Python Word批量删除指定段落：按关键词清理

用python-docx批量删除Word文档中包含指定关键词、匹配正则或为空的段落，清理模板里的占位符、备注、空行，适合批量清洗合同、报告。

场景痛点

从旧模板复制出来的合同里，总残留"【此处填写XXX】""（备注：略）"这种占位符，要一份份手动找出来删。空段落一堆，排版难看。用Python扫描所有段落，按关键词或正则批量删，另存干净版本，比手工快十倍。

用到的库

pip install python-docx

完整代码

# -*- coding: utf-8 -*-
"""
批量删除 Word 文档中：
1) 包含指定关键词的段落
2) 完全为空的段落
3) 匹配正则的段落
"""
import re
from pathlib import Path
from docx import Document


def delete_paragraph(paragraph):
"""真正从 XML 里删掉一个段落（python-docx 没有直接删除API）"""
p = paragraph._element
p.getparent().remove(p)
# 释放引用
paragraph._p = paragraph._element = None


def clean_docx(docx_path: Path, output_path: Path,
keywords=None,
patterns=None,
remove_empty=True):
"""
keywords: 段落包含任一关键词就删
patterns: 段落匹配任一正则就删
remove_empty: 是否删除空段落
"""
keywords = keywords or []
patterns = patterns or []
compiled = [re.compile(p) for p in patterns]

doc = Document(str(docx_path))
deleted_kw = 0
deleted_re = 0
deleted_empty = 0

# 从后往前删，避免索引错乱
paragraphs = list(doc.paragraphs)
for para in reversed(paragraphs):
text = para.text.strip()

# 空段落
if remove_empty and not text:
delete_paragraph(para)
deleted_empty += 1
continue

# 关键词命中
if any(kw in text for kw in keywords):
delete_paragraph(para)
deleted_kw += 1
continue

# 正则命中
if any(pat.search(text) for pat in compiled):
delete_paragraph(para)
deleted_re += 1
continue

# 表格单元格里的段落也清一遍
for table in doc.tables:
for row in table.rows:
for cell in row.cells:
for para in reversed(list(cell.paragraphs)):
text = para.text.strip()
if remove_empty and not text:
delete_paragraph(para)
elif any(kw in text for kw in keywords):
delete_paragraph(para)

doc.save(str(output_path))
return {
"关键词删除": deleted_kw,
"正则删除": deleted_re,
"空段落删除": deleted_empty,
}


def batch_clean(input_dir, output_dir):
input_dir = Path(input_dir)
output_dir = Path(output_dir)
output_dir.mkdir(parents=True, exist_ok=True)

docx_files = [f for f in input_dir.glob("*.docx") if not f.name.startswith("~$")]
print(f"共 {len(docx_files)} 个文档")

for docx_file in docx_files:
out = output_dir / docx_file.name
stats = clean_docx(
docx_file, out,
keywords=["【此处", "【填写", "（备注", "（注："],
patterns=[r"^第?\s*[一二三四五六七八九十\d]+\s*页\s*$",   # 页码行
r"^\s*第\s*[一二三四五六七八九十\d]+\s*节\s*$"],
remove_empty=True,
)
print(f"✓ {docx_file.name}: 关键词删{stats['关键词删除']} 正则删{stats['正则删除']} 空段删{stats['空段落删除']}")


if __name__ == "__main__":
batch_clean(
input_dir="./word_docs",
output_dir="./cleaned_docs",
)