Merge pull request matplotlib#29794 from QuLogic/text-language

QuLogic · web-flow · commit 44c17bc1d90f · 2025-09-20T00:13:51.000-04:00
Add language parameter to Text objects
diff --git a/doc/release/next_whats_new/text_language.rst b/doc/release/next_whats_new/text_language.rst
@@ -0,0 +1,37 @@
+Specifying text language
+------------------------
+
+OpenType fonts may support language systems which can be used to select different
+typographic conventions, e.g., localized variants of letters that share a single Unicode
+code point, or different default font features. The text API now supports setting a
+language to be used and may be set/get with:
+
+- `matplotlib.text.Text.set_language` / `matplotlib.text.Text.get_language`
+- Any API that creates a `.Text` object by passing the *language* argument (e.g.,
+  ``plt.xlabel(..., language=...)``)
+
+The language of the text must be in a format accepted by libraqm, namely `a BCP47
+language code <https://www.w3.org/International/articles/language-tags/>`_. If None or
+unset, then no particular language will be implied, and default font settings will be
+used.
+
+For example, Matplotlib's default font ``DejaVu Sans`` supports language-specific glyphs
+in the Serbian and Macedonian languages in the Cyrillic alphabet (vs Russian),
+or the Sámi family of languages in the Latin alphabet (vs English).
+
+.. plot::
+    :include-source:
+
+    fig = plt.figure(figsize=(7, 3))
+
+    char = '\U00000431'
+    fig.text(0.5, 0.8, f'\\U{ord(char):08x}', fontsize=40, horizontalalignment='center')
+    fig.text(0, 0.6, f'Serbian: {char}', fontsize=40, language='sr')
+    fig.text(1, 0.6, f'Russian: {char}', fontsize=40, language='ru',
+             horizontalalignment='right')
+
+    char = '\U0000014a'
+    fig.text(0.5, 0.3, f'\\U{ord(char):08x}', fontsize=40, horizontalalignment='center')
+    fig.text(0, 0.1, f'Inari Sámi: {char}', fontsize=40, language='smn')
+    fig.text(1, 0.1, f'English: {char}', fontsize=40, language='en',
+             horizontalalignment='right')
diff --git a/lib/matplotlib/_text_helpers.py b/lib/matplotlib/_text_helpers.py
@@ -26,7 +26,7 @@ def warn_on_missing_glyph(codepoint, fontnames):
         f"missing from font(s) {fontnames}.")
 
 
-def layout(string, font, *, kern_mode=Kerning.DEFAULT):
+def layout(string, font, *, kern_mode=Kerning.DEFAULT, language=None):
     """
     Render *string* with *font*.
 
@@ -41,14 +41,17 @@ def layout(string, font, *, kern_mode=Kerning.DEFAULT):
         The font.
     kern_mode : Kerning
         A FreeType kerning mode.
+    language : str, optional
+        The language of the text in a format accepted by libraqm, namely `a BCP47
+        language code <https://www.w3.org/International/articles/language-tags/>`_.
 
     Yields
     ------
     LayoutItem
     """
     x = 0
     prev_glyph_index = None
-    char_to_font = font._get_fontmap(string)
+    char_to_font = font._get_fontmap(string)  # TODO: Pass in language.
     base_font = font
     for char in string:
         # This has done the fallback logic
diff --git a/lib/matplotlib/backends/backend_agg.py b/lib/matplotlib/backends/backend_agg.py
@@ -190,7 +190,8 @@ def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
         font = self._prepare_font(prop)
         # We pass '0' for angle here, since it will be rotated (in raster
         # space) in the following call to draw_text_image).
-        font.set_text(s, 0, flags=get_hinting_flag())
+        font.set_text(s, 0, flags=get_hinting_flag(),
+                      language=mtext.get_language() if mtext is not None else None)
         font.draw_glyphs_to_bitmap(
             antialiased=gc.get_antialiased())
         d = font.get_descent() / 64.0
diff --git a/lib/matplotlib/backends/backend_pdf.py b/lib/matplotlib/backends/backend_pdf.py
@@ -2338,6 +2338,7 @@ def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
             return self.draw_mathtext(gc, x, y, s, prop, angle)
 
         fontsize = prop.get_size_in_points()
+        language = mtext.get_language() if mtext is not None else None
 
         if mpl.rcParams['pdf.use14corefonts']:
             font = self._get_font_afm(prop)
@@ -2348,7 +2349,7 @@ def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
             fonttype = mpl.rcParams['pdf.fonttype']
 
         if gc.get_url() is not None:
-            font.set_text(s)
+            font.set_text(s, language=language)
             width, height = font.get_width_height()
             self.file._annotations[-1][1].append(_get_link_annotation(
                 gc, x, y, width / 64, height / 64, angle))
@@ -2398,7 +2399,8 @@ def output_singlebyte_chunk(kerns_or_chars):
             prev_start_x = 0
             # Emit all the 1-byte characters in a BT/ET group.
             self.file.output(Op.begin_text)
-            for item in _text_helpers.layout(s, font, kern_mode=Kerning.UNFITTED):
+            for item in _text_helpers.layout(s, font, kern_mode=Kerning.UNFITTED,
+                                             language=language):
                 if _font_supports_glyph(fonttype, ord(item.char)):
                     if item.ft_object != prev_font:
                         if singlebyte_chunk:
diff --git a/lib/matplotlib/backends/backend_ps.py b/lib/matplotlib/backends/backend_ps.py
@@ -791,9 +791,10 @@ def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
                 thisx += width * scale
 
         else:
+            language = mtext.get_language() if mtext is not None else None
             font = self._get_font_ttf(prop)
             self._character_tracker.track(font, s)
-            for item in _text_helpers.layout(s, font):
+            for item in _text_helpers.layout(s, font, language=language):
                 ps_name = (item.ft_object.postscript_name
                            .encode("ascii", "replace").decode("ascii"))
                 glyph_name = item.ft_object.get_glyph_name(item.glyph_index)
diff --git a/lib/matplotlib/ft2font.pyi b/lib/matplotlib/ft2font.pyi
@@ -243,7 +243,12 @@ class FT2Font(Buffer):
     def set_charmap(self, i: int) -> None: ...
     def set_size(self, ptsize: float, dpi: float) -> None: ...
     def set_text(
-        self, string: str, angle: float = ..., flags: LoadFlags = ...
+        self,
+        string: str,
+        angle: float = ...,
+        flags: LoadFlags = ...,
+        *,
+        language: str | list[tuple[str, int, int]] | None = ...,
     ) -> NDArray[np.float64]: ...
     @property
     def ascender(self) -> int: ...
diff --git a/lib/matplotlib/mpl-data/matplotlibrc b/lib/matplotlib/mpl-data/matplotlibrc
@@ -292,6 +292,11 @@
 ## for more information on text properties
 #text.color: black
 
+## The language of the text in a format accepted by libraqm, namely `a BCP47 language
+## code <https://www.w3.org/International/articles/language-tags/>`_. If None, then no
+## particular language will be implied, and default font settings will be used.
+#text.language: None
+
 ## FreeType hinting flag ("foo" corresponds to FT_LOAD_FOO); may be one of the
 ## following (Proprietary Matplotlib-specific synonyms are given in parentheses,
 ## but their use is discouraged):
diff --git a/lib/matplotlib/rcsetup.py b/lib/matplotlib/rcsetup.py
@@ -1045,6 +1045,7 @@ def _convert_validator_spec(key, conv):
     "text.kerning_factor": validate_int_or_None,
     "text.antialiased":    validate_bool,
     "text.parse_math":     validate_bool,
+    "text.language":       validate_string_or_None,
 
     "mathtext.cal":            validate_font_properties,
     "mathtext.rm":             validate_font_properties,
diff --git a/lib/matplotlib/tests/test_ft2font.py b/lib/matplotlib/tests/test_ft2font.py
@@ -783,6 +783,37 @@ def test_ft2font_set_text():
     assert font.get_bitmap_offset() == (6, 0)
 
 
+@pytest.mark.parametrize(
+    'input',
+    [
+        [1, 2, 3],
+        [(1, 2)],
+        [('en', 'foo', 2)],
+        [('en', 1, 'foo')],
+    ],
+    ids=[
+        'nontuple',
+        'wrong length',
+        'wrong start type',
+        'wrong end type',
+    ],
+)
+def test_ft2font_language_invalid(input):
+    file = fm.findfont('DejaVu Sans')
+    font = ft2font.FT2Font(file, hinting_factor=1)
+    with pytest.raises(TypeError):
+        font.set_text('foo', language=input)
+
+
+def test_ft2font_language():
+    # This is just a smoke test.
+    file = fm.findfont('DejaVu Sans')
+    font = ft2font.FT2Font(file, hinting_factor=1)
+    font.set_text('foo')
+    font.set_text('foo', language='en')
+    font.set_text('foo', language=[('en', 1, 2)])
+
+
 def test_ft2font_loading():
     file = fm.findfont('DejaVu Sans')
     font = ft2font.FT2Font(file, hinting_factor=1)
diff --git a/lib/matplotlib/tests/test_text.py b/lib/matplotlib/tests/test_text.py
@@ -1202,3 +1202,60 @@ def test_ytick_rotation_mode():
         tick.set_rotation(angle)
 
     plt.subplots_adjust(left=0.4, right=0.6, top=.99, bottom=.01)
+
+
+@pytest.mark.parametrize(
+    'input, match',
+    [
+        ([1, 2, 3], 'must be list of tuple'),
+        ([(1, 2)], 'must be list of tuple'),
+        ([('en', 'foo', 2)], 'start location must be int'),
+        ([('en', 1, 'foo')], 'end location must be int'),
+    ],
+)
+def test_text_language_invalid(input, match):
+    with pytest.raises(TypeError, match=match):
+        Text(0, 0, 'foo', language=input)
+
+
+@image_comparison(baseline_images=['language.png'], remove_text=False, style='mpl20')
+def test_text_language():
+    fig = plt.figure(figsize=(5, 3))
+
+    t = fig.text(0, 0.8, 'Default', fontsize=32)
+    assert t.get_language() is None
+    t = fig.text(0, 0.55, 'Lang A', fontsize=32)
+    assert t.get_language() is None
+    t = fig.text(0, 0.3, 'Lang B', fontsize=32)
+    assert t.get_language() is None
+    t = fig.text(0, 0.05, 'Mixed', fontsize=32)
+    assert t.get_language() is None
+
+    # DejaVu Sans supports language-specific glyphs in the Serbian and Macedonian
+    # languages in the Cyrillic alphabet.
+    cyrillic = '\U00000431'
+    t = fig.text(0.4, 0.8, cyrillic, fontsize=32)
+    assert t.get_language() is None
+    t = fig.text(0.4, 0.55, cyrillic, fontsize=32, language='sr')
+    assert t.get_language() == 'sr'
+    t = fig.text(0.4, 0.3, cyrillic, fontsize=32)
+    t.set_language('ru')
+    assert t.get_language() == 'ru'
+    t = fig.text(0.4, 0.05, cyrillic * 4, fontsize=32,
+                 language=[('ru', 0, 1), ('sr', 1, 2), ('ru', 2, 3), ('sr', 3, 4)])
+    assert t.get_language() == (('ru', 0, 1), ('sr', 1, 2), ('ru', 2, 3), ('sr', 3, 4))
+
+    # Or the Sámi family of languages in the Latin alphabet.
+    latin = '\U0000014a'
+    t = fig.text(0.7, 0.8, latin, fontsize=32)
+    assert t.get_language() is None
+    with plt.rc_context({'text.language': 'en'}):
+        t = fig.text(0.7, 0.55, latin, fontsize=32)
+    assert t.get_language() == 'en'
+    t = fig.text(0.7, 0.3, latin, fontsize=32, language='smn')
+    assert t.get_language() == 'smn'
+    # Tuples are not documented, but we'll allow it.
+    t = fig.text(0.7, 0.05, latin * 4, fontsize=32)
+    t.set_language((('en', 0, 1), ('smn', 1, 2), ('en', 2, 3), ('smn', 3, 4)))
+    assert t.get_language() == (
+        ('en', 0, 1), ('smn', 1, 2), ('en', 2, 3), ('smn', 3, 4))
diff --git a/lib/matplotlib/text.py b/lib/matplotlib/text.py
@@ -2,6 +2,7 @@
 Classes for including text in a figure.
 """
 
+from collections.abc import Sequence
 import functools
 import logging
 import math
@@ -136,6 +137,7 @@ def __init__(self,
         super().__init__()
         self._x, self._y = x, y
         self._text = ''
+        self.set_language(None)
         self._reset_visual_defaults(
             text=text,
             color=color,
@@ -1422,6 +1424,42 @@ def _va_for_angle(self, angle):
             return 'baseline' if anchor_at_left else 'top'
         return 'top' if anchor_at_left else 'baseline'
 
+    def get_language(self):
+        """Return the language this Text is in."""
+        return self._language
+
+    def set_language(self, language):
+        """
+        Set the language of the text.
+
+        Parameters
+        ----------
+        language : str or None
+            The language of the text in a format accepted by libraqm, namely `a BCP47
+            language code <https://www.w3.org/International/articles/language-tags/>`_.
+
+            If None, then defaults to :rc:`text.language`.
+        """
+        _api.check_isinstance((Sequence, str, None), language=language)
+        language = mpl._val_or_rc(language, 'text.language')
+
+        if not cbook.is_scalar_or_string(language):
+            language = tuple(language)
+            for val in language:
+                if not isinstance(val, tuple) or len(val) != 3:
+                    raise TypeError('language must be list of tuple, not {language!r}')
+                sublang, start, end = val
+                if not isinstance(sublang, str):
+                    raise TypeError(
+                        'sub-language specification must be str, not {sublang!r}')
+                if not isinstance(start, int):
+                    raise TypeError('start location must be int, not {start!r}')
+                if not isinstance(end, int):
+                    raise TypeError('end location must be int, not {end!r}')
+
+        self._language = language
+        self.stale = True
+
 
 class OffsetFrom:
     """Callable helper class for working with `Annotation`."""
diff --git a/lib/matplotlib/text.pyi b/lib/matplotlib/text.pyi
@@ -14,7 +14,7 @@ from .transforms import (
     Transform,
 )
 
-from collections.abc import Iterable
+from collections.abc import Iterable, Sequence
 from typing import Any, Literal
 from .typing import ColorType, CoordsType
 
@@ -108,6 +108,8 @@ class Text(Artist):
     def set_antialiased(self, antialiased: bool) -> None: ...
     def _ha_for_angle(self, angle: Any) -> Literal['center', 'right', 'left'] | None: ...
     def _va_for_angle(self, angle: Any) -> Literal['center', 'top', 'baseline'] | None: ...
+    def get_language(self) -> str | tuple[tuple[str, int, int], ...] | None: ...
+    def set_language(self, language: str | Sequence[tuple[str, int, int]] | None) -> None: ...
 
 class OffsetFrom:
     def __init__(
diff --git a/lib/matplotlib/textpath.py b/lib/matplotlib/textpath.py
@@ -67,7 +67,7 @@ def get_text_width_height_descent(self, s, prop, ismath):
         d /= 64.0
         return w * scale, h * scale, d * scale
 
-    def get_text_path(self, prop, s, ismath=False):
+    def get_text_path(self, prop, s, ismath=False, *, language=None):
         """
         Convert text *s* to path (a tuple of vertices and codes for
         matplotlib.path.Path).
@@ -80,6 +80,9 @@ def get_text_path(self, prop, s, ismath=False):
             The text to be converted.
         ismath : {False, True, "TeX"}
             If True, use mathtext parser.  If "TeX", use tex for rendering.
+        language : str, optional
+            The language of the text in a format accepted by libraqm, namely `a BCP47
+            language code <https://www.w3.org/International/articles/language-tags/>`_.
 
         Returns
         -------
@@ -107,7 +110,8 @@ def get_text_path(self, prop, s, ismath=False):
             glyph_info, glyph_map, rects = self.get_glyphs_tex(prop, s)
         elif not ismath:
             font = self._get_font(prop)
-            glyph_info, glyph_map, rects = self.get_glyphs_with_font(font, s)
+            glyph_info, glyph_map, rects = self.get_glyphs_with_font(font, s,
+                                                                     language=language)
         else:
             glyph_info, glyph_map, rects = self.get_glyphs_mathtext(prop, s)
 
@@ -128,7 +132,7 @@ def get_text_path(self, prop, s, ismath=False):
         return verts, codes
 
     def get_glyphs_with_font(self, font, s, glyph_map=None,
-                             return_new_glyphs_only=False):
+                             return_new_glyphs_only=False, *, language=None):
         """
         Convert string *s* to vertices and codes using the provided ttf font.
         """
@@ -143,7 +147,7 @@ def get_glyphs_with_font(self, font, s, glyph_map=None,
 
         xpositions = []
         glyph_reprs = []
-        for item in _text_helpers.layout(s, font):
+        for item in _text_helpers.layout(s, font, language=language):
             glyph_repr = self._get_glyph_repr(item.ft_object, item.glyph_index)
             glyph_reprs.append(glyph_repr)
             xpositions.append(item.x)
diff --git a/lib/matplotlib/textpath.pyi b/lib/matplotlib/textpath.pyi
@@ -16,14 +16,17 @@ class TextToPath:
         self, s: str, prop: FontProperties, ismath: bool | Literal["TeX"]
     ) -> tuple[float, float, float]: ...
     def get_text_path(
-        self, prop: FontProperties, s: str, ismath: bool | Literal["TeX"] = ...
+        self, prop: FontProperties, s: str, ismath: bool | Literal["TeX"] = ..., *,
+        language: str | list[tuple[str, int, int]] | None = ...,
     ) -> list[np.ndarray]: ...
     def get_glyphs_with_font(
         self,
         font: FT2Font,
         s: str,
         glyph_map: dict[str, tuple[np.ndarray, np.ndarray]] | None = ...,
         return_new_glyphs_only: bool = ...,
+        *,
+        language: str | list[tuple[str, int, int]] | None = ...,
     ) -> tuple[
         list[tuple[str, float, float, float]],
         dict[str, tuple[np.ndarray, np.ndarray]],
diff --git a/src/ft2font.cpp b/src/ft2font.cpp
diff --git a/src/ft2font.h b/src/ft2font.h
diff --git a/src/ft2font_wrapper.cpp b/src/ft2font_wrapper.cpp
