added 'age' property

.gitignore

@@ -63,3 +63,7 @@ target/
 *14.0.0.*
 *15.0.0.*
 tests/data
+
+# Clinic scripts
+clinic*.py
+cpp.py

CHANGELOG.md

@@ -1,3 +1,6 @@
+## 15.0.0-2
+ - Add age property
+
 ## 15.0.0-1
  - Add vertical orientation property
 

README.md

@@ -15,6 +15,7 @@ Currently supported additional Unicode properties:
 * Indic Syllabic Category: `indic_syllabic_category(chr)`
 * Grapheme Cluster Break: `grapheme_cluster_break(chr)`
 * Vertical Orientation: `vertical_orientation(chr)`
+* Age: `age(chr)`
 * Total Strokes (CJK): `total_strokes(chr)`
 * Emoji: `is_emoji(chr)`
 * Emoji Presentation: `is_emoji_presentation(chr)`

makeunicodedata.py

@@ -67,6 +67,7 @@
 INDIC_SYLLABIC_CATEGORY = "IndicSyllabicCategory%s.txt"
 GRAPHEME_BREAK_PROPERTY = "auxiliary/GraphemeBreakProperty%s.txt"
 VERTICAL_ORIENTATION_PROPERTY = "VerticalOrientation%s.txt"
+AGE_PROPERTY = "DerivedAge%s.txt"
 EMOJI_DATA = "emoji/emoji-data%s.txt"
 
 # Private Use Areas -- in planes 1, 15, 16
@@ -977,7 +978,7 @@ def word_key(a):
 
 def makeunicodeprop(unicode, trace):
 
-    dummy = (0, 0, 0, 0, 0, 0, 0)
+    dummy = (0, 0, 0, 0, 0, 0, 0, 0)
     table = [dummy]
     cache = {0: dummy}
     index = [0] * len(unicode.chars)
@@ -996,7 +997,8 @@ def makeunicodeprop(unicode, trace):
             indic_syllabic = unicode.indic_syllabic.index(record.indic_syllabic)
             grapheme_cluster_break = unicode.grapheme_cluster_break.index(record.grapheme_cluster_break)
             vertical_orientation = unicode.vertical_orientation.index(record.vertical_orientation)
-            item = (script, block, script_extensions, indic_positional, indic_syllabic, grapheme_cluster_break, vertical_orientation)
+            age = unicode.age.index(record.age)
+            item = (script, block, script_extensions, indic_positional, indic_syllabic, grapheme_cluster_break, vertical_orientation, age)
             i = cache.get(item)
             if i is None:
                 cache[item] = i = len(table)
@@ -1016,7 +1018,7 @@ def makeunicodeprop(unicode, trace):
         fprint("/* a list of unique unicode property sets */")
         fprint("static const _PyUnicodePlus_PropertySet _PyUnicodePlus_Property_Sets[] = {")
         for item in table:
-            fprint("    {%d, %d, %d, %d, %d, %d, %d}," % item)
+            fprint("    {%d, %d, %d, %d, %d, %d, %d, %d}," % item)
         fprint("};")
         fprint()
 
@@ -1063,6 +1065,12 @@ def makeunicodeprop(unicode, trace):
         fprint("    NULL")
         fprint("};")
 
+        fprint("static const char *_PyUnicodePlus_AgeNames[] = {")
+        for name in unicode.age:
+            fprint("    \"%s\"," % name)
+        fprint("    NULL")
+        fprint("};")
+
         # split property set index table
         index1, index2, shift = splitbins(index, trace)
 
@@ -1521,6 +1529,16 @@ def __init__(self, version, cjk_check=True):
                 if table[i] is not None:
                     table[i].vertical_orientation = vertical_orientation[i]
 
+            age = ["Unassigned"] * 0x110000
+            for char, (ag, ) in UcdFile(AGE_PROPERTY, version).expanded():
+                age[char] = ag
+
+            self.age = ["Unassigned"] + sorted(set(age) - {"Unassigned"})
+
+            for i in range(0, 0x110000):
+                if table[i] is not None:
+                    table[i].age = age[i]
+
             for char, (p,) in UcdFile(EMOJI_DATA, version).expanded():
                 if table[char]:
                     table[char].binary_properties.add(p)

setup.py

@@ -20,7 +20,7 @@
 
 setup(
     name="unicodedataplus",
-    version="15.0.0-1",
+    version="15.0.0-2",
     description="Unicodedata with extensions for additional properties.",
     ext_modules=[main_module],
     author="Ben Yang",

tests/test_unicodedataplus.py

@@ -281,6 +281,24 @@ def test_vertical_orientation(self):
         self.assertEqual(self.db.vertical_orientation('\U0001E040'), 'R')
         self.assertEqual(self.db.vertical_orientation('\U0001F200'), 'Tu')
 
+    def test_age(self):
+        self.assertEqual(self.db.age('\u03DA'), '1.1')
+        self.assertEqual(self.db.age('\u20AB'), '2.0')
+        self.assertEqual(self.db.age('\u20AC'), '2.1')
+        self.assertEqual(self.db.age('\u058A'), '3.0')
+        self.assertEqual(self.db.age('\U00010423'), '3.1')
+        self.assertEqual(self.db.age('\u07B1'), '3.2')
+        self.assertEqual(self.db.age('\U00010083'), '4.0')
+        self.assertEqual(self.db.age('\u131F'), '4.1')
+        self.assertEqual(self.db.age('\U0001D363'), '5.0')
+        self.assertEqual(self.db.age('\uA95F'), '5.1')
+        self.assertEqual(self.db.age('\u0C34'), '7.0')
+        self.assertEqual(self.db.age('\U0001F6F8'), '10.0')
+        self.assertEqual(self.db.age('\u0EAC'), '12.0')
+        self.assertEqual(self.db.age('\U0002A6D9'), '13.0')
+        self.assertEqual(self.db.age('\u170D'), '14.0')
+
+
     def test_total_strokes(self):
         self.assertEqual(self.db.total_strokes('P'), 0)
         self.assertEqual(self.db.total_strokes('\u694A'), 13)

unicodedataplus/unicodedata.c

@@ -73,7 +73,8 @@ typedef struct {
                                            _PyUnicodePlus_Grapheme_Cluster_Break */
     const int vertical_orientation;   /* index into
                                            _PyUnicodePlus_Vertical_Orientation */
-
+    const int age;            /* index into
+                                           _PyUnicodePlus_Age */
 } _PyUnicodePlus_PropertySet;
 
 typedef struct {
@@ -95,6 +96,7 @@ typedef struct change_record {
     const unsigned char indic_syllabic_category_changed;
     const unsigned char grapheme_cluster_break_changed;
     const unsigned char vertical_orientation_changed;
+    const unsigned char age_changed;
     const unsigned char total_strokes_changed;
 } change_record;
 
@@ -707,6 +709,33 @@ unicodedata_UCD_vertical_orientation_impl(PyObject *self, int chr)
     return PyUnicode_FromString(_PyUnicodePlus_VerticalOrientationNames[index]);
 }
 
+/*[clinic input]
+unicodedata.UCD.age
+
+    self: self
+    chr: int(accept={str})
+    /
+
+Returns the Age property of the character chr as string.
+[clinic start generated code]*/
+
+static PyObject *
+unicodedata_UCD_age_impl(PyObject *self, int chr)
+/*[clinic end generated code: output=65b9ca0dc56b5516 input=57aa81559ef3dc45]*/
+{
+    int index;
+    Py_UCS4 c = (Py_UCS4)chr;
+    index = (int) _getpropset_ex(c)->age;
+    if (UCD_Check(self)) {
+        const change_record *old = get_old_record(self, c);
+        if (old->category_changed == 0)
+            index = 0; /* unassigned */
+        else if (old->age_changed != 0xFF)
+            index = old->age_changed;
+    }
+    return PyUnicode_FromString(_PyUnicodePlus_AgeNames[index]);
+}
+
 /*[clinic input]
 unicodedata.UCD.total_strokes
 
@@ -1987,6 +2016,7 @@ static PyMethodDef unicodedata_functions[] = {
     UNICODEDATA_UCD_INDIC_SYLLABIC_CATEGORY_METHODDEF
     UNICODEDATA_UCD_GRAPHEME_CLUSTER_BREAK_METHODDEF
     UNICODEDATA_UCD_VERTICAL_ORIENTATION_METHODDEF
+    UNICODEDATA_UCD_AGE_METHODDEF
     UNICODEDATA_UCD_TOTAL_STROKES_METHODDEF
     UNICODEDATA_UCD_DECOMPOSITION_METHODDEF
     UNICODEDATA_UCD_NAME_METHODDEF

unicodedataplus/unicodedata.c.37.h

@@ -444,6 +444,33 @@ unicodedata_UCD_vertical_orientation(PyObject *self, PyObject *arg)
     return return_value;
 }
 
+PyDoc_STRVAR(unicodedata_UCD_age__doc__,
+"age($self, chr, /)\n"
+"--\n"
+"\n"
+"Returns the Age property of the character chr as string.");
+
+#define UNICODEDATA_UCD_AGE_METHODDEF    \
+    {"age", (PyCFunction)unicodedata_UCD_age, METH_O, unicodedata_UCD_age__doc__},
+
+static PyObject *
+unicodedata_UCD_age_impl(PyObject *self, int chr);
+
+static PyObject *
+unicodedata_UCD_age(PyObject *self, PyObject *arg)
+{
+    PyObject *return_value = NULL;
+    int chr;
+
+    if (!PyArg_Parse(arg, "C:age", &chr)) {
+        goto exit;
+    }
+    return_value = unicodedata_UCD_age_impl(self, chr);
+
+exit:
+    return return_value;
+}
+
 PyDoc_STRVAR(unicodedata_UCD_total_strokes__doc__,
 "total_strokes($self, chr, /)\n"
 "--\n"
@@ -791,4 +818,4 @@ unicodedata_UCD_is_extended_pictographic(PyObject *self, PyObject *arg)
 exit:
     return return_value;
 }
-/*[clinic end generated code: output=48287109f3866503 input=a9049054013a1b77]*/
+/*[clinic end generated code: output=786b068a16dce3a5 input=a9049054013a1b77]*/

unicodedataplus/unicodedata.c.h

@@ -600,6 +600,42 @@ unicodedata_UCD_vertical_orientation(PyObject *self, PyObject *arg)
     return return_value;
 }
 
+PyDoc_STRVAR(unicodedata_UCD_age__doc__,
+"age($self, chr, /)\n"
+"--\n"
+"\n"
+"Returns the Age property of the character chr as string.");
+
+#define UNICODEDATA_UCD_AGE_METHODDEF    \
+    {"age", (PyCFunction)unicodedata_UCD_age, METH_O, unicodedata_UCD_age__doc__},
+
+static PyObject *
+unicodedata_UCD_age_impl(PyObject *self, int chr);
+
+static PyObject *
+unicodedata_UCD_age(PyObject *self, PyObject *arg)
+{
+    PyObject *return_value = NULL;
+    int chr;
+
+    if (!PyUnicode_Check(arg)) {
+        _PyArg_BadArgument("age", "argument", "a unicode character", arg);
+        goto exit;
+    }
+    if (PyUnicode_READY(arg)) {
+        goto exit;
+    }
+    if (PyUnicode_GET_LENGTH(arg) != 1) {
+        _PyArg_BadArgument("age", "argument", "a unicode character", arg);
+        goto exit;
+    }
+    chr = PyUnicode_READ_CHAR(arg, 0);
+    return_value = unicodedata_UCD_age_impl(self, chr);
+
+exit:
+    return return_value;
+}
+
 PyDoc_STRVAR(unicodedata_UCD_total_strokes__doc__,
 "total_strokes($self, chr, /)\n"
 "--\n"
@@ -1065,4 +1101,4 @@ unicodedata_UCD_is_extended_pictographic(PyObject *self, PyObject *arg)
 exit:
     return return_value;
 }
-/*[clinic end generated code: output=03097ed6016ce64a input=a9049054013a1b77]*/
+/*[clinic end generated code: output=652110d3ec494e7a input=a9049054013a1b77]*/

unicodedataplus/unicodedata.c.pypy.h

@@ -429,6 +429,32 @@ unicodedata_UCD_vertical_orientation(PyObject *self, PyObject *arg)
     return return_value;
 }
 
+PyDoc_STRVAR(unicodedata_UCD_age__doc__,
+"age($self, chr, /)\n"
+"--\n"
+"\n"
+"Returns the Age property of the character chr as string.");
+
+#define UNICODEDATA_UCD_AGE_METHODDEF    \
+    {"age", (PyCFunction)unicodedata_UCD_age, METH_O, unicodedata_UCD_age__doc__},
+
+static PyObject *
+unicodedata_UCD_age_impl(PyObject *self, int chr);
+
+static PyObject *
+unicodedata_UCD_age(PyObject *self, PyObject *arg)
+{
+    PyObject *return_value = NULL;
+    int chr;
+
+    if (!PyArg_Parse(arg, "C:age", &chr))
+        goto exit;
+    return_value = unicodedata_UCD_age_impl(self, chr);
+
+exit:
+    return return_value;
+}
+
 PyDoc_STRVAR(unicodedata_UCD_total_strokes__doc__,
 "total_strokes($self, chr, /)\n"
 "--\n"
@@ -764,4 +790,4 @@ unicodedata_UCD_is_extended_pictographic(PyObject *self, PyObject *arg)
 exit:
     return return_value;
 }
-/*[clinic end generated code: output=dd3db21bb6c026ae input=a9049054013a1b77]*/
+/*[clinic end generated code: output=b84333aeb6f0ccb5 input=a9049054013a1b77]*/