#!/usr/bin/env python
# Generate a unicode sampler from an Xmodmap file

ENCODING = 'utf-8'
XKEYSIMDEF_H = "/usr/include/X11/keysymdef.h"


def sample(lines, keysyms):
    """Conver lines from an Xmodmap file to sampler lines."""
    for line in lines:
        # parse line
        line = line.strip()
        if not line.startswith('keycode '):
            continue
        fields = line.split()
        # keycode  10 = 1 exclam U2081 onehalf
        fields = fields + [None] * (7-len(fields))
        k,keycode,eq,a,b,c,d = fields
        assert k == 'keycode', k
        assert eq == '=', eq
        keycode = int(keycode)
        if keycode > 100:
            continue  # control code, not very interesting

        # convert keys to unicode
        keys = [a,b,c,d]
        for i,key in enumerate(keys):
            if key == None:
                continue
            if key in keysyms:
                keys[i] = keysyms[key]
                continue
            assert key.startswith('U'), key
            key = r'\u'+key[len('U'):]
            keys[i] = unicode(key, 'unicode escape')

        line = (' '.join([key or ' ' for key in keys])).rstrip()
        if len(line) > 0:
            yield line

def read_keysyms(file=None):
    """Read keysymdef.h and extract keysym -> unicode mappings."""
    if file == None:
        file = XKEYSIMDEF_H
    keysyms = {}
    for line in open(file, 'r').readlines():
        if not line.startswith('#define XK_'):
            continue
        # #define XK_space                         0x0020  /* U+0020 SPACE */
        fields = line.split()
        assert len(fields) >= 3, fields
        keysym = fields[1]
        code = fields[2]
        assert keysym.startswith('XK_'), keysym
        assert code.startswith('0x'), code
        keysym = keysym[len('XK_'):]
        if len(fields) >= 5:
            codepoint = fields[4]
            if not codepoint.startswith('U+'):
                codepoint = None
            else:
                codepoint = r'\u'+codepoint[len('U+'):]
                codepoint = unicode(codepoint, 'unicode escape')
        else:
            codepoint = None  # keysym to unicode mapping not well defined
        keysyms[keysym] = codepoint
    return keysyms


if __name__ == '__main__':
    import sys
    import time

    xmodmap_filename = sys.argv[1]

    keysyms = read_keysyms()
    lines = open(xmodmap_filename, 'r').readlines()
    #lines = sample(lines, keysyms)
    lines = sorted(sample(lines, keysyms))
    print 'These characters are bound in my current .Xmodmap'
    print '(%s)' % time.asctime()
    print ''
    for line in lines:
        print line.encode(ENCODING)