feedparser.py

来自「Harvestman-最新版本」· Python 代码 · 共 1,630 行 · 第 1/5 页

                  'http://madskills.com/public/xml/rss/module/pingback/': 'pingback',
                  'http://prismstandard.org/namespaces/1.2/basic/':       'prism',
                  'http://www.w3.org/1999/02/22-rdf-syntax-ns#':          'rdf',
                  'http://www.w3.org/2000/01/rdf-schema#':                'rdfs',
                  'http://purl.org/rss/1.0/modules/reference/':           'ref',
                  'http://purl.org/rss/1.0/modules/richequiv/':           'reqv',
                  'http://purl.org/rss/1.0/modules/search/':              'search',
                  'http://purl.org/rss/1.0/modules/slash/':               'slash',
                  'http://schemas.xmlsoap.org/soap/envelope/':            'soap',
                  'http://purl.org/rss/1.0/modules/servicestatus/':       'ss',
                  'http://hacks.benhammersley.com/rss/streaming/':        'str',
                  'http://purl.org/rss/1.0/modules/subscription/':        'sub',
                  'http://purl.org/rss/1.0/modules/syndication/':         'sy',
                  'http://purl.org/rss/1.0/modules/taxonomy/':            'taxo',
                  'http://purl.org/rss/1.0/modules/threading/':           'thr',
                  'http://purl.org/rss/1.0/modules/textinput/':           'ti',
                  'http://madskills.com/public/xml/rss/module/trackback/':'trackback',
                  'http://wellformedweb.org/commentAPI/':                 'wfw',
                  'http://purl.org/rss/1.0/modules/wiki/':                'wiki',
                  'http://www.w3.org/1999/xhtml':                         'xhtml',
                  'http://www.w3.org/XML/1998/namespace':                 'xml',
                  'http://schemas.pocketsoap.com/rss/myDescModule/':      'szf'
}
    _matchnamespaces = {}

    can_be_relative_uri = ['link', 'id', 'wfw_comment', 'wfw_commentrss', 'docs', 'url', 'href', 'comments', 'license', 'icon', 'logo']
    can_contain_relative_uris = ['content', 'title', 'summary', 'info', 'tagline', 'subtitle', 'copyright', 'rights', 'description']
    can_contain_dangerous_markup = ['content', 'title', 'summary', 'info', 'tagline', 'subtitle', 'copyright', 'rights', 'description']
    html_types = ['text/html', 'application/xhtml+xml']
    
    def __init__(self, baseuri=None, baselang=None, encoding='utf-8'):
        if _debug: sys.stderr.write('initializing FeedParser\n')
        if not self._matchnamespaces:
            for k, v in self.namespaces.items():
                self._matchnamespaces[k.lower()] = v
        self.feeddata = FeedParserDict() # feed-level data
        self.encoding = encoding # character encoding
        self.entries = [] # list of entry-level data
        self.version = '' # feed type/version, see SUPPORTED_VERSIONS
        self.namespacesInUse = {} # dictionary of namespaces defined by the feed

        # the following are used internally to track state;
        # this is really out of control and should be refactored
        self.infeed = 0
        self.inentry = 0
        self.incontent = 0
        self.intextinput = 0
        self.inimage = 0
        self.inauthor = 0
        self.incontributor = 0
        self.inpublisher = 0
        self.insource = 0
        self.sourcedata = FeedParserDict()
        self.contentparams = FeedParserDict()
        self._summaryKey = None
        self.namespacemap = {}
        self.elementstack = []
        self.basestack = []
        self.langstack = []
        self.baseuri = baseuri or ''
        self.lang = baselang or None
        if baselang:
            self.feeddata['language'] = baselang

    def unknown_starttag(self, tag, attrs):
        if _debug: sys.stderr.write('start %s with %s\n' % (tag, attrs))
        # normalize attrs
        attrs = [(k.lower(), v) for k, v in attrs]
        attrs = [(k, k in ('rel', 'type') and v.lower() or v) for k, v in attrs]
        
        # track xml:base and xml:lang
        attrsD = dict(attrs)
        baseuri = attrsD.get('xml:base', attrsD.get('base')) or self.baseuri
        self.baseuri = _urljoin(self.baseuri, baseuri)
        lang = attrsD.get('xml:lang', attrsD.get('lang'))
        if lang == '':
            # xml:lang could be explicitly set to '', we need to capture that
            lang = None
        elif lang is None:
            # if no xml:lang is specified, use parent lang
            lang = self.lang
        if lang:
            if tag in ('feed', 'rss', 'rdf:RDF'):
                self.feeddata['language'] = lang
        self.lang = lang
        self.basestack.append(self.baseuri)
        self.langstack.append(lang)
        
        # track namespaces
        for prefix, uri in attrs:
            if prefix.startswith('xmlns:'):
                self.trackNamespace(prefix[6:], uri)
            elif prefix == 'xmlns':
                self.trackNamespace(None, uri)

        # track inline content
        if self.incontent and self.contentparams.has_key('type') and not self.contentparams.get('type', 'xml').endswith('xml'):
            # element declared itself as escaped markup, but it isn't really
            self.contentparams['type'] = 'application/xhtml+xml'
        if self.incontent and self.contentparams.get('type') == 'application/xhtml+xml':
            # Note: probably shouldn't simply recreate localname here, but
            # our namespace handling isn't actually 100% correct in cases where
            # the feed redefines the default namespace (which is actually
            # the usual case for inline content, thanks Sam), so here we
            # cheat and just reconstruct the element based on localname
            # because that compensates for the bugs in our namespace handling.
            # This will horribly munge inline content with non-empty qnames,
            # but nobody actually does that, so I'm not fixing it.
            tag = tag.split(':')[-1]
            return self.handle_data('<%s%s>' % (tag, ''.join([' %s="%s"' % t for t in attrs])), escape=0)

        # match namespaces
        if tag.find(':') <> -1:
            prefix, suffix = tag.split(':', 1)
        else:
            prefix, suffix = '', tag
        prefix = self.namespacemap.get(prefix, prefix)
        if prefix:
            prefix = prefix + '_'

        # special hack for better tracking of empty textinput/image elements in illformed feeds
        if (not prefix) and tag not in ('title', 'link', 'description', 'name'):
            self.intextinput = 0
        if (not prefix) and tag not in ('title', 'link', 'description', 'url', 'href', 'width', 'height'):
            self.inimage = 0
        
        # call special handler (if defined) or default handler
        methodname = '_start_' + prefix + suffix
        try:
            method = getattr(self, methodname)
            return method(attrsD)
        except AttributeError:
            return self.push(prefix + suffix, 1)

    def unknown_endtag(self, tag):
        if _debug: sys.stderr.write('end %s\n' % tag)
        # match namespaces
        if tag.find(':') <> -1:
            prefix, suffix = tag.split(':', 1)
        else:
            prefix, suffix = '', tag
        prefix = self.namespacemap.get(prefix, prefix)
        if prefix:
            prefix = prefix + '_'

        # call special handler (if defined) or default handler
        methodname = '_end_' + prefix + suffix
        try:
            method = getattr(self, methodname)
            method()
        except AttributeError:
            self.pop(prefix + suffix)

        # track inline content
        if self.incontent and self.contentparams.has_key('type') and not self.contentparams.get('type', 'xml').endswith('xml'):
            # element declared itself as escaped markup, but it isn't really
            self.contentparams['type'] = 'application/xhtml+xml'
        if self.incontent and self.contentparams.get('type') == 'application/xhtml+xml':
            tag = tag.split(':')[-1]
            self.handle_data('</%s>' % tag, escape=0)

        # track xml:base and xml:lang going out of scope
        if self.basestack:
            self.basestack.pop()
            if self.basestack and self.basestack[-1]:
                self.baseuri = self.basestack[-1]
        if self.langstack:
            self.langstack.pop()
            if self.langstack: # and (self.langstack[-1] is not None):
                self.lang = self.langstack[-1]

    def handle_charref(self, ref):
        # called for each character reference, e.g. for '&#160;', ref will be '160'
        if not self.elementstack: return
        ref = ref.lower()
        if ref in ('34', '38', '39', '60', '62', 'x22', 'x26', 'x27', 'x3c', 'x3e'):
            text = '&#%s;' % ref
        else:
            if ref[0] == 'x':
                c = int(ref[1:], 16)
            else:
                c = int(ref)
            text = unichr(c).encode('utf-8')
        self.elementstack[-1][2].append(text)

    def handle_entityref(self, ref):
        # called for each entity reference, e.g. for '&copy;', ref will be 'copy'
        if not self.elementstack: return
        if _debug: sys.stderr.write('entering handle_entityref with %s\n' % ref)
        if ref in ('lt', 'gt', 'quot', 'amp', 'apos'):
            text = '&%s;' % ref
        else:
            # entity resolution graciously donated by Aaron Swartz
            def name2cp(k):
                import htmlentitydefs
                if hasattr(htmlentitydefs, 'name2codepoint'): # requires Python 2.3
                    return htmlentitydefs.name2codepoint[k]
                k = htmlentitydefs.entitydefs[k]
                if k.startswith('&#') and k.endswith(';'):
                    return int(k[2:-1]) # not in latin-1
                return ord(k)
            try: name2cp(ref)
            except KeyError: text = '&%s;' % ref
            else: text = unichr(name2cp(ref)).encode('utf-8')
        self.elementstack[-1][2].append(text)

    def handle_data(self, text, escape=1):
        # called for each block of plain text, i.e. outside of any tag and
        # not containing any character or entity references
        if not self.elementstack: return
        if escape and self.contentparams.get('type') == 'application/xhtml+xml':
            text = _xmlescape(text)
        self.elementstack[-1][2].append(text)

    def handle_comment(self, text):
        # called for each comment, e.g. <!-- insert message here -->
        pass

    def handle_pi(self, text):
        # called for each processing instruction, e.g. <?instruction>
        pass

    def handle_decl(self, text):
        pass

    def parse_declaration(self, i):
        # override internal declaration handler to handle CDATA blocks
        if _debug: sys.stderr.write('entering parse_declaration\n')
        if self.rawdata[i:i+9] == '<![CDATA[':
            k = self.rawdata.find(']]>', i)
            if k == -1: k = len(self.rawdata)
            self.handle_data(_xmlescape(self.rawdata[i+9:k]), 0)
            return k+3
        else:
            k = self.rawdata.find('>', i)
            return k+1

    def mapContentType(self, contentType):
        contentType = contentType.lower()
        if contentType == 'text':
            contentType = 'text/plain'
        elif contentType == 'html':
            contentType = 'text/html'
        elif contentType == 'xhtml':
            contentType = 'application/xhtml+xml'
        return contentType
    
    def trackNamespace(self, prefix, uri):
        loweruri = uri.lower()
        if (prefix, loweruri) == (None, 'http://my.netscape.com/rdf/simple/0.9/') and not self.version:
            self.version = 'rss090'
        if loweruri == 'http://purl.org/rss/1.0/' and not self.version:
            self.version = 'rss10'
        if loweruri == 'http://www.w3.org/2005/atom' and not self.version:
            self.version = 'atom10'
        if loweruri.find('backend.userland.com/rss') <> -1:
            # match any backend.userland.com namespace
            uri = 'http://backend.userland.com/rss'
            loweruri = uri
        if self._matchnamespaces.has_key(loweruri):
            self.namespacemap[prefix] = self._matchnamespaces[loweruri]
            self.namespacesInUse[self._matchnamespaces[loweruri]] = uri
        else:
            self.namespacesInUse[prefix or ''] = uri

    def resolveURI(self, uri):
        return _urljoin(self.baseuri or '', uri)
    
    def decodeEntities(self, element, data):
        return data

    def push(self, element, expectingText):
        self.elementstack.append([element, expectingText, []])

    def pop(self, element, stripWhitespace=1):
        if not self.elementstack: return
        if self.elementstack[-1][0] != element: return
        
        element, expectingText, pieces = self.elementstack.pop()
        output = ''.join(pieces)
        if stripWhitespace:
            output = output.strip()
        if not expectingText: return output

        # decode base64 content
        if base64 and self.contentparams.get('base64', 0):
            try:
                output = base64.decodestring(output)
            except binascii.Error:
                pass
            except binascii.Incomplete:
                pass
                
        # resolve relative URIs
        if (element in self.can_be_relative_uri) and output:
            output = self.resolveURI(output)
        
        # decode entities within embedded markup
        if not self.contentparams.get('base64', 0):
            output = self.decodeEntities(element, output)

        # remove temporary cruft from contentparams
        try:
            del self.contentparams['mode']
        except KeyError:
            pass
        try:
            del self.contentparams['base64']
        except KeyError:
            pass

        # resolve relative URIs within embedded markup
        if self.mapContentType(self.contentparams.get('type', 'text/html')) in self.html_types:
            if element in self.can_contain_relative_uris:
                output = _resolveRelativeURIs(output, self.baseuri, self.encoding)
        
        # sanitize embedded markup
        if self.mapContentType(self.contentparams.get('type', 'text/html')) in self.html_types:
            if element in self.can_contain_dangerous_markup:
                output = _sanitizeHTML(output, self.encoding)

        if self.encoding and type(output) != type(u''):
            try:
                output = unicode(output, self.encoding)
            except:
                pass