D7net Mini Sh3LL v1
Current File : //../../lib/ssl/../grub/../galera/../python3.8/urllib/robotparser.py |
""" robotparser.py
Copyright (C) 2000 Bastian Kleineidam
You can choose between two licenses when using this package:
1) GNU GPLv2
2) PSF license for Python 2.2
The robots.txt Exclusion Protocol is implemented as specified in
http://www.robotstxt.org/norobots-rfc.txt
"""
import collections
import urllib.parse
import urllib.request
__all__ = ["RobotFileParser"]
RequestRate = collections.namedtuple("RequestRate", "requests seconds")
class RobotFileParser:
""" This class provides a set of methods to read, parse and answer
questions about a single robots.txt file.
"""
def __init__(self, url=''):
self.entries = []
self.sitemaps = []
self.default_entry = None
self.disallow_all = False
self.allow_all = False
self.set_url(url)
self.last_checked = 0
def mtime(self):
"""Returns the time the robots.txt file was last fetched.
This is useful for long-running web spiders that need to
check for new robots.txt files periodically.
"""
return self.last_checked
def modified(self):
"""Sets the time the robots.txt file was last fetched to the
current time.
"""
import time
self.last_checked = time.time()
def set_url(self, url):
"""Sets the URL referring to a robots.txt file."""
self.url = url
self.host, self.path = urllib.parse.urlparse(url)[1:3]
def read(self):
"""Reads the robots.txt URL and feeds it to the parser."""
try:
f = urllib.request.urlopen(self.url)
except urllib.error.HTTPError as err:
if err.code in (401, 403):
self.disallow_all = True
elif err.code >= 400 and err.code < 500:
self.allow_all = True
else:
raw = f.read()
self.parse(raw.decode("utf-8").splitlines())
def _add_entry(self, entry):
if "*" in entry.useragents:
# the default entry is considered last
if self.default_entry is None:
# the first default entry wins
self.default_entry = entry
else:
self.entries.append(entry)
def parse(self, lines):
"""Parse the input lines from a robots.txt file.
We allow that a user-agent: line is not preceded by
one or more blank lines.
"""
# states:
# 0: start state
# 1: saw user-agent line
# 2: saw an allow or disallow line
state = 0
entry = Entry()
self.modified()
for line in lines:
if not line:
if state == 1:
entry = Entry()
state = 0
elif state == 2:
self._add_entry(entry)
entry = Entry()
state = 0
# remove optional comment and strip line
i = line.find('#')
if i >= 0:
line = line[:i]
line = line.strip()
if not line:
continue
line = line.split(':', 1)
if len(line) == 2:
line[0] = line[0].strip().lower()
line[1] = urllib.parse.unquote(line[1].strip())
if line[0] == "user-agent":
if state == 2:
self._add_entry(entry)
entry = Entry()
entry.useragents.append(line[1])
state = 1
elif line[0] == "disallow":
if state != 0:
entry.rulelines.append(RuleLine(line[1], False))
state = 2
elif line[0] == "allow":
if state != 0:
entry.rulelines.append(RuleLine(line[1], True))
state = 2
elif line[0] == "crawl-delay":
if state != 0:
# before trying to convert to int we need to make
# sure that robots.txt has valid syntax otherwise
# it will crash
if line[1].strip().isdigit():
entry.delay = int(line[1])
state = 2
elif line[0] == "request-rate":
if state != 0:
numbers = line[1].split('/')
# check if all values are sane
if (len(numbers) == 2 and numbers[0].strip().isdigit()
and numbers[1].strip().isdigit()):
entry.req_rate = RequestRate(int(numbers[0]), int(numbers[1]))
state = 2
elif line[0] == "sitemap":
# According to http://www.sitemaps.org/protocol.html
# "This directive is independent of the user-agent line,
# so it doesn't matter where you place it in your file."
# Therefore we do not change the state of the parser.
self.sitemaps.append(line[1])
if state == 2:
self._add_entry(entry)
def can_fetch(self, useragent, url):
"""using the parsed robots.txt decide if useragent can fetch url"""
if self.disallow_all:
return False
if self.allow_all:
return True
# Until the robots.txt file has been read or found not
# to exist, we must assume that no url is allowable.
# This prevents false positives when a user erroneously
# calls can_fetch() before calling read().
if not self.last_checked:
return False
# search for given user agent matches
# the first match counts
parsed_url = urllib.parse.urlparse(urllib.parse.unquote(url))
url = urllib.parse.urlunparse(('','',parsed_url.path,
parsed_url.params,parsed_url.query, parsed_url.fragment))
url = urllib.parse.quote(url)
if not url:
url = "/"
for entry in self.entries:
if entry.applies_to(useragent):
return entry.allowance(url)
# try the default entry last
if self.default_entry:
return self.default_entry.allowance(url)
# agent not found ==> access granted
return True
def crawl_delay(self, useragent):
if not self.mtime():
return None
for entry in self.entries:
if entry.applies_to(useragent):
return entry.delay
if self.default_entry:
return self.default_entry.delay
return None
def request_rate(self, useragent):
if not self.mtime():
return None
for entry in self.entries:
if entry.applies_to(useragent):
return entry.req_rate
if self.default_entry:
return self.default_entry.req_rate
return None
def site_maps(self):
if not self.sitemaps:
return None
return self.sitemaps
def __str__(self):
entries = self.entries
if self.default_entry is not None:
entries = entries + [self.default_entry]
return '\n\n'.join(map(str, entries))
class RuleLine:
"""A rule line is a single "Allow:" (allowance==True) or "Disallow:"
(allowance==False) followed by a path."""
def __init__(self, path, allowance):
if path == '' and not allowance:
# an empty value means allow all
allowance = True
path = urllib.parse.urlunparse(urllib.parse.urlparse(path))
self.path = urllib.parse.quote(path)
self.allowance = allowance
def applies_to(self, filename):
return self.path == "*" or filename.startswith(self.path)
def __str__(self):
return ("Allow" if self.allowance else "Disallow") + ": " + self.path
class Entry:
"""An entry has one or more user-agents and zero or more rulelines"""
def __init__(self):
self.useragents = []
self.rulelines = []
self.delay = None
self.req_rate = None
def __str__(self):
ret = []
for agent in self.useragents:
ret.append(f"User-agent: {agent}")
if self.delay is not None:
ret.append(f"Crawl-delay: {self.delay}")
if self.req_rate is not None:
rate = self.req_rate
ret.append(f"Request-rate: {rate.requests}/{rate.seconds}")
ret.extend(map(str, self.rulelines))
return '\n'.join(ret)
def applies_to(self, useragent):
"""check if this entry applies to the specified agent"""
# split the name token and make it lower case
useragent = useragent.split("/")[0].lower()
for agent in self.useragents:
if agent == '*':
# we have the catch-all agent
return True
agent = agent.lower()
if agent in useragent:
return True
return False
def allowance(self, filename):
"""Preconditions:
- our agent applies to this entry
- filename is URL decoded"""
for line in self.rulelines:
if line.applies_to(filename):
return line.allowance
return True
AnonSec - 2021 | Recode By D7net