Python正则表达式捕获组扩展

问题描述

我正在使用(?P<name>)命名捕获组，其中包含与冠状病毒大流行有关的动词和词干列表。

import regex
import pandas as pd


data = {'id':[1,2,3,4,5],'text':['The pandemy is spreading','He is fighting Covid-19','The pandemic virus spreads','This sentence is about a different topic','How do we stop the virus ?']}
df = pd.DataFrame(data)

def covid_lang(text):    
    predicates = ['avoid','contain','track','spread','contact','stop','combat','fight']
    subjects = ['Corona','corona','Covid-19','epidem','infect','virus','pandem','disease','outbreak']

    p1 = fr'(?<=\b(?P<predicate>{"|".join(predicates)}))[^\.]*(?P<subject>{"|".join(subjects)}[a-z]*)'

    result = []
    for m in regex.finditer(p1,text,regex.S):
        result.append([m.group('predicate'),m.group('subject')])

    p2 = fr'\b(?P<subject>{"|".join(subjects)})[^\.]*(?<=\b(?P<predicate>{"|".join(predicates)}))'
    for m in regex.finditer(p2,regex.S):
        result.append([m.group('subject'),m.group('predicate')])

    return result

df['result'] = df['text'].apply(covid_lang)

如果有匹配项，我想返回的主题不仅是词的词干，还包括整个词（即“大流行”和“大流行”，而不是“大流行”）。我尝试在单词列表之后添加[a-z]*，以便捕获组在单词结束时停止，但不会改变任何内容。

此外，是否可以在单个查询中加入两个查询（谓词先于主语，主题先于谓语）？我尝试使用(p1)|(p2)，但不适用于已命名的已捕获组。

最后，是否可以在单个单词中包含Corona和corona之类的大写和小写字母？

解决方法

这应该完成所有三个操作：

from xml.etree.ElementPath import prepare_descendant

import regex
import pandas as pd

data = {'id':[1,2,3,4,5],'text':['The pandemy is spreading','He is fighting Covid-19','The pandemic virus spreads','This sentence is about a different topic','How do we stop the virus ?']}
df = pd.DataFrame(data)

def expand_word(word):
    return f'({word}[a-z]*)'

def construct_named_group_from_list_of_words(word_type,word_list):
    expanded_word_regex_list = [expand_word(stem) for stem in word_list]
    word_in_named_group = fr'(?P<{word_type}>{"|".join(expanded_word_regex_list)})'
    return word_in_named_group

def covid_lang(text):
    predicates = ['avoid','contain','track','spread','contact','stop','combat','fight']
    subjects = ['corona','covid-19','epidem','infect','virus','pandem','disease','outbreak']

    predicate_in_named_group = construct_named_group_from_list_of_words("predicate",predicates)
    subject_in_named_group = construct_named_group_from_list_of_words("subject",subjects)

    result = []

    p1 = fr'(?<=\b{predicate_in_named_group})[^\.]*{subject_in_named_group}'
    p2 = fr'\b{subject_in_named_group}[^\.]*(?<=\b{predicate_in_named_group})'

    p = fr'({p1})|({p2})'

    for m in regex.finditer(p,text,regex.S | regex.IGNORECASE):
        result.append([m.group('predicate'),m.group('subject')])


    return result


df['result'] = df['text'].apply(covid_lang)

print(df)

输出：

   id                                      text                  result
0   1                  The pandemy is spreading  [[spreading,pandemy]]
1   2                   He is fighting Covid-19     [[fight,Covid-19]]
2   3                The pandemic virus spreads   [[spreads,pandemic]]
3   4  This sentence is about a different topic                      []
4   5                How do we stop the virus ?         [[stop,virus]]

但是我不确定您是否始终要先输出谓词？如果没有，应该这样做：

from xml.etree.ElementPath import prepare_descendant

import regex
import pandas as pd


data = {'id':[1,subjects)

    result = []

    p1 = fr'(?<=\b{predicate_in_named_group})[^\.]*{subject_in_named_group}'
    p2 = fr'\b{subject_in_named_group}[^\.]*(?<=\b{predicate_in_named_group})'

    for m in regex.finditer(p1,m.group('subject')])

    for m in regex.finditer(p2,regex.S | regex.IGNORECASE):
        result.append([m.group('subject'),m.group('predicate')])

    return result


df['result'] = df['text'].apply(covid_lang)

print(df)

输出：

   id                                      text                  result
0   1                  The pandemy is spreading  [[pandemy,spreading]]
1   2                   He is fighting Covid-19     [[fight,Covid-19]]
2   3                The pandemic virus spreads   [[pandemic,spreads]]
3   4  This sentence is about a different topic                      []
4   5                How do we stop the virus ?         [[stop,virus]]