "weight": weight,
             "italic": italic,
             "allowed_formats": allowed_formats,
+            "subset": subset,
         },
         sort_keys=True,
     )

pyfonts/google.py

@@ -13,6 +13,7 @@ def load_google_font(
     weight: Optional[Union[int, str]] = None,
     italic: Optional[bool] = None,
     allowed_formats: List[str] = ["woff2", "woff", "ttf", "otf"],
+    subset: str = "latin",
     use_cache: bool = True,
     danger_not_verify_ssl: bool = False,
 ) -> FontProperties:
@@ -29,6 +30,7 @@ def load_google_font(
             'regular', 'medium', 'semi-bold', 'bold', 'extra-bold', 'black'. Default is `None`.
         italic: Whether to use the italic variant. Default is `None`.
         allowed_formats: List of acceptable font file formats. Defaults to ["woff2", "woff", "ttf", "otf"].
+        subset: Unicode subset to select (e.g., "latin", "thai"). Defaults to "latin".
         use_cache: Whether or not to cache fonts (to make pyfonts faster). Default to `True`.
         danger_not_verify_ssl: Whether or not to to skip SSL certificate on
             `ssl.SSLCertVerificationError`. If `True`, it's a **security risk** (such as data breaches or
@@ -56,6 +58,7 @@ def load_google_font(
         weight=weight,
         allowed_formats=allowed_formats,
         use_cache=use_cache,
+        subset=subset,
     )
 
     font = load_font(
@@ -69,6 +72,7 @@ def load_google_font(
         endpoint=_GOOGLE_ENDPOINT,
         family=family,
         allowed_formats=allowed_formats,
+        subset=subset,
         use_cache=use_cache,
         danger_not_verify_ssl=danger_not_verify_ssl,
     )

pyfonts/main.py

@@ -228,6 +228,7 @@ def _get_font_variant_files(font: FontProperties) -> set[str]:
                     italic=italic,
                     allowed_formats=provider_metadata["allowed_formats"],
                     use_cache=provider_metadata["use_cache"],
+                    subset=provider_metadata["subset"],
                 )
                 variant_font = load_font(
                     font_url=font_url,

pyfonts/utils.py

@@ -16,13 +16,26 @@
 _FONT_PROVIDER_METADATA_ATTR = "_pyfonts_provider_metadata"
 
 
+def _parse_css_subsets(css_text: str) -> dict[str, str]:
+    parts = re.split(r"/\*\s*(\S+)\s*\*/", css_text)
+    if len(parts) < 3:
+        return {"": css_text}
+    subsets: dict[str, str] = {}
+    for i in range(1, len(parts), 2):
+        name = parts[i]
+        block = parts[i + 1] if i + 1 < len(parts) else ""
+        subsets[name] = block
+    return subsets
+
+
 def _get_fonturl(
     endpoint: str,
     family: str,
     weight: Optional[Union[int, str]],
     italic: Optional[bool],
     allowed_formats: list,
     use_cache: bool,
+    subset: str = "latin",
 ) -> Optional[str]:
     """
     Construct the URL for a given endpoint, font family and style parameters,
@@ -35,14 +48,15 @@ def _get_fonturl(
         weight: Numeric font weight (e.g., 400, 700). If None, no weight axis is set.
         allowed_formats: List of acceptable font file extensions (e.g., ["woff2", "ttf"]).
         use_cache: Whether or not to cache fonts (to make pyfonts faster).
+        subset: Unicode subset to select from the CSS (e.g., "latin", "thai"). Defaults to "latin".
 
     Returns:
         Direct URL to the font file matching the requested style and format.
     """
     if isinstance(weight, str):
         weight: int = _map_weight_to_numeric(weight)
 
-    cache_key: str = _cache_key(family, weight, italic, allowed_formats)
+    cache_key: str = _cache_key(family, weight, italic, allowed_formats, subset)
     if use_cache:
         if not _MEMORY_CACHE and os.path.exists(_CACHE_FILE):
             _MEMORY_CACHE.update(_load_cache_from_disk())
@@ -76,10 +90,19 @@ def _get_fonturl(
             " does not exist."
         )
 
+    subsets = _parse_css_subsets(css_text)
+    search_text = subsets.get(subset, css_text)
+
     formats_pattern = "|".join(map(re.escape, allowed_formats))
     font_urls: list = re.findall(
-        rf"url\((https://[^)]+\.({formats_pattern}))\)", css_text
+        rf"url\((https://[^)]+\.({formats_pattern}))\)", search_text
     )
+
+    if not font_urls:
+        font_urls = re.findall(
+            rf"url\((https://[^)]+\.({formats_pattern}))\)", css_text
+        )
+
     if not font_urls:
         raise RuntimeError(
             f"No font files found in formats {allowed_formats} for '{family}'"
@@ -125,6 +148,7 @@ def _attach_font_provider_metadata(
     endpoint: str,
     family: str,
     allowed_formats: list[str],
+    subset: str,
     use_cache: bool,
     danger_not_verify_ssl: bool,
 ) -> FontProperties:
@@ -135,6 +159,7 @@ def _attach_font_provider_metadata(
             "endpoint": endpoint,
             "family": family,
             "allowed_formats": list(allowed_formats),
+            "subset": subset,
             "use_cache": use_cache,
             "danger_not_verify_ssl": danger_not_verify_ssl,
         },

tests/test_bunny.py

@@ -56,6 +56,21 @@ def test_load_bunny_font(family, weight, italic, use_cache):
     assert font.get_name() == family
 
 
+@pytest.mark.parametrize("subset", ["latin", "thai"])
+def test_get_fonturl_subset(subset):
+    url = _get_fonturl(
+        endpoint="https://fonts.bunny.net/css",
+        family="Chakra Petch",
+        weight=400,
+        italic=False,
+        allowed_formats=["woff", "ttf", "otf"],
+        use_cache=False,
+        subset=subset,
+    )
+    assert isinstance(url, str)
+    assert subset in url
+
+
 def test_weird_api_error():
     with pytest.raises(ValueError, match="No font available for the request at URL*"):
         load_bunny_font("Alice", italic=True)

tests/test_cache.py

@@ -1,7 +1,7 @@
 import pytest
 import json
 from pyfonts import clear_pyfonts_cache, load_font, load_google_font
-from pyfonts.cache import _load_cache_from_disk
+from pyfonts.cache import _cache_key, _load_cache_from_disk
 import sys
 
 pytestmark = pytest.mark.skipif(
@@ -10,6 +10,12 @@
 )
 
 
+def test_cache_key_differs_by_subset():
+    key_latin = _cache_key("Roboto", 400, False, ["woff"], "latin")
+    key_thai = _cache_key("Roboto", 400, False, ["woff"], "thai")
+    assert key_latin != key_thai
+
+
 def test_load_cache_when_file_missing(tmp_path, monkeypatch):
     monkeypatch.setattr("pyfonts.cache._CACHE_FILE", tmp_path / "missing.json")
     assert _load_cache_from_disk() == {}

tests/test_utils.py

@@ -1,7 +1,31 @@
-from pyfonts.utils import _map_weight_to_numeric
+from pyfonts.utils import _map_weight_to_numeric, _parse_css_subsets
 import pytest
 
 
+def test_parse_css_subsets_multi():
+    css = (
+        "/* thai */\n"
+        "@font-face { src: url(https://example.com/font-thai.woff); }\n"
+        "/* latin */\n"
+        "@font-face { src: url(https://example.com/font-latin.woff); }\n"
+        "/* latin-ext */\n"
+        "@font-face { src: url(https://example.com/font-latin-ext.woff); }\n"
+    )
+    result = _parse_css_subsets(css)
+    assert "thai" in result
+    assert "latin" in result
+    assert "latin-ext" in result
+    assert "font-thai" in result["thai"]
+    assert "font-latin.woff" in result["latin"]
+    assert "font-latin-ext" in result["latin-ext"]
+
+
+def test_parse_css_subsets_no_comments():
+    css = "@font-face { src: url(https://example.com/font.woff); }"
+    result = _parse_css_subsets(css)
+    assert result == {"": css}
+
+
 def test_map_weight_to_numeric():
     assert _map_weight_to_numeric("thin") == 100
     assert _map_weight_to_numeric("extra-light") == 200