postgresql/contrib/unaccent/generate_unaccent_rules.py

124 lines
5.4 KiB
Python
Raw Normal View History

#!/usr/bin/python
#
# This script builds unaccent.rules on standard output when given the
# contents of UnicodeData.txt[1] on standard input. Optionally includes
# ligature expansion, if --expand-ligatures is given on the command line.
#
# The approach is to use the Unicode decomposition data to identify
# precomposed codepoints that are equivalent to a ligature of several
# letters, or a base letter with any number of diacritical marks.
# There is also a small set of special cases for codepoints that we
# traditionally support even though Unicode doesn't consider them to
# be ligatures or letters with marks.
#
# [1] http://unicode.org/Public/7.0.0/ucd/UnicodeData.txt
import re
import sys
def print_record(codepoint, letter):
print (unichr(codepoint) + "\t" + letter).encode("UTF-8")
class Codepoint:
def __init__(self, id, general_category, combining_ids):
self.id = id
self.general_category = general_category
self.combining_ids = combining_ids
def is_plain_letter(codepoint):
"""Return true if codepoint represents a plain ASCII letter."""
return (codepoint.id >= ord('a') and codepoint.id <= ord('z')) or \
(codepoint.id >= ord('A') and codepoint.id <= ord('Z'))
def is_mark(codepoint):
"""Returns true for diacritical marks (combining codepoints)."""
return codepoint.general_category in ("Mn", "Me", "Mc")
def is_letter_with_marks(codepoint, table):
"""Returns true for plain letters combined with one or more marks."""
# See http://www.unicode.org/reports/tr44/tr44-14.html#General_Category_Values
return len(codepoint.combining_ids) > 1 and \
is_plain_letter(table[codepoint.combining_ids[0]]) and \
all(is_mark(table[i]) for i in codepoint.combining_ids[1:])
def is_letter(codepoint, table):
"""Return true for letter with or without diacritical marks."""
return is_plain_letter(codepoint) or is_letter_with_marks(codepoint, table)
def get_plain_letter(codepoint, table):
"""Return the base codepoint without marks."""
if is_letter_with_marks(codepoint, table):
return table[codepoint.combining_ids[0]]
elif is_plain_letter(codepoint):
return codepoint
else:
raise "mu"
def is_ligature(codepoint, table):
"""Return true for letters combined with letters."""
return all(is_letter(table[i], table) for i in codepoint.combining_ids)
def get_plain_letters(codepoint, table):
"""Return a list of plain letters from a ligature."""
assert(is_ligature(codepoint, table))
return [get_plain_letter(table[id], table) for id in codepoint.combining_ids]
def main(expand_ligatures):
# http://www.unicode.org/reports/tr44/tr44-14.html#Character_Decomposition_Mappings
decomposition_type_pattern = re.compile(" *<[^>]*> *")
table = {}
all = []
# read everything we need into memory
for line in sys.stdin.readlines():
fields = line.split(";")
if len(fields) > 5:
# http://www.unicode.org/reports/tr44/tr44-14.html#UnicodeData.txt
general_category = fields[2]
decomposition = fields[5]
decomposition = re.sub(decomposition_type_pattern, ' ', decomposition)
id = int(fields[0], 16)
combining_ids = [int(s, 16) for s in decomposition.split(" ") if s != ""]
codepoint = Codepoint(id, general_category, combining_ids)
table[id] = codepoint
all.append(codepoint)
# walk through all the codepoints looking for interesting mappings
for codepoint in all:
if codepoint.general_category.startswith('L') and \
len(codepoint.combining_ids) > 1:
if is_letter_with_marks(codepoint, table):
print_record(codepoint.id,
chr(get_plain_letter(codepoint, table).id))
elif expand_ligatures and is_ligature(codepoint, table):
print_record(codepoint.id,
"".join(unichr(combining_codepoint.id)
for combining_codepoint \
in get_plain_letters(codepoint, table)))
# some special cases
print_record(0x00d8, "O") # LATIN CAPITAL LETTER O WITH STROKE
print_record(0x00f8, "o") # LATIN SMALL LETTER O WITH STROKE
print_record(0x0110, "D") # LATIN CAPITAL LETTER D WITH STROKE
print_record(0x0111, "d") # LATIN SMALL LETTER D WITH STROKE
print_record(0x0131, "i") # LATIN SMALL LETTER DOTLESS I
print_record(0x0126, "H") # LATIN CAPITAL LETTER H WITH STROKE
print_record(0x0127, "h") # LATIN SMALL LETTER H WITH STROKE
print_record(0x0141, "L") # LATIN CAPITAL LETTER L WITH STROKE
print_record(0x0142, "l") # LATIN SMALL LETTER L WITH STROKE
print_record(0x0149, "'n") # LATIN SMALL LETTER N PRECEDED BY APOSTROPHE
print_record(0x0166, "T") # LATIN CAPITAL LETTER T WITH STROKE
print_record(0x0167, "t") # LATIN SMALL LETTER t WITH STROKE
print_record(0x0401, u"\u0415") # CYRILLIC CAPITAL LETTER IO
print_record(0x0451, u"\u0435") # CYRILLIC SMALL LETTER IO
if expand_ligatures:
print_record(0x00c6, "AE") # LATIN CAPITAL LETTER AE
print_record(0x00df, "ss") # LATIN SMALL LETTER SHARP S
print_record(0x00e6, "ae") # LATIN SMALL LETTER AE
print_record(0x0152, "OE") # LATIN CAPITAL LIGATURE OE
print_record(0x0153, "oe") # LATIN SMALL LIGATURE OE
if __name__ == "__main__":
main(len(sys.argv) == 2 and sys.argv[1] == "--expand-ligatures")