Недопустимый байт продолжения при попытке прочитать файл graphml - PullRequest
Новая
       43

Недопустимый байт продолжения при попытке прочитать файл graphml

0 голосов
/ 11 октября 2018

Я создал файл graphml с помощью networkx.Когда я пытаюсь прочитать: 

graph = nx.read_graphml(graph_path)

Я получаю эту ошибку: 

File "/usr/src/app/application/datasources/preprocess_sources.py", line 96, in create_similarities_dictionary
backend_1   |     graph = nx.read_graphml(graph_path)
backend_1   |   File "<decorator-gen-622>", line 2, in read_graphml
backend_1   |   File "/usr/local/lib/python3.6/site-packages/networkx/utils/decorators.py", line 240, in _open_file
backend_1   |     result = func_to_be_decorated(*new_args, **kwargs)
backend_1   |   File "/usr/local/lib/python3.6/site-packages/networkx/readwrite/graphml.py", line 239, in read_graphml
backend_1   |     glist = list(reader(path=path))
backend_1   |   File "/usr/local/lib/python3.6/site-packages/networkx/readwrite/graphml.py", line 728, in __call__
backend_1   |     self.xml = ElementTree(file=path)
backend_1   |   File "/usr/local/lib/python3.6/xml/etree/ElementTree.py", line 557, in __init__
backend_1   |     self.parse(file)
backend_1   |   File "/usr/local/lib/python3.6/xml/etree/ElementTree.py", line 597, in parse
backend_1   |     self._root = parser._parse_whole(source)
backend_1   | UnicodeDecodeError: 'utf-8' codec can't decode byte 0xd5 in position 1: invalid continuation byte

Я уверен, что в файле должны быть какие-то странные символы, но все же это нене работаетНужно ли декодировать файл перед чтением?Это вообще возможно?

Спасибо за вашу помощь!

Редактировать: Вот как строится график: 

def skill_graph_from_df(self, sx_dataframe, path_of_existing=""):
        """Builds directed graph from data frame, where the weight of the edges is the confidence, as used in associaton analysis.
        :param sx_dataframe: Pandas Dataframe - columns: tags, postid, page, alltext.
        :param path_of_existing: str - path of an existing skill graph in GraphML format.
            New data is added to this graph. New graph is built if string is empty.
        :return: void
        """

        self.df_all = sx_dataframe
        self.pagelist = self.df_all.page.unique()

        len_df = len(self.df_all)

        # directed graph with confidence of the rule keyword 1 => keyword 2 as weight for edges (google association analysis for explanation)
        if path_of_existing is not "":
            # import GraphML graph
            self.read_graph(path_of_existing)
            self.keywords_di.graph['pages'] = self.keywords_di.graph['pages'] + ", " + ", ".join(self.pagelist)
        else:
            self.keywords_di.graph['pages'] = ", ".join(self.pagelist)

        for i in range(len_df):
            taglist = nltk.word_tokenize(self.df_all.iloc[i, 0])
            pairs = findsubsets(taglist, 2)  # pairs of keywords
            for word in taglist:  # adds nodes
                if word in self.keywords_di.nodes:
                    self.keywords_di.nodes[word]['count'] += 1
                else:
                    self.keywords_di.add_node(word, count=1)

            for pair in pairs:  # adds edges
                if pair in self.keywords_di.edges:
                    self.keywords_di.edges[pair]['paircount'] += 1
                    self.keywords_di.edges[pair[::-1]]['paircount'] += 1
                else:
                    self.keywords_di.add_edge(*pair, paircount=1)
                    self.keywords_di.add_edge(*pair[::-1], paircount=1)
        for node in self.keywords_di:
            for edge in self.keywords_di.out_edges([node]):
                self.keywords_di.edges[edge]['confidence'] = self.keywords_di.edges[edge]['paircount'] / self.keywords_di.nodes[node]['count']
...