linkchecker/linkcheck/checker/httpurl.py
2010-11-21 19:50:51 +01:00

719 lines
30 KiB
Python

# -*- coding: iso-8859-1 -*-
# Copyright (C) 2000-2010 Bastian Kleineidam
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License along
# with this program; if not, write to the Free Software Foundation, Inc.,
# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
"""
Handle http links.
"""
import urlparse
import urllib
import re
import errno
import zlib
import socket
from cStringIO import StringIO
import Cookie
from .. import (log, LOG_CHECK, gzip2 as gzip, strformat, url as urlutil,
httplib2 as httplib, LinkCheckerError, configuration)
from . import (internpaturl, proxysupport, httpheaders as headers, urlbase,
get_url_from)
# import warnings
from .const import WARN_HTTP_ROBOTS_DENIED, \
WARN_HTTP_WRONG_REDIRECT, WARN_HTTP_MOVED_PERMANENT, \
WARN_HTTP_EMPTY_CONTENT, WARN_HTTP_COOKIE_STORE_ERROR, \
WARN_HTTP_DECOMPRESS_ERROR, WARN_HTTP_UNSUPPORTED_ENCODING, \
WARN_HTTP_AUTH_UNKNOWN
# helper alias
unicode_safe = strformat.unicode_safe
supportHttps = hasattr(httplib, "HTTPSConnection")
_supported_encodings = ('gzip', 'x-gzip', 'deflate')
# Amazon blocks all HEAD requests
_is_amazon = re.compile(r'^www\.amazon\.(com|de|ca|fr|co\.(uk|jp))').search
class HttpUrl (internpaturl.InternPatternUrl, proxysupport.ProxySupport):
"""
Url link with http scheme.
"""
def reset (self):
"""
Initialize HTTP specific variables.
"""
super(HttpUrl, self).reset()
self.max_redirects = 5
self.has301status = False
# flag if check had to fallback from HEAD to GET method
self.fallback_get = False
# flag if connection is persistent
self.persistent = False
# URLs seen through 301/302 redirections
self.aliases = []
# initialize check data
self.headers = None
self.auth = None
self.cookies = []
# temporary data filled when reading redirections
self._data = None
# flag indicating connection reuse
self.reused_connection = False
# flag telling if GET method is allowed; determined by robots.txt
self.method_get_allowed = True
def allows_robots (self, url):
"""
Fetch and parse the robots.txt of given url. Checks if LinkChecker
can get the requested resource content. HEAD requests however are
still allowed.
@param url: the url to be requested
@type url: string
@return: True if access is granted, otherwise False
@rtype: bool
"""
roboturl = self.get_robots_txt_url()
user, password = self.get_user_password()
rb = self.aggregate.robots_txt
callback = self.aggregate.connections.host_wait
return rb.allows_url(roboturl, url, self.proxy, user, password,
callback=callback)
def add_size_info (self):
"""Get size of URL content from HTTP header."""
if self.headers and "Content-Length" in self.headers and \
"Transfer-Encoding" not in self.headers:
# Note that content-encoding causes size differences since
# the content data is always decoded.
try:
self.size = int(self.headers["Content-Length"])
if self.dlsize == -1:
self.dlsize = self.size
except (ValueError, OverflowError):
pass
else:
self.size = -1
def check_connection (self):
"""
Check a URL with HTTP protocol.
Here is an excerpt from RFC 1945 with common response codes:
The first digit of the Status-Code defines the class of response. The
last two digits do not have any categorization role. There are 5
values for the first digit:
- 1xx: Informational - Not used, but reserved for future use
- 2xx: Success - The action was successfully received,
understood, and accepted.
- 3xx: Redirection - Further action must be taken in order to
complete the request
- 4xx: Client Error - The request contains bad syntax or cannot
be fulfilled
- 5xx: Server Error - The server failed to fulfill an apparently
valid request
"""
# set the proxy, so a 407 status after this is an error
self.set_proxy(self.aggregate.config["proxy"].get(self.scheme))
# check robots.txt
if not self.allows_robots(self.url):
# remove all previously stored results
self.add_warning(
_("Access denied by robots.txt, skipping content checks."),
tag=WARN_HTTP_ROBOTS_DENIED)
self.method_get_allowed = False
# first try with HEAD
self.method = "HEAD"
# check for amazon server quirk
if _is_amazon(self.urlparts[1]):
self.add_info(_("Amazon servers block HTTP HEAD requests."))
if self.method_get_allowed:
self.add_info(_("Using GET method for Amazon server."))
self.method = "GET"
# check the http connection
response = self.check_http_connection()
if self.headers and "Server" in self.headers:
server = self.headers['Server']
else:
server = _("unknown")
if self.fallback_get:
self.add_info(_("Server `%(name)s' did not support HEAD request; "
"a GET request was used instead.") %
{"name": server})
# redirections might have changed the URL
self.url = urlparse.urlunsplit(self.urlparts)
# check response
if response:
self.check_response(response)
response.close()
def check_http_connection (self):
"""
Check HTTP connection and return get response and a flag
if the check algorithm had to fall back to the GET method.
@return: response or None if url is already handled
@rtype: HttpResponse or None
"""
response = None
while True:
if response is not None:
response.close()
try:
response = self._try_http_response()
except httplib.BadStatusLine, msg:
# some servers send empty HEAD replies
if self.method == "HEAD" and self.method_get_allowed:
log.debug(LOG_CHECK, "Bad status line %r: falling back to GET", msg)
self.method = "GET"
self.aliases = []
self.fallback_get = True
continue
raise
except socket.error, msg:
# some servers reset the connection on HEAD requests
if self.method == "HEAD" and self.method_get_allowed and \
msg[0] == errno.ECONNRESET:
self.method = "GET"
self.aliases = []
self.fallback_get = True
continue
raise
if response.reason:
response.reason = unicode_safe(response.reason)
log.debug(LOG_CHECK,
"Response: %s %s", response.status, response.reason)
log.debug(LOG_CHECK, "Headers: %s", self.headers)
# proxy enforcement (overrides standard proxy)
if response.status == 305 and self.headers:
oldproxy = (self.proxy, self.proxyauth)
newproxy = self.headers.getheader("Location")
self.add_info(_("Enforced proxy `%(name)s'.") %
{"name": newproxy})
self.set_proxy(newproxy)
if not self.proxy:
self.set_result(
_("Enforced proxy `%(name)s' ignored, aborting.") %
{"name": newproxy},
valid=False)
return response
response.close()
response = self._try_http_response()
# restore old proxy settings
self.proxy, self.proxyauth = oldproxy
try:
tries, response = self.follow_redirections(response)
except httplib.BadStatusLine, msg:
# some servers send empty HEAD replies
if self.method == "HEAD" and self.method_get_allowed:
log.debug(LOG_CHECK, "Bad status line %r: falling back to GET", msg)
self.method = "GET"
self.aliases = []
self.fallback_get = True
continue
raise
if tries == -1:
log.debug(LOG_CHECK, "already handled")
response.close()
self.do_check_content = False
return None
if tries >= self.max_redirects:
if self.method == "HEAD" and self.method_get_allowed:
# Microsoft servers tend to recurse HEAD requests
self.method = "GET"
self.aliases = []
self.fallback_get = True
continue
self.set_result(_("more than %d redirections, aborting") %
self.max_redirects, valid=False)
return response
# user authentication
if response.status == 401:
authenticate = self.headers.get('WWW-Authenticate')
if not authenticate or not authenticate.startswith("Basic"):
# LinkChecker only supports Basic authorization
args = {"auth": authenticate}
self.add_warning(
_("Unsupported HTTP authentication `%(auth)s', " \
"only `Basic' authentication is supported.") % args,
tag=WARN_HTTP_AUTH_UNKNOWN)
return
if not self.auth:
import base64
_user, _password = self.get_user_password()
self.auth = "Basic " + \
base64.encodestring("%s:%s" % (_user, _password))
log.debug(LOG_CHECK,
"Authentication %s/%s", _user, _password)
continue
elif response.status >= 400:
# retry with GET (but do not set fallback flag)
if self.method == "HEAD" and self.method_get_allowed:
self.method = "GET"
self.aliases = []
continue
elif self.headers and self.method == "HEAD" and self.method_get_allowed:
# test for HEAD support
mime = headers.get_content_type(self.headers)
poweredby = self.headers.get('X-Powered-By', '')
server = self.headers.get('Server', '')
if mime in ('application/octet-stream', 'text/plain') and \
(poweredby.startswith('Zope') or server.startswith('Zope')):
# Zope server could not get Content-Type with HEAD
self.method = "GET"
self.aliases = []
self.fallback_get = True
continue
break
return response
def follow_redirections (self, response, set_result=True):
"""
Follow all redirections of http response.
"""
log.debug(LOG_CHECK, "follow all redirections")
redirected = self.url
tries = 0
while response.status in [301, 302] and self.headers and \
tries < self.max_redirects:
newurl = self.headers.getheader("Location",
self.headers.getheader("Uri", ""))
# make new url absolute and unicode
newurl = unicode_safe(newurl)
newurl = urlparse.urljoin(redirected, newurl)
log.debug(LOG_CHECK, "Redirected to %r", newurl)
self.add_info(_("Redirected to `%(url)s'.") % {'url': newurl})
# norm base url - can raise UnicodeError from url.idna_encode()
redirected, is_idn = urlbase.url_norm(newurl)
if is_idn:
pass # XXX warn about idn use
log.debug(LOG_CHECK, "Norm redirected to %r", redirected)
urlparts = strformat.url_unicode_split(redirected)
if urlparts[1] != self.urlparts[1]:
# check extern filter again
self.set_extern(redirected)
if self.extern[0] and self.extern[1]:
if set_result:
self.check301status(response)
self.add_info(
_("The redirected URL is outside of the domain "
"filter, checked only syntax."))
self.set_result(u"filtered")
return -1, response
# check robots.txt allowance again
if not self.allows_robots(redirected):
if set_result:
self.add_warning(
_("Access to redirected URL denied by robots.txt, "
"checked only syntax."),
tag=WARN_HTTP_ROBOTS_DENIED)
self.set_result(u"syntax OK")
return -1, response
# see about recursive redirect
all_seen = [self.cache_url_key] + self.aliases
if redirected in all_seen:
if self.method == "HEAD" and self.method_get_allowed:
# Microsoft servers tend to recurse HEAD requests
# fall back to the original url and use GET
return self.max_redirects, response
recursion = all_seen + [redirected]
if set_result:
self.set_result(
_("recursive redirection encountered:\n %(urls)s") %
{"urls": "\n => ".join(recursion)}, valid=False)
return -1, response
if urlparts[0] == self.scheme or urlparts[0] in ('http', 'https'):
# remember redirected url as alias
self.aliases.append(redirected)
else:
# in case of changed scheme make new URL object
newobj = get_url_from(
redirected, self.recursion_level, self.aggregate,
parent_url=self.parent_url, base_ref=self.base_ref,
line=self.line, column=self.column, name=self.name)
if set_result:
self.add_warning(
_("Redirection to URL `%(newurl)s' with different scheme"
" found; the original URL was `%(url)s'.") %
{"url": self.url, "newurl": newobj.url},
tag=WARN_HTTP_WRONG_REDIRECT)
self.set_result(u"syntax OK")
# append new object to queue
self.aggregate.urlqueue.put(newobj)
# pretend to be finished and logged
return -1, response
# note: urlparts has to be a list
self.urlparts = urlparts
if set_result:
self.check301status(response)
# check cache again on the changed URL
if self.aggregate.urlqueue.checked_redirect(redirected, self):
return -1, response
# new response data
response.close()
response = self._try_http_response()
tries += 1
return tries, response
def check301status (self, response):
"""If response page has been permanently moved add a warning."""
if response.status == 301 and not self.has301status:
self.add_warning(_("HTTP 301 (moved permanent) encountered: you"
" should update this link."),
tag=WARN_HTTP_MOVED_PERMANENT)
self.has301status = True
def get_alias_cache_data (self):
"""
Return all data values that should be put in the cache,
minus redirection warnings.
"""
data = self.get_cache_data()
data["warnings"] = [
x for x in self.warnings if x[0] != "http-moved-permanent"]
data["info"] = self.info
return data
def check_response (self, response):
"""Check final result and log it."""
if response.status >= 400:
self.set_result(u"%r %s" % (response.status, response.reason),
valid=False)
else:
if response.status == 204:
# no content
self.add_warning(unicode_safe(response.reason),
tag=WARN_HTTP_EMPTY_CONTENT)
# store cookies for valid links
if self.aggregate.config['storecookies']:
for c in self.cookies:
self.add_info(_("Sent cookie: %(cookie)s.") %
{"cookie": c})
try:
out = self.aggregate.cookies.add(self.headers,
self.urlparts[0],
self.urlparts[1],
self.urlparts[2])
except Cookie.CookieError, msg:
self.add_warning(_("Could not store cookies: %(msg)s.") %
{'msg': str(msg)},
tag=WARN_HTTP_COOKIE_STORE_ERROR)
if response.status >= 200:
self.set_result(u"%r %s" % (response.status, response.reason))
else:
self.set_result(u"OK")
modified = self.headers.get('Last-Modified', '')
if modified:
self.add_info(_("Last modified %(date)s.") % {"date": modified})
def _try_http_response (self):
"""Try to get a HTTP response object. For reused persistent
connections that the server closed unexpected, a new connection
will be opened.
"""
try:
return self._get_http_response()
except socket.error, msg:
if msg.args[0] == 32 and self.reused_connection:
# server closed persistent connection - retry
log.debug(LOG_CHECK, "Server closed connection: retry")
self.persistent = False
return self._get_http_response()
raise
except httplib.BadStatusLine, msg:
if not msg and self.reused_connection:
# server closed connection - retry
log.debug(LOG_CHECK, "Empty status line: retry")
self.persistent = False
return self._get_http_response()
raise
def _get_http_response (self):
"""
Send HTTP request and get response object.
"""
if self.proxy:
host = self.proxy
scheme = self.proxytype
else:
host = self.urlparts[1]
scheme = self.urlparts[0]
log.debug(LOG_CHECK, "Connecting to %r", host)
# close/release a previous connection
self.close_connection()
self.url_connection = self.get_http_object(host, scheme)
# the anchor fragment is not part of a HTTP URL, see
# http://tools.ietf.org/html/rfc2616#section-3.2.2
anchor = ''
if self.proxy:
path = urlparse.urlunsplit((self.urlparts[0], self.urlparts[1],
self.urlparts[2], self.urlparts[3], anchor))
else:
path = urlparse.urlunsplit(('', '', self.urlparts[2],
self.urlparts[3], anchor))
self.url_connection.putrequest(self.method, path, skip_host=True,
skip_accept_encoding=True)
# be sure to use the original host as header even for proxies
self.url_connection.putheader("Host", self.urlparts[1])
# userinfo is from http://user@pass:host/
if self.userinfo:
self.url_connection.putheader("Authorization", self.userinfo)
# auth is the -u and -p configuration options
elif self.auth:
self.url_connection.putheader("Authorization", self.auth)
if self.proxyauth:
self.url_connection.putheader("Proxy-Authorization",
self.proxyauth)
if (self.parent_url and
self.parent_url.startswith(('http://', 'https://'))):
self.url_connection.putheader("Referer", self.parent_url)
self.url_connection.putheader("User-Agent", configuration.UserAgent)
self.url_connection.putheader("Accept-Encoding",
"gzip;q=1.0, deflate;q=0.9, identity;q=0.5")
if self.aggregate.config['sendcookies']:
scheme = self.urlparts[0]
host = self.urlparts[1]
port = urlutil.default_ports.get(scheme, 80)
host, port = urllib.splitnport(host, port)
path = self.urlparts[2]
self.cookies = self.aggregate.cookies.get(scheme, host, port, path)
for c in self.cookies:
name = c.client_header_name()
value = c.client_header_value()
self.url_connection.putheader(name, value)
self.url_connection.endheaders()
response = self.url_connection.getresponse(True)
self.timeout = headers.http_timeout(response)
self.headers = response.msg
self.persistent = not response.will_close
if self.persistent and self.method == "HEAD":
# Some servers send page content after a HEAD request,
# but only after making the *next* request. This breaks
# protocol synchronisation. Workaround here is to close
# the connection after HEAD.
# Example: http://www.empleo.gob.mx (Apache/1.3.33 (Unix) mod_jk)
self.persistent = False
if self.persistent and (self.method == "GET" or
self.headers.getheader("Content-Length") != "0"):
# always read content from persistent connections
self._read_content(response)
assert not response.will_close
# If possible, use official W3C HTTP response name
if response.status in httplib.responses:
response.reason = httplib.responses[response.status]
return response
def get_http_object (self, host, scheme):
"""
Open a HTTP connection.
@param host: the host to connect to
@type host: string of the form <host>[:<port>]
@param scheme: 'http' or 'https'
@type scheme: string
@return: open HTTP(S) connection
@rtype: httplib.HTTP(S)Connection
"""
_user, _password = self.get_user_password()
key = (scheme, self.urlparts[1], _user, _password)
conn = self.aggregate.connections.get(key)
if conn is not None:
log.debug(LOG_CHECK, "reuse cached HTTP(S) connection %s", conn)
self.reused_connection = True
return conn
self.aggregate.connections.wait_for_host(host)
if scheme == "http":
h = httplib.HTTPConnection(host)
elif scheme == "https" and supportHttps:
h = httplib.HTTPSConnection(host)
else:
msg = _("Unsupported HTTP url scheme `%(scheme)s'") % {"scheme": scheme}
raise LinkCheckerError(msg)
if log.is_debug(LOG_CHECK):
h.set_debuglevel(1)
h.connect()
return h
def read_content (self):
"""Get content of the URL target. The content data is cached after
the first call to this method.
@return: URL content, decompressed and decoded
@rtype: string
"""
assert self.method_get_allowed, 'unallowed content read'
self.method = "GET"
response = self._try_http_response()
response = self.follow_redirections(response, set_result=False)[1]
self.headers = response.msg
# Re-read size info, since the GET request result could be different
# than a former HEAD request.
self.add_size_info()
if self._data is None:
self._read_content(response)
data, size = self._data, self._size
self._data = self._size = None
return data, size
def _read_content (self, response):
"""Read URL contents and store then in self._data.
This way, the method can be called by other functions than
read_content()"""
data = response.read()
self._size = len(data)
encoding = headers.get_content_encoding(self.headers)
if encoding in _supported_encodings:
try:
if encoding == 'deflate':
f = StringIO(zlib.decompress(data))
else:
f = gzip.GzipFile('', 'rb', 9, StringIO(data))
except zlib.error, msg:
self.add_warning(_("Decompress error %(err)s") %
{"err": str(msg)},
tag=WARN_HTTP_DECOMPRESS_ERROR)
f = StringIO(data)
try:
data = f.read()
finally:
f.close()
# store temporary data
self._data = data
def encoding_supported (self):
"""Check if page encoding is supported."""
encoding = headers.get_content_encoding(self.headers)
if encoding and encoding not in _supported_encodings and \
encoding != 'identity':
self.add_warning(_("Unsupported content encoding `%(encoding)s'.") %
{"encoding": encoding},
tag=WARN_HTTP_UNSUPPORTED_ENCODING)
return False
return True
def set_title_from_content (self):
"""Check if it's allowed to read content before execution."""
if self.method_get_allowed:
super(HttpUrl, self).set_title_from_content()
def get_anchors (self):
"""Check if it's allowed to read content before execution."""
if self.method_get_allowed:
super(HttpUrl, self).get_anchors()
def content_allows_robots (self):
"""Check if it's allowed to read content before execution."""
if not self.method_get_allowed:
return False
return super(HttpUrl, self).content_allows_robots()
def check_warningregex (self):
"""Check if it's allowed to read content before execution."""
if self.method_get_allowed:
super(HttpUrl, self).check_warningregex()
def is_html (self):
"""
See if this URL points to a HTML file by looking at the
Content-Type header, file extension and file content.
@return: True if URL points to HTML file
@rtype: bool
"""
if not (self.valid and self.headers):
return False
mime = headers.get_content_type(self.headers)
if self.ContentMimetypes.get(mime) != "html":
return False
return self.encoding_supported()
def is_css (self):
"""Return True iff content of this url is CSS stylesheet."""
if not (self.valid and self.headers):
return False
mime = headers.get_content_type(self.headers)
if self.ContentMimetypes.get(mime) != "css":
return False
return self.encoding_supported()
def is_http (self):
"""
This is a HTTP file.
@return: True
@rtype: bool
"""
return True
def is_parseable (self):
"""
Check if content is parseable for recursion.
@return: True if content is parseable
@rtype: bool
"""
if not (self.valid and self.headers):
return False
if headers.get_content_type(self.headers) not in self.ContentMimetypes:
return False
return self.encoding_supported()
def parse_url (self):
"""
Parse file contents for new links to check.
"""
ctype = headers.get_content_type(self.headers)
if self.is_html():
self.parse_html()
elif self.is_css():
self.parse_css()
elif ctype == "application/x-shockwave-flash":
self.parse_swf()
elif ctype == "application/msword":
self.parse_word()
def get_robots_txt_url (self):
"""
Get the according robots.txt URL for this URL.
@return: robots.txt URL
@rtype: string
"""
return "%s://%s/robots.txt" % tuple(self.urlparts[0:2])
def close_connection (self):
"""
If connection is persistent, add it to the connection pool.
Else close the connection. Errors on closing are ignored.
"""
if self.url_connection is None:
# no connection is open
return
# add to cached connections
_user, _password = self.get_user_password()
key = ("http", self.urlparts[1], _user, _password)
if self.persistent and self.url_connection.is_idle():
self.aggregate.connections.add(
key, self.url_connection, self.timeout)
else:
try:
self.url_connection.close()
except Exception:
# ignore close errors
pass
self.url_connection = None