如何用Python Regex提取多组合同字段的全部匹配结果

Core Fix for Multiple Matches #

• Replace regex methods that only return the first match (like search()) with finditer() or findall() to capture all contract groups.
• Design regex patterns to target complete contract blocks instead of isolated fields, ensuring each group of related variables is captured as a single entry.

Sample Implementation (Python) #

Assuming you've extracted text from your Word document into a string doc_text:

Step 1: Define Block-Focused Regex Pattern #

Adjust this pattern to match the exact structure of your contract text:

import re
import pandas as pd

# Regex to capture full contract blocks (tweak lookaheads/quantifiers for your text)
contract_pattern = re.compile(
    r"Contract Number:\s*(\S+)\s*"
    r"Location:\s*(.+?)\s*(?=Contract Items|Contract Code|Federal Aid|$)\s*"
    r"Contract Items:\s*(.+?)\s*(?=Contract Code|Federal Aid|$)\s*"
    r"Contract Code:\s*(\S+)\s*"
    r"Federal Aid:\s*(\S+)",
    re.DOTALL | re.IGNORECASE
)

# Extract all matching contract blocks
matches = contract_pattern.finditer(doc_text)

# Convert matches to structured data
contract_list = []
for match in matches:
    contract_list.append({
        "contract_number": match.group(1),
        "location": match.group(2).strip(),
        "contract_items": match.group(3).strip(),
        "contract_code": match.group(4),
        "federal_aid": match.group(5)
    })

# Generate tabular output
df = pd.DataFrame(contract_list)
print(df)

Pattern Adjustment Notes #

• Use re.DOTALL to let . match newlines (critical for multi-line contract blocks).
• Non-greedy quantifiers (+?) prevent the pattern from spanning across multiple contract entries.
• Update lookaheads ((?=...)) to match the actual delimiters between contracts in your text.

Loop Over Multiple Documents #

Wrap the extraction logic in a function to process batches of Word files:

from docx import Document
import os

def extract_contracts(doc_path):
    # Read Word document text
    doc = Document(doc_path)
    doc_text = "\n".join([para.text for para in doc.paragraphs])
    
    # Extract contracts
    matches = contract_pattern.finditer(doc_text)
    entries = []
    for match in matches:
        entries.append({
            "source_doc": doc_path,
            "contract_number": match.group(1),
            "location": match.group(2).strip(),
            "contract_items": match.group(3).strip(),
            "contract_code": match.group(4),
            "federal_aid": match.group(5)
        })
    return entries

# Process all .docx files in a directory
all_entries = []
target_dir = "/path/to/your/contract_docs"

for filename in os.listdir(target_dir):
    if filename.endswith(".docx"):
        full_path = os.path.join(target_dir, filename)
        all_entries.extend(extract_contracts(full_path))

# Save combined results to CSV
final_df = pd.DataFrame(all_entries)
final_df.to_csv("all_contracts.csv", index=False)

Expected Output Example #

Key Tips #

• Test your regex pattern against sample text first to refine capturing groups.
• Add optional groups (e.g., (?:Federal Aid:\s*(\S+))?) to handle missing fields gracefully.
• Install required libraries with pip install python-docx pandas.

内容的提问来源于stack exchange，提问作者Pepa