def fetch(self):
"""Attempt to fetch the url (if isyanked is not True) and fill in link
attributes (based on isinternal)."""
# fully ignore links that should not be feteched
if self.isyanked:
debugio.info(' %s' % self.url)
debugio.info(' ' + self.isyanked)
return
# see if we can import the proper module for this scheme
schememodule = schemes.get_schememodule(self.scheme)
if schememodule is None:
self.isyanked = 'unsupported scheme (' + self.scheme + ')'
self._ischanged = True
debugio.info(' %s' % self.url)
debugio.info(' ' + self.isyanked)
return
debugio.info(' %s' % self.url)
content = schememodule.fetch(self, parsers.get_mimetypes())
self.isfetched = True
self._ischanged = True
# skip parsing of content if we were returned nothing
if content is None:
return
# find a parser for the content-type
parsermodule = parsers.get_parsermodule(self.mimetype)
if parsermodule is None:
debugio.debug('crawler.Link.fetch(): unsupported content-type: %s' % self.mimetype)
return
# parse the content
parsermodule.parse(content, self)
def check_for_whole_start_tag(self, i):
"""Override to catch assertion exception."""
try:
return HTMLParser.HTMLParser.check_for_whole_start_tag(self, i)
except AssertionError:
debugio.debug('parsers.html._MyHTMLParser.check_for_whole_start_tag(): caught assertion error')
return None
def _get_robotparser(self, link):
"""Return the proper robots parser for the given url or None if one
cannot be constructed. Robot parsers are cached per scheme and
netloc."""
# only some schemes have a meaningful robots.txt file
if link.scheme != 'http' and link.scheme != 'https':
debugio.debug(
'crawler._get_robotparser() called with unsupported scheme (%s)'
% link.scheme)
return None
# split out the key part of the url
location = urlparse.urlunsplit((link.scheme, link.netloc, '', '', ''))
# try to create a new robotparser if we don't already have one
if not self._robotparsers.has_key(location):
import httplib
debugio.info(' getting robots.txt for %s' % location)
self._robotparsers[location] = None
try:
rp = robotparser.RobotFileParser()
rp.set_url(
urlparse.urlunsplit(
(link.scheme, link.netloc, '/robots.txt', '', '')))
rp.read()
self._robotparsers[location] = rp
except (TypeError, IOError, httplib.HTTPException):
# ignore any problems setting up robot parser
pass
return self._robotparsers[location]
def _get_robotparser(self, link):
"""Return the proper robots parser for the given url or None if one
cannot be constructed. Robot parsers are cached per scheme and
netloc."""
# only some schemes have a meaningful robots.txt file
if link.scheme != 'http' and link.scheme != 'https':
debugio.debug('crawler._get_robotparser() called with unsupported scheme (%s)' % link.scheme)
return None
# split out the key part of the url
location = urlparse.urlunsplit((link.scheme, link.netloc, '', '', ''))
# try to create a new robotparser if we don't already have one
if not self._robotparsers.has_key(location):
import httplib
debugio.info(' getting robots.txt for %s' % location)
self._robotparsers[location] = None
try:
rp = robotparser.RobotFileParser()
rp.set_url(urlparse.urlunsplit(
(link.scheme, link.netloc, '/robots.txt', '', '') ))
rp.read()
self._robotparsers[location] = rp
except (TypeError, IOError, httplib.HTTPException):
# ignore any problems setting up robot parser
pass
return self._robotparsers[location]
def fetch(self):
"""Attempt to fetch the url (if isyanked is not True) and fill in link
attributes (based on isinternal)."""
# fully ignore links that should not be feteched
if self.isyanked:
debugio.info(' %s' % self.url)
debugio.info(' ' + self.isyanked)
return
# see if we can import the proper module for this scheme
schememodule = schemes.get_schememodule(self.scheme)
if schememodule is None:
self.isyanked = 'unsupported scheme (' + self.scheme + ')'
self._ischanged = True
debugio.info(' %s' % self.url)
debugio.info(' ' + self.isyanked)
return
debugio.info(' %s' % self.url)
content = schememodule.fetch(self, parsers.get_mimetypes())
self.isfetched = True
self._ischanged = True
# skip parsing of content if we were returned nothing
if content is None:
return
# find a parser for the content-type
parsermodule = parsers.get_parsermodule(self.mimetype)
if parsermodule is None:
debugio.debug(
'crawler.Link.fetch(): unsupported content-type: %s' %
self.mimetype)
return
# parse the content
parsermodule.parse(content, self)
def crawl(self, serfp=None):
"""Crawl the website based on the urls specified with
add_internal(). If the serialization file pointer
is specified the crawler writes out updated links to
the file while crawling the site."""
# TODO: have some different scheme to crawl a site (e.g. separate
# internal and external queues, threading, etc)
tocheck = []
# add all unfetched site urls
for link in self.linkMap.values():
if not link.isyanked and not link.isfetched:
tocheck.append(link)
# add all internal urls
for url in self._internal_urls:
tocheck.append(self.get_link(url))
# repeat until we have nothing more to check
fetchedlinks = 0
while len(tocheck) > 0:
debugio.debug('crawler.crawl(): items left to check: %d' %
len(tocheck))
# choose a link from the tocheck list
link = tocheck.pop(0)
# skip link it there is nothing to check
if link.isyanked or link.isfetched:
continue
# fetch the link's contents
link.fetch()
# add children to tocheck
for child in link.children:
if not child.isyanked and not child.isfetched and not child in tocheck:
tocheck.append(child)
# add embedded content
for embed in link.embedded:
if not embed.isyanked and not embed.isfetched and not embed in tocheck:
tocheck.append(embed)
# serialize all as of yet unserialized links
fetchedlinks += 1
# TODO: make this configurable
if serfp and fetchedlinks >= 5:
fetchedlinks = 0
import serialize
for link in self.linkMap.values():
if link._ischanged:
serialize.serialize_link(serfp, link)
link._ischanged = False
serfp.flush()
# sleep between requests if configured
if config.WAIT_BETWEEN_REQUESTS > 0:
debugio.debug('crawler.crawl(): sleeping %s seconds' %
config.WAIT_BETWEEN_REQUESTS)
time.sleep(config.WAIT_BETWEEN_REQUESTS)
# serialize remaining changed links
if serfp:
import serialize
for link in self.linkMap.values():
if link._ischanged:
serialize.serialize_link(serfp, link)
link._ischanged = False
serfp.flush()
def crawl(self, serfp=None):
"""Crawl the website based on the urls specified with
add_internal(). If the serialization file pointer
is specified the crawler writes out updated links to
the file while crawling the site."""
# TODO: have some different scheme to crawl a site (e.g. separate
# internal and external queues, threading, etc)
tocheck = []
# add all unfetched site urls
for link in self.linkMap.values():
if not link.isyanked and not link.isfetched:
tocheck.append(link)
# add all internal urls
for url in self._internal_urls:
tocheck.append(self.get_link(url))
# repeat until we have nothing more to check
fetchedlinks = 0
while len(tocheck) > 0:
debugio.debug('crawler.crawl(): items left to check: %d' % len(tocheck))
# choose a link from the tocheck list
link = tocheck.pop(0)
# skip link it there is nothing to check
if link.isyanked or link.isfetched:
continue
# fetch the link's contents
link.fetch()
# add children to tocheck
for child in link.children:
if not child.isyanked and not child.isfetched and not child in tocheck:
tocheck.append(child)
# add embedded content
for embed in link.embedded:
if not embed.isyanked and not embed.isfetched and not embed in tocheck:
tocheck.append(embed)
# serialize all as of yet unserialized links
fetchedlinks += 1
# TODO: make this configurable
if serfp and fetchedlinks >= 5:
fetchedlinks = 0
import serialize
for link in self.linkMap.values():
if link._ischanged:
serialize.serialize_link(serfp, link)
link._ischanged = False
serfp.flush()
# sleep between requests if configured
if config.WAIT_BETWEEN_REQUESTS > 0:
debugio.debug('crawler.crawl(): sleeping %s seconds' % config.WAIT_BETWEEN_REQUESTS)
time.sleep(config.WAIT_BETWEEN_REQUESTS)
# serialize remaining changed links
if serfp:
import serialize
for link in self.linkMap.values():
if link._ischanged:
serialize.serialize_link(serfp, link)
link._ischanged = False
serfp.flush()
def check_for_whole_start_tag(self, i):
"""Override to catch assertion exception."""
try:
return HTMLParser.HTMLParser.check_for_whole_start_tag(self, i)
except AssertionError:
debugio.debug(
'parsers.html._MyHTMLParser.check_for_whole_start_tag(): caught assertion error'
)
return None
def parse(content, link):
"""Parse the specified content and extract an url list, a list of images a
title and an author. The content is assumed to contain HMTL."""
# create parser and feed it the content
parser = _MyHTMLParser(link)
try:
parser.feed(content)
parser.close()
except Exception, e:
# ignore (but log) all errors
debugio.debug('parsers.html.parse(): caught exception: '+str(e))
def parse(content, link):
"""Parse the specified content and extract an url list, a list of images a
title and an author. The content is assumed to contain HMTL."""
# create parser and feed it the content
parser = _MyHTMLParser(link)
try:
parser.feed(content)
parser.close()
except Exception, e:
# ignore (but log) all errors
debugio.debug('parsers.html.parse(): caught exception: ' + str(e))
def error(self, message):
"""Override superclass' error() method to ignore errors."""
# construct error message
message += ', ' + self._location()
# store error message
debugio.debug('parsers.html._MyHTMLParser.error(): problem parsing html: '+message)
if self.errmsg is None:
self.errmsg = message
# increment error count
self.errcount += 1
if self.errcount > 10:
raise HTMLParser.HTMLParseError(message, self.getpos())
def _deserialize_site(site, key, value):
"""The data in the key value pair is fed into the site."""
debugio.debug("%s=%s" % (key, value))
if key == 'internal_url':
site.add_internal(_readstring(value, False))
elif key == 'internal_re':
site.add_internal_re(_readstring(value))
elif key == 'external_re':
site.add_external_re(_readstring(value))
elif key == 'yanked_re':
site.add_yanked_re(_readstring(value))
else:
raise DeSerializeException('parse error')
def _deserialize_site(site, key, value):
"""The data in the key value pair is fed into the site."""
debugio.debug("%s=%s" % (key, value))
if key == 'internal_url':
site.add_internal(_readstring(value, False))
elif key == 'internal_re':
site.add_internal_re(_readstring(value))
elif key == 'external_re':
site.add_external_re(_readstring(value))
elif key == 'yanked_re':
site.add_yanked_re(_readstring(value))
else:
raise DeSerializeException('parse error')
def error(self, message):
"""Override superclass' error() method to ignore errors."""
# construct error message
message += ', ' + self._location()
# store error message
debugio.debug(
'parsers.html._MyHTMLParser.error(): problem parsing html: ' +
message)
if self.errmsg is None:
self.errmsg = message
# increment error count
self.errcount += 1
if self.errcount > 10:
raise HTMLParser.HTMLParseError(message, self.getpos())
def _cwd(ftp, path):
"""Go down the path on the ftp server returning the part that cannot be
changed into."""
# split the path into directories
dirs = path.split('/')
try:
# decend down the tree
while len(dirs) > 0:
d = dirs[0]
if d != '':
debugio.debug('schemes.ftp._cwd(): CWD '+d+': '+ftp.cwd(d))
dirs.pop(0)
return None
except ftplib.error_perm, e:
debugio.debug('schemes.ftp._cwd(): CWD '+d+': '+str(e))
return '/'.join(dirs)
def _getconnection(netloc):
"""Return a FTP connection object to the specified server."""
# NOTE: this method is not thread safe
if _ftpconnections.has_key(netloc):
return _ftpconnections[netloc]
# split url into useful parts
(userpass, host) = urllib.splituser(netloc)
if userpass is not None:
(user, passwd) = urllib.splitpasswd(userpass)
else:
(user, passwd) = ('anonymous', '')
(host, port) = urllib.splitnport(host, ftplib.FTP_PORT)
# initialize a new connection
ftp = ftplib.FTP()
debugio.debug('schemes.ftp._getconnection(): CONNECT: '+ftp.connect(host, port))
debugio.debug('schemes.ftp._getconnection(): LOGIN: '+ftp.login(user, passwd))
_ftpconnections[netloc] = ftp
return ftp
def fetch(link, acceptedtypes):
"""Fetch the specified link."""
# try to fetch the document
try:
ftp = _getconnection(link.netloc)
debugio.debug('schemes.ftp.fetch(): CWD /: '+ftp.cwd('/'))
# descend down the directory tree as far as we can go
path = urllib.unquote(link.path)
path = _cwd(ftp, path)
# check if we are dealing with an (exising) directory
if path is None:
return _fetch_directory(link, ftp, acceptedtypes)
else:
return _fetch_file(link, ftp, path, acceptedtypes)
except ftplib.all_errors, e:
debugio.debug('schemes.ftp.fetch(): CAUGHT '+str(e))
link.add_linkproblem(str(e))
return None
def _fetch_directory(link, path, acceptedtypes):
"""Retrieve some basic information about the directory.
This checks that the directory has a trailing slash and
returns a list of files in the directory, unless a configured
filename is found in the directory (in which case this function
acts as if the file was fetched)."""
# if the name does not end with a slash, redirect
if path[-1:] != os.path.sep:
debugio.debug('directory referenced without trailing slash')
link.redirect(urlparse.urljoin(link.url, link.path+'/'))
return None
# check contents of directory for some common files
for fname in config.FILE_INDEXES:
if os.path.isfile(os.path.join(path, fname)):
debugio.debug('pick up %s from directory' % fname)
# the the directory contains an index.html, use that
return _fetch_file(link, os.path.join(path, fname), acceptedtypes)
# otherwise add the directory's files as children
debugio.debug('add files as children of this page')
try:
link.ispage = True
for f in os.listdir(path):
link.add_child(urlparse.urljoin(link.url, urllib.pathname2url(f)))
except os.error, e:
link.add_linkproblem(str(e))
def _fetch_directory(link, path, acceptedtypes):
"""Retrieve some basic information about the directory.
This checks that the directory has a trailing slash and
returns a list of files in the directory, unless a configured
filename is found in the directory (in which case this function
acts as if the file was fetched)."""
# if the name does not end with a slash, redirect
if path[-1:] != os.path.sep:
debugio.debug('directory referenced without trailing slash')
link.redirect(urlparse.urljoin(link.url, link.path+'/'))
return None
# check contents of directory for some common files
for fname in config.FILE_INDEXES:
if os.path.isfile(os.path.join(path, fname)):
debugio.debug('pick up %s from directory' % fname)
# the the directory contains an index.html, use that
return _fetch_file(link, os.path.join(path, fname), acceptedtypes)
# otherwise add the directory's files as children
debugio.debug('add files as children of this page')
try:
link.ispage = True
for f in os.listdir(path):
link.add_child(urlparse.urljoin(link.url, urllib.pathname2url(f)))
except os.error, e:
link.add_linkproblem(str(e))
def postprocess(self):
# build the list of urls that were set up with add_internal() that
# do not have a parent (they form the base for the site)
for url in self._internal_urls:
link = self.linkMap[url].follow_link()
if link == None:
debugio.warn('base link %s redirects to nowhere' % url)
continue
# add the link to bases
debugio.debug('crawler.postprocess(): adding %s to bases' % link.url)
self.bases.append(link)
# if we got no bases, just use the first internal one
if len(self.bases) == 0:
debugio.debug('crawler.postprocess(): fallback to adding %s to bases' % self._internal_urls[0])
self.bases.append(self.linkMap[self._internal_urls[0]])
# do a breadth first traversal of the website to determin depth and
# figure out page children
tocheck = []
for link in self.bases:
link.depth = 0
tocheck.append(link)
# repeat until we have nothing more to check
while len(tocheck) > 0:
debugio.debug('crawler.postprocess(): items left to examine: %d' % len(tocheck))
# choose a link from the tocheck list
link = tocheck.pop(0)
# figure out page children
for child in link._pagechildren():
# skip children already in our list or the wrong depth
if child in tocheck or child.depth != link.depth+1:
continue
tocheck.append(child)
# set some compatibility properties
# TODO: figure out a better way to get to this to the plugins
self.base = self.bases[0].url
def _fetch_directory(link, ftp, acceptedtypes):
"""Handle the ftp directory."""
# check that the url ends with a slash
if link.path[-1:] != '/':
debugio.debug('schemes.ftp._fetch_directory(): directory referenced without trailing slash')
link.redirect(urlparse.urljoin(link.url, link.path+'/'))
return None
# retreive the contents of the directory
# FIXME: this raises an exception for empty directories, probably replace with own command
contents = ftp.nlst()
# check contents of directory for some common files
for f in config.FTP_INDEXES:
if f in contents:
debugio.debug('schemes.ftp._fetch_directory(): pick up %s from directory' % f)
# the the directory contains an index.html, use that
return _fetch_file(link, ftp, f, acceptedtypes)
# just add files in directory as children
debugio.debug('schemes.ftp._fetch_directory(): add files as children of this page')
link.ispage = True
debugio.debug('schemes.ftp._fetch_directory(): TYPE A: '+ftp.voidcmd('TYPE A'))
# FIXME: this raises an exception for empty directories
for f in contents:
link.add_child(urlparse.urljoin(link.url, urllib.quote(f)))
return None
def _fetch_file(link, ftp, path, acceptedtypes):
"""Try to download the file in path that should be in the current
directory of the ftp instance. The path can also point to a non-existant
file or directory."""
# figure out the size of the document
link.size = ftp.size(path)
debugio.debug('schemes.ftp.fetch(): size='+str(link.size))
# guess the mimetype of the document
if link.mimetype is None:
link.mimetype = mimetypes.guess_type(path)[0]
# try to fetch file
if link.mimetype in acceptedtypes:
debugio.debug('schemes.ftp.fetch(): TYPE I: '+ftp.voidcmd('TYPE I'))
(conn, size) = ftp.ntransfercmd('RETR ' + path)
if size:
content = conn.makefile().read(size)
else:
content = conn.makefile().read()
debugio.debug('schemes.ftp.fetch(): fetched, size=%d' % len(content))
return content
return None
def postprocess(self):
# build the list of urls that were set up with add_internal() that
# do not have a parent (they form the base for the site)
for url in self._internal_urls:
link = self.linkMap[url].follow_link()
if link == None:
debugio.warn('base link %s redirects to nowhere' % url)
continue
# add the link to bases
debugio.debug('crawler.postprocess(): adding %s to bases' %
link.url)
self.bases.append(link)
# if we got no bases, just use the first internal one
if len(self.bases) == 0:
debugio.debug(
'crawler.postprocess(): fallback to adding %s to bases' %
self._internal_urls[0])
self.bases.append(self.linkMap[self._internal_urls[0]])
# do a breadth first traversal of the website to determin depth and
# figure out page children
tocheck = []
for link in self.bases:
link.depth = 0
tocheck.append(link)
# repeat until we have nothing more to check
while len(tocheck) > 0:
debugio.debug('crawler.postprocess(): items left to examine: %d' %
len(tocheck))
# choose a link from the tocheck list
link = tocheck.pop(0)
# figure out page children
for child in link._pagechildren():
# skip children already in our list or the wrong depth
if child in tocheck or child.depth != link.depth + 1:
continue
tocheck.append(child)
# set some compatibility properties
# TODO: figure out a better way to get to this to the plugins
self.base = self.bases[0].url
def fetch(link, acceptedtypes):
"""Open connection to url and report information given by GET command."""
# TODO: HTTP connection pooling?
# TODO: implement proxy requests for https
# split netloc in user:pass part and host:port part
(userpass, netloc) = urllib.splituser(link.netloc)
proxyuserpass = None
scheme = link.scheme
# check which host to connect to (if using proxies)
if config.PROXIES and config.PROXIES.has_key(link.scheme):
# pass the complete url in the request, connecting to the proxy
path = urlparse.urlunsplit((link.scheme, netloc, link.path, link.query, ''))
(scheme, netloc) = urlparse.urlsplit(config.PROXIES[link.scheme])[0:2]
(proxyuserpass, netloc) = urllib.splituser(netloc)
else:
# otherwise direct connect to the server with partial url
path = urlparse.urlunsplit(('', '', link.path, link.query, ''))
# remove trailing : from netloc
if netloc[-1] == ':':
netloc = netloc[:-1]
conn = None
try:
try:
# create the connection
debugio.debug('schemes.http.fetch: connecting to %s' % netloc)
if scheme == 'http':
conn = httplib.HTTPConnection(netloc)
elif scheme == 'https':
conn = httplib.HTTPSConnection(netloc)
# the requests adds a correct host header for us
conn.putrequest('GET', path)
if userpass is not None:
(user, passwd) = urllib.splitpasswd(userpass)
conn.putheader(
'Authorization',
'Basic '+base64.encodestring(str(user)+':'+str(passwd)).strip() )
if proxyuserpass is not None:
(user, passwd) = urllib.splitpasswd(proxyuserpass)
conn.putheader(
'Proxy-Authorization',
'Basic '+base64.encodestring(str(user)+':'+str(passwd)).strip() )
# bypass proxy cache
if config.BYPASSHTTPCACHE:
conn.putheader('Cache-control', 'no-cache')
conn.putheader('Pragma', 'no-cache')
conn.putheader('User-Agent','webcheck %s' % config.VERSION)
conn.endheaders()
# wait for the response
response = conn.getresponse()
link.status = '%s %s' % (response.status, response.reason)
debugio.debug('schemes.http.fetch(): HTTP response: %s' % link.status)
# dump proxy hit/miss debugging info
if config.PROXIES and config.PROXIES.has_key(link.scheme):
try:
debugio.debug('schemes.http.fetch(): X-Cache: %s' % str(response.getheader('X-Cache')))
except AttributeError:
pass
# retrieve some information from the headers
try:
link.mimetype = response.msg.gettype()
debugio.debug('schemes.http.fetch(): mimetype: %s' % str(link.mimetype))
except AttributeError:
pass
try:
link.set_encoding(_charsetpattern.search(response.getheader('Content-type')).group(1))
except (AttributeError, TypeError):
pass
try:
link.size = int(response.getheader('Content-length'))
debugio.debug('schemes.http.fetch(): size: %s' % str(link.size))
except (KeyError, TypeError):
pass
try:
link.mtime = time.mktime(response.msg.getdate('Last-Modified'))
debugio.debug('schemes.http.fetch(): mtime: %s' % time.strftime('%c', time.localtime(link.mtime)))
except (OverflowError, TypeError, ValueError):
pass
# handle redirects
# 301=moved permanently, 302=found, 303=see other, 307=temporary redirect
if response.status in (301, 302, 303, 307):
# consider a 301 (moved permanently) a problem
if response.status == 301:
link.add_linkproblem(str(response.status)+': '+response.reason)
# find url that is redirected to
location = urlparse.urljoin(link.url, response.getheader('Location', ''))
# create the redirect
link.redirect(location)
return None
elif response.status != 200:
# handle error responses
link.add_linkproblem(str(response.status)+': '+response.reason)
return None
elif link.mimetype in acceptedtypes:
# return succesful responses
# TODO: support gzipped content
# TODO: add checking for size
return response.read()
except httplib.HTTPException, e:
debugio.debug('error reading HTTP response: '+str(e))
link.add_linkproblem('error reading HTTP response: '+str(e))
return None
except (socket.error, socket.sslerror), e:
if hasattr(e, 'args') and len(e.args) == 2:
debugio.debug("error reading HTTP response: "+str(e.args[1]))
link.add_linkproblem("error reading HTTP response: "+str(e.args[1]))
else:
debugio.debug("error reading HTTP response: "+str(e))
link.add_linkproblem("error reading HTTP response: "+str(e))
return None
def _fetch_file(link, path, acceptedtypes):
"""Retrieve some basic information of the specified file and return
the contents of the file."""
# get stats of file
try:
stats = os.stat(path)
link.size = stats.st_size
link.mtime = stats.st_mtime
except os.error, e:
link.add_linkproblem(str(e))
return None
# guess mimetype
if link.mimetype is None:
link.mimetype = mimetypes.guess_type(path)[0]
debugio.debug('mimetype='+str(link.mimetype))
debugio.debug('acceptedtypes='+str(acceptedtypes))
# fetch the document if there is any point
if link.mimetype in acceptedtypes:
debugio.debug('FETCH')
try:
# TODO: add size checking
return open(path, 'r').read()
except IOError, e:
debugio.debug('PROBLEM: '+str(e))
link.add_linkproblem(str(e))
return None
def fetch(link, acceptedtypes):
"""Retreive some basic information about the file.
Store the results in the link object."""
def parse(content, link):
"""Parse the specified content and extract an url list, a list of images a
title and an author. The content is assumed to contain HMTL."""
# create parser and feed it the content
parser = _MyHTMLParser(link)
try:
parser.feed(content)
parser.close()
except Exception, e:
# ignore (but log) all errors
debugio.debug('parsers.html.parse(): caught exception: ' + str(e))
# check for parser errors
if parser.errmsg is not None:
debugio.debug('parsers.html.parse(): problem parsing html: ' +
parser.errmsg)
link.add_pageproblem('problem parsing html: %s' % parser.errmsg)
# dump encoding
debugio.debug('parsers.html.parse(): html encoding: %s' %
str(link.encoding))
# flag that the link contains a valid page
link.ispage = True
# save the title
if parser.title is not None:
link.title = _maketxt(parser.title, link.encoding).strip()
# save the author
if parser.author is not None:
link.author = _maketxt(parser.author, link.encoding).strip()
# figure out the base of the document (for building the other urls)
base = link.url
if parser.base is not None:
def fetch(link, acceptedtypes):
"""Open connection to url and report information given by GET command."""
# TODO: HTTP connection pooling?
# TODO: implement proxy requests for https
# split netloc in user:pass part and host:port part
(userpass, netloc) = urllib.splituser(link.netloc)
proxyuserpass = None
scheme = link.scheme
# check which host to connect to (if using proxies)
if config.PROXIES and config.PROXIES.has_key(link.scheme):
# pass the complete url in the request, connecting to the proxy
path = urlparse.urlunsplit(
(link.scheme, netloc, link.path, link.query, ''))
(scheme, netloc) = urlparse.urlsplit(config.PROXIES[link.scheme])[0:2]
(proxyuserpass, netloc) = urllib.splituser(netloc)
else:
# otherwise direct connect to the server with partial url
path = urlparse.urlunsplit(('', '', link.path, link.query, ''))
# remove trailing : from netloc
if netloc[-1] == ':':
netloc = netloc[:-1]
conn = None
try:
try:
# create the connection
debugio.debug('schemes.http.fetch: connecting to %s' % netloc)
if scheme == 'http':
conn = httplib.HTTPConnection(netloc)
elif scheme == 'https':
conn = httplib.HTTPSConnection(netloc)
# the requests adds a correct host header for us
conn.putrequest('GET', path)
if userpass is not None:
(user, passwd) = urllib.splitpasswd(userpass)
conn.putheader(
'Authorization', 'Basic ' +
base64.encodestring(str(user) + ':' + str(passwd)).strip())
if proxyuserpass is not None:
(user, passwd) = urllib.splitpasswd(proxyuserpass)
conn.putheader(
'Proxy-Authorization', 'Basic ' +
base64.encodestring(str(user) + ':' + str(passwd)).strip())
# bypass proxy cache
if config.BYPASSHTTPCACHE:
conn.putheader('Cache-control', 'no-cache')
conn.putheader('Pragma', 'no-cache')
conn.putheader('User-Agent', 'webcheck %s' % config.VERSION)
conn.endheaders()
# wait for the response
response = conn.getresponse()
link.status = '%s %s' % (response.status, response.reason)
debugio.debug('schemes.http.fetch(): HTTP response: %s' %
link.status)
# dump proxy hit/miss debugging info
if config.PROXIES and config.PROXIES.has_key(link.scheme):
try:
debugio.debug('schemes.http.fetch(): X-Cache: %s' %
str(response.getheader('X-Cache')))
except AttributeError:
pass
# retrieve some information from the headers
try:
link.mimetype = response.msg.gettype()
debugio.debug('schemes.http.fetch(): mimetype: %s' %
str(link.mimetype))
except AttributeError:
pass
try:
link.set_encoding(
_charsetpattern.search(
response.getheader('Content-type')).group(1))
except (AttributeError, TypeError):
pass
try:
link.size = int(response.getheader('Content-length'))
debugio.debug('schemes.http.fetch(): size: %s' %
str(link.size))
except (KeyError, TypeError):
pass
try:
link.mtime = time.mktime(response.msg.getdate('Last-Modified'))
debugio.debug('schemes.http.fetch(): mtime: %s' %
time.strftime('%c', time.localtime(link.mtime)))
except (OverflowError, TypeError, ValueError):
pass
# handle redirects
# 301=moved permanently, 302=found, 303=see other, 307=temporary redirect
if response.status in (301, 302, 303, 307):
# consider a 301 (moved permanently) a problem
if response.status == 301:
link.add_linkproblem(
str(response.status) + ': ' + response.reason)
# find url that is redirected to
location = urlparse.urljoin(link.url,
response.getheader('Location', ''))
# create the redirect
link.redirect(location)
return None
elif response.status != 200:
# handle error responses
link.add_linkproblem(
str(response.status) + ': ' + response.reason)
return None
elif link.mimetype in acceptedtypes:
# return succesful responses
# TODO: support gzipped content
# TODO: add checking for size
return response.read()
except httplib.HTTPException, e:
debugio.debug('error reading HTTP response: ' + str(e))
link.add_linkproblem('error reading HTTP response: ' + str(e))
return None
except (socket.error, socket.sslerror), e:
if hasattr(e, 'args') and len(e.args) == 2:
debugio.debug("error reading HTTP response: " + str(e.args[1]))
link.add_linkproblem("error reading HTTP response: " +
str(e.args[1]))
else:
debugio.debug("error reading HTTP response: " + str(e))
link.add_linkproblem("error reading HTTP response: " + str(e))
return None
link.add_linkproblem(str(response.status)+': '+response.reason)
return None
elif link.mimetype in acceptedtypes:
# return succesful responses
# TODO: support gzipped content
# TODO: add checking for size
return response.read()
except httplib.HTTPException, e:
debugio.debug('error reading HTTP response: '+str(e))
link.add_linkproblem('error reading HTTP response: '+str(e))
return None
except (socket.error, socket.sslerror), e:
if hasattr(e, 'args') and len(e.args) == 2:
debugio.debug("error reading HTTP response: "+str(e.args[1]))
link.add_linkproblem("error reading HTTP response: "+str(e.args[1]))
else:
debugio.debug("error reading HTTP response: "+str(e))
link.add_linkproblem("error reading HTTP response: "+str(e))
return None
except Exception, e:
# handle all other exceptions
debugio.debug('unknown exception caught: '+str(e))
link.add_linkproblem('error reading HTTP response: '+str(e))
import traceback
traceback.print_exc()
return None
finally:
# close the connection before returning
if conn is not None:
conn.close()
def _fetch_file(link, path, acceptedtypes):
"""Retrieve some basic information of the specified file and return
the contents of the file."""
# get stats of file
try:
stats = os.stat(path)
link.size = stats.st_size
link.mtime = stats.st_mtime
except os.error, e:
link.add_linkproblem(str(e))
return None
# guess mimetype
if link.mimetype is None:
link.mimetype = mimetypes.guess_type(path)[0]
debugio.debug('mimetype='+str(link.mimetype))
debugio.debug('acceptedtypes='+str(acceptedtypes))
# fetch the document if there is any point
if link.mimetype in acceptedtypes:
debugio.debug('FETCH')
try:
# TODO: add size checking
return open(path, 'r').read()
except IOError, e:
debugio.debug('PROBLEM: '+str(e))
link.add_linkproblem(str(e))
return None
def fetch(link, acceptedtypes):
"""Retreive some basic information about the file.
Store the results in the link object."""
return htmlunescape(unicode(txt, errors='replace'))
def parse(content, link):
"""Parse the specified content and extract an url list, a list of images a
title and an author. The content is assumed to contain HMTL."""
# create parser and feed it the content
parser = _MyHTMLParser(link)
try:
parser.feed(content)
parser.close()
except Exception, e:
# ignore (but log) all errors
debugio.debug('parsers.html.parse(): caught exception: '+str(e))
# check for parser errors
if parser.errmsg is not None:
debugio.debug('parsers.html.parse(): problem parsing html: '+parser.errmsg)
link.add_pageproblem('problem parsing html: %s' % parser.errmsg)
# dump encoding
debugio.debug('parsers.html.parse(): html encoding: %s' % str(link.encoding))
# flag that the link contains a valid page
link.ispage = True
# save the title
if parser.title is not None:
link.title = _maketxt(parser.title, link.encoding).strip()
# save the author
if parser.author is not None:
link.author = _maketxt(parser.author, link.encoding).strip()
# figure out the base of the document (for building the other urls)
base = link.url
if parser.base is not None:
base = parser.base
return None
elif link.mimetype in acceptedtypes:
# return succesful responses
# TODO: support gzipped content
# TODO: add checking for size
return response.read()
except httplib.HTTPException, e:
debugio.debug('error reading HTTP response: ' + str(e))
link.add_linkproblem('error reading HTTP response: ' + str(e))
return None
except (socket.error, socket.sslerror), e:
if hasattr(e, 'args') and len(e.args) == 2:
debugio.debug("error reading HTTP response: " + str(e.args[1]))
link.add_linkproblem("error reading HTTP response: " +
str(e.args[1]))
else:
debugio.debug("error reading HTTP response: " + str(e))
link.add_linkproblem("error reading HTTP response: " + str(e))
return None
except Exception, e:
# handle all other exceptions
debugio.debug('unknown exception caught: ' + str(e))
link.add_linkproblem('error reading HTTP response: ' + str(e))
import traceback
traceback.print_exc()
return None
finally:
# close the connection before returning
if conn is not None:
conn.close()
def crawl(self):
"""Crawl the website based on the urls specified with
add_internal()."""
# TODO: have some different scheme to crawl a site (e.g. separate
# internal and external queues, threading, etc)
tocheck = []
for u in self._internal_urls:
tocheck.append(self._get_link(u))
# repeat until we have nothing more to check
while len(tocheck) > 0:
debugio.debug("crawler.crawl(): items left to check: %d" % len(tocheck))
# choose a link from the tocheck list
link=tocheck.pop(0)
# skip link it there is nothing to check
if link.isyanked or link.isfetched:
continue
# fetch the link's contents
link.fetch()
# add children to tocheck
for child in link.children:
if not child.isyanked and not child.isfetched and not child in tocheck:
tocheck.append(child)
# add embedded content
for embed in link.embedded:
if not embed.isyanked and not embed.isfetched and not embed in tocheck:
tocheck.append(embed)
# sleep between requests if configured
if config.WAIT_BETWEEN_REQUESTS > 0:
debugio.debug('sleeping %s seconds' % config.WAIT_BETWEEN_REQUESTS)
time.sleep(config.WAIT_BETWEEN_REQUESTS)
# build the list of urls that were set up with add_internal() that
# do not have a parent (they form the base for the site)
bases = [ ]
for u in self._internal_urls:
l = self.linkMap[u].follow_link()
if l == None:
debugio.warn('base link %s redirects to nowhere' % u)
continue
# if the link has no parent add it to the result list unless it is the first one
if len(l.parents) == 0 or len(bases) == 0:
debugio.debug('crawler.crawl(): adding %s to bases' % l.url)
bases.append(l)
# if we got no bases, just use the first internal one
if len(bases) == 0:
debugio.debug('crawler.crawl(): fallback to adding %s to bases' % self._internal_urls[0])
bases.append(self.linkMap[self._internal_urls[0]])
# do a breadth first traversal of the website to determin depth and
# figure out page children
tocheck = []
for link in bases:
link.depth = 0
tocheck.append(link)
# repeat until we have nothing more to check
while len(tocheck) > 0:
debugio.debug("crawler.crawl(): items left to examine: %d" % len(tocheck))
# choose a link from the tocheck list
link = tocheck.pop(0)
# figure out page children
for child in link._pagechildren():
# skip children already in our list or the wrong depth
if child in tocheck or child.depth != link.depth+1:
continue
tocheck.append(child)
# set some compatibility properties
# TODO: figure out a better way to get to this to the plugins
self.base = bases[0].url
