symbian-qemu-0.9.1-12/python-win32-2.6.1/lib/robotparser.py
author martin.trojer@nokia.com
Fri, 31 Jul 2009 15:01:17 +0100
changeset 1 2fb8b9db1c86
permissions -rw-r--r--
Initial QEMU (symbian-qemu-0.9.1-12) import
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     1
""" robotparser.py
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     3
    Copyright (C) 2000  Bastian Kleineidam
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     4
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     5
    You can choose between two licenses when using this package:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     6
    1) GNU GPLv2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     7
    2) PSF license for Python 2.2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     8
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
     9
    The robots.txt Exclusion Protocol is implemented as specified in
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    10
    http://info.webcrawler.com/mak/projects/robots/norobots-rfc.html
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    11
"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    12
import urlparse
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    13
import urllib
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    14
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    15
__all__ = ["RobotFileParser"]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    16
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    17
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    18
class RobotFileParser:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    19
    """ This class provides a set of methods to read, parse and answer
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    20
    questions about a single robots.txt file.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    21
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    22
    """
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    23
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    24
    def __init__(self, url=''):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    25
        self.entries = []
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    26
        self.default_entry = None
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    27
        self.disallow_all = False
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    28
        self.allow_all = False
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    29
        self.set_url(url)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    30
        self.last_checked = 0
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    31
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    32
    def mtime(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    33
        """Returns the time the robots.txt file was last fetched.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    34
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    35
        This is useful for long-running web spiders that need to
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    36
        check for new robots.txt files periodically.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    37
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    38
        """
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    39
        return self.last_checked
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    40
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    41
    def modified(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    42
        """Sets the time the robots.txt file was last fetched to the
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    43
        current time.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    44
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    45
        """
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    46
        import time
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    47
        self.last_checked = time.time()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    48
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    49
    def set_url(self, url):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    50
        """Sets the URL referring to a robots.txt file."""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    51
        self.url = url
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    52
        self.host, self.path = urlparse.urlparse(url)[1:3]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    53
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    54
    def read(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    55
        """Reads the robots.txt URL and feeds it to the parser."""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    56
        opener = URLopener()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    57
        f = opener.open(self.url)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    58
        lines = [line.strip() for line in f]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    59
        f.close()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    60
        self.errcode = opener.errcode
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    61
        if self.errcode in (401, 403):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    62
            self.disallow_all = True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    63
        elif self.errcode >= 400:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    64
            self.allow_all = True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    65
        elif self.errcode == 200 and lines:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    66
            self.parse(lines)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    67
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    68
    def _add_entry(self, entry):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    69
        if "*" in entry.useragents:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    70
            # the default entry is considered last
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    71
            self.default_entry = entry
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    72
        else:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    73
            self.entries.append(entry)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    74
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    75
    def parse(self, lines):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    76
        """parse the input lines from a robots.txt file.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    77
           We allow that a user-agent: line is not preceded by
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    78
           one or more blank lines."""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    79
        # states:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    80
        #   0: start state
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    81
        #   1: saw user-agent line
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    82
        #   2: saw an allow or disallow line
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    83
        state = 0
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    84
        linenumber = 0
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    85
        entry = Entry()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    86
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    87
        for line in lines:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    88
            linenumber += 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    89
            if not line:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    90
                if state == 1:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    91
                    entry = Entry()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    92
                    state = 0
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    93
                elif state == 2:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    94
                    self._add_entry(entry)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    95
                    entry = Entry()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    96
                    state = 0
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    97
            # remove optional comment and strip line
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    98
            i = line.find('#')
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
    99
            if i >= 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   100
                line = line[:i]
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   101
            line = line.strip()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   102
            if not line:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   103
                continue
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   104
            line = line.split(':', 1)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   105
            if len(line) == 2:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   106
                line[0] = line[0].strip().lower()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   107
                line[1] = urllib.unquote(line[1].strip())
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   108
                if line[0] == "user-agent":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   109
                    if state == 2:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   110
                        self._add_entry(entry)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   111
                        entry = Entry()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   112
                    entry.useragents.append(line[1])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   113
                    state = 1
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   114
                elif line[0] == "disallow":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   115
                    if state != 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   116
                        entry.rulelines.append(RuleLine(line[1], False))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   117
                        state = 2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   118
                elif line[0] == "allow":
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   119
                    if state != 0:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   120
                        entry.rulelines.append(RuleLine(line[1], True))
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   121
                        state = 2
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   122
        if state == 2:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   123
            self.entries.append(entry)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   124
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   125
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   126
    def can_fetch(self, useragent, url):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   127
        """using the parsed robots.txt decide if useragent can fetch url"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   128
        if self.disallow_all:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   129
            return False
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   130
        if self.allow_all:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   131
            return True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   132
        # search for given user agent matches
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   133
        # the first match counts
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   134
        url = urllib.quote(urlparse.urlparse(urllib.unquote(url))[2]) or "/"
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   135
        for entry in self.entries:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   136
            if entry.applies_to(useragent):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   137
                return entry.allowance(url)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   138
        # try the default entry last
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   139
        if self.default_entry:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   140
            return self.default_entry.allowance(url)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   141
        # agent not found ==> access granted
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   142
        return True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   143
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   144
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   145
    def __str__(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   146
        return ''.join([str(entry) + "\n" for entry in self.entries])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   147
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   148
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   149
class RuleLine:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   150
    """A rule line is a single "Allow:" (allowance==True) or "Disallow:"
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   151
       (allowance==False) followed by a path."""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   152
    def __init__(self, path, allowance):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   153
        if path == '' and not allowance:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   154
            # an empty value means allow all
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   155
            allowance = True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   156
        self.path = urllib.quote(path)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   157
        self.allowance = allowance
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   158
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   159
    def applies_to(self, filename):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   160
        return self.path == "*" or filename.startswith(self.path)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   161
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   162
    def __str__(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   163
        return (self.allowance and "Allow" or "Disallow") + ": " + self.path
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   164
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   165
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   166
class Entry:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   167
    """An entry has one or more user-agents and zero or more rulelines"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   168
    def __init__(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   169
        self.useragents = []
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   170
        self.rulelines = []
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   171
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   172
    def __str__(self):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   173
        ret = []
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   174
        for agent in self.useragents:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   175
            ret.extend(["User-agent: ", agent, "\n"])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   176
        for line in self.rulelines:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   177
            ret.extend([str(line), "\n"])
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   178
        return ''.join(ret)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   179
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   180
    def applies_to(self, useragent):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   181
        """check if this entry applies to the specified agent"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   182
        # split the name token and make it lower case
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   183
        useragent = useragent.split("/")[0].lower()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   184
        for agent in self.useragents:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   185
            if agent == '*':
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   186
                # we have the catch-all agent
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   187
                return True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   188
            agent = agent.lower()
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   189
            if agent in useragent:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   190
                return True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   191
        return False
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   192
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   193
    def allowance(self, filename):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   194
        """Preconditions:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   195
        - our agent applies to this entry
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   196
        - filename is URL decoded"""
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   197
        for line in self.rulelines:
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   198
            if line.applies_to(filename):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   199
                return line.allowance
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   200
        return True
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   201
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   202
class URLopener(urllib.FancyURLopener):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   203
    def __init__(self, *args):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   204
        urllib.FancyURLopener.__init__(self, *args)
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   205
        self.errcode = 200
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   206
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   207
    def prompt_user_passwd(self, host, realm):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   208
        ## If robots.txt file is accessible only with a password,
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   209
        ## we act as if the file wasn't there.
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   210
        return None, None
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   211
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   212
    def http_error_default(self, url, fp, errcode, errmsg, headers):
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   213
        self.errcode = errcode
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   214
        return urllib.FancyURLopener.http_error_default(self, url, fp, errcode,
2fb8b9db1c86 Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff changeset
   215
                                                        errmsg, headers)