такая функция: def call(self, razbor): line = " ".join([["text"] for in razbor])…
- User
такая функция: def call(self, razbor): line = " ".join([["text"] for _ in razbor]) print('line: ', line) processed = self.pipeline.process(line, self.error) print('processed: ', processed) answer = '\n'.join(re.findall("^\d+.+\n", processed, re.M)) print('answer: ', answer) answer = f"Id\tForm\tLemma\tUPosTag\tXPosTag\tFeats\tHead\tDepRel\tDeps\tMisc\n{answer}" print('answer: ', answer) df = pd.read_csv(StringIO(answer), sep="\t", quoting=3) df.reset_index(drop=True).to_csv('/home/user/MyProjects/ocas_idom_logic/df.csv') print('StringIO(answer): ', StringIO(answer)) print('df: ', df) print('all rasbor ', razbor) for (, row), d in zip(df.iterrows(), razbor): print('!!! _ : ', ) print('!!! row : ', row) print('d: ', d) feats = row.Feats.split("|") if "|" in row.Feats else [] print('feats', feats) print('head_id', int(row.Head)) print('rel', row.DepRel) print('pos', row.UPosTag) print('dict', {.split("=")[0]: _.split("=")[1] for _ in feats}) self.update_token(d, { "head_id": int(row.Head), "rel": row.DepRel, "pos": row.UPosTag, **UD2OpenCorpora({.split("=")[0]: _.split("=")[1] for _ in feats},) }) выводит: line: Набережные Челны processed: # newdoc
newpar
sent_id = 1
1 Набережные Набережный ADJ _ Case=Nom|Degree=Pos|Number=Plur 2 nsubj _ _ 2 Челны Челный ADJ _ Degree=Pos|Number=Plur|Variant=Short 0 root _ _
answer: 1 Набережные Набережный ADJ _ Case=Nom|Degree=Pos|Number=Plur 2 nsubj _ _
2 Челны Челный ADJ _ Degree=Pos|Number=Plur|Variant=Short 0 root _ _
answer: Id Form Lemma UPosTag XPosTag Feats Head DepRel Deps Misc 1 Набережные Набережный ADJ _ Case=Nom|Degree=Pos|Number=Plur 2 nsubj _ _
2 Челны Челный ADJ _ Degree=Pos|Number=Plur|Variant=Short 0 root _ _
StringIO(answer): <_io.StringIO object at 0x7f53fb981280> df: Id Form Lemma UPosTag XPosTag Feats Head DepRel Deps Misc 0 1 Набережные Набережный ADJ _ Case=Nom|Degree=Pos|Number=Plur 2 nsubj _ _ 1 2 Челны Челный ADJ _ Degree=Pos|Number=Plur|Variant=Short 0 root _ _ all rasbor Набережные Челны !!! _ : 0 !!! row : Id 1 Form Набережные Lemma Набережный UPosTag ADJ XPosTag _ Feats Case=Nom|Degree=Pos|Number=Plur Head 2 DepRel nsubj Deps _ Misc _ Name: 0, dtype: object d: {'id': 1, 'place': (0, 10), 'text': 'Набережные', 'postfix': ' '} feats ['Case=Nom', 'Degree=Pos', 'Number=Plur'] head_id 2 rel nsubj pos ADJ dict {'Case': 'Nom', 'Degree': 'Pos', 'Number': 'Plur'} !!! _ : 1 !!! row : Id 2 Form Челны Lemma Челный UPosTag ADJ XPosTag _ Feats Degree=Pos|Number=Plur|Variant=Short Head 0 DepRel root Deps _ Misc _ Name: 1, dtype: object d: {'id': 2, 'place': (11, 16), 'text': 'Челны'} feats ['Degree=Pos', 'Number=Plur', 'Variant=Short'] head_id 0 rel root pos ADJ dict {'Degree': 'Pos', 'Number': 'Plur', 'Variant': 'Short'} , перепиши эту функцию без использования pandas
- Assistant
Вот переписанная функция без использования pandas:
from io import StringIO import re def __call__(self, razbor): line = " ".join([_["text"] for _ in razbor]) print('line: ', line) processed = self.pipeline.process(line, self.error) print('processed: ', processed) answer = '\n'.join(re.findall("^\d+.+\n", processed, re.M)) print('answer: ', answer) answer = f"Id\tForm\tLemma\tUPosTag\tXPosTag\tFeats\tHead\tDepRel\tDeps\tMisc\n{answer}" print('answer: ', answer) output = StringIO(answer) df = [] for i, row in enumerate(output): if i > 0: # Skip the header line columns = row.strip().split('\t') df.append(columns) print('df: ', df) print('all razbor ', razbor) for i, d in enumerate(razbor): print('!!! row : ', df[i]) print('d: ', d) feats = df[i][5].split("|") if "|" in df[i][5] else [] print('feats', feats) print('head_id', int(df[i][6])) print('rel', df[i][7]) print('pos', df[i][3]) print('dict', {_.split("=")[0]: _.split("=")[1] for _ in feats}) self._update_token(d, { "head_id": int(df[i][6]), "rel": df[i][7], "pos": df[i][3], **UD2OpenCorpora({_.split("=")[0]: _.split("=")[1] for _ in feats},) })Надо отметить, что в этой функции не используется загрузка и сохранение DataFrame, вместо этого данные из ответа сохраняются в списке df.