Add Python wrapper from Martin Hosken

diff --git a/contrib/python/lib/harfbuzz.pyx b/contrib/python/lib/harfbuzz.pyx
new file mode 100644 (file)
index 0000000..1f254ff
--- /dev/null
+++ b/contrib/python/lib/harfbuzz.pyx
@@ -0,0 +1,306 @@
+cdef extern from "stdlib.h":
+    ctypedef int size_t
+    void *malloc(size_t size)
+    void free(void* ptr)
+
+cdef extern from "ft2build.h" :
+    pass
+
+cdef extern from "freetype/freetype.h" :
+    ctypedef void *FT_Library 
+    ctypedef void *FT_Face
+    ctypedef int FT_Error
+
+    FT_Error FT_Init_FreeType (FT_Library *alib)
+    FT_Error FT_Done_FreeType (FT_Library alib)
+    FT_Error FT_Done_Face (FT_Face alib)
+    FT_Error FT_New_Face( FT_Library library, char *path, unsigned long index, FT_Face *aface)
+    FT_Error FT_Set_Char_Size (FT_Face aFace, unsigned int size_x, unsigned int size_y, unsigned int res_x, unsigned int res_y)
+
+cdef extern from "hb-common.h" :
+    cdef enum hb_direction_t :
+        HB_DIRECTION_LTR
+        HB_DIRECTION_RTL
+        HB_DIRECTION_TTB
+        HB_DIRECTION_BTT
+    ctypedef unsigned long hb_codepoint_t
+    ctypedef long hb_position_t
+    ctypedef unsigned long hb_mask_t
+    ctypedef unsigned long hb_tag_t
+    ctypedef void (*hb_destroy_func_t) (void *user_data)
+
+cdef extern from "hb-unicode.h" :
+# there must be a better way of syncing this list with the true source
+    ctypedef enum hb_script_t :
+        HB_SCRIPT_INVALID_CODE = -1,
+        HB_SCRIPT_COMMON = 0,
+        HB_SCRIPT_INHERITED,
+        HB_SCRIPT_ARABIC,
+        HB_SCRIPT_ARMENIAN,
+        HB_SCRIPT_BENGALI,
+        HB_SCRIPT_BOPOMOFO,
+        HB_SCRIPT_CHEROKEE,
+        HB_SCRIPT_DESERET,
+        HB_SCRIPT_DEVANAGARI,
+        HB_SCRIPT_ETHIOPIC,
+        HB_SCRIPT_GEORGIAN,
+        HB_SCRIPT_GOTHIC,
+        HB_SCRIPT_GREEK,
+        HB_SCRIPT_GUJARATI,
+        HB_SCRIPT_GURMUKHI,
+        HB_SCRIPT_HAN,
+        HB_SCRIPT_HANGUL,
+        HB_SCRIPT_HEBREW,
+        HB_SCRIPT_HIRAGANA,
+        HB_SCRIPT_KANNADA,
+        HB_SCRIPT_KATAKANA,
+        HB_SCRIPT_KHMER,
+        HB_SCRIPT_LAO,
+        HB_SCRIPT_LATIN,
+        HB_SCRIPT_MALAYALAM,
+        HB_SCRIPT_MONGOLIAN,
+        HB_SCRIPT_MYANMAR,
+        HB_SCRIPT_OGHAM,
+        HB_SCRIPT_OLD_ITALIC,
+        HB_SCRIPT_ORIYA,
+        HB_SCRIPT_RUNIC,
+        HB_SCRIPT_SINHALA,
+        HB_SCRIPT_SYRIAC,
+        HB_SCRIPT_TAMIL,
+        HB_SCRIPT_TELUGU,
+        HB_SCRIPT_THAANA,
+        HB_SCRIPT_THAI,
+        HB_SCRIPT_TIBETAN,
+        HB_SCRIPT_CANADIAN_ABORIGINAL,
+        HB_SCRIPT_YI,
+        HB_SCRIPT_TAGALOG,
+        HB_SCRIPT_HANUNDO,
+        HB_SCRIPT_BUHID,
+        HB_SCRIPT_TAGBANWA,
+# Unicode 4.0
+        HB_SCRIPT_BRAILLE,
+        HB_SCRIPT_CYPRIOT,
+        HB_SCRIPT_LIMBU,
+        HB_SCRIPT_OSMANYA,
+        HB_SCRIPT_SHAVIAN,
+        HB_SCRIPT_LINEAR_B,
+        HB_SCRIPT_TAI_LE,
+        HB_SCRIPT_UGARITIC,
+# Unicode 4.1
+        HB_SCRIPT_NEW_TAI_LUE,
+        HB_SCRIPT_BUGINESE,
+        HB_SCRIPT_GLAGOLITIC,
+        HB_SCRIPT_TIFINAGH,
+        HB_SCRIPT_SYLOTI_NAGRI,
+        HB_SCRIPT_OLD_PERSIAN,
+        HB_SCRIPT_KHAROSHTHI,
+# Unicode 5.0
+        HB_SCRIPT_UNKNOWN,
+        HB_SCRIPT_BALINESE,
+        HB_SCRIPT_CUNEIFORM,
+        HB_SCRIPT_PHOENICIAN,
+        HB_SCRIPT_PHAGS_PA,
+        HB_SCRIPT_NKO,
+# Unicode 5.1
+        HB_SCRIPT_KAYAH_LI,
+        HB_SCRIPT_LEPCHA,
+        HB_SCRIPT_REJANG,
+        HB_SCRIPT_SUNDANESE,
+        HB_SCRIPT_SAURASHTRA,
+        HB_SCRIPT_CHAM,
+        HB_SCRIPT_OL_CHIKI,
+        HB_SCRIPT_VAI,
+        HB_SCRIPT_CARIAN,
+        HB_SCRIPT_LYCIAN,
+        HB_SCRIPT_LYDIAN
+
+cdef extern from "hb-language.h" :
+    ctypedef void *hb_language_t
+    hb_language_t hb_language_from_string(char *str)
+    char * hb_language_to_string(hb_language_t language)
+
+cdef extern from "hb-ot-tag.h" :
+    hb_script_t hb_ot_string_to_script (char *sname)
+
+cdef extern from "hb-buffer.h" :
+    ctypedef struct hb_buffer_t :
+        pass
+
+    ctypedef struct hb_glyph_info_t :
+        hb_codepoint_t codepoint
+        hb_mask_t mask
+        unsigned long cluster
+
+    ctypedef struct hb_glyph_position_t :
+        hb_position_t x_advance
+        hb_position_t y_advance
+        hb_position_t x_offset
+        hb_position_t y_offset
+        unsigned long internal
+
+    hb_buffer_t *hb_buffer_create(unsigned int size)
+    hb_buffer_t *hb_buffer_reference(hb_buffer_t *buffer)
+    unsigned int hb_buffer_get_reference_count(hb_buffer_t *buffer)
+    void hb_buffer_destroy(hb_buffer_t *buffer)
+    void hb_buffer_set_direction(hb_buffer_t *buffer, hb_direction_t direction)
+    hb_direction_t hb_buffer_get_direction(hb_buffer_t *buffer)
+    void hb_buffer_set_script(hb_buffer_t *buffer, hb_script_t script)
+    hb_script_t hb_buffer_get_script(hb_buffer_t *buffer)
+    void hb_buffer_set_language(hb_buffer_t *buffer, hb_language_t language)
+    hb_language_t hb_buffer_get_language(hb_buffer_t *buffer)
+    void hb_buffer_clear(hb_buffer_t *)
+    void hb_buffer_clear_positions(hb_buffer_t *buffer)
+    void hb_buffer_ensure(hb_buffer_t *buffer, unsigned int size)
+    void hb_buffer_reverse(hb_buffer_t *buffer)
+    void hb_buffer_reverse_clusters(hb_buffer_t *buffer)
+    void hb_buffer_add_glyph(hb_buffer_t *buffer, hb_codepoint_t codepoint, hb_mask_t mask, unsigned int cluster)
+    void hb_buffer_add_utf8(hb_buffer_t *buffer, char *text, unsigned int text_length, unsigned int item_offset, unsigned int item_length)
+    unsigned int hb_buffer_get_length(hb_buffer_t *buffer)
+    hb_glyph_info_t *hb_buffer_get_glyph_infos(hb_buffer_t *buffer)
+    hb_glyph_position_t *hb_buffer_get_glyph_positions(hb_buffer_t *buffer)
+
+cdef extern from "hb-blob.h" :
+    cdef struct hb_blob_t :
+        pass
+# do I need blob functions here?
+
+cdef extern from "hb-font.h" :
+    ctypedef struct hb_face_t :
+        pass
+    ctypedef struct hb_font_t :
+        pass
+
+    ctypedef hb_blob_t * (*hb_get_table_func_t) (hb_tag_t tag, void *user_data)
+    hb_face_t * hb_face_create_for_data(hb_blob_t *blob, unsigned int index)
+    hb_face_t * hb_face_create_for_tables(hb_get_table_func_t get_table, hb_destroy_func_t destroy, void *user_data)
+    hb_face_t * hb_face_reference(hb_face_t *face)
+    unsigned int hb_face_get_reference_count(hb_face_t *face)
+    void hb_face_destroy(hb_face_t *face)
+    void hb_font_destroy(hb_font_t *font)
+    hb_blob_t * hb_face_get_table(hb_face_t *face, hb_tag_t tag)
+    void hb_font_set_tracecallback(hb_font_t *font, void (*cb)(char *type, int index, hb_buffer_t *))
+
+
+cdef extern from "hb-shape.h" :
+    ctypedef struct hb_feature_t :
+        int tag
+        unsigned int value
+        unsigned int start
+        unsigned int end
+
+    void hb_shape (hb_font_t *font, hb_face_t *face, hb_buffer_t *buffer, hb_feature_t *features, unsigned int num_features)
+
+cdef extern from "hb-ft.h" :
+    hb_face_t *hb_ft_face_create (FT_Face ft_face, hb_destroy_func_t destroy)
+    hb_font_t *hb_ft_font_create (FT_Face ft_face, hb_destroy_func_t destroy)
+
+class glyphinfo :
+ 
+    def __init__(self, gid, cluster, advance, offset, internal = 0) :
+        self.gid = gid
+        self.cluster = cluster
+        self.advance = advance
+        self.offset = offset
+        self.internal = internal
+
+    def __repr__(self) :
+        res = "{0:d}>{1:d}@({2:.2f},{3:.2f})+({4:.2f},{5:.2f})".format(self.gid, self.cluster, self.offset[0], self.offset[1], self.advance[0], self.advance[1])
+        if self.internal : res += "/i=" + str(self.internal)
+        return res
+
+gcb = None
+cdef void hb_python_trace(char *aType, int i, hb_buffer_t *aBuffer) :
+    gcb(aType, i)
+
+cdef class buffer :
+    cdef hb_buffer_t *buffer
+
+    def __init__(self, char *text, unsigned int length) :
+        """Note text must be a utf-8 string and length is number of chars"""
+        self.buffer = hb_buffer_create(length)
+        hb_buffer_add_utf8(self.buffer, text, length, 0, len(text))
+
+    def set_scriptlang(self, char *script, char *lang) :
+        cdef hb_language_t language
+        cdef hb_script_t scriptnum
+
+        language = hb_language_from_string(lang)
+        scriptnum = hb_ot_string_to_script(script)
+        hb_buffer_set_script(self.buffer, scriptnum)
+        hb_buffer_set_language(self.buffer, language)
+
+    def get_info(self, scale = 1) :
+        cdef hb_glyph_info_t *infos
+        cdef hb_glyph_position_t *positions
+        cdef unsigned int num
+        cdef unsigned int i
+        res = []
+
+        num = hb_buffer_get_length(self.buffer)
+        infos = hb_buffer_get_glyph_infos(self.buffer)
+        positions = hb_buffer_get_glyph_positions(self.buffer)
+        for 0 <= i < num :
+            temp = glyphinfo(infos[i].codepoint, infos[i].cluster, (positions[i].x_advance / scale, positions[i].y_advance / scale), (positions[i].x_offset / scale, positions[i].y_offset / scale), positions[i].internal)
+            res.append(temp)
+        return res
+
+    def get_settings(self) :
+        cdef hb_script_t script
+        cdef hb_language_t lang
+
+        script = hb_buffer_get_script(self.buffer)
+        lang = hb_buffer_get_language(self.buffer)
+        return {'script' : script, 'language' : hb_language_to_string(lang)}
+
+    def __del__(self) :
+        hb_buffer_destroy(self.buffer)
+
+cdef class ft :
+    cdef FT_Library engine
+    cdef FT_Face face
+    cdef hb_face_t *hbface
+    cdef hb_font_t *hbfont
+    cdef object tracefn
+
+    def __init__(self, char *fname, size, trace = None) :
+        global gcb
+        cdef FT_Library engine
+        FT_Init_FreeType(&engine)
+        self.engine = engine
+        cdef FT_Face face
+        FT_New_Face(engine, fname, 0, &face)
+        FT_Set_Char_Size(face, size << 6, size << 6, 144, 144)
+        self.face = face
+        self.hbface = hb_ft_face_create(face, <void (*)(void *)>hb_face_destroy)
+        self.hbfont = hb_ft_font_create(face, <void (*)(void *)>hb_font_destroy)
+        if trace :
+            self.tracefn = trace
+            gcb = self.trace
+            hb_font_set_tracecallback(self.hbfont, hb_python_trace)
+
+    def trace(self, aType, index) :
+        self.tracefn(self, aType, index)
+
+    def __del__(self) :
+        cdef FT_Library engine
+        engine = self.engine
+        cdef FT_Face face
+        face = self.face
+        FT_Done_Face(face)
+        FT_Done_FreeType(engine)
+
+    def shape(self, buffer aBuffer, features = {}) :
+        cdef hb_feature_t *feats
+        cdef hb_feature_t *aFeat
+        feats = <hb_feature_t *>malloc(sizeof(hb_feature_t) * len(features))
+        aFeat = feats
+        for k,v in features.items() :
+            k = k + "    "
+            aFeat.tag = (ord(k[0]) << 24) + (ord(k[1]) << 16) + (ord(k[2]) << 8) + ord(k[3])
+            aFeat.value = int(v)
+            aFeat.start = 0
+            aFeat.end = -1
+            aFeat += 1
+        hb_shape(self.hbfont, self.hbface, aBuffer.buffer, feats, len(features))
+
+

diff --git a/contrib/python/runpy b/contrib/python/runpy
new file mode 100755 (executable)
index 0000000..b39db1b
--- /dev/null
+++ b/contrib/python/runpy
@@ -0,0 +1,2 @@
+#!/bin/sh
+LD_LIBRARY_PATH=../../src/.libs PYTHONPATH=build/lib.`python -c 'import distutils.util, sys; print distutils.util.get_platform()+"-"+str(sys.version_info[0])+"."+str(sys.version_info[1])'` "$@"

diff --git a/contrib/python/scripts/hbtestfont b/contrib/python/scripts/hbtestfont
new file mode 100755 (executable)
index 0000000..4d1134d
--- /dev/null
+++ b/contrib/python/scripts/hbtestfont
@@ -0,0 +1,35 @@
+#!/usr/bin/python
+
+import harfbuzz, optparse
+
+buffer = None
+
+def tracefn(ft, aType, index) :
+    print aType + "(" + str(index) + "): " + str(buffer.get_info())
+
+usage = '''usage: %prog [options] fontfile "codepoints"
+    Generates output of glyphs and positions. Each entry is of the form:
+        glyphid>cluster@(offsetx,offsety)+(advancex,advancey)
+
+    codepoints is a space separated list of hex values of Unicode codepoints'''
+p = optparse.OptionParser(usage=usage)
+p.add_option('-s', '--size', default=12, type="int", help="point size")
+p.add_option('-l', '--lang', help="language code")
+p.add_option('-c', '--script', help="script code")
+p.add_option('-f', '--feature', action='append', help="define a feature key=val")
+p.add_option('-d', '--debug', action='store_true', help="Output trace info")
+(opts, args) = p.parse_args()
+
+ft = harfbuzz.ft(args[0], opts.size, trace = tracefn if opts.debug else None)
+text = "".join(unichr(int(c, 16)) for c in args[1].split(" "))
+bytes = text.encode('utf_8')
+buffer = harfbuzz.buffer(bytes, len(text))
+if (opts.lang or opts.script) : buffer.set_scriptlang(opts.script if opts.script else "", opts.lang if opts.lang else "")
+features = {}
+if opts.feature :
+    for f in opts.feature :
+        k, v = f.split("=")
+        features[k] = v
+ft.shape(buffer, features = features)
+res = buffer.get_info(64)       # scale for 26.6
+print res

diff --git a/contrib/python/setup.py b/contrib/python/setup.py
new file mode 100755 (executable)
index 0000000..ef2412b
--- /dev/null
+++ b/contrib/python/setup.py
@@ -0,0 +1,24 @@
+#!/usr/bin/python
+
+from distutils.core import setup
+from glob import glob
+from Pyrex.Distutils.extension import Extension
+from Pyrex.Distutils import build_ext
+
+setup(name='harfbuzz',
+    version='0.0.1',
+    description='Harfbuzz compatibility layer',
+    long_description='Harfbuzz python integration modules and supporting scripts',
+    maintainer='Martin Hosken',
+    maintainer_email='martin_hosken@sil.org',
+    packages=['harfbuzz'],
+    ext_modules = [
+        Extension("harfbuzz", ["lib/harfbuzz.pyx"], libraries=["harfbuzz"], library_dirs=["../../src/.libs"], include_dirs=["/usr/include/freetype2", "../../src"])
+        ],
+    cmdclass = {'build_ext' : build_ext},
+    scripts = glob('scripts/*'),
+    license = 'LGPL',
+    platforms = ['Linux', 'Win32', 'Mac OS X'],
+    package_dir = {'harfbuzz' : 'lib'}
+)
+