def __init__(self, charset):
self.encoder = getencoder(charset)
return
def usage():
print 'usage: htmldom.py [-d] [-b base_href] [-c charset_in] [-C codec_out] [url ...]'
sys.exit(2)
try:
(opts, args) = getopt.getopt(sys.argv[1:], 'dc:C:b:')
except getopt.GetoptError:
usage()
(charset_in, codec_out, base_href) = ('iso-8859-1', 'euc-jp', '')
for (k, v) in opts:
if k == '-d': debug += 1
elif k == '-c': charset_in = v
elif k == '-C': codec_out = v
elif k == '-b': base_href = v
encoder = getencoder(codec_out)
for url in args:
agent = agent.Agent(debug=sys.stdout)
(fp, content_type, charset) = agent.get(url)
stylesheet = ActiveStyleSheet(agent, base_href=(base_href or url))
root = parse(fp,
base_href=url,
charset=(charset or charset_in),
stylesheet=stylesheet)
fp.close()
stylesheet.dump()
validate(root, root)
if debug:
for (i, e) in root.walk():
sys.stdout.write(' ' * i + repr(e) + '\n')
else:
def __init__(self, charset): self.encoder = getencoder(charset) self.output_text = None return
def __init__(self, charset): self.encoder = getencoder(charset) return
if __name__ == '__main__':
import getopt, agent
def usage():
print 'usage: htmldom.py [-d] [-b base_href] [-c charset_in] [-C codec_out] [url ...]'
sys.exit(2)
try:
(opts, args) = getopt.getopt(sys.argv[1:], 'dc:C:b:')
except getopt.GetoptError:
usage()
(charset_in, codec_out, base_href) = ('iso-8859-1', 'euc-jp', '')
for (k, v) in opts:
if k == '-d': debug += 1
elif k == '-c': charset_in = v
elif k == '-C': codec_out = v
elif k == '-b': base_href = v
encoder = getencoder(codec_out)
for url in args:
agent = agent.Agent(debug=sys.stdout)
(fp, content_type, charset) = agent.get(url)
stylesheet = ActiveStyleSheet(agent, base_href=(base_href or url))
root = parse(fp, base_href=url, charset=(charset or charset_in), stylesheet=stylesheet)
fp.close()
stylesheet.dump()
validate(root, root)
if debug:
for (i,e) in root.walk():
sys.stdout.write(' '*i+repr(e)+'\n')
else:
for c in root.dump():
sys.stdout.write(encoder(c, 'xmlcharrefreplace')[0])
