1 files changed, 97 insertions, 0 deletions
diff --git a/po/htm2po.py b/po/htm2po.py
new file mode 100755
index 0000000..683240e
--- /dev/null
+++ b/po/htm2po.py
@@ -0,0 +1,97 @@
+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+import sys, re
+from BeautifulSoup import BeautifulSoup
+
+help = """
+arg 1 : input file
+arg 2 : the locale field which you wanna extract
+arg 3 : output po file
+"""
+
+DESIGN_ID = "Design ID"
+dString = {}
+
+def findindexof( word, tds ):
+	for i, td in enumerate(tds):
+		try:
+			if td.contents[0]==word:
+				return i
+		except IndexError:
+			pass
+	return -1
+
+if __name__=="__main__":
+	if len(sys.argv)>3:
+		in_file = sys.argv[1]
+		locale = sys.argv[2]
+		po_file = sys.argv[3]
+	else:
+		print help
+		sys.exit()
+
+	print 'input file :', in_file
+	print 'locale :', locale
+	print 'po file :', po_file
+
+	f = open( in_file, 'rt' )
+	soup = BeautifulSoup( f.read() )
+	f.close()
+
+	for i, tr in enumerate(soup('tr')):
+		tds = tr('td')
+		if i==0: 
+			iID = findindexof( DESIGN_ID, tds )
+			iLocale = findindexof( locale, tds )
+			print 'index of design id :',iID, ", index of",locale,":", iLocale
+			if iID<0 or iLocale<0: 
+				print 'index failed'
+				sys.exit()
+			continue
+
+		bPass = True;
+		for j,td in enumerate(tds):
+			try:
+				if j==iID: ID = td.contents[0]	
+				elif j==iLocale: szLocale = td.contents[0]
+			except IndexError:
+				bPass = False;
+				pass
+
+		if bPass: 
+			dString[ID] = szLocale
+			#print i, '\t', ID, '\t' ,szLocale
+
+	#modify po file
+	f = open( po_file, 'rt' )
+	po = f.read().decode('utf-8')
+	f.close()
+
+	for k, v in dString.items():
+		#find key
+		msg = 'msgid "'+ k + '"'
+		sp = po.find( msg )
+
+		#if yes, replace
+		if sp>-1:
+			lines = po.split('\n')
+			for i, line in enumerate(lines):
+				if line.find( msg )>-1:
+					#replace next line
+					print 'replace :',k
+					v = v.replace('\r', '')			#remove linefeed
+					v = v.replace('\n ', '\\n') #remove carriage return
+					lines[i+1] = 'msgstr "' + v + '"'
+			po = "\n".join( lines )
+
+		#if no, append
+		#if sp<0:	
+		#	print 'append', k
+		#	po += 'msgid "' + k + '"\n'
+		#	po += 'msgstr "' + v + '"\n\n' 
+
+	newfile = po_file+'.new'
+	f = open( newfile , 'wt' )
+	f.write( po.encode('utf-8') )
+	f. close()
+	print newfile, 'is created'