Rework mapping of dvi glyph indices to freetype indices.

anntzer · anntzer · commit 7a760d185f3a · 2025-04-01T07:37:54.000+02:00
In 89a7e19, an API for converting "dvi glyph indices" (as stored in a dvi file) to FreeType-compatible keys (either "indices into the native charmap" or "glyph names") was introduced. It was intended that end users (i.e., backends) would check the type of `text.glyph_name_or_index` ((A) int or (B) str) and load the glyph accordingly ((A) `FT_Set_Charmap(native_cmap); FT_Load_Char(index);` or (B) `FT_Load_Glyph(FT_Get_Name_Index(name));`); however, with the future introduction of {xe,lua}tex support, this kind of type checking becomes inconvenient, because {xe,lua}tex's "dvi glyph indices", which are directly equal to FreeType glyph indices (i.e. they would be loaded with `FT_Load_Glyph(index);`), would normally also be converted to ints. This PR introduces a new API (`_index_dvi_to_freetype`) to perform this mapping, always mapping to FreeType glyph indices (i.e. one can always just call `FT_Load_Glyph` on the result). To do so, in case (A) it loads itself the native charmap (something the end user needed to do by themselves previously) and performs the cmap-to-index conversion (`FT_Get_Char_Index`) previously implicit in `FT_Load_Char`; in case (B) it performs itself the name-to-index conversion (`FT_Get_Name_Index`). When {xe,lua}tex support is introduced in the future, `_index_dvi_to_freetype` will just return the index as is. Note that this API is intentionally kept private for now (even though it is used by textpath) and the old APIs are not deprecated yet; I intend to wait until {xe,lua}tex support is actually merged to do so, to avoid possible future back-and-forth changes on the public APIs. In case (A), this PR also improves on the detection of the native charmap, which was previously detected via heuristics (`_select_native_charmap`), but is now read by directly accessing the Type 1 font "encoding vector".
diff --git a/lib/matplotlib/dviread.py b/lib/matplotlib/dviread.py
@@ -31,7 +31,7 @@
 
 import numpy as np
 
-from matplotlib import _api, cbook
+from matplotlib import _api, cbook, font_manager
 
 _log = logging.getLogger(__name__)
 
@@ -579,7 +579,7 @@ class DviFont:
        Size of the font in Adobe points, converted from the slightly
        smaller TeX points.
     """
-    __slots__ = ('texname', 'size', '_scale', '_vf', '_tfm')
+    __slots__ = ('texname', 'size', '_scale', '_vf', '_tfm', '_encoding')
 
     def __init__(self, scale, tfm, texname, vf):
         _api.check_isinstance(bytes, texname=texname)
@@ -588,6 +588,7 @@ def __init__(self, scale, tfm, texname, vf):
         self.texname = texname
         self._vf = vf
         self.size = scale * (72.0 / (72.27 * 2**16))
+        self._encoding = None
 
     widths = _api.deprecated("3.11")(property(lambda self: [
         (1000 * self._tfm.width.get(char, 0)) >> 20
@@ -630,6 +631,35 @@ def _height_depth_of(self, char):
             hd[-1] = 0
         return hd
 
+    # TODO: Make this public when {xe,lua}tex support is merged; simultaneously
+    # deprecate Text.glyph_name_or_index.
+    def _index_dvi_to_freetype(self, idx):
+        """Convert dvi glyph indices to FreeType ones."""
+        # Glyphs indices stored in the dvi file map to FreeType glyph indices
+        # (i.e., which can be passed to FT_Load_Glyph) in various ways:
+        # - if pdftex.map specifies an ".enc" file for the font, that file maps
+        #   dvi indices to Adobe glyph names, which can then be converted to
+        #   FreeType glyph indices with FT_Get_Name_Index.
+        # - if no ".enc" file is specified, then the font must be a Type 1
+        #   font, and dvi indices directly index into the font's CharStrings
+        #   vector.
+        # - (xetex & luatex, currently unsupported, can also declare "native
+        #   fonts", for which dvi indices are equal to FreeType indices.)
+        if self._encoding is None:
+            psfont = PsfontsMap(find_tex_file("pdftex.map"))[self.texname]
+            if psfont.filename is None:
+                raise ValueError("No usable font file found for {} ({}); "
+                                 "the font may lack a Type-1 version"
+                                 .format(psfont.psname.decode("ascii"),
+                                         psfont.texname.decode("ascii")))
+            face = font_manager.get_font(psfont.filename)
+            if psfont.encoding:
+                self._encoding = [face.get_name_index(name)
+                                  for name in _parse_enc(psfont.encoding)]
+            else:
+                self._encoding = face._get_type1_encoding_vector()
+        return self._encoding[idx]
+
 
 class Vf(Dvi):
     r"""
@@ -1023,8 +1053,7 @@ def _parse_enc(path):
     Returns
     -------
     list
-        The nth entry of the list is the PostScript glyph name of the nth
-        glyph.
+        The nth list item is the PostScript glyph name of the nth glyph.
     """
     no_comments = re.sub("%.*", "", Path(path).read_text(encoding="ascii"))
     array = re.search(r"(?s)\[(.*)\]", no_comments).group(1)
@@ -1155,15 +1184,10 @@ def _print_fields(*args):
                 print(f"font: {font.texname.decode('latin-1')} "
                       f"(scale: {font._scale / 2 ** 20}) at {fontpath}")
                 face = FT2Font(fontpath)
-                TextToPath._select_native_charmap(face)
                 _print_fields("x", "y", "glyph", "chr", "w")
                 for text in group:
-                    if psfont.encoding:
-                        glyph_name = _parse_enc(psfont.encoding)[text.glyph]
-                    else:
-                        glyph_name = face.get_glyph_name(
-                            face.get_char_index(text.glyph))
-                    glyph_str = fontTools.agl.toUnicode(glyph_name)
+                    glyph_str = fontTools.agl.toUnicode(
+                        face.get_glyph_name(font._index_dvi_to_freetype(text.glyph)))
                     _print_fields(text.x, text.y, text.glyph, glyph_str, text.width)
             if page.boxes:
                 print("--- BOXES ---")
diff --git a/lib/matplotlib/textpath.py b/lib/matplotlib/textpath.py
@@ -238,17 +238,8 @@ def get_glyphs_tex(self, prop, s, glyph_map=None,
             if char_id not in glyph_map:
                 font.clear()
                 font.set_size(self.FONT_SCALE, self.DPI)
-                glyph_name_or_index = text.glyph_name_or_index
-                if isinstance(glyph_name_or_index, str):
-                    index = font.get_name_index(glyph_name_or_index)
-                    font.load_glyph(index, flags=LoadFlags.TARGET_LIGHT)
-                elif isinstance(glyph_name_or_index, int):
-                    self._select_native_charmap(font)
-                    font.load_char(
-                        glyph_name_or_index, flags=LoadFlags.TARGET_LIGHT)
-                else:  # Should not occur.
-                    raise TypeError(f"Glyph spec of unexpected type: "
-                                    f"{glyph_name_or_index!r}")
+                idx = text.font._index_dvi_to_freetype(text.glyph)
+                font.load_glyph(idx, flags=LoadFlags.TARGET_LIGHT)
                 glyph_map_new[char_id] = font.get_path()
 
             glyph_ids.append(char_id)
@@ -269,23 +260,6 @@ def get_glyphs_tex(self, prop, s, glyph_map=None,
         return (list(zip(glyph_ids, xpositions, ypositions, sizes)),
                 glyph_map_new, myrects)
 
-    @staticmethod
-    def _select_native_charmap(font):
-        # Select the native charmap. (we can't directly identify it but it's
-        # typically an Adobe charmap).
-        for charmap_code in [
-                1094992451,  # ADOBE_CUSTOM.
-                1094995778,  # ADOBE_STANDARD.
-        ]:
-            try:
-                font.select_charmap(charmap_code)
-            except (ValueError, RuntimeError):
-                pass
-            else:
-                break
-        else:
-            _log.warning("No supported encoding in font (%s).", font.fname)
-
 
 text_to_path = TextToPath()
 
diff --git a/src/ft2font_wrapper.cpp b/src/ft2font_wrapper.cpp
@@ -1431,6 +1431,32 @@ PyFT2Font_get_image(PyFT2Font *self)
     return py::array_t<unsigned char>(dims, im.get_buffer());
 }
 
+const char *PyFT2Font__get_type1_encoding_vector__doc__ = R"""(
+    Return a list mapping CharString indices of a Type 1 font to FreeType glyph indices.
+
+    Returns
+    -------
+    list[int]
+)""";
+
+static std::array<FT_UInt, 256>
+PyFT2Font__get_type1_encoding_vector(PyFT2Font *self)
+{
+    auto face = self->x->get_face();
+    auto indices = std::array<FT_UInt, 256>{};
+    for (auto i = 0; i < indices.size(); ++i) {
+        auto len = FT_Get_PS_Font_Value(face, PS_DICT_ENCODING_ENTRY, i, nullptr, 0);
+        if (len == -1) {
+            throw std::runtime_error{
+                "FT_Get_PS_Font_Value tried to access a non-existent value"};
+        }
+        auto buf = std::unique_ptr<char[]>{new char[len]};
+        FT_Get_PS_Font_Value(face, PS_DICT_ENCODING_ENTRY, i, buf.get(), len);
+        indices[i] = FT_Get_Name_Index(face, buf.get());
+    }
+    return indices;
+}
+
 static const char *
 PyFT2Font_postscript_name(PyFT2Font *self)
 {
@@ -1761,6 +1787,8 @@ PYBIND11_MODULE(ft2font, m, py::mod_gil_not_used())
              PyFT2Font_get_sfnt_table__doc__)
         .def("get_path", &PyFT2Font_get_path, PyFT2Font_get_path__doc__)
         .def("get_image", &PyFT2Font_get_image, PyFT2Font_get_image__doc__)
+        .def("_get_type1_encoding_vector", &PyFT2Font__get_type1_encoding_vector,
+             PyFT2Font__get_type1_encoding_vector__doc__)
 
         .def_property_readonly("postscript_name", &PyFT2Font_postscript_name,
                                "PostScript name of the font.")
