解析 xml 文件并将其读取到 pandas 数据帧中

Question

我有一个文件夹和其他几个文件夹。这些文件夹都包含 talkbank CHILDES xml 文件

我编写了代码将数据提取到 pandas 数据框中。此代码的工作原理是要求用户输入包含 xml 文件的文件夹的目录，但我希望它以这样的方式工作，即用户在想要提取 xml 文件时不必一直输入目录。我可以以这样的方式编写：在代码中指定所有文件目录，并且一旦 .py 文件运行，它就会从所有文件夹中提取 xml？

import nltk
import os
import pandas as pd
from lxml import etree
from nltk.corpus.reader import CHILDESCorpusReader

# Function to get user input for the directory

def get_input_directory():
dir_childes_corpus = input("Enter the directory containing CHILDES XML files: ")
return dir_childes_corpus

# Path containing CHILDES XML files

dir_childes_corpus = get_input_directory()

# Empty lists to store speaker and utterance data

speakers_data = [ ]
utterance_data = [ ]

# Define namespaces

namespaces = {
'tb': 'http://www.talkbank.org/ns/talkbank',
'xsi': 'http://www.w3.org/2001/XMLSchema-instance'
}

for filename in os.listdir(dir_childes_corpus):
if filename.endswith('.xml'):
file_path = os.path.join(dir_childes_corpus, filename)

        # Parse XML data
        tree = etree.parse(file_path)
        
        # Extract dialogue_id from folder name and filename
        folder_name = os.path.basename(os.path.dirname(file_path))
        dialogue_id = f"{folder_name}/{os.path.splitext(filename)[0]}.xml"
    
    
        # Extract participant information
        speakers_info = []
        for participant in tree.xpath("//tb:Participants/tb:participant", namespaces=namespaces):
            speaker_info = {
                'dialogue_id': dialogue_id,
                'speaker_id': participant.get('id'),
                'speaker_name': participant.get('name'),
                'role': participant.get('role'),
                'age': participant.get('age'),
                'sex': participant.get('sex')
            }
            speakers_info.append(speaker_info)
        
        df_speakers = pd.DataFrame(speakers_info)
    
        # Extract utterance information including morphemes count
        utts_info = []
        for utt in tree.xpath("//tb:u", namespaces=namespaces):
            speaker = utt.get('who')
            uID = utt.get('uID')  # Extract uID
            utterance_text = ' '.join(utt.xpath(".//tb:w/text()", namespaces=namespaces))
            
           # Count morphemes
            utterance_length = len(utt.xpath(".//tb:w/tb:mor", namespaces=namespaces))
            
            utt_info = {
                'dialogue_id': dialogue_id,
                'uID': uID,
                'speaker': speaker,
                'utterance': utterance_text,
                'utterance_length': utterance_length
            }
            utts_info.append(utt_info)
        df_utts = pd.DataFrame(utts_info)
    
        # Append data to lists
        speakers_data.append(df_speakers)
        utterance_data.append(df_utts)

# Concatenate dataframes

speakers_data = pd.concat(speakers_data, ignore_index=True)
utterance_data = pd.concat(utterance_data, ignore_index=True)

Answer 1

您也可以使用

os.walk()

搜索目录或子目录中的所有 .xml：

import os

for root, dirs, files in os.walk(".", topdown=False):
   for name in files:
       if name.endswith('.xml'):
           # do something ...
          print(os.path.join(root, name))