# -*- coding: iso-8859-1 -*-
# Copyright (C) 2000-2005  Bastian Kleineidam
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
"""
Base URL handler.
"""

import sys
import os
import urlparse
import urllib2
import urllib
import time
import errno
import socket
import select
import traceback

import linkcheck
import linkcheck.linkparse
import linkcheck.checker
import linkcheck.strformat
import linkcheck.containers
import linkcheck.log
import linkcheck.httplib2
import linkcheck.HtmlParser.htmlsax


def urljoin (parent, url, scheme):
    """
    If url is relative, join parent and url. Else leave url as-is.

    @return joined url
    """
    if url.startswith(scheme+":"):
        return url
    return urlparse.urljoin(parent, url)


class UrlBase (object):
    """
    An URL with additional information like validity etc.
    """

    def __init__ (self, base_url, recursion_level, consumer,
                  parent_url = None, base_ref = None,
                  line = -1, column = -1, name = u""):
        """
        Initialize check data, and store given variables.

        @param base_url: unquoted and possibly unnormed url
        @param recursion_level: on what check level lies the base url
        @param consumer: consumer instance
        @param parent_url: quoted and normed url of parent or None
        @param base_ref: quoted and normed url of <base href=""> or None
        @param line: line number of url in parent content
        @param column: column number of url in parent content
        @param name: name of url or empty
        """
        self.init(base_ref, base_url, parent_url, recursion_level,
                  consumer, line, column, name)
        self.reset()
        self.check_syntax()


    def init (self, base_ref, base_url, parent_url, recursion_level,
              consumer, line, column, name):
        """
        Initialize internal data.
        """
        self.base_ref = base_ref
        # note that self.base_url must not be modified
        self.base_url = base_url
        self.parent_url = parent_url
        self.recursion_level = recursion_level
        self.consumer = consumer
        self.line = line
        self.column = column
        self.name = name
        if self.base_ref:
            assert not linkcheck.url.url_needs_quoting(self.base_ref), \
                   "unquoted %r" % self.base_ref
        if self.parent_url:
            assert not linkcheck.url.url_needs_quoting(self.parent_url), \
                   "unquoted %r" % self.parent_url
        url = linkcheck.checker.absolute_url(base_url, base_ref, parent_url)
        # assume file link if no scheme is found
        self.scheme = url.split(":", 1)[0] or "file"

    def reset (self):
        """
        Reset all variables to default values.
        """
        # self.url is constructed by self.build_url() out of base_url
        # and (base_ref or parent) as absolute and normed url.
        # This the real url we use when checking so it also referred to
        # as 'real url'
        self.url = None
        # a splitted version of url for convenience
        self.urlparts = None
        # the anchor part of url
        self.anchor = None
        # the result message string and flag
        self.result = u""
        self.has_result = False
        # cached or not
        self.cached = False
        # valid or not
        self.valid = True
        # list of warnings (without duplicates)
        self.warnings = linkcheck.containers.SetList()
        # list of infos (without duplicates)
        self.info = linkcheck.containers.SetList()
        # download time
        self.dltime = -1
        # download size
        self.dlsize = -1
        # check time
        self.checktime = 0
        # connection object
        self.url_connection = None
        # data of url content
        self.data = None
        # if data is filled
        self.has_content = False
        # cache keys, are set by build_url() calling set_cache_keys()
        self.cache_url_key = None
        self.cache_content_key = None
        # extern flags (is_extern, is_strict), both enabled as default
        self.extern = (1, 1)
        # flag if the result should be cached
        self.caching = True

    def set_result (self, msg, valid=True):
        """
        Set result string and validity.
        """
        if self.has_result:
            linkcheck.log.warn(linkcheck.LOG_CHECK,
              "Double result %r (previous %r) for %s", msg, self.result, self)
        else:
            self.has_result = True
        if not isinstance(msg, unicode):
            linkcheck.log.warn(linkcheck.LOG_CHECK,
                               "Non-unicode result for %s: %r", self, msg)
        elif not msg:
            linkcheck.log.warn(linkcheck.LOG_CHECK,
                               "Empty result for %s", self)
        self.result = msg
        self.valid = valid

    def is_parseable (self):
        """
        Return True iff content of this url is parseable.
        """
        return False

    def is_html (self):
        """
        Return True iff content of this url is HTML formatted.
        """
        return False

    def is_http (self):
        """
        Return True for http:// URLs.
        """
        return False

    def is_file (self):
        """
        Return True for file:// URLs.
        """
        return False

    def add_warning (self, s, tag=None):
        """
        Add a warning string.
        """
        self.warnings.append((tag, s))

    def add_info (self, s):
        """
        Add an info string.
        """
        self.info.append(s)

    def copy_from_cache (self, cache_data):
        """
        Fill attributes from cache data.
        """
        self.result = cache_data["result"]
        self.warnings.extend(cache_data["warnings"])
        self.info.extend(cache_data["info"])
        self.valid = cache_data["valid"]
        self.dltime = cache_data["dltime"]
        self.dlsize = cache_data["dlsize"]
        self.cached = True

    def get_cache_data (self):
        """
        Return all data values that should be put in the cache.
        """
        return {"result": self.result,
                "warnings": self.warnings,
                "info": self.info,
                "valid": self.valid,
                "dltime": self.dltime,
                "dlsize": self.dlsize,
               }

    def set_cache_keys (self):
        """
        Set keys for URL checking and content recursion.
        """
        # remove anchor from content cache key since we assume
        # URLs with different anchors to have the same content
        self.cache_content_key = urlparse.urlunsplit(self.urlparts[:4]+[u''])
        assert isinstance(self.cache_content_key, unicode), self
        linkcheck.log.debug(linkcheck.LOG_CACHE, "Content cache key %r",
                            self.cache_content_key)
        # construct cache key
        if self.consumer.config("anchorcaching") and \
           self.consumer.config("anchors"):
            # do not ignore anchor
            parts = self.urlparts[:]
            parts[4] = self.anchor
            self.cache_url_key = urlparse.urlunsplit(parts)
        else:
            # no anchor caching
            self.cache_url_key = self.cache_content_key
        assert isinstance(self.cache_url_key, unicode), self
        linkcheck.log.debug(linkcheck.LOG_CACHE, "URL cache key %r",
                            self.cache_url_key)

    def check_syntax (self):
        """
        Called before self.check(), this function inspects the
        url syntax. Success enables further checking, failure
        immediately logs this url. Syntax checks must not
        use any network resources.

        @return: True if syntax is correct, else False.
        @rtype: bool
        """
        linkcheck.log.debug(linkcheck.LOG_CHECK, "checking syntax")
        if (self.base_url is None) or \
           (not self.base_url and not self.parent_url):
            self.set_result(_("URL is empty"), valid=False)
            return
        try:
            self.build_url()
            # check url warnings
            effectiveurl = urlparse.urlunsplit(self.urlparts)
            if self.url != effectiveurl:
                self.add_warning(_("Effective URL %r.") % effectiveurl,
                                 tag="url-effective-url")
                self.url = effectiveurl
        except tuple(linkcheck.checker.ExcSyntaxList), msg:
            self.set_result(linkcheck.strformat.unicode_safe(msg),
                            valid=False)
            return
        self.set_cache_keys()

    def build_url (self):
        """
        Construct self.url and self.urlparts out of the given base
        url information self.base_url, self.parent_url and self.base_ref.
        """
        # norm base url
        base_url, is_idn = linkcheck.url.url_norm(self.base_url)
        if is_idn:
            self.add_warning(_("""URL %r has a unicode domain name which
                          is not yet widely supported. You should use
                          the URL %r instead.""") % (self.base_url, base_url),
                          tag="url-unicode-domain")
        elif self.base_url != base_url:
            self.add_warning(
              _("Base URL is not properly normed. Normed URL is %(url)s.") % \
               {'url': base_url}, tag="url-unnormed")
        # make url absolute
        if self.base_ref:
            # use base reference as parent url
            if ":" not in self.base_ref:
                # some websites have a relative base reference
                self.base_ref = urljoin(self.parent_url, self.base_ref,
                                        self.scheme)
            self.url = urljoin(self.base_ref, base_url, self.scheme)
        elif self.parent_url:
            # strip the parent url query and anchor
            urlparts = list(urlparse.urlsplit(self.parent_url))
            urlparts[3] = urlparts[4] = ""
            parent_url = urlparse.urlunsplit(urlparts)
            self.url = urljoin(parent_url, base_url, self.scheme)
        else:
            self.url = base_url
        # note: urljoin can unnorm the url path, so norm it again
        urlparts = list(urlparse.urlsplit(self.url))
        if urlparts[2]:
            urlparts[2] = linkcheck.url.collapse_segments(urlparts[2])
        self.url = urlparse.urlunsplit(urlparts)
        # split into (modifiable) list
        self.urlparts = linkcheck.strformat.url_unicode_split(self.url)
        # and unsplit again
        self.url = urlparse.urlunsplit(self.urlparts)
        # check userinfo@host:port syntax
        self.userinfo, host = urllib.splituser(self.urlparts[1])
        # set host lowercase
        if self.userinfo:
            self.urlparts[1] = "%s@%s" % (self.userinfo, host.lower())
        else:
            self.urlparts[1] = host.lower()
        # safe anchor for later checking
        self.anchor = self.urlparts[4]
        self.host, self.port = urllib.splitport(host)
        if self.port is not None:
            if not linkcheck.url.is_numeric_port(self.port):
                raise linkcheck.LinkCheckerError, \
                         _("URL has invalid port %r") % str(self.port)
            self.port = int(self.port)

    def check (self):
        """
        Main check function for checking this URL.
        """
        if self.consumer.config("trace"):
            linkcheck.trace.trace_on()
        try:
            self.local_check()
            self.consumer.checked(self)
        except (socket.error, select.error):
            self.consumer.interrupted(self)
            # on Unix, ctrl-c can raise
            # error: (4, 'Interrupted system call')
            etype, value = sys.exc_info()[:2]
            if etype == 4:
                raise KeyboardInterrupt, value
            else:
                raise
        except KeyboardInterrupt:
            self.consumer.interrupted(self)
            raise
        except:
            self.consumer.interrupted(self)
            linkcheck.checker.internal_error()

    def add_country_info (self):
        """
        Try to ask GeoIP database for country info.
        """
        country = self.consumer.get_country_name(self.host)
        if country is not None:
            self.add_info(_("URL is located in %s.") % _(country))

    def local_check (self):
        """
        Local check function can be overridden in subclasses.
        """
        linkcheck.log.debug(linkcheck.LOG_CHECK, "Checking %s", self)
        if self.recursion_level and self.consumer.config('wait'):
            linkcheck.log.debug(linkcheck.LOG_CHECK,
                                "sleeping for %d seconds",
                                self.consumer.config('wait'))
            time.sleep(self.consumer.config('wait'))
        t = time.time()
        self.set_extern(self.url)
        if self.extern[0] and self.extern[1]:
            self.add_info(_("Outside of domain filter, checked only syntax."))
            return

        # check connection
        linkcheck.log.debug(linkcheck.LOG_CHECK, "checking connection")
        try:
            self.check_connection()
            self.add_country_info()
            if self.consumer.config("anchors"):
                self.check_anchors()
        except tuple(linkcheck.checker.ExcList):
            value = self.handle_exception()
            # make nicer error msg for unknown hosts
            if isinstance(value, socket.error) and value[0] == -2:
                value = _('Hostname not found')
            # make nicer error msg for bad status line
            if isinstance(value, linkcheck.httplib2.BadStatusLine):
                value = _('Bad HTTP response %r') % str(value)
            self.set_result(linkcheck.strformat.unicode_safe(value),
                            valid=False)

        # check content
        warningregex = self.consumer.config("warningregex")
        if warningregex and self.valid:
            linkcheck.log.debug(linkcheck.LOG_CHECK, "checking content")
            try:
                self.check_content(warningregex)
            except tuple(linkcheck.checker.ExcList):
                value = self.handle_exception()
                self.set_result(linkcheck.strformat.unicode_safe(value),
                                valid=False)

        self.checktime = time.time() - t
        # check recursion
        try:
            if self.allows_recursion():
                self.parse_url()
            # check content size
            self.check_size()
        except tuple(linkcheck.checker.ExcList):
            value = self.handle_exception()
            self.add_warning(_("could not get content: %r") % str(value),
                            tag="url-error-getting-content")
        # close
        self.close_connection()

    def close_connection (self):
        """
        Close an opened url connection.
        """
        if self.url_connection is None:
            # no connection is open
            return
        try:
            self.url_connection.close()
        except:
            # ignore close errors
            pass
        self.url_connection = None

    def handle_exception (self):
        """
        An exception occurred. Log it and set the cache flag.
        """
        etype, value, tb = sys.exc_info()
        linkcheck.log.debug(linkcheck.LOG_CHECK, "exception %s",
                            traceback.format_tb(tb))
        # note: etype must be the exact class, not a subclass
        if (etype in linkcheck.checker.ExcNoCacheList) or \
           (etype == socket.error and value[0]==errno.EBADF) or \
            not value:
            # EBADF occurs when operating on an already socket
            self.caching = False
        return value

    def check_connection (self):
        """
        The basic connection check uses urllib2.urlopen to initialize
        a connection object.
        """
        self.url_connection = urllib2.urlopen(self.url)

    def allows_recursion (self):
        """
        Return True iff we can recurse into the url's content.
        """
        linkcheck.log.debug(linkcheck.LOG_CHECK,
                            "checking recursion of %r ...", self.url)
        # Test self.valid before self.is_parseable().
        if not self.valid:
            linkcheck.log.debug(linkcheck.LOG_CHECK, "... no, invalid.")
            return False
        if not self.is_parseable():
            linkcheck.log.debug(linkcheck.LOG_CHECK, "... no, not parseable.")
            return False
        if not self.can_get_content():
            linkcheck.log.debug(linkcheck.LOG_CHECK,
                                "... no, cannot get content.")
            return False
        if self.consumer.config("recursionlevel") >= 0 and \
           self.recursion_level >= self.consumer.config("recursionlevel"):
            linkcheck.log.debug(linkcheck.LOG_CHECK,
                                "... no, maximum recursion level reached.")
            return False
        if self.extern[0]:
            linkcheck.log.debug(linkcheck.LOG_CHECK, "... no, extern.")
            return False
        if not self.content_allows_robots():
            linkcheck.log.debug(linkcheck.LOG_CHECK, "... no, robots.")
            return False
        linkcheck.log.debug(linkcheck.LOG_CHECK, "... yes, recursion.")
        return True

    def content_allows_robots (self):
        """
        Return True if the content of this URL forbids robots to
        search for recursive links.
        """
        if not self.is_html():
            return True
        if not (self.is_http() or self.is_file()):
            return True
        h = linkcheck.linkparse.MetaRobotsFinder(self.get_content())
        p = linkcheck.HtmlParser.htmlsax.parser(h)
        h.parser = p
        p.feed(self.get_content())
        p.flush()
        h.parser = None
        p.handler = None
        return h.follow

    def check_anchors (self):
        """
        If URL was valid and a HTML resource, check the anchors and
        log a warning when an anchor was not found.
        """
        if not (self.valid and self.anchor and self.is_html() and \
                self.can_get_content()):
            # do not bother
            return
        linkcheck.log.debug(linkcheck.LOG_CHECK, "checking anchor %r",
                            self.anchor)
        h = linkcheck.linkparse.LinkFinder(self.get_content(),
                                   tags={'a': [u'name'], None: [u'id']})
        p = linkcheck.HtmlParser.htmlsax.parser(h)
        h.parser = p
        p.feed(self.get_content())
        p.flush()
        h.parser = None
        p.handler = None
        for cur_anchor, line, column, name, base in h.urls:
            if cur_anchor == self.anchor:
                return
        self.add_warning(_("Anchor #%s not found.") % self.anchor,
                         tag="url-anchor-not-found")

    def set_extern (self, url):
        """
        Match URL against extern and intern link patterns. If no pattern
        matches the URL is extern. Sets self.extern to a tuple (bool,
        bool) with content (is_extern, is_strict).

        @return: None
        """
        for entry in self.consumer.config("externlinks"):
            match = entry['pattern'].search(url)
            if (entry['negate'] and not match) or \
               (match and not entry['negate']):
                linkcheck.log.debug(linkcheck.LOG_CHECK, "Extern URL %r", url)
                self.extern = (1, entry['strict'])
                return
        for entry in self.consumer.config("internlinks"):
            match = entry['pattern'].search(url)
            if (entry['negate'] and not match) or \
               (match and not entry['negate']):
                linkcheck.log.debug(linkcheck.LOG_CHECK, "Intern URL %r", url)
                self.extern = (0, 0)
                return
        linkcheck.log.debug(linkcheck.LOG_CHECK,
                            "Explicit extern URL %r", url)
        self.extern = (1, 0)
        return

    def can_get_content (self):
        """
        Indicate wether url get_content() can be called.
        """
        return True

    def get_content (self):
        """
        Precondition: url_connection is an opened URL.
        """
        if not self.has_content:
            t = time.time()
            self.data = self.url_connection.read()
            self.dltime = time.time() - t
            self.dlsize = len(self.data)
            self.has_content = True
        return self.data

    def check_content (self, warningregex):
        """
        If a warning expression was given, call this function to check it
        against the content of this url.
        """
        if not self.can_get_content():
            return
        match = warningregex.search(self.get_content())
        if match:
            self.add_warning(_("Found %r in link contents.") % match.group(),
                             tag="url-warnregex-found")

    def check_size (self):
        """
        If a maximum size was given, call this function to check it
        against the content size of this url.
        """
        maxbytes = self.consumer.config("warnsizebytes")
        if maxbytes is not None and self.dlsize >= maxbytes:
            self.add_warning(_("Content size %s is larger than %s.") % \
                         (linkcheck.strformat.strsize(self.dlsize),
                          linkcheck.strformat.strsize(maxbytes)),
                          tag="url-content-too-large")

    def parse_url (self):
        """
        Parse url content and search for recursive links.
        Default parse type is html.
        """
        self.parse_html()

    def get_user_password (self):
        """
        Get tuple (user, password) from configured authentication.
        Both user and password can be None if not specified.
        """
        for auth in self.consumer.config("authentication"):
            if auth['pattern'].match(self.url):
                return auth['user'], auth['password']
        return None, None

    def parse_html (self):
        """
        Parse into HTML content and search for URLs to check.
        Found URLs are added to the URL queue.
        """
        linkcheck.log.debug(linkcheck.LOG_CHECK, "Parsing HTML %s", self)
        h = linkcheck.linkparse.LinkFinder(self.get_content())
        p = linkcheck.HtmlParser.htmlsax.parser(h)
        h.parser = p
        p.feed(self.get_content())
        p.flush()
        h.parser = None
        p.handler = None
        for url, line, column, name, codebase in h.urls:
            if codebase:
                base_ref = codebase
            else:
                base_ref = h.base_ref
            url_data = linkcheck.checker.get_url_from(url,
                  self.recursion_level+1, self.consumer, parent_url=self.url,
                  base_ref=base_ref, line=line, column=column, name=name,
                  cmdline=False)
            self.consumer.append_url(url_data)

    def parse_opera (self):
        """
        Parse an opera bookmark file.
        """
        linkcheck.log.debug(linkcheck.LOG_CHECK,
                            "Parsing Opera bookmarks %s", self)
        name = ""
        lineno = 0
        lines = self.get_content().splitlines()
        for line in lines:
            lineno += 1
            line = line.strip()
            if line.startswith("NAME="):
                name = line[5:]
            elif line.startswith("URL="):
                url = line[4:]
                if url:
                    url_data = linkcheck.checker.get_url_from(url,
                              self.recursion_level+1, self.consumer,
                              parent_url=self.url, line=lineno, name=name,
                              cmdline=False)
                    self.consumer.append_url(url_data)
                name = ""

    def parse_text (self):
        """
        Parse a text file with on url per line; comment and blank
        lines are ignored.
        """
        linkcheck.log.debug(linkcheck.LOG_CHECK, "Parsing text %s", self)
        lineno = 0
        for line in self.get_content().splitlines():
            lineno += 1
            line = line.strip()
            if not line or line.startswith('#'):
                continue
            url_data = linkcheck.checker.get_url_from(line,
                              self.recursion_level+1, self.consumer,
                              parent_url=self.url, line=lineno,
                              cmdline=False)
            self.consumer.append_url(url_data)

    def parse_css (self):
        """
        Parse a CSS file for url() patterns.
        """
        linkcheck.log.debug(linkcheck.LOG_CHECK, "Parsing CSS %s", self)
        lineno = 0
        for line in self.get_content().splitlines():
            lineno += 1
            for mo in linkcheck.linkparse.css_url_re.finditer(line):
                column = mo.start("url")
                url = linkcheck.strformat.unquote(mo.group("url").strip())
                url_data = linkcheck.checker.get_url_from(url,
                             self.recursion_level+1, self.consumer,
                             parent_url=self.url, line=lineno, column=column,
                             cmdline=False)
                self.consumer.append_url(url_data)

    def serialized (self):
        """
        Return serialized url check data as unicode string.
        """
        sep = linkcheck.strformat.unicode_safe(os.linesep)
        if self.base_url is not None:
            assert isinstance(self.base_url, unicode), self
        if self.parent_url is not None:
            assert isinstance(self.parent_url, unicode), self
        if self.base_ref is not None:
            assert isinstance(self.base_ref, unicode), self
        assert isinstance(self.name, unicode), self
        return sep.join([
            u"%s link" % self.scheme,
            u"base_url=%r" % self.base_url,
            u"parent_url=%r" % self.parent_url,
            u"base_ref=%r" % self.base_ref,
            u"recursion_level=%s" % self.recursion_level,
            u"url_connection=%s" % self.url_connection,
            u"line=%d" % self.line,
            u"column=%d" % self.column,
            u"name=%r" % self.name,
           ])

    def get_intern_pattern (self):
        """
        Get pattern for intern URL matching.

        @return non-empty regex pattern or None
        @rtype String or None
        """
        return None

    def __str__ (self):
        """
        Get URL info.

        @return: URL info, encoded with the output logger encoding
        @rtype: string
        """
        s = self.serialized()
        return self.consumer.config('logger').encode(s)

    def __repr__ (self):
        """
        Get URL info.

        @return: URL info
        @rtype: unicode
        """
        return u"<%s >" % self.serialized()