wip

fonts.py

@@ -0,0 +1,121 @@
+import unicodedata
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Generator, List, Optional, Set, Tuple
+
+from fontTools.ttLib import TTFont, TTLibFileIsCollectionError  # type: ignore
+from fontTools.ttLib.ttCollection import TTCollection  # type: ignore
+
+# https://github.com/googlefonts/noto-cjk/raw/main/Sans/Variable/OTC/NotoSansCJK-VF.ttf.ttc
+
+
+@dataclass
+class Font:
+    family: str
+    name: str
+    codepoints: Set[int]
+    is_bitmap: bool
+    size: int
+
+
+def make_fonts(path: Path, is_bitmap: bool, size: int):
+    for tt_font in get_tt_fonts(path):
+        yield Font(
+            family=font_family(tt_font),
+            name=font_name(tt_font),
+            codepoints=set(),
+            is_bitmap=is_bitmap,
+            size=size,
+        )
+
+
+def get_tt_fonts(path: Path) -> Generator[TTFont, None, None]:
+    try:
+        yield TTFont(path)
+    except TTLibFileIsCollectionError:
+        collection = TTCollection(path)
+        for font in collection.fonts:
+            yield font
+
+
+def load_fonts(paths: List[str]) -> List[TTFont]:
+    return list(fonts_generator(paths))
+
+
+def fonts_generator(paths: List[str]) -> Generator[TTFont, None, None]:
+    for path in paths:
+        try:
+            yield TTFont(path)
+        except TTLibFileIsCollectionError:
+            collection = TTCollection(path)
+            for font in collection.fonts:
+                yield font
+
+
+def font_family(font: TTFont) -> str:
+    return font["name"].getDebugName(1)
+
+
+def font_name(font: TTFont) -> str:
+    return font["name"].getDebugName(4)
+
+
+def can_render(font: TTFont, char: str) -> bool:
+    for table in font["cmap"].tables:
+        if table.isUnicode() and ord(char) in table.cmap:
+            return True
+    return False
+
+
+def get_font_for_char(fonts: List[TTFont], char: str) -> Optional[TTFont]:
+    for font in fonts:
+        if can_render(font, char):
+            return font
+
+
+def group_by_font(text: str, fonts: List[TTFont]) -> Generator[Tuple[str, TTFont], None, None]:
+    if not text:
+        return
+
+    buffer = ""
+    font = None
+
+    for char in text:
+        char_font = get_font_for_char(fonts, char)
+        if not char_font:
+            print(f"Cannot render char: {char} {char_name(char)} {ord(char)}")
+            continue
+
+        if not font:
+            font = char_font
+
+        if font == char_font:
+            buffer += char
+        else:
+            yield buffer, font
+            font = char_font
+            buffer = char
+
+    if buffer and font:
+        yield buffer, font
+
+
+def char_name(char: str):
+    try:
+        return unicodedata.name(char)
+    except ValueError:
+        return "NO NAME"
+
+
+text = "Hello 😊! Let's grab a 能ばぜ表聴ゆぞ coffee ☕️ and go for a ride 🚗."
+
+if __name__ == "__main__":
+    fonts = make_fonts(Path("NotoSans-Regular.ttf"), False, 20)
+    fonts = make_fonts(Path("NotoSansMath-Regular.ttf"), False, 20)
+    fonts = make_fonts(Path("NotoSansCJK-Light.ttf"), False, 20)
+    fonts = make_fonts(Path("NotoColorEmoji.ttf"), True, 109)
+
+    # with open("chat.200.txt") as f:
+    #     while line := f.readline().strip():
+    #         for fragment, font in group_by_font(line, fonts):
+    #             print(font_name(font).ljust(25), fragment)