notepad-plus-plus-legacy/scintilla/scripts/GenerateCharacterCategory.py
Don Ho cb7806e94c [UPDATE] Update Scintilla to 3.3.4
[BUG_FIXED] (Author: Dave Brotherstone) Fix scintilla crash bug while closing a document.

git-svn-id: svn://svn.tuxfamily.org/svnroot/notepadplus/repository/trunk@1104 f5eea248-9336-0410-98b8-ebc06183d4e3
2013-08-28 00:44:27 +00:00

36 lines
1.3 KiB
Python

# Script to generate CharacterCategory.cxx from Python's Unicode data
# Should be run rarely when a Python with a new version of Unicode data is available.
# Should not be run with old versions of Python.
import codecs, os, platform, sys, unicodedata
from FileGenerator import Regenerate
def findCategories(filename):
with codecs.open(filename, "r", "UTF-8") as infile:
lines = [x.strip() for x in infile.readlines() if "\tcc" in x]
values = "".join(lines).replace(" ","").split(",")
print(values)
return [v[2:] for v in values]
def updateCharacterCategory(filename):
values = ["// Created with Python %s, Unicode %s" % (
platform.python_version(), unicodedata.unidata_version)]
category = unicodedata.category(chr(0))
startRange = 0
for ch in range(sys.maxunicode):
uch = chr(ch)
if unicodedata.category(uch) != category:
value = startRange * 32 + categories.index(category)
values.append("%d," % value)
category = unicodedata.category(uch)
startRange = ch
value = startRange * 32 + categories.index(category)
values.append("%d," % value)
Regenerate(filename, "//", values)
categories = findCategories("../lexlib/CharacterCategory.h")
updateCharacterCategory("../lexlib/CharacterCategory.cxx")