Example #1
0
 def __fetch_remote(self, unid, contents, format=None, time=None,
         digest=None): # {{{
     "Like __fetch, but accept contents for-, or fetch contents from remote ID. "
     if contents:
         # XXX: Accept given data as-is
         pass
     else: 
         # Query server for contents and metadata
         source_id = self.__remote_id(unid, format)
         res = model.Resource.get_or_insert(
                 hashlib.md5(source_id).hexdigest(), remote_id=source_id )
         self.__srcnfo = self.store.getinfo(self.alias, unid).next()
         etag = None
         if self.__srcnfo:
             # XXX: or allow local update, see assert format later on
             assert not time and not digest, \
                     "Re-using exiting source."
             # Fetch updated only
             remote_update = util.fetch_uriref(
                     source_id, self.__srcnfo.format,
                     self.__srcnfo.time, res.etag,
                     self.__srcnfo.digest)
             if remote_update:
                 logger.info('Remote updated')
                 contents, format, time, etag, digest, charset = \
                         remote_update
                 doctree = None
                 if format and self.__srcnfo.format:
                     assert format == self.__srcnfo.format, \
                             "Remote format changed."
             else:
                 if format and self.__srcnfo.format:
                     assert format == self.__srcnfo.format, \
                             "Indicated format does not match record."
                 if not self.__src:
                     self.__src = self.__srcnfo.parent()
                 contents, format, doctree, time, digest, charset = \
                         self.__src.contents, \
                         self.__srcnfo.format, \
                         self.__src.doctree, \
                         self.__srcnfo.time, \
                         self.__srcnfo.digest, \
                         self.__srcnfo.charset
             self.__doctree = doctree                            
         else:
             # Fetch new Source
             logging.info('Retrieving new remote content for %s. ', unid)
             contents, format, time, etag, digest, charset = \
                                     util.fetch_uriref(source_id, format,
                                             time, None, digest)
         if etag:
             res.etag = etag
             res.put()
     self.__assert_contents(contents)
     self.__assert_charset(charset)
     self.__assert_digest(digest)
     if time:
         self.__assert_datetime(time)
Example #2
0
    def stat(self, unid, digest=None): # {{{
        """
        Returns True value when UNID is in storage and up to date, or false when
        it needs (re)processing.

        Passing a digest will assure failure when known digest is different.
        Otherwise a remote HTTP request may be needed for remote content.
        """
        self._initialize()
        self.__assert_alias()
        self.__srcnfo = self.store.getinfo(self.alias, unid).next()
        if self.__srcnfo and digest:
            if unid_digest != self.__srcnfo.digest:
                logger.info("%s invalidated by digest", unid)
                v = 0
            else:
                v = 1
        elif hasattr(self.alias, 'remote_path'):
            if self.__srcnfo:
                format = self.__srcnfo.format
            else:                
                format = self.alias.default_source_format
            source_id = self.__remote_id(unid, format)
            logger.debug("Checking remote path for %s (%s)", unid, source_id)
            res = None
            if self.__srcnfo:
                # Known remote UNID
                res = model.Resource.get_or_insert(
                        hashlib.md5(source_id).hexdigest(), remote_id=source_id)
                src = util.fetch_uriref(source_id, format, 
                        self.__srcnfo.time, res.etag, self.__srcnfo.digest)
            else:
                # Check for existince of remote source
                src = util.fetch_uriref(source_id)
            if src:
                #v = src
                contents, format, time, etag, digest, charset = src
                src = None
                if self.__srcnfo:
                    logger.info("Remote update (%s): %s",
                            self.alias.handle, source_id)
                else:                    
                    logger.info("New remote source (%s): %s ", 
                            self.alias.handle, source_id)
                v = 0
            elif self.__srcnfo:
                v = 1
            else:
                raise exception.NotFound(unid)
        else:
            v = 0
        return self.__r_stat(unid, digest, v)            
Example #3
0
    def stat(self, unid, digest=None):
        """
        Returns True value when UNID is in storage and up to date, or false when
        it needs (re)processing.

        Passing a digest will assure failure when known digest is different.
        Otherwise a remote HTTP request may be needed for remote content.
        """
        assert ALIAS_re.match(unid), "Invalid aliased ID: %s " % unid
        info = self.store.getinfo(self.alias, unid).next()
        if info and digest:
            if unid_digest != info.digest:
                logger.info("%s invalidated by digest", unid)
                return False
            else:
                return True
        elif hasattr(self.alias, 'remote_path'):
            source_id = self.__remote_id(unid)
            logger.info("Checking remote path for %s (%s)", unid, source_id)
            res = None
            if info:
                # Known remote UNID
                res = model.Resource.get_or_insert(
                        hashlib.md5(source_id).hexdigest(), remote_id=source_id)
                rst = fetch_uriref(source_id, info.time, res.etag,
                        info.digest)
            else:
                # Check for existince of remote source
                rst = fetch_uriref(source_id)
            if rst:
                return False
                contents, time, etag, digest = rst
                src = None
                if info:
                    src = info.parent()
                    logger.warning("Remote update (%s): %s",
                            self.alias.handle, source_id)
                else:                    
                    logger.warning("New remote source (%s): %s ", 
                            self.alias.handle, source_id)
                    src, info = self.store.add(self.alias, unid, contents,
                            digest=digest, time=time)
                    #srcdigest = 'Content-MD5'
                    res = model.Resource.get_or_insert(
                            digest, remote_id=source_id)
                assert isinstance(contents, unicode)
                #src = info.parent()
                #src.source = contents
                #src.doctree = None
                #src.put()
                #info.digest = digest
                #info.time = time
                #info.put()
                #res.etag = etag
                #res.put()
                return False
            elif info:
                return True
            else:
                raise exception.NotFound(unid)
        else:
            return False