def run(self):
""" Main entry point """
# Read command line
usage = "%prog [options] <url> <address> [<address2>] [<address...>]"
parser = OptionParser(usage=usage,
version=self.NAME + ' ' + self.VERSION)
parser.add_option("--http-user",
dest="http_user",
help="Username for HTTP POST authentication")
parser.add_option("--http-pass",
dest="http_pass",
help="Password for HTTP POST authentication")
parser.add_option("-s",
"--smtp",
dest="smtp",
help="SMTP server address. Default: localhost",
default='localhost')
parser.add_option("--smtp-user",
dest="smtp_user",
help="Username for SMTP authentication")
parser.add_option("--smtp-pass",
dest="smtp_pass",
help="Password for SMTP authentication")
parser.add_option("-c",
"--cc",
dest="cc",
help="Carbon Copy recipient")
parser.add_option("-f",
"--from",
dest="sender",
help="eMail sender. Default: emailer@localhost",
default="emailer@localhost")
parser.add_option(
"-j",
"--subject",
dest="subject",
help="eMail Subject. Default: MailWWW Autogenerated Mail",
default="MailWWW Autogenerated Mail")
parser.add_option("-n",
"--no-css",
dest="nocss",
help="Disable embedding of linked Style Sheets",
default=False,
action="store_true")
parser.add_option(
"-m",
"--multiple",
dest="multiple",
help=
"Send multiple emails: one for each recipient (Cc field is ignored)",
default=False,
action="store_true")
parser.add_option("-v",
"--verbose",
dest="verbose",
help="Show progress information",
default=False,
action="store_true")
(options, args) = parser.parse_args()
# Parse mandatory arguments
if len(args) < 2:
parser.error("unvalid number of arguments")
dest = []
i = 0
for a in args:
if i == 0:
url = a
else:
dest.append(a)
i += 1
# Parse optional arguments
http_user = options.http_user
http_pass = options.http_pass
cc = []
if options.cc:
cc.append(options.cc)
host = options.smtp
port = 25
user = options.smtp_user
pwd = options.smtp_pass
sender = options.sender
subject = options.subject
nocss = options.nocss
multiple = options.multiple
verbose = options.verbose
logging.basicConfig(level=logging.DEBUG if verbose else logging.INFO)
# Opens URL
logging.info('Fetching url %s', url)
data = None
if http_user or http_pass:
# Use POST authentication
data = urllib.urlencode({
'username': http_user,
'password': http_pass,
'login': True
})
f = urllib.urlopen(url, data)
html = f.read()
# Search for meta content-type tag, use this encoding when found
encre = re.compile(
r'<meta\s+http-equiv=(?:"|\')Content-Type(?:"|\')\s+content=(?:"|\')([^\'"]*)(?:"|\')\s*/>',
re.I | re.M)
match = encre.search(html)
if match:
encoding = self.__parseEncoding(match.group(1))
try:
html = unicode(html, encoding, errors='replace')
except LookupError as e:
encoding = self.__parseEncoding(f.headers['content-type'])
html = unicode(html, encoding, errors='replace')
else:
encoding = self.__parseEncoding(f.headers['content-type'])
html = unicode(html, encoding, errors='replace')
logging.info('Detected charset: %s', encoding)
f.close()
# Retrieve linked style sheets
if not nocss:
logging.info('Fetching Style Sheets...')
parser = CSSLister(url)
parser.feed(html)
parser.close()
for search, replace in parser.get_replacements():
html = html.replace(search, replace, 1)
# Prepare mail
msg = MIMEMultipart()
msg['Date'] = formatdate(localtime=True)
msg['Message-ID'] = make_msgid('emailer')
msg['Subject'] = subject
msg['From'] = sender
if cc and not multiple:
msg['Cc'] = ', '.join(cc)
msg.preamble = 'This is a milti-part message in MIME format.'
txt = MIMEText(html.encode('utf-8'), 'html', 'utf-8')
msg.attach(txt)
if not multiple:
msg['To'] = ', '.join(dest)
# Sends message
smtp = smtplib.SMTP()
smtp.connect(host, port)
if user:
smtp.login(user, pwd)
if multiple:
for d in dest:
del msg['To']
msg['To'] = d
logging.info('Sending mail to: %s', d)
smtp.sendmail(sender, d, msg.as_string())
else:
logging.info('Sending mail to: %s, Cc: %s', dest, cc)
smtp.sendmail(sender, dest + cc, msg.as_string())
smtp.quit()
options.path = '/'+options.path
conn = httplib.HTTPConnection("localhost:29392")
conn.request(method,options.action+options.path)
resp = conn.getresponse()
data = resp.read()
if options.action == '/download':
print sys.getsizeof(data)
exit()
parser = ET.XMLTreeBuilder()
parser.feed(data)
tree = ET.ElementTree(parser.close())
root = tree.getroot()
#print root.text
ET.dump(tree)
if options.action == '/list' and method == "GET":
print "File listing: "
file_listing(options.path, root)
if options.action == '/tagdata' and method == "GET":
print "Tag Data: "
tag_data("/", root, options)
print "\n\n\f"
except IOError:
sys.exit(0) # broken stdout => broken pipe
except Exception as e:
sys.stderr.write("\nError extracting article: " + str(e) + "\n")
if __name__ == "__main__":
try:
parser = OptionParser(usage="usage: %s [-d output-directory] <stdin>" % sys.argv[0])
parser.add_option("-d",
action="store", type="string", dest="directory",
help="directory where to store the articles")
(options, args) = parser.parse_args()
if len(args) > 0:
print parser.usage
exit(1)
htmlparser = HTMLParser.HTMLParser()
if options.directory != None:
do_article = lambda a: save_article(htmlparser, a, options.directory)
else:
do_article = lambda a: print_article(htmlparser, a)
parser = WikiParser(do_article)
parser.process()
parser.close()
except KeyboardInterrupt:
sys.stderr.write("\n\nCancelled. Partial results may have been generated.\n")
exit(1)
def run(self):
""" Main entry point """
# Read command line
usage = "%prog [options] <url> <address> [<address2>] [<address...>]"
parser = OptionParser(usage=usage, version=self.NAME + ' ' + self.VERSION)
parser.add_option("--http-user", dest="http_user",
help="Username for HTTP POST authentication")
parser.add_option("--http-pass", dest="http_pass",
help="Password for HTTP POST authentication")
parser.add_option("-s", "--smtp", dest="smtp",
help="SMTP server address. Default: localhost",
default='localhost')
parser.add_option("--smtp-user", dest="smtp_user",
help="Username for SMTP authentication")
parser.add_option("--smtp-pass", dest="smtp_pass",
help="Password for SMTP authentication")
parser.add_option("-c", "--cc", dest="cc",
help="Carbon Copy recipient")
parser.add_option("-f", "--from", dest="sender",
help="eMail sender. Default: emailer@localhost",
default="emailer@localhost")
parser.add_option("-j", "--subject", dest="subject",
help="eMail Subject. Default: MailWWW Autogenerated Mail",
default="MailWWW Autogenerated Mail")
parser.add_option("-n", "--no-css", dest="nocss",
help="Disable embedding of linked Style Sheets",
default=False, action="store_true")
parser.add_option("-m", "--multiple", dest="multiple",
help="Send multiple emails: one for each recipient (Cc field is ignored)",
default=False, action="store_true")
parser.add_option("-v", "--verbose", dest="verbose",
help="Show progress information",
default=False, action="store_true")
(options, args) = parser.parse_args()
# Parse mandatory arguments
if len(args) < 2:
parser.error("unvalid number of arguments")
dest = []
i = 0
for a in args:
if i == 0:
url = a
else:
dest.append(a)
i += 1
# Parse optional arguments
http_user = options.http_user
http_pass = options.http_pass
cc = []
if options.cc:
cc.append(options.cc)
host = options.smtp
port = 25
user = options.smtp_user
pwd = options.smtp_pass
sender = options.sender
subject = options.subject
nocss = options.nocss
multiple = options.multiple
verbose = options.verbose
logging.basicConfig(level=logging.DEBUG if verbose else logging.INFO)
# Opens URL
logging.info('Fetching url %s', url)
data = None
if http_user or http_pass:
# Use POST authentication
data = urllib.urlencode({ 'username': http_user, 'password': http_pass, 'login': True })
f = urllib.urlopen(url, data)
html = f.read()
# Search for meta content-type tag, use this encoding when found
encre = re.compile(r'<meta\s+http-equiv=(?:"|\')Content-Type(?:"|\')\s+content=(?:"|\')([^\'"]*)(?:"|\')\s*/>',
re.I | re.M)
match = encre.search(html)
if match:
encoding = self.__parseEncoding(match.group(1))
try:
html = unicode(html, encoding, errors='replace')
except LookupError as e:
encoding = self.__parseEncoding(f.headers['content-type'])
html = unicode(html, encoding, errors='replace')
else:
encoding = self.__parseEncoding(f.headers['content-type'])
html = unicode(html, encoding, errors='replace')
logging.info('Detected charset: %s', encoding)
f.close()
# Retrieve linked style sheets
if not nocss:
logging.info('Fetching Style Sheets...')
parser = CSSLister(url)
parser.feed(html)
parser.close()
for search, replace in parser.get_replacements():
html = html.replace(search, replace, 1)
# Prepare mail
msg = MIMEMultipart()
msg['Date'] = formatdate(localtime=True)
msg['Message-ID'] = make_msgid('emailer')
msg['Subject'] = subject
msg['From'] = sender
if cc and not multiple:
msg['Cc'] = ', '.join(cc)
msg.preamble = 'This is a milti-part message in MIME format.'
txt = MIMEText(html.encode('utf-8'), 'html', 'utf-8')
msg.attach(txt)
if not multiple:
msg['To'] = ', '.join(dest)
# Sends message
smtp = smtplib.SMTP()
smtp.connect(host, port)
if user:
smtp.login(user, pwd)
if multiple:
for d in dest:
del msg['To']
msg['To'] = d
logging.info('Sending mail to: %s', d)
smtp.sendmail(sender, d, msg.as_string())
else:
logging.info('Sending mail to: %s, Cc: %s', dest, cc)
smtp.sendmail(sender, dest+cc, msg.as_string())
smtp.quit()
