Stroke getter

getstrokes.py

#!/usr/bin/env python3

import sys
import binascii
import string
import codecs
import xml.etree.ElementTree as ET
import collections

stroketypes = {
	"31d0": "H",
	"31d1": "S",
	"31d2": "P",
	"31c4": "SW",
	"31c6": "HZG",
	"31df": "SWG"
}

def realord(s, pos = 0):
	if s == None: return None
	code = ord(s[pos])
	if code >= 0xD800 and code < 0xDC00:
		if (len(s) <= pos + 1):
			print("realord warning: missing surrogate character")
			return 0
		code2 = ord(s[pos + 1])
		if code2 >= 0xDC00 and code < 0xE000:
			code = 0x10000 + ((code - 0xD800) << 10) + (code2 - 0xDC00)
	return code

class Stroke:
	def __init__(self, attrs):
		self.id = attrs['id'].split(':')[1]
		self.type = attrs['{http://kanjivg.tagaini.net}type']

	def __repr__(self):
		if stroketypes.get(hex(realord(self.type))[2:]):
			return stroketypes[hex(realord(self.type))[2:]]
		return hex(self.type)[2:]

class Kanji:
	def __init__(self, char):
		root = getTree(char).getroot()

		self.parts = collections.OrderedDict()
		
		for child in root.find('*').findall('*'):
			if child.type == '{http://www.w3.org/2000/svg}path':


def getTree(char):
	return ET.parse('/sdcard/Kanji/kanji/' + hex(realord(char))[2:].zfill(5) + '.svg')

def printTree(char):
	ET.dump(getTree(char))

def printStrokes(char):
	tree = ET.parse('/sdcard/Kanji/kanji/' + hex(realord(char))[2:].zfill(5) + '.svg')
	root = tree.getroot()

	strokes = collections.OrderedDict()

	for child in root.findall('.//{http://www.w3.org/2000/svg}path'):
		stroke = Stroke(child.attrib)
		strokes[stroke.id] = Stroke(child.attrib)

	print(char + ':', end=' ')
	for key in strokes:
		print('\t', end='')
		print(strokes[key], end='')
		print('('+strokes[key].type+')')
	print()
	
if __name__ == '__main__':
 for char in sys.argv[1:]:
  printStrokes(char)