def is_page_robot_scannable(self):
"""
Returns a boolean that tells whether the page is robot scrapeable.
"""
robotcheck = RobotFileParser()
robotcheck.set_url(self.urlparse[0]+'://'+self.urlparse[1]+'/robots.txt')
robotcheck.read()
return robotcheck.can_fetch(settings.SPIDER_USER_AGENT, self.url)
def test_parse(self):
from robotparser import RobotFileParser
rules = RobotFileParser()
rules.set_url("http://www.sogou.com/robots.txt")
rules.read()
self.assertEqual(
rules.can_fetch("mozilla", "http://www.sogou.com/sohu/robots.txt"),
False)
def _allowed_to_open(self, url):
host = urlparse.urlsplit(url)[1]
robots_url = urlparse.urlunsplit(('http', host, '/robots.txt', '', ''))
rp = RobotFileParser(robots_url)
try:
rp.read()
except:
return False
return rp.can_fetch(self._agent_name, url)
def get_robots(url):
'''
Initialize robots parser for this domain
:param url:
:return:
'''
rp = RobotFileParser()
rp.set_url(urlparse.urljoin(url, '/robots.txt'))
rp.read()
return rp
def http_open(self, request):
#request -- urllib2.Request
url = request.get_full_url()
host = urlsplit(url)[1]
robots_url = urlunsplit(('http', host, '/robots.txt', '', ''))
rp = RobotFileParser(robots_url)
rp.read()
if not rp.can_fetch(self.agentname, url):
raise RuntimeError('Forbidden by robots.txt')
return urllib2.HTTPHandler.http_open(self, request)
def checkRobots(URL):
time.sleep(1)
parsed = urlparse(URL)
robotsUrl = parsed.scheme + "://" + parsed.netloc + "/robots.txt"
robotParser = RobotFileParser()
robotParser.set_url(robotsUrl)
robotParser.read()
result = robotParser.can_fetch("*", URL)
return result
def http_open(self, req):
url = req.get_full_url()
host = urlsplit(url)[1]
robots_url = urlunsplit(('http', host, '/robots.txt', '', ''))
robotfileparser = RobotFileParser(robots_url)
robotfileparser.read()
if not robotfileparser.can_fetch(self.crawlername, url):
raise RuntimeError('Forbidden by robots.txt')
return urllib2.HTTPHandler.http_open(self, req)
def checkRobots(URL):
time.sleep(1)
parsed = urlparse(URL)
robotsUrl = parsed.scheme + "://"+ parsed.netloc+"/robots.txt"
robotParser = RobotFileParser()
robotParser.set_url(robotsUrl)
robotParser.read()
result = robotParser.can_fetch("*",URL)
return result
def can_fetch(self,url): host,path=urlparse.urlparse(url)[1:3] if (self.rules.has_key(host)): return self.rules[host].can_fetch(self.agent,url) else: rp=RobotFileParser() robot_url="http://"+host+"/robots.txt" rp.set_url(robot_url) rp.read() self.rules[host]=rp return rp.can_fetch(self.agent,url)
def can_fetch(self, url):
host, path = urlparse.urlparse(url)[1:3]
if (self.rules.has_key(host)):
return self.rules[host].can_fetch(self.agent, url)
else:
rp = RobotFileParser()
robot_url = "http://" + host + "/robots.txt"
rp.set_url(robot_url)
rp.read()
self.rules[host] = rp
return rp.can_fetch(self.agent, url)
def check_robots(self, url):
'''check the robots.txt in this url's domain'''
hostname = urlparse(url).netloc
if hostname not in self.domain_list.keys(): # no records in domain_list
rp = RobotFileParser('http://%s/robots.txt' % hostname)
print("%s: fetching %s" % (url, rp.url))
try:
rp.read() # get new robots.txt
except IOError, e: # url's server not available(connection timeout)
log.error(str(e))
rp.disallow_all = True # reject all request
self.domain_list[hostname] = rp # add domain entry into domain_list
class HolidayScrapper:
def __init__(self):
self.rp = RobotFileParser()
self.rp.set_url('https://www.timeanddate.com/robots.txt')
self.rp.read()
if not self.rp.can_fetch('WasThereAHoliday', init_url):
raise RuntimeError('Scrapping forbidden due to robots.txt file')
self.countries = self.get_countries(self.get_page(init_url))
try:
# removing entries which are not countries
self.countries.remove('un')
except ValueError:
pass
try:
# removing entries which are not countries
self.countries.remove('world')
except ValueError:
pass
def get_data(self):
all_data = pd.DataFrame()
for cntr in self.countries:
print 'Fetching data for ' + cntr
try:
url = 'https://www.timeanddate.com/holidays/' + cntr + '/2016#!hol=8389401'
if not self.rp.can_fetch('WasThereAHoliday', url):
raise RuntimeError(
'Scrapping forbidden due to robots.txt file')
soup = self.get_page('https://www.timeanddate.com/holidays/' +
cntr + '/2016#!hol=8389401')
html_table = soup.find('table')
df_table = pd.read_html(str(html_table))[0]
df_table['country'] = cntr
all_data = all_data.append(df_table)
except ValueError:
print 'Problem occured when fetching data for ' + cntr
pass
return all_data
@staticmethod
def get_page(url):
page = requests.get(url, headers=headers)
soup = BeautifulSoup(page.text, 'lxml')
return soup
@staticmethod
def get_countries(soup):
countries = []
select_list = soup.find(id="co")
for cntr in select_list.children:
countries.append(cntr['value'])
return countries
def robots_check(url):
# creating url for robots.txt
root_url = tld.get_tld(url)
prefix = "http://www."
suffix = "/robots.txt"
robots_url = prefix + root_url + suffix
# checking url validity
rp = RobotFileParser()
rp.set_url(robots_url)
rp.read()
return rp.can_fetch("*", url)
def try_add_robot(self, url):
parsed_url = urlparse(url)
if parsed_url.netloc not in self.robots:
try:
robot_url = parsed_url.scheme + '://' + parsed_url.netloc + \
'/robots.txt'
rp = RobotFileParser(robot_url)
rp.read()
self.robots[parsed_url.netloc] = rp
except IOError as e:
print str(e)
except Exception as e:
print str(e)
def check_robots(self, url):
'''check the robots.txt in this url's domain'''
hostname = urlparse(url).netloc
if hostname not in self.domain_list.keys(
): # no records in domain_list
rp = RobotFileParser('http://%s/robots.txt' % hostname)
print("%s: fetching %s" % (url, rp.url))
try:
rp.read() # get new robots.txt
except IOError, e: # url's server not available(connection timeout)
log.error(str(e))
rp.disallow_all = True # reject all request
self.domain_list[
hostname] = rp # add domain entry into domain_list
def can_read(url):
domain = domain_name(url)
if domain not in Permissions:
rp = RobotFileParser()
rp.set_url(urljoin('http://' + domain, 'robots.txt'))
try:
rp.read()
except:
return False
Permissions[domain] = rp
res = False
try:
res = Permissions[domain].can_fetch("*", url)
except:
return False
return res
def urlopen(self, host):
robo_url = host.get_robots_url()
print self.robotdict
cached_parser = self.robotdict.get(robo_url)
if cached_parser:
logging.info("Found in Cache: " + robo_url)
else:
logging.info("Fetching: " + robo_url)
cached_parser = RobotFileParser()
self.robotdict.put(robo_url, cached_parser)
cached_parser.set_url(robo_url)
cached_parser.read()
if cached_parser.can_fetch('*', host. get_url()):
print 'Going to fetch:', host.get_url()
return self.fetch_file(host.get_url())
else:
logging.info("Forbidden by Robots.txt")
return None
def _get_soup(path):
"""Gets soup from the given path, respecting robots.txt"""
full_path = BASE_URL + path
# Set a user-agent
user_agent = 'dcnotify/%s' % __version__
http_headers = {'User-Agent': '%s' % user_agent}
# Honor robots.txt
robots = RobotFileParser()
robots.set_url("%s/robots.txt" % BASE_URL)
robots.read()
if not robots.can_fetch(user_agent, full_path):
raise ValueError("Path disallowed by robots.txt")
# Make a make a request, raising any HTTP errors that might occur
request = get(full_path, headers=http_headers)
request.raise_for_status()
return bs(request.text)
def _get_soup(path):
"""Gets soup from the given path, respecting robots.txt"""
full_path = BASE_URL + path
# Set a user-agent
user_agent = 'dcnotify/%s' % __version__
http_headers = {'User-Agent': '%s' % user_agent}
# Honor robots.txt
robots = RobotFileParser()
robots.set_url("%s/robots.txt" % BASE_URL)
robots.read()
if not robots.can_fetch(user_agent, full_path):
raise ValueError("Path disallowed by robots.txt")
# Make a make a request, raising any HTTP errors that might occur
request = get(full_path, headers=http_headers)
request.raise_for_status()
return bs(request.text)
class spider(object):
CurLink = ""
linknText = []
headings = []
def __init__(self, link):
self.CurLink = link
self.r = RobotFileParser()
def crawl(self):
self.r.set_url(urlparse.unquote(self.CurLink))
self.r.read()
self.html = urlopen(self.CurLink).read()
self.bs = BeautifulSoup(self.html, "lxml")
for i in self.bs.findAll("h1", text=True):
self.headings.append(i.text)
for i in self.bs.findAll("h2", text=True):
self.headings.append(i.text)
for i in self.bs.findAll("h3", text=True):
self.headings.append(i.text)
for i in self.bs.findAll("h4", text=True):
self.headings.append(i.text)
for i in self.bs.findAll("h5", text=True):
self.headings.append(i.text)
for i in self.bs.findAll("h6", text=True):
self.headings.append(i.text)
for link in self.bs.findAll('a', href=True):
aLink = urlparse.urljoin(self.CurLink, link['href'])
if (self.r.can_fetch("*", aLink)):
self.linknText.append({
"URL": aLink,
"AnchorText": link.string
})
def __init__(self, url):
self.url = urlManip.cleanURL(url)
self.pages = []
self.suggestions = set()
self.loaded = False
logger.info("Loading %s..." % (self.url))
try:
requests.get(self.url)
self.loaded = True
except IOError as e:
logger.error("%s cannot be loaded: %s" % (self.url, e))
# if the website can be loaded
if self.loaded == True:
logger.info("Load successful. Generating suggestions...")
# get robots.txt
rp = RobotFileParser(self.url + "robots.txt")
try:
rp.read()
except IOError:
logger.warning("robots.txt cannot be found.")
# get home page
self.pages.append(Page(self.url))
# get all pages on homepage
self.pages[0].load()
for link in self.pages[0].internalLinks:
if rp.can_fetch("*", link):
if link[:4] == 'http':
self.pages.append(Page(link))
else:
self.pages.append(Page(self.url + link))
else:
logger.debug("Ignoring %s based on robots.txt" % link)
class Crawler():
# Variables
parserobots = False
output = None
report = False
config = None
domain = ""
exclude = []
skipext = []
drop = []
debug = False
tocrawl = set([])
crawled = set([])
excluded = set([])
marked = {}
# TODO also search for window.location={.*?}
linkregex = re.compile(b'<a href=[\'|"](.*?)[\'"].*?>')
rp = None
response_code={}
nb_url=1 # Number of url.
nb_rp=0 # Number of url blocked by the robots.txt
nb_exclude=0 # Number of url excluded by extension or word
output_file = None
target_domain = ""
def __init__(self, parserobots=False, output=None, report=False ,domain="", exclude=[], skipext=[], drop=[], debug=False):
self.parserobots = parserobots
self.output = output
self.report = report
self.domain = domain
self.exclude = exclude
self.skipext = skipext
self.drop = drop
self.debug = debug
if self.debug:
logging.basicConfig(level=logging.DEBUG)
self.tocrawl = set([domain])
try:
self.target_domain = urlparse.urlparse(domain)[1]
except:
raise ValueError("Invalid domain")
if self.output:
try:
self.output_file = open(self.output, 'w')
except:
logging.debug ("Output file not available.")
exit(255)
def run(self):
print (config.xml_header, file if file else self.output_file)
logging.debug("Start the crawling process")
while len(self.tocrawl) != 0:
self.__crawling()
logging.debug("Crawling as reach the end of all found link")
print (config.xml_footer, file if file else self.output_file)
def __crawling(self):
crawling = self.tocrawl.pop()
url = urlparse.urlparse(crawling)
self.crawled.add(crawling)
request = Request(crawling, headers={"User-Agent":config.crawler_user_agent})
try:
response = urlopen(request)
except Exception as e:
if hasattr(e,'code'):
if e.code in self.response_code:
self.response_code[e.code]+=1
else:
self.response_code[e.code]=1
# Gestion des urls marked pour le reporting
if self.report:
if e.code in self.marked:
self.marked[e.code].append(crawling)
else:
self.marked[e.code] = [crawling]
logging.debug ("{1} ==> {0}".format(e, crawling))
return self.__continue_crawling()
# Read the response
try:
msg = response.read()
if response.getcode() in self.response_code:
self.response_code[response.getcode()]+=1
else:
self.response_code[response.getcode()]=1
response.close()
# Get the last modify date
if 'last-modified' in response.headers:
date = response.headers['Last-Modified']
else:
date = response.headers['Date']
date = datetime.strptime(date, '%a, %d %b %Y %H:%M:%S %Z')
except Exception as e:
logging.debug ("{1} ===> {0}".format(e, crawling))
return None
print ("<url><loc>"+url.geturl()+"</loc><lastmod>"+date.strftime('%Y-%m-%dT%H:%M:%S+00:00')+"</lastmod></url>", file if file else self.output_file)
if self.output_file:
self.output_file.flush()
# Found links
links = self.linkregex.findall(msg)
for link in links:
link = link.decode("utf-8")
#logging.debug("Found : {0}".format(link))
if link.startswith('/'):
link = 'http://' + url[1] + link
elif link.startswith('#'):
link = 'http://' + url[1] + url[2] + link
elif not link.startswith('http'):
link = 'http://' + url[1] + '/' + link
# Remove the anchor part if needed
if "#" in link:
link = link[:link.index('#')]
# Drop attributes if needed
for toDrop in self.drop:
link=re.sub(toDrop,'',link)
# Parse the url to get domain and file extension
parsed_link = urlparse.urlparse(link)
domain_link = parsed_link.netloc
target_extension = os.path.splitext(parsed_link.path)[1][1:]
if (link in self.crawled):
continue
if (link in self.tocrawl):
continue
if (link in self.excluded):
continue
if (domain_link != self.target_domain):
continue
if ("javascript" in link):
continue
# Count one more URL
self.nb_url+=1
# Check if the navigation is allowed by the robots.txt
if (not self.can_fetch(link)):
self.exclude_link(link)
self.nb_rp+=1
continue
# Check if the current file extension is allowed or not.
if (target_extension in self.skipext):
self.exclude_link(link)
self.nb_exclude+=1
continue
# Check if the current url doesn't contain an excluded word
if (not self.exclude_url(link)):
self.exclude_link(link)
self.nb_exclude+=1
continue
self.tocrawl.add(link)
return None
def __continue_crawling(self):
if self.tocrawl:
self.__crawling()
def exclude_link(self,link):
if link not in self.excluded:
self.excluded.add(link)
def checkRobots(self):
if self.domain[len(self.domain)-1] != "/":
self.domain += "/"
request = Request(self.domain+"robots.txt", headers={"User-Agent":config.crawler_user_agent})
self.rp = RobotFileParser()
self.rp.set_url(self.domain+"robots.txt")
self.rp.read()
def can_fetch(self, link):
try:
if self.parserobots:
if self.rp.can_fetch("*", link):
return True
else:
logging.debug ("Crawling of {0} disabled by robots.txt".format(link))
return False
if not self.parserobots:
return True
return True
except:
# On error continue!
logging.debug ("Error during parsing robots.txt")
return True
def exclude_url(self, link):
for ex in self.exclude:
if ex in link:
return False
return True
def make_report(self):
print ("Number of found URL : {0}".format(self.nb_url))
print ("Number of link crawled : {0}".format(len(self.crawled)))
if self.parserobots:
print ("Number of link block by robots.txt : {0}".format(self.nb_rp))
if self.skipext or self.exclude:
print ("Number of link exclude : {0}".format(self.nb_exclude))
for code in self.response_code:
print ("Nb Code HTTP {0} : {1}".format(code, self.response_code[code]))
for code in self.marked:
print ("Link with status {0}:".format(code))
for uri in self.marked[code]:
print ("\t- {0}".format(uri))
class SiteMap():
def __init__(self, main_page=None, robotrules=True):
"""
Constuctor method that initializes the members that are used during crawling process
:param main_page: The root page that needs to be crawled for generation of sitemap
"""
logging.info("Consider Robot.txt ? ==> "+str(robotrules))
self.robotrules = robotrules
self.site_map = {} # map that records the visits of urls, datemodified and assets
self.network = {} # map that maintains the network/graph of webpages visited
# The intention of this map is for visual rendering using d3.js
self.unvisited = set([]) # a set to keep the list of urls yet to be visited
self.start_page = None # the root page, this is used to avoid cycle and keeping crawl
# process limited to single domain.
self.robot_txt_rules = None
if main_page:
self.unvisited.add(main_page)
try:
self.start_page = urlparse(main_page).netloc
except:
logging.error("Improper URL, Please provide a Valid Url:"+main_page)
exit(0)
if self.robotrules == "True":
try:
logging.info("robot.txt respected")
self.robot_txt_rules = RobotFileParser()
self.robot_txt_rules.set_url(main_page + "/robots.txt")
self.robot_txt_rules.read()
except:
logging.error("Unable to read the robot.txt file")
self.robotrules = False # error reading robot.txt, ignore it forever
@timeit
def generate(self, site_map=None):
"""
This method holds the invoking control of the crawler method and drives the crawling process.
Basically a BFS style method that keeps popping the elements from the queue [self.unvisited set]
and scraping the urls.
Once the crawling process is done, this creates sitemap using the self.site_map dictionary with
just url, date-modified tags with dummy frequency and priorities.
:param site_map: name of the site_map file so as to create xml entries.
:return:
"""
while self.unvisited:
self.crawl()
# create xml from the site_map dictionary
header = """<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
xmlns:xhtml="http://www.w3.org/1999/xhtml"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9
http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd">
"""
footer = """\n</urlset>\n"""
entry = "\t<url>\n\
\t\t<loc>%s</loc>\n\
\t\t<lastmod>%s</lastmod>\n\
\t\t<changefreq>monthly</changefreq>\n\
\t\t<priority> 1 </priority>\n\
\t</url>\
"
xml = header
for url in self.site_map.keys():
xml += entry % (url, self.site_map[url]['date']) + "\n"
xml += footer
if site_map != None:
self.write_to_file(site_map, xml)
else:
self.write_to_file("sitemap.xml", xml)
return xml
def write_to_file(self, file_name, content):
"""
A utility method to just write the contents of the file into a given file name.
Alert: This overwrites if the file does exist in the current directory.
:param file_name: name of the file, sitemap in our case.
:param content: contents of the file
:return: None
"""
f = open(file_name, 'w')
f.write(content)
f.close()
def compose_url_from_href(self, url, href):
"""
There are different ways a href could specify a location and it varies in different ways based on how
the page is designed. This method takes few styles into consideration and ignores some, cleans and creates
a valid url link so as to keep it ready for the crawl method.
:param url: basae url of the current page
:param href: one of the hyper links of the page
:return: a well formed and valid http link
"""
if href.startswith('/'):
return "http://%s%s"%(url.netloc, href)
elif href.startswith('#'):
return "http://%s%s%s"%(url.netloc, url.path, href)
elif href.startswith('./'):
return "http://%s%s"%(url.netloc, href[1:])
elif not href.startswith('http'):
return "http://" + url.netloc + '/' + href
return href
def get_out_going_edges(self, url, html_body):
"""
This method encompasses the BFS along with the coupling with crawl and generator as it changes the state
of the unvisited map. Basically this method extracts the links that belong to the same domain as the start
page, cleans them with compose_url_from_href method and updates the map. This also avoids unnecessary traps
like href links pointing to 'javascript', 'mailto' etc.
:param url: current page url
:param html_body: current page's html content
:return: returns all the valid and wellformed out going links from this page
"""
soup = BeautifulSoup(html_body, "html.parser")
valid_links_for_this_page = []
for a in soup.find_all('a', href=True):
href = a['href']
href = self.compose_url_from_href(url, href.decode("utf-8"))
# clean the href so that it will have legitimate urls instead of #cluttered ones and q=param prints
href = urldefrag(href)[0] # skip intra links [this took time to find out !] ##1
# remove query params as only the path matters
if href.find('?') != -1:
href = href[:href.find('?')] ##2
new_page = urlparse(href)
# add to the queue only it it doesn't cause a cycle
# assumption: if a link ends with domain.com, assuming it can be crawled to make sitemap complete
if not str(new_page.netloc).endswith(self.start_page): # doesn't belong to domain
continue
if self.robot_allows(href) and \
not href in self.site_map.keys() and \
not href in self.unvisited and \
not 'javascript:' in href and \
not 'mailto:' in href:
self.unvisited.add(href)
valid_links_for_this_page.append(href)
return valid_links_for_this_page
def record_visit(self, url, headers, html_body):
"""
Any time a specific url of a site is changed, its last-modified date and time are kept in the page headers.
This info helps bots and crawlers to not to crawl the page if it has not been updated since last crawl.
This method is used to preserve the url crawled and its last-modified time along with assets scraped into
the container dictionary for later usage to generate sitemap and visualization network.
:param url: url of the just finished crawling page
:param headers: header information of the crawled page
:param html_body: html content of the page
:return: None
"""
if 'last-modified' in headers:
date = headers['Last-Modified']
else:
date = headers['Date']
self.site_map[url] = {
'date': date,
'assets': self.get_static_assets(html_body)
}
def get_static_assets(self, html_body):
"""
A html page could contain other links such as .css, .img. .mp4 and .js. All these files are not dynamic
though they could produce dynamic results. The code or text that exists in these files is constant and
static. These files are referred as static assets and for the definition of this challenge, I have chosen
to keep all the info in a single dictionary and extract them at the end for reports, results and stats.
:param html_body: html content of the page.
:return: returns a dictionary that encompasses .css, .img, ijs files as lists.
"""
# add static assets of the page .css, .js and image urls may be ?
soup = BeautifulSoup(html_body, "html.parser")
img = soup.findAll("img")
css = soup.findAll("link", {"rel": "stylesheet"})
# js is tricky: I faced an issue with inline javascript and ignoring it for the time being.
# an extract like html_body with just needed parts is a must for excluding inline scripts and styles.
jss = []
for x in soup.findAll('script'):
try:
list.append(x['src'])
except KeyError:
pass
csss = []
imgs = []
jss = []
for link in css:
csss.append(link['href'])
for link in img:
imgs.append(link['src'])
for link in jss:
jss.append(link['src'])
return {
'css': csss,
'img': imgs,
'js': jss
}
def crawl(self):
"""
The main driver method that crawls the pages. This main does below steps:
for every unvisited [vertex|page] that belongs to the requested domain:
crawl the page
record valid links and their last-modified-dates
:return: None
"""
page = self.unvisited.pop()
# if robot.txt is defined, use Disallow to avoid pages. domain.robot.txt doesn't exist so the crawler
# must find all the pages for report.
logging.info("Starting to Crawl Page: " + page)
url = urlparse(page)
try:
response = urlopen(page)
except:
logging.debug("Issue with the url: " + page)
return None
try:
html_body = response.read() # response.getcode()
response.close()
# record visit ans assets
self.record_visit(page, response.headers, html_body)
logging.debug("Queued Pages: {0}, Crawled Pages: {1}".format(len(self.unvisited), len(self.site_map)))
except:
logging.debug("Issue while opening url: " + page)
return None
connects = self.get_out_going_edges(url, html_body)
# simple Graph that keeps the order of the pages crawled.
for i, url in enumerate(connects):
self.network[page] = {
'to': connects,
'assets': {
'css': self.site_map[page]['assets']['css'],
'js': self.site_map[page]['assets']['js'],
'img': self.site_map[page]['assets']['img']
}
}
return None
def get_site_map(self):
"""
Returns the compiled sitemap structure
:return: sitemap data structure
"""
return self.site_map
def get_network_graph(self):
"""
Returns the compiled network in the order of the crawled pages
:return: network graph
"""
return self.network
def get_network_json_format(self):
"""
Returns the crawl traverse order sequence in json format
:return: network in json format
"""
return json.dumps(self.network)
def set_start_page(self, url):
"""
This could be useful if one is testing
:param url: start page to start the crawling.
:return:
"""
self.start_page = url
def robot_allows(self, link):
if not self.robotrules: return True
try:
if self.robot_txt_rules.can_fetch("*", link):
return True
return False
except:
return True
class spider(object):
CurLink = ""
linkURI = []
texts = []
Meta = {}
def __init__(self, link):
self.CurLink = link
self.r = RobotFileParser()
def crawl(self):
self.r.set_url(urlparse.unquote(self.CurLink))
self.r.read()
self.html = urlopen(self.CurLink).read()
self.bs = BeautifulSoup(self.html, "lxml")
for script in self.bs(["script", "style"]):
script.extract()
text = self.bs.get_text()
lines = (line.strip() for line in text.splitlines())
chunks = (phrase.strip() for line in lines
for phrase in line.split(" "))
for chunk in chunks:
if chunk:
self.texts.append(chunk)
# site = urlparse.urlsplit(self.CurLink).scheme + "://" + urlparse.urlsplit(self.CurLink).netloc + "/sitemap.aspx"
# r = requests.get(site)
if requests.get(
urlparse.urlsplit(self.CurLink).scheme + "://" +
urlparse.urlsplit(self.CurLink).netloc +
"/sitemap.aspx").ok == True:
root = etree.fromstring(
requests.get(
urlparse.urlsplit(self.CurLink).scheme + "://" +
urlparse.urlsplit(self.CurLink).netloc +
"/sitemap.xml").content)
for sitemap in root:
children = sitemap.getchildren()
self.linkURI.append(children[0].text)
elif requests.get(
urlparse.urlsplit(self.CurLink).scheme + "://" +
urlparse.urlsplit(self.CurLink).netloc +
"/sitemap.xml").ok == True:
root = etree.fromstring(
requests.get(
urlparse.urlsplit(self.CurLink).scheme + "://" +
urlparse.urlsplit(self.CurLink).netloc +
"/sitemap.xml").content)
for sitemap in root:
children = sitemap.getchildren()
self.linkURI.append(children[0].text)
else:
for link in self.bs.findAll('a', href=True):
aLink = urlparse.urljoin(self.CurLink, link['href'])
if (self.r.can_fetch("*", aLink)):
self.linkURI.append(aLink)
page = metadata_parser.MetadataParser(url=self.CurLink)
meta = page.metadata
keyw = "null"
descr = "null"
if (meta.get('meta').get('Keywords')):
keyw = meta['meta']['Keywords'].split(', ')
if (meta.get('meta').get('Description')):
descr = meta['meta']['Description']
self.Meta = {
'title': meta['page']['title'],
'url': meta['_internal']['url_actual'],
'description': descr,
'keyword': keyw
}
class SimpleCrawler:
USER_AGENT = 'SimpleCrawler/0.1'
HEADERS = {
'User-Agent': USER_AGENT,
'Accept-Encoding': 'gzip',
'Connection': 'keep-alive'
}
CONTENT_TYPE_PAT = re.compile(r'([^\s;]+)(.*charset=([^\s;]+))?', re.I)
def __init__(self, starturl, index_html='', maxlevel=1,
cookie_file=None, acldb=None, urldb=None, default_charset=None,
delay=0, timeout=300, debug=0):
(proto, self.hostport, _x, _y, _z) = urlsplit(starturl)
assert proto == 'http'
#Thread.__init__(self)
self.debug = debug
self.index_html = index_html
if cookie_file:
self.cookiejar = MozillaCookieJar(cookie_file)
self.cookiejar.load()
else:
self.cookiejar = None
self.robotstxt = RobotFileParser()
self.robotstxt.set_url(urljoin(starturl, '/robots.txt'))
self.robotstxt.read()
self.conn = None
self.urldb = urldb
self.acldb = acldb
self.curlevel = 0
self.delay = delay
self.timeout = timeout
self.default_charset = default_charset
if starturl.endswith('/'):
starturl += self.index_html
self.urls = [(starturl, maxlevel)]
self.crawled = {} # 1:injected, 2:crawled
return
def accept_url(self, url):
if url.endswith('/'):
url += self.index_html
if self.acldb and not self.acldb.allowed(url):
return None
return url
def inject_url(self, url):
if (not self.curlevel) or (not url) or (url in self.crawled): return False
if not self.robotstxt.can_fetch(self.USER_AGENT, url):
if self.debug:
print >>stderr, 'DISALLOW: %r' % url
return None
if self.debug:
print >>stderr, 'INJECT: %r' % url
self.crawled[url] = 1
self.urls.append((url, self.curlevel-1))
return True
def get1(self, url, maxretry=3, maxredirect=3):
if self.debug:
print >>stderr, 'GET: %r' % url
# loop
for rtry in range(maxredirect):
# forge urllib2.Request object.
req = Request(url)
# add cookie headers if necessary.
if self.cookiejar:
self.cookiejar.add_cookie_header(req)
headers = req.unredirected_hdrs
headers.update(self.HEADERS)
else:
headers = self.HEADERS
# get response.
for ctry in range(maxretry):
try:
if not self.conn:
print >>stderr, 'Making connection: %r...' % (self.hostport,)
self.conn = HTTPConnection(self.hostport)
self.conn.request('GET', req.get_selector().replace(' ',''), '', headers)
self.conn.sock.settimeout(self.timeout)
resp = self.conn.getresponse()
break
except BadStatusLine, x:
# connection closed unexpectedly
print >>stderr, 'Connection closed unexpectedly.'
# it restarts the connection...
self.conn.close()
self.conn = None
except socket.error, x:
# connection closed unexpectedly
print >>stderr, 'Socket error:', x
self.conn.close()
self.conn = None
else:
class Webpage(object):
"""
Objects that refer to individual webpages. If the url is scrapeable the
object will be filled with that data, indexed, and inserted into a database
to be searched.
"""
number_of_scraped_pages = 0
def __init__(self, url):
"""
Creates a webpage object and assigns it the provided url.
"""
self.url = url
if self.url not in black_list and self.url not in scraped_urls:
self.needs_to_be_scraped = True
else:
self.needs_to_be_scraped = False
def page_robot_scannable(self):
"""
Checks whether the page is allowed to be crawled
"""
if self.need_to_be_scraped is True:
# REFACTOR to remove try statement.
try:
headers = {'User-agent':settings.SPIDER_USER_AGENT}
self.urlparse = urlparse.urlparse(self.url)
self.robotcheck = RobotFileParser()
self.robotcheck.set_url('http://'+self.urlparse[1]+'/robots.txt') # Only works with http right now.
self.robotcheck.read()
self.need_to_be_scraped = self.robotcheck.can_fetch(settings.SPIDER_USER_AGENT, self.url)
except:
self.need_to_be_scraped = False
def get_page(self):
"""
The url is requested with a GET request. The page html is scraped
directly, while elements of it aee scraped in parse_page
"""
self.headers = {'User-agent':settings.SPIDER_USER_AGENT}
#REFACTOR to remove try
try:
self.request = requests.get(self.url, headers=headers)
self.pagehtml = BeautifulSoup(self.request.text) #REFACTOR, don't use BeautifulSoup
self.count = self.instanceID.next()
Webpage.number_of_scraped_pages += 1
except:
raise Exception
def get_visible_elements(self, element):
"""
Checks that the element is not contained in <style>, <script>, <head>,
<title> or [document]. It also cannot be commented out.
"""
if element.parent.name in ['style', 'script', '[document]', 'head', 'title']:
return False
elif re.match('<!--.*-->', str(element)):
return False
return True
def parse_page(self):
"""
This method parses the HTML page and extracts the title of the page,
the outgoing links, the number of outgoing links, and the text.
"""
self.title = self.pagehtml.find('title').text
self.page_text = self.pagehtml.findAll(text=true)
for item in filter(get_visible_elements, self.pagetext):
if item != '\n':
self.pagetext+= item
self.pagelinks = {}
for link in soup.findAll('a'):
self.pagelinks[link.get('href')] = 1
for link in self.pagehtml:
pass
# determine if link is relative or absolute. if relative, change it to absolute
def inverted_index_page_text(self):
"""
Iterates through the words in the page text and creates and adds them
to an index.
"""
self.pagetextlist = self.pagetext.split(' ') #Noted error: This catches punctuation along with words.
for index, word in enumerate(self.pagetextlist):
if word not in STOP_WORDS:
if not inverted_index.get(word):
inverted_index[word]={'url':self.url,'offsets':[index]}
else:
inverted_index[word]['offsets'].append(index)
def set_page_scraped(self):
"""
Once the page is scraped it is flagged as such
"""
self.needs_to_be_scraped = False
def _create_robot_file_parser(self, url):
host = urlparse.urlsplit(url)[1]
robots_url = urlparse.urlunsplit(('http', host, '/robots.txt', '', ''))
rp = RobotFileParser(robots_url)
rp.read()
return rp
class MarioDepth:
def __init__(self, starturl, callback, callpre=None, callfail=None, concount=MAXCONCOUNT, depth=2, accept_url_patterns=None, reject_url_patterns=None):
self.concount = concount
self.callback = callback
self.callpre = callpre
self.callfail = callfail
self.depth = depth
self.starturl = starturl
self.baseurl = URL.baseurl(starturl)
self.urls = []
self.crawled = {}
self.link_title_db = LinkTitleDB()
self.accept_url_patterns = accept_url_patterns
self.reject_url_patterns = reject_url_patterns
self.robotstxt = RobotFileParser()
self.robotstxt.set_url(urljoin(starturl, '/robots.txt'))
self.referer = starturl
try:
self.robotstxt.read()
except:
logger.debug(Traceback())
#self.lightcloud = LightCloud.connect('n0')
def __call__(self, n=None):
if n: self.concount = n
current_depth = self.depth
self.urls.append((self.starturl, current_depth))
while self.urls:
self.depth_get()
logger.debug('%d unprocessed urls'%(len(self.urls)))
def depth_get(self):
mario = MarioBatch(callback=self.next_depth, callpre=self.callpre, callfail=self.callfail)
pool = coros.CoroutinePool(max_size=len(self.urls))
while self.urls:
waiters = []
#self.add_job(mario)
counter = 0
while self.urls:
if counter > 9: break;
counter += 1
waiters.append(pool.execute(self.add_job, mario))
logger.debug('Depth break')
for waiter in waiters:
waiter.wait()
mario(self.concount)
def add_job(self, mario):
if not self.urls: return
url, depth = self.urls.pop()
if self.visited(url, depth): return
mario.add_job(url, args=depth)
def visited(self, url, depth):
#is_duplicate = URL.is_duplicate(url, self.lightcloud)
return depth==0 and is_duplicate or depth < self.depth and self.crawled.has_key(url) and self.crawled[url] == 2
def next_depth(self, response):
#with_timeout(1, self.lightcloud.set, LightCloud.crawled_url_key(response.effective_url), response.url, timeout_value=None)
for link, title in URL.link_title(response.body, response.effective_url):
if not self.inject_url(link, response.args):continue
self.link_title_db.add(link, response.effective_url, title)
if callable(self.callback): self.callback(response)
self.crawled[response.effective_url] = 2
if response.effective_url != response.url:
self.crawled[response.url] = 2
self.referer = response.effective_url
def inject_url(self, url, depth):
if not (depth and url and url not in self.crawled):
#logger.debug('IGNORE(%d): %r'%(depth, url))
return None
if isinstance(url, unicode): url = url.encode('utf-8')
if self.reject_url(url):
logger.debug('REJECT: %r' % url)
return None
try:
can_fetch = self.robotstxt.can_fetch(USER_AGENT['safari'], url)
except:
can_fetch = True
if self.baseurl!='http://hi.baidu.com/' and not can_fetch:
logger.debug('DISALLOW: %r' % url)
return None
logger.debug('INJECT(%d): %r' % (depth-1, url))
self.crawled[url] = 1
self.urls.append((url, depth-1))
return True
def reject_url(self, url):
return self.baseurl != URL.baseurl(url) and (not self.accept_url_patterns or not re.match('|'.join(self.accept_url_patterns), url) or self.reject_url_patterns or re.match('|'.join(self.reject_url_patterns), url))
class WebPage(object):
def __init__(self, url):
self.page_url = url
self.parsed_url = urlparse.urlparse(url)
self.lang = ""
self.isDownload = False
self.title = ""
self.text = ""
self.soup = None
self.robot = RobotFileParser()
def __normalize_link__(self, link):
if not link:
return None
if link.startswith('//'):
return self.parsed_url.scheme + ':' + link
elif link.startswith('/'):
return self.parsed_url.scheme + '://' + self.parsed_url.hostname + link
elif link.startswith('http://') or link.startswith('https://'):
return link
elif link.startswith("irc://"):
return None
elif link.startswith('#') or link.startswith('javascript:'):
return None
else:
return urlparse.urljoin(self.page_url, link)
def __delete_unnecessary_tags(self):
if self.soup is None:
return
if self.soup.title is None:
self.title = ""
else:
self.title = self.soup.title.string
for tag in self.soup(
['style', 'script', '[document]', 'head', 'title']):
tag.decompose()
def __get_stems(self, text):
if self.lang in LANGUAGES:
stemer = snowballstemmer.stemmer(LANGUAGES[self.lang])
else:
raise NotImplementedError("That lang not implemented")
stems_dict = dict()
for char in [",", ". ", "!", "?", " - ", "/n"]:
text = text.replace(char, " ")
for word in text.split():
stem_word = stemer.stemWord(word.lower())
if stem_word in stems_dict:
stems_dict[stem_word] += 1
else:
stems_dict[stem_word] = 1
return stems_dict
def download_page(self):
try:
self.robot.set_url("{0}://{1}/robots.txt".format(
self.parsed_url.scheme, self.parsed_url.hostname))
self.robot.read()
if self.robot.can_fetch("*", self.page_url):
response = requests.get(self.page_url, verify=False)
else:
return False
except requests.exceptions.InvalidSchema:
return False
except KeyError:
return False
except Exception:
return False
if response.status_code == 200:
self.soup = BeautifulSoup(response.text, "html.parser")
self.__delete_unnecessary_tags()
self.text = "".join(self.soup.strings)
try:
self.lang = detect(self.text)
except Exception:
self.lang = "en"
self.isDownload = True
return True
else:
return False
def get_links(self):
if not self.isDownload:
raise Exception("You should download page")
def get_links_generator():
for link in self.soup.find_all("a"):
normalized_link = self.__normalize_link__(link.get("href"))
if normalized_link is None:
continue
else:
yield normalized_link
return get_links_generator()
def get_text_stems(self):
if not self.isDownload:
raise Exception("You should download page")
return self.__get_stems(self.text)
def get_title_stems(self):
if not self.isDownload:
raise Exception("You should download page")
return self.__get_stems(self.title)
def get_domain(self):
return self.parsed_url.hostname
def test_parse(self):
from robotparser import RobotFileParser
rules=RobotFileParser()
rules.set_url("http://www.sogou.com/robots.txt")
rules.read()
self.assertEqual(rules.can_fetch("mozilla","http://www.sogou.com/sohu/robots.txt"),False)
class SiteMap():
def __init__(self, main_page=None, robotrules=True):
"""
Constuctor method that initializes the members that are used during crawling process
:param main_page: The root page that needs to be crawled for generation of sitemap
"""
logging.info("Consider Robot.txt ? ==> " + str(robotrules))
self.robotrules = robotrules
self.site_map = {
} # map that records the visits of urls, datemodified and assets
self.network = {
} # map that maintains the network/graph of webpages visited
# The intention of this map is for visual rendering using d3.js
self.unvisited = set(
[]) # a set to keep the list of urls yet to be visited
self.start_page = None # the root page, this is used to avoid cycle and keeping crawl
# process limited to single domain.
self.robot_txt_rules = None
if main_page:
self.unvisited.add(main_page)
try:
self.start_page = urlparse(main_page).netloc
except:
logging.error("Improper URL, Please provide a Valid Url:" +
main_page)
exit(0)
if self.robotrules == "True":
try:
logging.info("robot.txt respected")
self.robot_txt_rules = RobotFileParser()
self.robot_txt_rules.set_url(main_page + "/robots.txt")
self.robot_txt_rules.read()
except:
logging.error("Unable to read the robot.txt file")
self.robotrules = False # error reading robot.txt, ignore it forever
@timeit
def generate(self, site_map=None):
"""
This method holds the invoking control of the crawler method and drives the crawling process.
Basically a BFS style method that keeps popping the elements from the queue [self.unvisited set]
and scraping the urls.
Once the crawling process is done, this creates sitemap using the self.site_map dictionary with
just url, date-modified tags with dummy frequency and priorities.
:param site_map: name of the site_map file so as to create xml entries.
:return:
"""
while self.unvisited:
self.crawl()
# create xml from the site_map dictionary
header = """<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
xmlns:xhtml="http://www.w3.org/1999/xhtml"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9
http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd">
"""
footer = """\n</urlset>\n"""
entry = "\t<url>\n\
\t\t<loc>%s</loc>\n\
\t\t<lastmod>%s</lastmod>\n\
\t\t<changefreq>monthly</changefreq>\n\
\t\t<priority> 1 </priority>\n\
\t</url>\
"
xml = header
for url in self.site_map.keys():
xml += entry % (url, self.site_map[url]['date']) + "\n"
xml += footer
if site_map != None:
self.write_to_file(site_map, xml)
else:
self.write_to_file("sitemap.xml", xml)
return xml
def write_to_file(self, file_name, content):
"""
A utility method to just write the contents of the file into a given file name.
Alert: This overwrites if the file does exist in the current directory.
:param file_name: name of the file, sitemap in our case.
:param content: contents of the file
:return: None
"""
f = open(file_name, 'w')
f.write(content)
f.close()
def compose_url_from_href(self, url, href):
"""
There are different ways a href could specify a location and it varies in different ways based on how
the page is designed. This method takes few styles into consideration and ignores some, cleans and creates
a valid url link so as to keep it ready for the crawl method.
:param url: basae url of the current page
:param href: one of the hyper links of the page
:return: a well formed and valid http link
"""
if href.startswith('/'):
return "http://%s%s" % (url.netloc, href)
elif href.startswith('#'):
return "http://%s%s%s" % (url.netloc, url.path, href)
elif href.startswith('./'):
return "http://%s%s" % (url.netloc, href[1:])
elif not href.startswith('http'):
return "http://" + url.netloc + '/' + href
return href
def get_out_going_edges(self, url, html_body):
"""
This method encompasses the BFS along with the coupling with crawl and generator as it changes the state
of the unvisited map. Basically this method extracts the links that belong to the same domain as the start
page, cleans them with compose_url_from_href method and updates the map. This also avoids unnecessary traps
like href links pointing to 'javascript', 'mailto' etc.
:param url: current page url
:param html_body: current page's html content
:return: returns all the valid and wellformed out going links from this page
"""
soup = BeautifulSoup(html_body, "html.parser")
valid_links_for_this_page = []
for a in soup.find_all('a', href=True):
href = a['href']
href = self.compose_url_from_href(url, href.decode("utf-8"))
# clean the href so that it will have legitimate urls instead of #cluttered ones and q=param prints
href = urldefrag(href)[
0] # skip intra links [this took time to find out !] ##1
# remove query params as only the path matters
if href.find('?') != -1:
href = href[:href.find('?')] ##2
new_page = urlparse(href)
# add to the queue only it it doesn't cause a cycle
# assumption: if a link ends with domain.com, assuming it can be crawled to make sitemap complete
if not str(new_page.netloc).endswith(
self.start_page): # doesn't belong to domain
continue
if self.robot_allows(href) and \
not href in self.site_map.keys() and \
not href in self.unvisited and \
not 'javascript:' in href and \
not 'mailto:' in href:
self.unvisited.add(href)
valid_links_for_this_page.append(href)
return valid_links_for_this_page
def record_visit(self, url, headers, html_body):
"""
Any time a specific url of a site is changed, its last-modified date and time are kept in the page headers.
This info helps bots and crawlers to not to crawl the page if it has not been updated since last crawl.
This method is used to preserve the url crawled and its last-modified time along with assets scraped into
the container dictionary for later usage to generate sitemap and visualization network.
:param url: url of the just finished crawling page
:param headers: header information of the crawled page
:param html_body: html content of the page
:return: None
"""
if 'last-modified' in headers:
date = headers['Last-Modified']
else:
date = headers['Date']
self.site_map[url] = {
'date': date,
'assets': self.get_static_assets(html_body)
}
def get_static_assets(self, html_body):
"""
A html page could contain other links such as .css, .img. .mp4 and .js. All these files are not dynamic
though they could produce dynamic results. The code or text that exists in these files is constant and
static. These files are referred as static assets and for the definition of this challenge, I have chosen
to keep all the info in a single dictionary and extract them at the end for reports, results and stats.
:param html_body: html content of the page.
:return: returns a dictionary that encompasses .css, .img, ijs files as lists.
"""
# add static assets of the page .css, .js and image urls may be ?
soup = BeautifulSoup(html_body, "html.parser")
img = soup.findAll("img")
css = soup.findAll("link", {"rel": "stylesheet"})
# js is tricky: I faced an issue with inline javascript and ignoring it for the time being.
# an extract like html_body with just needed parts is a must for excluding inline scripts and styles.
jss = []
for x in soup.findAll('script'):
try:
list.append(x['src'])
except KeyError:
pass
csss = []
imgs = []
jss = []
for link in css:
csss.append(link['href'])
for link in img:
imgs.append(link['src'])
for link in jss:
jss.append(link['src'])
return {'css': csss, 'img': imgs, 'js': jss}
def crawl(self):
"""
The main driver method that crawls the pages. This main does below steps:
for every unvisited [vertex|page] that belongs to the requested domain:
crawl the page
record valid links and their last-modified-dates
:return: None
"""
page = self.unvisited.pop()
# if robot.txt is defined, use Disallow to avoid pages. domain.robot.txt doesn't exist so the crawler
# must find all the pages for report.
logging.info("Starting to Crawl Page: " + page)
url = urlparse(page)
try:
response = urlopen(page)
except:
logging.debug("Issue with the url: " + page)
return None
try:
html_body = response.read() # response.getcode()
response.close()
# record visit ans assets
self.record_visit(page, response.headers, html_body)
logging.debug("Queued Pages: {0}, Crawled Pages: {1}".format(
len(self.unvisited), len(self.site_map)))
except:
logging.debug("Issue while opening url: " + page)
return None
connects = self.get_out_going_edges(url, html_body)
# simple Graph that keeps the order of the pages crawled.
for i, url in enumerate(connects):
self.network[page] = {
'to': connects,
'assets': {
'css': self.site_map[page]['assets']['css'],
'js': self.site_map[page]['assets']['js'],
'img': self.site_map[page]['assets']['img']
}
}
return None
def get_site_map(self):
"""
Returns the compiled sitemap structure
:return: sitemap data structure
"""
return self.site_map
def get_network_graph(self):
"""
Returns the compiled network in the order of the crawled pages
:return: network graph
"""
return self.network
def get_network_json_format(self):
"""
Returns the crawl traverse order sequence in json format
:return: network in json format
"""
return json.dumps(self.network)
def set_start_page(self, url):
"""
This could be useful if one is testing
:param url: start page to start the crawling.
:return:
"""
self.start_page = url
def robot_allows(self, link):
if not self.robotrules: return True
try:
if self.robot_txt_rules.can_fetch("*", link):
return True
return False
except:
return True
class SimpleCrawler:
USER_AGENT = 'SimpleCrawler/0.1'
HEADERS = {
'User-Agent': USER_AGENT,
'Accept-Encoding': 'gzip',
'Connection': 'keep-alive'
}
CONTENT_TYPE_PAT = re.compile(r'([^\s;]+)(.*charset=([^\s;]+))?', re.I)
def __init__(self,
starturl,
index_html='',
maxlevel=1,
cookie_file=None,
acldb=None,
urldb=None,
default_charset=None,
delay=0,
timeout=300,
debug=0):
(proto, self.hostport, _x, _y, _z) = urlsplit(starturl)
assert proto == 'http'
#Thread.__init__(self)
self.debug = debug
self.index_html = index_html
if cookie_file:
self.cookiejar = MozillaCookieJar(cookie_file)
self.cookiejar.load()
else:
self.cookiejar = None
self.robotstxt = RobotFileParser()
self.robotstxt.set_url(urljoin(starturl, '/robots.txt'))
self.robotstxt.read()
self.conn = None
self.urldb = urldb
self.acldb = acldb
self.curlevel = 0
self.delay = delay
self.timeout = timeout
self.default_charset = default_charset
if starturl.endswith('/'):
starturl += self.index_html
self.urls = [(starturl, maxlevel)]
self.crawled = {} # 1:injected, 2:crawled
return
def accept_url(self, url):
if url.endswith('/'):
url += self.index_html
if self.acldb and not self.acldb.allowed(url):
return None
return url
def inject_url(self, url):
if (not self.curlevel) or (not url) or (url in self.crawled):
return False
if not self.robotstxt.can_fetch(self.USER_AGENT, url):
if self.debug:
print >> stderr, 'DISALLOW: %r' % url
return None
if self.debug:
print >> stderr, 'INJECT: %r' % url
self.crawled[url] = 1
self.urls.append((url, self.curlevel - 1))
return True
def get1(self, url, maxretry=3, maxredirect=3):
if self.debug:
print >> stderr, 'GET: %r' % url
# loop
for rtry in range(maxredirect):
# forge urllib2.Request object.
req = Request(url)
# add cookie headers if necessary.
if self.cookiejar:
self.cookiejar.add_cookie_header(req)
headers = req.unredirected_hdrs
headers.update(self.HEADERS)
else:
headers = self.HEADERS
# get response.
for ctry in range(maxretry):
try:
if not self.conn:
print >> stderr, 'Making connection: %r...' % (
self.hostport, )
self.conn = HTTPConnection(self.hostport)
self.conn.request('GET',
req.get_selector().replace(' ', ''), '',
headers)
self.conn.sock.settimeout(self.timeout)
resp = self.conn.getresponse()
break
except BadStatusLine, x:
# connection closed unexpectedly
print >> stderr, 'Connection closed unexpectedly.'
# it restarts the connection...
self.conn.close()
self.conn = None
except socket.error, x:
# connection closed unexpectedly
print >> stderr, 'Socket error:', x
self.conn.close()
self.conn = None
else:
class CrawlerProcess:
def __init__(self, index, lock):
self.current_process_id = index
self.lock = lock
number_of_retries = 0
#print("[CREATED CRAWLER PROCESS]", self.current_process_id)
# Create the chrome driver with which we will fetch and parse sites
chrome_options = webdriver.ChromeOptions()
chrome_options.add_argument("headless")
self.driver = webdriver.Chrome(chrome_options=chrome_options)
"""
site is a dictionary with all the fields from the database (domain, robots_content, sitemap_content)
"""
self.site = None
"""
robots_parser is an object which allows us to use the robots.txt file
"""
self.robots_parser = None
"""
Holds all the pages which will be added to the frontier at the end of each run
"""
self.pages_to_add_to_frontier = []
"""
current_page is a dictionary with an id (database id for updating) and url field
"""
self.current_page = database_handler.get_page_from_frontier(self.lock)
"""
If a page was fetched from the frontier the crawler can continue, otherwise try again in DELAY seconds
If the frontier is still empty after MAX_NUMBER_OF_RETRIES was reached, we can assume that the frontier is
really empty and no crawler process is going to insert new pages
"""
while self.current_page or number_of_retries < MAX_NUMBER_OF_RETRIES:
if self.current_page:
number_of_retries = 0
try:
self.crawl()
except Exception as error:
print(
"[CRAWLER PROCESS] An unhandled error occurred while parsing page: {}"
.format(self.current_page["url"]), error)
else:
# No page was fetched from the frontier, try again in DELAY seconds
number_of_retries += 1
print(
"[CRAWLER PROCESS] Frontier is empty, retrying in 10 seconds",
self.current_process_id)
time.sleep(DELAY)
# Reset all variables after a page was successfully transferred from the frontier
self.current_page = database_handler.get_page_from_frontier(
self.lock)
self.site = None
self.robots_parser = None
self.pages_to_add_to_frontier = []
self.quit()
print(
"[STOPPED CRAWLER PROCESS] Frontier is empty after several tries",
self.current_process_id)
def crawl(self):
#print(" {} - [CRAWLING PAGE]".format(self.current_process_id), self.current_page["url"])
domain = self.get_domain_url(self.current_page["url"])
self.site = database_handler.get_site(domain)
if self.site is None:
self.create_site(domain)
else:
if self.site["robots_content"] is not None:
# Create robots_parser from robots.txt saved in the database
self.parse_robots(self.site["robots_content"])
self.current_page["site_id"] = self.site["id"]
self.current_page["accessed_time"] = datetime.now()
if self.allowed_to_crawl_current_page(
self.current_page["url"]) is False:
#print(" [CRAWLING] Robots do not allow this page to be crawled: {}".format(self.current_page["url"]))
self.current_page["page_type_code"] = PAGE_TYPES["disallowed"]
self.current_page["http_status_code"] = 500
database_handler.remove_page_from_frontier(self.current_page)
return
else:
# If a crawl delay is available in robots wait until the page can be crawled then continue
self.wait_for_crawl_delay_to_elapse()
# The crawler is allowed to crawl the current site, therefore we can perform a request
page_response = self.fetch_response(self.current_page["url"])
if page_response:
# No errors while fetching the response
content_type = ""
if "content-type" in page_response.headers:
# Content type is not necessarily always present (e. g. when Transfer-Encoding is set)
content_type = page_response.headers['content-type']
self.current_page["http_status_code"] = page_response.status_code
if CONTENT_TYPES["HTML"] in content_type:
# We got an HTML page
html_content = self.fetch_rendered_page_source(
self.current_page["url"])
if html_content is not None:
if self.is_duplicate_page(html_content):
print(
" [CRAWLING] Found page duplicate, that has already been parsed: ",
self.current_page["url"])
self.current_page["page_type_code"] = PAGE_TYPES[
"duplicate"]
self.current_page[
"hash_content"] = hash_driver.create_content_hash(
html_content)
else:
# page is not treated as duplicate page - insert hash signature to db
database_handler.insert_page_signatures(
self.current_page["id"],
self.current_page["hash_signature"])
self.current_page["page_type_code"] = PAGE_TYPES[
"html"]
self.current_page["html_content"] = html_content
self.current_page[
"hash_content"] = hash_driver.create_content_hash(
html_content)
parsed_page = self.parse_page(
self.current_page["html_content"])
if len(parsed_page['links']):
for link in parsed_page['links']:
self.add_page_to_frontier_array(link)
if len(parsed_page['images']):
for image_url in parsed_page['images']:
self.add_page_to_frontier_array(image_url)
else:
# An error occurred while rendering page
self.current_page["page_type_code"] = PAGE_TYPES["error"]
self.current_page["http_status_code"] = 500
elif CONTENT_TYPES["IMG"] in content_type:
# We can be pretty sure that we have an image
self.current_page["page_type_code"] = PAGE_TYPES["image"]
filename = self.get_image_filename(self.current_page["url"])
image_data = {
"page_id": self.current_page["id"],
"content_type": content_type,
"data": page_response.content,
"data_size": len(page_response.content),
"accessed_time": datetime.now(),
"filename": filename
}
database_handler.insert_image_data(image_data)
else:
# The crawler detected a non-image binary file
self.current_page["page_type_code"] = PAGE_TYPES["binary"]
data_type_code = None
# Find the correct data_type_code from all the content types
for code, value in CONTENT_TYPES.items():
if content_type == value:
data_type_code = code
if data_type_code is None:
# The content type is not in the allowed values, therefore we can ignore it
testing = None
#print(" [CRAWLING] Page response content-type is not in CONTENT_TYPES: ", content_type)
else:
page_data = {
"page_id": self.current_page["id"],
"data_type_code": data_type_code,
"data": page_response.content,
"data_size": len(page_response.content)
}
database_handler.insert_page_data(page_data)
else:
# An error occurred while fetching page (SSL certificate error, timeout, etc.)
self.current_page["page_type_code"] = PAGE_TYPES["error"]
self.current_page["http_status_code"] = 500
# Update the page in the database, remove FRONTIER type and replace it with the correct one
database_handler.remove_page_from_frontier(self.current_page)
# Add all the links from the page and sitemap to the frontier
database_handler.add_pages_to_frontier(self.pages_to_add_to_frontier)
#print(" {} - [CRAWLING] Finished crawling".format(self.current_process_id))
"""
Fetch a response from the url, so that we get the status code and find out if any errors occur while fetching
(some sites for example require a certificate to connect, some sites timeout, etc.)
"""
def fetch_response(self, url):
try:
response = requests.get(url)
return response
except requests.exceptions.RequestException as exception:
print(" [CRAWLING - ERROR]", exception)
return None
"""
Create a new site object and insert it into the database
"""
def create_site(self, domain):
# We need to create a new site object
self.site = {"domain": domain}
robots_content = self.fetch_robots(domain)
sitemap_content = None
if robots_content is not None:
# Create robots_parser from fetched robots.txt
self.parse_robots(robots_content)
sitemaps = self.robots_parser.get_sitemaps()
if len(sitemaps) > 0:
for sitemap_url in sitemaps:
sitemap_content = self.fetch_sitemap(sitemap_url)
if sitemap_content is not None:
self.parse_sitemap(sitemap_content)
self.site["robots_content"] = robots_content
self.site["sitemap_content"] = sitemap_content
# Insert the new site into database and return the id
self.site["id"] = database_handler.insert_site(self.site)
"""
Fetch and render the site in the chrome driver then return the resulting html so that it can be saved in the
current page html_content
"""
def fetch_rendered_page_source(self, url):
try:
self.driver.get(url)
return self.driver.page_source
except Exception as error:
print(" [CRAWLING] Error while fetching rendered page source",
error)
return None
"""
Get the domain name of the current site so that we can check if the site is already in the database or if we
have to create it
"""
def get_domain_url(self, url):
parsed_uri = urlparse(url)
return '{uri.scheme}://{uri.netloc}/'.format(uri=parsed_uri)
"""
Get the filename from an online image resource
https://stackoverflow.com/questions/10552188/python-split-url-to-find-image-name-and-extension
"""
def get_image_filename(self, image_url):
filename = image_url.split('/')[-1]
return filename
def fetch_robots(self, domain):
response = self.fetch_response(domain + "/robots.txt")
# We need to check if the returned file is actually a txt file, because some sites route back to the index page
if response and response.status_code is 200 and "text/plain" in response.headers[
'content-type']:
return response.text
return None
def fetch_sitemap(self, sitemap_url):
response = self.fetch_response(sitemap_url)
if response and response.status_code is 200:
# Sitemap found
return response.text
return None
"""
This function parses the robots.txt from memory using the modified robotparser class
The self.robots_parser includes functions to check if the parser is allowed to parse a certain site
"""
def parse_robots(self, robots_text):
self.robots_parser = RobotFileParser(robots_text)
self.robots_parser.read()
"""
https://stackoverflow.com/questions/31276001/parse-xml-sitemap-with-python
This only works for the standard XML sitemap
"""
def parse_sitemap(self, sitemap_xml):
try:
soup = BeautifulSoup(sitemap_xml, 'lxml')
sitemap_tags = soup.find_all("loc")
if sitemap_tags is None:
return
for sitemap_tag in sitemap_tags:
url = self.get_parsed_url(sitemap_tag.text)
if url:
self.add_page_to_frontier_array(url)
except Exception as error:
print(error)
"""
Checks if robots are set for the current site and if they allow the crawling of the current page
"""
def allowed_to_crawl_current_page(self, url):
if self.robots_parser is not None:
return self.robots_parser.can_fetch('*', url)
return True
"""
Checks if crawl-delay property is set and if it exists check if the required time has elapsed
"""
def wait_for_crawl_delay_to_elapse(self):
try:
if self.robots_parser is not None:
crawl_delay = self.robots_parser.crawl_delay('*')
if crawl_delay is not None:
if "last_crawled_at" in self.site and self.site[
"last_crawled_at"] is not None:
site_last_crawled_at = self.site["last_crawled_at"]
can_crawl_again_at = site_last_crawled_at + timedelta(
seconds=crawl_delay)
current_time = datetime.now()
time_difference = (can_crawl_again_at -
current_time).total_seconds()
if time_difference > 0:
#print(" [CRAWLING] Crawl delay has not yet elapsed for site: {}".format(
# self.site["domain"]))
time.sleep(crawl_delay)
except Exception as error:
print(" [CRAWLING] Error while handling crawl delay", error)
"""
Use the chrome driver to fetch all links and image sources in the rendered page (the driver already returns
absolute urls)
Note: Sometimes throws StaleElementReferenceException, need to check what that's about. The exception itself
just means that the desired element is no longer rendered in DOM. Maybe the memory was getting low, since I got the
error when I was running 10 crawler processes.
"""
def parse_page(self, html_content):
links = []
images = []
try:
browser = self.driver
anchor_tags = browser.find_elements_by_tag_name("a")
for anchor_tag in anchor_tags:
href = anchor_tag.get_attribute("href")
url = self.get_parsed_url(href)
if url:
links.append(url)
image_tags = browser.find_elements_by_tag_name("img")
for image_tag in image_tags:
src = image_tag.get_attribute("src")
if src:
image_url = self.get_parsed_image_url(src)
if image_url:
images.append(image_url)
soup = BeautifulSoup(html_content, 'html.parser')
script_tags = soup.findAll('script')
for script_tag in script_tags:
links_from_javascript = self.parse_links_from_javacript(
script_tag.text)
for link in links_from_javascript:
links.append(self.get_parsed_url(link))
return {"links": links, "images": images}
except Exception as error:
print("[ERROR WHILE RENDERING WITH WEB DRIVER]", error)
return {"links": links, "images": images}
"""
Find all the hrefs that are set in javascript code (window.location changes)
"""
def parse_links_from_javacript(self, javascript_text):
links = []
try:
links = re.findall(
r'(http://|https://)([\w_-]+(?:(?:\.[\w_-]+)+))([\w.,@?^=%&:/~+#-]*[\w@?^=%&/~+#-])?',
javascript_text)
if not links:
return []
links = [''.join(link) for link in links]
except Exception as error:
print(" [CRAWLING] Error while parsing links from Javascript",
error)
return links
"""
Create a parsed url (ignore javascript and html actions, remove hashes, fix relative urls etc.)
"""
# TODO: remove index.html index.php
def get_parsed_url(self, url):
if url is None or url is "":
return None
domain = self.site["domain"]
if not url.startswith("http"):
# Since the chrome driver returns absolute urls, the url is most likely javascript or action
if 'javascript:' in url:
# This is just javascript code inside a href
return None
if ('mailto:' in url) or ('tel:' in url):
# This is an action inside a href
return None
if url[0] is "#":
# Link starts with a # (it's a target link)
return None
if url is "/":
# This is the index page, which we already have in the frontier
return None
if url.startswith("www"):
url = "http://{}".format(url).strip()
"""
Fix relative urls just in case
This function might not work correctly since it's almost impossible to know which root url the link
takes when it's added to the site
"""
if url[0] == "/":
if domain[-1] == "/":
# Make sure only one slash is present
url = url[1:]
else:
if domain[-1] != "/":
url = "/{}".format(url)
url = "{}{}".format(domain, url).strip()
# Remove everything after the hash
if "#" in url:
url = url.split("#")[0]
# Encode special characters (the second parameter are characters that the encoder will not encode)
url = quote(url.encode("UTF-8"), ':/-_.~&?+=')
return url
"""
Parse image urls
"""
def get_parsed_image_url(self, url):
if url is None or url is "":
return None
# Do not parse base64 images
if url.startswith("data:image"):
return None
if not url.startswith("http"):
# This is very unlikely, since the chrome driver returns all the image sources with absolute urls
domain = self.site["domain"]
"""
Fix relative urls just in case
This function might not work correctly since it's almost impossible to know which root url the link
takes when it's added to the site
"""
if url[0] == "/":
if domain[-1] == "/":
# Make sure only one slash is present
url = url[1:]
else:
if domain[-1] != "/":
url = "/{}".format(url)
# Create an absolute url
url = "{}{}".format(domain, url).strip()
return url
"""
The duplicate page should not have the html_content value set, page_type_code should be DUPLICATE and
that's it
"""
def is_duplicate_page(self, html_content):
# sha256 digest of complete html_content
h = hash_driver.create_content_hash(html_content)
# first check if page is exact copy of already parsed documents
if database_handler.find_page_duplicate(h):
return True
else:
# create set of hash shingles
# in order to prevent pages using lots of same tags to be treated as similar, remove html tags
hash_set = hash_driver.text_to_shingle_set(
self.remove_markups(html_content))
# hash signature will be inserted to db later
self.current_page["hash_signature"] = hash_set
# calculate similarity between current document and already parsed documents using Jaccard similarity
similarity = database_handler.calculate_biggest_similarity(
hash_set)
#print("SIMILARITY: ", similarity)
return similarity > MAX_SIMILARITY
"""
Remove markup tags from html content
"""
def remove_markups(self, html_content):
return BeautifulSoup(html_content, "html.parser").text
def add_page_to_frontier_array(self, page_url):
page_domain = self.get_domain_url(page_url)
if ALLOWED_DOMAIN in page_domain:
# Only add pages in the allowed domain
self.pages_to_add_to_frontier.append({
"from":
self.current_page["id"],
"to":
page_url
})
def quit(self):
self.driver.quit()
# update time is old, we update it
if rp.mtime() < (time.time() - settings.ROBOTS_TXT_CACHE):
self.logger.info("Refresh %s/robots.txt cache" % hostname)
try:
rp.read()
except Exception, e:
print(e)
self.logger.info("Unable to get or parse %s/robots.txt" % hostname)
rp.disallow_all = False
rp.allow_all = True
else:
self.logger.debug("Retrieve cached %s/robots.txt" % hostname)
else:
# First (or very long) time we see this domain, create a new
# RobotFileParser and read it once
self.logger.info("First hit on %s/robots.txt" % hostname)
rp = RobotFileParser(url="%s://%s/robots.txt" % (scheme, hostname))
try:
rp.read()
except Exception, e:
print(e)
self.logger.info("Unable to get or parse %s/robots.txt" % hostname)
rp.disallow_all = False
rp.allow_all = True
# In any case, we update the last robotstxt fetched time
rp.modified()
self.robotstxt[hostname] = rp
return rp
def _create_robot_file_parser(self, url):
host = urlparse.urlsplit(url)[1]
robots_url = urlparse.urlunsplit(('http', host, '/robots.txt', '', ''))
rp = RobotFileParser(robots_url)
rp.read()
return rp