Python reader.XMLCorpusView方法代码示例

# 需要导入模块: from nltk.corpus import reader [as 别名]
# 或者: from nltk.corpus.reader import XMLCorpusView [as 别名]
def _buildrelationindex(self):
        #print('building relation index...', file=sys.stderr)
        freltypes = PrettyList(x for x in XMLCorpusView(self.abspath("frRelation.xml"),
                                            'frameRelations/frameRelationType',
                                            self._handle_framerelationtype_elt))
        self._freltyp_idx = {}
        self._frel_idx = {}
        self._frel_f_idx = defaultdict(set)
        self._ferel_idx = {}

        for freltyp in freltypes:
            self._freltyp_idx[freltyp.ID] = freltyp
            for frel in freltyp.frameRelations:
                supF = frel.superFrame = frel[freltyp.superFrameName] = Future((lambda fID: lambda: self.frame_by_id(fID))(frel.supID))
                subF = frel.subFrame = frel[freltyp.subFrameName] = Future((lambda fID: lambda: self.frame_by_id(fID))(frel.subID))
                self._frel_idx[frel.ID] = frel
                self._frel_f_idx[frel.supID].add(frel.ID)
                self._frel_f_idx[frel.subID].add(frel.ID)
                for ferel in frel.feRelations:
                    ferel.superFrame = supF
                    ferel.subFrame = subF
                    ferel.superFE = Future((lambda fer: lambda: fer.superFrame.FE[fer.superFEName])(ferel))
                    ferel.subFE = Future((lambda fer: lambda: fer.subFrame.FE[fer.subFEName])(ferel))
                    self._ferel_idx[ferel.ID] = ferel
        #print('...done building relation index', file=sys.stderr) 

# 需要导入模块: from nltk.corpus import reader [as 别名]
# 或者: from nltk.corpus.reader import XMLCorpusView [as 别名]
def _lu_file(self, lu, ignorekeys=[]):
        """
        Augment the LU information that was loaded from the frame file
        with additional information from the LU file.
        """
        fn_luid = lu.ID

        fname = "lu{0}.xml".format(fn_luid)
        locpath = os.path.join("{0}".format(self._root), self._lu_dir, fname)
        #print(locpath, file=sys.stderr)
        if not self._lu_idx:
            self._buildluindex()

        try:
            elt = XMLCorpusView(locpath, 'lexUnit')[0]
        except IOError:
            raise FramenetError('Unknown LU id: {0}'.format(fn_luid))

        lu2 = self._handle_lexunit_elt(elt, ignorekeys)
        lu.subCorpus = lu2.subCorpus

        return lu.subCorpus 

# 需要导入模块: from nltk.corpus import reader [as 别名]
# 或者: from nltk.corpus.reader import XMLCorpusView [as 别名]
def _buildframeindex(self):
        # The total number of Frames in Framenet is fairly small (~1200) so
        # this index should not be very large
        if not self._frel_idx:
            self._buildrelationindex()  # always load frame relations before frames,
            # otherwise weird ordering effects might result in incomplete information
        self._frame_idx = {}
        for f in XMLCorpusView(self.abspath("frameIndex.xml"),
                               'frameIndex/frame', self._handle_elt):
            self._frame_idx[f['ID']] = f 

# 需要导入模块: from nltk.corpus import reader [as 别名]
# 或者: from nltk.corpus.reader import XMLCorpusView [as 别名]
def _buildcorpusindex(self):
        # The total number of fulltext annotated documents in Framenet
        # is fairly small (~90) so this index should not be very large
        self._fulltext_idx = {}
        for doclist in XMLCorpusView(self.abspath("fulltextIndex.xml"),
                                     'fulltextIndex/corpus',
                                     self._handle_fulltextindex_elt):
            for doc in doclist:
                self._fulltext_idx[doc.ID] = doc 

# 需要导入模块: from nltk.corpus import reader [as 别名]
# 或者: from nltk.corpus.reader import XMLCorpusView [as 别名]
def _buildluindex(self):
        # The number of LUs in Framenet is about 13,000 so this index
        # should not be very large
        self._lu_idx = {}
        for lu in XMLCorpusView(self.abspath("luIndex.xml"),
                                'luIndex/lu', self._handle_elt):
            self._lu_idx[lu['ID']] = lu # populate with LU index entries. if any of these
            # are looked up they will be replaced by full LU objects. 

# 需要导入模块: from nltk.corpus import reader [as 别名]
# 或者: from nltk.corpus.reader import XMLCorpusView [as 别名]
def _lu_file(self, lu, ignorekeys=[]):
        """
        Augment the LU information that was loaded from the frame file
        with additional information from the LU file.
        """
        fn_luid = lu.ID

        fname = "lu{0}.xml".format(fn_luid)
        locpath = os.path.join("{0}".format(self._root), self._lu_dir, fname)
        #print(locpath, file=sys.stderr)
        if not self._lu_idx:
            self._buildluindex()

        try:
            elt = XMLCorpusView(locpath, 'lexUnit')[0]
        except IOError:
            raise FramenetError('Unknown LU id: {0}'.format(fn_luid))

        lu2 = self._handle_lexunit_elt(elt, ignorekeys)
        lu.URL = self._fnweb_url + '/' + self._lu_dir + '/' + fname
        lu.subCorpus = lu2.subCorpus
        lu.exemplars = SpecialList('luexemplars', 
                                   [sent for subc in lu.subCorpus for sent in subc.sentence])
        for sent in lu.exemplars:
            sent['LU'] = lu
            sent['frame'] = lu.frame
            for aset in sent.annotationSet:
                aset['LU'] = lu
                aset['frame'] = lu.frame

        return lu