[or-cvs] r15717: First commit of tools to convert between wml and po formats. (in translation/trunk: . tools)

[ampleyfly@xxxxxxxx]

Author: ampleyfly
Date: 2008-07-07 06:23:53 -0400 (Mon, 07 Jul 2008)
New Revision: 15717

Added:
   translation/trunk/tools/
   translation/trunk/tools/po2wml.py
   translation/trunk/tools/wml2po.py
Log:
First commit of tools to convert between wml and po formats. Only handles paragraphs for now.

Added: translation/trunk/tools/po2wml.py
===================================================================
--- translation/trunk/tools/po2wml.py	                        (rev 0)
+++ translation/trunk/tools/po2wml.py	2008-07-07 10:23:53 UTC (rev 15717)
@@ -0,0 +1,90 @@
+import sys
+import getopt
+import re
+
+def usage():
+	print 'Usage: po2wml [-h | --help] [-i | --input] [-o | --output] [-t | --template]'
+
+def unescape(string):
+	return string.replace('\\n', '\n')
+
+def replace_string(indata, matchiter, string):
+	stack = []
+
+	while True:
+		try:
+			match = matchiter.next()
+			if match.group(1):
+				print 'paragraph closed'
+				start = stack.pop()
+				end = match.start()
+				return indata[:start] + string + indata[end:], matchiter
+			else:
+				print 'paragraph opened'
+				stack.append(match.end())
+		except StopIteration:
+			print 'couldn not find string to replace with %s' % string
+			break
+
+def parse(infile, outfile, template):
+	templatedata = template.read()
+	regex = re.compile('<(/?)p\s*>')
+	matchiter = regex.finditer(templatedata)
+
+	for line in infile:
+		if line.startswith(('#', 'msgid')):
+			continue
+		elif line.startswith('msgstr'):
+			string = line[7:].lstrip(' "').rstrip(' "\n')
+			string = unescape(string)
+			if string != '':
+				print 'trying to substitute for %s from %s' % (string, line)
+				templatedata, matchiter = replace_string(templatedata, matchiter, string)
+			else:
+				# ugly way of stepping past current paragraph
+				matchiter = replace_string(templatedata, matchiter, string)[1]
+	
+	outfile.write(templatedata)
+			
+
+def main(argv):
+	try:
+		opts, args = getopt.getopt(argv, 'hi:o:t:', ['help', 'input=', 'output=', 'template='])
+	except getopt.GetoptError:
+		usage()
+		sys.exit(2)
+
+	infile = outfile = template = None
+
+	for opt, arg in opts:
+		if opt in ('-h', '--help'):
+			usage()
+			sys.exit()
+		elif opt in ('-i', '--input'):
+			try:
+				infile = open(arg, 'r')
+			except IOError:
+				print 'The file %s could not be opened for reading, exiting...' % arg
+				sys.exit(-1)
+		elif opt in ('-o', '--output'):
+			try:
+				outfile = open(arg, 'w')
+			except IOError:
+				print 'The file %s could not be opened for writing, exiting...' % arg
+				sys.exit(-1)
+		elif opt in ('-t', '--template'):
+			try:
+				template = open(arg, 'r')
+			except IOError:
+				print 'The file %s could not be opened for reading, exiting...' % arg
+				sys.exit(-1)
+	
+	if infile == None: infile = sys.stdin
+	if outfile == None: outfile = sys.stdout
+	if template == None: template = sys.stdin
+
+	parse(infile, outfile, template)
+
+if __name__ == '__main__':
+	main(sys.argv[1:])
+

Added: translation/trunk/tools/wml2po.py
===================================================================
--- translation/trunk/tools/wml2po.py	                        (rev 0)
+++ translation/trunk/tools/wml2po.py	2008-07-07 10:23:53 UTC (rev 15717)
@@ -0,0 +1,75 @@
+import sys
+import getopt
+import re
+
+string_no = 1
+
+def usage():
+	print 'Usage: wml2po [-h | --help] [-i | --input] [-o | --output]'
+
+def escape(string):
+	return string.replace('\n', '\\n')
+
+def add_po_string(filename, outfile, string):
+	global string_no
+	outdata =  '#: %s:%d\n' % (filename, string_no)
+	outdata += 'msgid "%s"\n' % escape(string)
+	outdata += 'msgstr ""\n'
+	outdata += '\n'
+	outfile.write(outdata)
+	string_no = string_no + 1
+
+def parse(infile, outfile):
+	indata = infile.read()
+	regex = re.compile('<(/?)p\s*>')
+	matchiter = regex.finditer(indata)
+
+	stack = []
+
+	for match in matchiter:
+		if match.group(1):
+			#print 'paragraph closed'
+			start = stack.pop()
+			end = match.start()
+			contents = indata[start:end]
+			#print 'start %d, end %d, contents: %s' % (start, end, contents)
+
+			add_po_string(infile.name, outfile, contents)
+		else:
+			#print 'paragraph opened'
+			stack.append(match.end())
+
+def main(argv):
+	try:
+		opts, args = getopt.getopt(argv, 'hi:o:', ['help', 'input=', 'output='])
+	except getopt.GetoptError:
+		usage()
+		sys.exit(2)
+
+	infile = outfile = None
+
+	for opt, arg in opts:
+		if opt in ('-h', '--help'):
+			usage()
+			sys.exit()
+		elif opt in ('-i', '--input'):
+			try:
+				infile = open(arg, 'r')
+			except IOError:
+				print 'The file %s could not be opened for reading, exiting...' % arg
+				sys.exit(-1)
+		elif opt in ('-o', '--output'):
+			try:
+				outfile = open(arg, 'w')
+			except IOError:
+				print 'The file %s could not be opened for writing, exiting...' % arg
+				sys.exit(-1)
+
+	if infile == None: infile = sys.stdin 
+	if outfile == None: outfile = sys.stdout 
+
+	parse(infile, outfile)
+
+if __name__ == '__main__':
+	main(sys.argv[1:])
+