[font_parser] Fallback to freetype if fontTools raise an exception wh…

…en trying to read the cmap

Fix #44

font_collector/font/abc_font_face.py

@@ -266,7 +266,7 @@ def get_missing_glyphs(
         if error: raise FT_Exception(error)
 
         ttFont = TTFont(self.font_file.filename, fontNumber=self.font_index)
-        supported_cmaps = FontParser.get_supported_cmaps(ttFont)
+        supported_cmaps = FontParser.get_supported_cmaps(ttFont, self.font_file.filename, self.font_index)
         supported_charmaps = []
         for i in range(face.contents.num_charmaps):
             charmap = face.contents.charmaps[i]

font_collector/font/factory_abc_font_face.py

@@ -75,7 +75,7 @@ def __create_font(ttFont: TTFont, font_path: Path, font_index: int) -> NormalFon
         Returns:
             An FontFace instance that represent the ttFont.
         """
-        cmaps = FontParser.get_supported_cmaps(ttFont)
+        cmaps = FontParser.get_supported_cmaps(ttFont, font_path, font_index)
         if len(cmaps) == 0:
              raise InvalidNormalFontFaceException(f"The font doesn't contain any valid cmap.")
 

font_collector/font/font_parser.py

@@ -496,12 +496,16 @@ def get_symbol_cmap_encoding(face: FT_Face) -> Optional[str]:
 
 
     @staticmethod
-    def get_supported_cmaps(font: TTFont) -> List[CMap]:
+    def get_supported_cmaps(
+        font: TTFont, font_path: Path, font_index: int
+    ) -> List[CMap]:
         """
         Retrieve supported CMaps from a TrueType font.
 
         Args:
             font: A fontTools object representing the font.
+            font_path: Font path.
+            font_index: Font index.
         Returns:
             A list of supported CMaps.
             - To determine which CMaps are supported, refer to FontParser.get_cmap_encoding().
@@ -511,16 +515,29 @@ def get_supported_cmaps(font: TTFont) -> List[CMap]:
         microsoft_cmaps: List[CMap] = []
         macintosh_cmaps: List[CMap] = []
 
-        cmap_tables: List[CmapSubtable] = font["cmap"].tables
-
-        for table in cmap_tables:
-            encoding = FontParser.get_cmap_encoding(table.platformID, table.platEncID)
-            if encoding is not None:
-                cmap = CMap(table.platformID, table.platEncID)
-                if table.platformID == PlatformID.MICROSOFT:
-                    microsoft_cmaps.append(cmap)
-                elif table.platformID == PlatformID.MACINTOSH:
-                    macintosh_cmaps.append(cmap)
+        try:
+            cmap_tables: List[CmapSubtable] = font["cmap"].tables
+
+            for table in cmap_tables:
+                encoding = FontParser.get_cmap_encoding(table.platformID, table.platEncID)
+                if encoding is not None:
+                    cmap = CMap(table.platformID, table.platEncID)
+                    if table.platformID == PlatformID.MICROSOFT:
+                        microsoft_cmaps.append(cmap)
+                    elif table.platformID == PlatformID.MACINTOSH:
+                        macintosh_cmaps.append(cmap)
+        except Exception:
+            with font_path.open("rb") as f:
+                face = Face(f, font_index)
+
+            for charmap in face.charmaps:
+                encoding = FontParser.get_cmap_encoding(charmap.platform_id, charmap.encoding_id)
+                if encoding is not None:
+                    cmap = CMap(charmap.platform_id, charmap.encoding_id)
+                    if charmap.platform_id == PlatformID.MICROSOFT:
+                        microsoft_cmaps.append(cmap)
+                    elif charmap.platform_id == PlatformID.MACINTOSH:
+                        macintosh_cmaps.append(cmap)
         return macintosh_cmaps if len(microsoft_cmaps) == 0 else microsoft_cmaps
 
 

font_collector/font/variable_font_face.py

@@ -187,7 +187,7 @@ def variable_font_to_collection(self, save_path: Path, cache_generated_font: boo
         if len(fonts_face) == 0:
             raise ValueError(f"There is no valid font at the index {self.font_index}")
 
-        cmaps = FontParser.get_supported_cmaps(ttFont)
+        cmaps = FontParser.get_supported_cmaps(ttFont, self.font_file.filename, self.font_index)
 
         for font_face in fonts_face:
             generated_font_face = instancer.instantiateVariableFont(ttFont, font_face.named_instance_coordinates)

tests/font/test_font_loader.py

@@ -29,6 +29,7 @@ def test_load_additional_fonts():
         Path(os.path.join(font_directory, "font_mac.TTF")),
         Path(os.path.join(font_directory, "font_with_invalid_os2_table.ttf")),
         Path(os.path.join(font_directory, "font_without axis_value.ttf")),
+        Path(os.path.join(font_directory, "invalid_cmap.ttf")),
         Path(os.path.join(font_directory, "opentype_font_collection.ttc")),
         Path(os.path.join(font_directory, "PENBOX.otf")),
         Path(os.path.join(font_directory, "SFProDisplay-Bold.ttf")),

tests/font/test_font_parser.py

@@ -204,23 +204,29 @@ def test_get_symbol_cmap_encoding():
 
 def test_get_supported_cmaps():
     # This font contain 1 valid mac cmap
-    font_path = os.path.join(os.path.dirname(dir_path), "file", "fonts", "font_mac.TTF")
+    font_path = Path(os.path.join(os.path.dirname(dir_path), "file", "fonts", "font_mac.TTF"))
     font = TTFont(font_path)
-    cmaps = FontParser.get_supported_cmaps(font)
+    cmaps = FontParser.get_supported_cmaps(font, font_path, 0)
     assert cmaps == [CMap(1, 0)]
 
     # This font contain unicode cmap and multiple microsoft cmap
-    font_path = os.path.join(os.path.dirname(dir_path), "file", "fonts", "font_cmap_encoding_1.ttf")
+    font_path = Path(os.path.join(os.path.dirname(dir_path), "file", "fonts", "font_cmap_encoding_1.ttf"))
     font = TTFont(font_path)
-    cmaps = FontParser.get_supported_cmaps(font)
+    cmaps = FontParser.get_supported_cmaps(font, font_path, 0)
     assert cmaps == [CMap(3, 1), CMap(3, 10)]
 
     # This font contain 1 microsoft cmap, 1 valid mac cmap and 1 unicode cmap
-    font_path = os.path.join(os.path.dirname(dir_path), "file", "fonts", "font_cmap_encoding_0.ttf")
+    font_path = Path(os.path.join(os.path.dirname(dir_path), "file", "fonts", "font_cmap_encoding_0.ttf"))
     font = TTFont(font_path)
-    cmaps = FontParser.get_supported_cmaps(font)
+    cmaps = FontParser.get_supported_cmaps(font, font_path, 0)
     assert cmaps == [CMap(3, 0)]
 
+    # This font contain a "invalid" cmap. fontTools raise a exception, so we fallback to freetype which doesn't raise an exception
+    font_path = Path(os.path.join(os.path.dirname(dir_path), "file", "fonts", "invalid_cmap.ttf"))
+    font = TTFont(font_path)
+    cmaps = FontParser.get_supported_cmaps(font, font_path, 0)
+    assert cmaps == [CMap(3, 1)]
+
 
 def test_get_cmap_encoding():
     # It could be any format