ENH: add c++ code + Python API to get char -> font mapping

tacaswell · tacaswell · commit 8a26ff952704 · 2022-08-09T17:46:03.000-04:00
diff --git a/lib/matplotlib/tests/test_ft2font.py b/lib/matplotlib/tests/test_ft2font.py
@@ -76,3 +76,30 @@ def test_font_fallback_chinese(fig_test, fig_ref, family_name, file_name):
     ):
         fig_ref.text(0.05, .85 - 0.15*j, txt, family=ref_font)
         fig_test.text(0.05, .85 - 0.15*j, txt, family=test_font)
+
+
+@pytest.mark.parametrize(
+    "family_name, file_name",
+    [
+        ("WenQuanYi Zen Hei", "wqy-zenhei.ttc"),
+        ("Noto Sans CJK JP", "NotoSansCJK-Regular.ttc"),
+    ],
+)
+def test__get_fontmap(family_name, file_name):
+    fp = fm.FontProperties(family=[family_name])
+    if Path(fm.findfont(fp)).name != file_name:
+        pytest.skip(f"Font {family_name} ({file_name}) is missing")
+
+    text = "There are 几个汉字 in between!"
+    ft = fm.get_font(
+        fm.fontManager._find_fonts_by_props(
+            fm.FontProperties(family=["DejaVu Sans", family_name])
+        )
+    )
+
+    fontmap = ft._get_fontmap(text)
+    for char, font in fontmap.items():
+        if ord(char) > 127:
+            assert Path(font.fname).name == file_name
+        else:
+            assert Path(font.fname).name == "DejaVuSans.ttf"
diff --git a/src/ft2font.cpp b/src/ft2font.cpp
@@ -603,6 +603,31 @@ void FT2Font::load_char(long charcode, FT_Int32 flags, FT2Font *&ft_object, bool
     }
 }
 
+
+bool FT2Font::get_char_fallback_index(FT_ULong charcode, int& index) const
+{
+    FT_UInt glyph_index = FT_Get_Char_Index(face, charcode);
+    if (glyph_index) {
+        // -1 means the host has the char and we do not need to fallback
+        index = -1;
+        return true;
+    } else {
+        int inner_index = 0;
+        bool was_found;
+
+        for (size_t i = 0; i < fallbacks.size(); ++i) {
+            // TODO handle recursion somehow!
+            was_found = fallbacks[i]->get_char_fallback_index(charcode, inner_index);
+            if (was_found) {
+                index = i;
+                return true;
+            }
+        }
+    }
+    return false;
+}
+
+
 bool FT2Font::load_char_with_fallback(FT2Font *&ft_object_with_glyph,
                                       FT_UInt &final_glyph_index,
                                       std::vector<FT_Glyph> &parent_glyphs,
diff --git a/src/ft2font.h b/src/ft2font.h
@@ -108,6 +108,7 @@ class FT2Font
     FT_UInt get_char_index(FT_ULong charcode, bool fallback);
     void get_cbox(FT_BBox &bbox);
     PyObject* get_path();
+    bool get_char_fallback_index(FT_ULong charcode, int& index) const;
 
     FT_Face const &get_face() const
     {
diff --git a/src/ft2font_wrapper.cpp b/src/ft2font_wrapper.cpp
@@ -7,6 +7,9 @@
 // From Python
 #include <structmember.h>
 
+#include <set>
+#include <algorithm>
+
 #define STRINGIFY(s) XSTRINGIFY(s)
 #define XSTRINGIFY(s) #s
 
@@ -552,6 +555,76 @@ static PyObject *PyFT2Font_get_kerning(PyFT2Font *self, PyObject *args)
     return PyLong_FromLong(result);
 }
 
+const char *PyFT2Font_get_fontmap__doc__ =
+    "_get_fontmap(self, string)\n"
+    "--\n\n"
+    "Get a mapping between characters and the font that includes them.\n"
+    "A dictionary mapping unicode characters to PyFT2Font objects.";
+static PyObject *PyFT2Font_get_fontmap(PyFT2Font *self, PyObject *args, PyObject *kwds)
+{
+    PyObject *textobj;
+    const char *names[] = { "string", NULL };
+
+    if (!PyArg_ParseTupleAndKeywords(
+            args, kwds, "O:_get_fontmap", (char **)names, &textobj)) {
+        return NULL;
+    }
+
+    std::set<FT_ULong> codepoints;
+    size_t size;
+
+    if (PyUnicode_Check(textobj)) {
+        size = PyUnicode_GET_LENGTH(textobj);
+#if defined(PYPY_VERSION) && (PYPY_VERSION_NUM < 0x07040000)
+        // PyUnicode_ReadChar is available from PyPy 7.3.2, but wheels do not
+        // specify the micro-release version, so put the version bound at 7.4
+        // to prevent generating wheels unusable on PyPy 7.3.{0,1}.
+        Py_UNICODE *unistr = PyUnicode_AsUnicode(textobj);
+        for (size_t i = 0; i < size; ++i) {
+            codepoints.insert(unistr[i]);
+        }
+#else
+        for (size_t i = 0; i < size; ++i) {
+            codepoints.insert(PyUnicode_ReadChar(textobj, i));
+        }
+#endif
+    } else {
+        PyErr_SetString(PyExc_TypeError, "String must be str");
+        return NULL;
+    }
+    PyObject *char_to_font;
+    if (!(char_to_font = PyDict_New())) {
+        return NULL;
+    }
+    for (auto it = codepoints.begin(); it != codepoints.end(); ++it) {
+        auto x = *it;
+        PyObject* target_font;
+        int index;
+        if (self->x->get_char_fallback_index(x, index)) {
+            if (index >= 0) {
+                target_font = self->fallbacks[index];
+            } else {
+                target_font = (PyObject *)self;
+            }
+        } else {
+            // TODO Handle recursion!
+            target_font = (PyObject *)self;
+        }
+
+        PyObject *key = NULL;
+        bool error = (!(key = PyUnicode_FromFormat("%c", x))
+                      || (PyDict_SetItem(char_to_font, key, target_font) == -1));
+        Py_XDECREF(key);
+        if (error) {
+            Py_DECREF(char_to_font);
+            PyErr_SetString(PyExc_ValueError, "Something went very wrong");
+            return NULL;
+        }
+    }
+    return char_to_font;
+}
+
+
 const char *PyFT2Font_set_text__doc__ =
     "set_text(self, string, angle, flags=32)\n"
     "--\n\n"
@@ -1525,6 +1598,7 @@ static PyTypeObject *PyFT2Font_init_type()
         {"select_charmap", (PyCFunction)PyFT2Font_select_charmap, METH_VARARGS, PyFT2Font_select_charmap__doc__},
         {"get_kerning", (PyCFunction)PyFT2Font_get_kerning, METH_VARARGS, PyFT2Font_get_kerning__doc__},
         {"set_text", (PyCFunction)PyFT2Font_set_text, METH_VARARGS|METH_KEYWORDS, PyFT2Font_set_text__doc__},
+        {"_get_fontmap", (PyCFunction)PyFT2Font_get_fontmap, METH_VARARGS|METH_KEYWORDS, PyFT2Font_get_fontmap__doc__},
         {"get_num_glyphs", (PyCFunction)PyFT2Font_get_num_glyphs, METH_NOARGS, PyFT2Font_get_num_glyphs__doc__},
         {"load_char", (PyCFunction)PyFT2Font_load_char, METH_VARARGS|METH_KEYWORDS, PyFT2Font_load_char__doc__},
         {"load_glyph", (PyCFunction)PyFT2Font_load_glyph, METH_VARARGS|METH_KEYWORDS, PyFT2Font_load_glyph__doc__},

Original file line number	Diff line number	Diff line change
`@@ -108,6 +108,7 @@ class FT2Font`
`108`	`108`	`FT_UInt get_char_index(FT_ULong charcode, bool fallback);`
`109`	`109`	`void get_cbox(FT_BBox &bbox);`
`110`	`110`	`PyObject* get_path();`
	`111`	`+ bool get_char_fallback_index(FT_ULong charcode, int& index) const;`
`111`	`112`
`112`	`113`	`FT_Face const &get_face() const`
`113`	`114`	`{`