329 lines
14 KiB
Python
329 lines
14 KiB
Python
|
import re
|
|||
|
|
import json
|
|||
|
|
from pathlib import Path
|
|||
|
|
|
|||
|
|
from docx import Document
|
|||
|
|
from typing import Dict, List, Any, Tuple
|
|||
|
|
from collections import defaultdict
|
|||
|
|
|
|||
|
|
class EnhancedDocxExtractor:
|
|||
|
|
def __init__(self):
|
|||
|
|
# 定义字段名称的多种变体
|
|||
|
|
self.field_variants = {
|
|||
|
|
'姓名': ['姓名', '姓 名', '姓 名', '姓名：', '姓 名：','姓 名'],
|
|||
|
|
'性别': ['性别', '性 别', '性 别', '性别：', '性 别：','性 别'],
|
|||
|
|
'出生年月': ['出生年月', '出生年月：', '出生日期', '出生日期：'],
|
|||
|
|
'民族': ['民族', '民族：', '民 族'],
|
|||
|
|
'政治面貌': ['政治面貌', '政治面貌：', '政治面貌：'],
|
|||
|
|
'现任职单位及部门': ['现任职单位及部门', '单位及部门', '工作单位', '现任职单位'],
|
|||
|
|
'联系电话': ['联系电话', '电话', '手机', '联系电话：', '手机号'],
|
|||
|
|
'联系地址': ['联系地址', '地址', '联系地址：', '家庭地址'],
|
|||
|
|
'学历学位': ['学历', '学历：', '学 历', '学历\n学位','学位','学位：','学 位'],
|
|||
|
|
'毕业院校': ['毕业院校', '毕业学校', '毕业院校：','毕业院校系及专业'],
|
|||
|
|
'专业': ['专业', '专业：', '系及专业', '所学专业'],
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
def extract_with_table_structure(self, docx_path: str) -> Dict[str, Any]:
|
|||
|
|
"""
|
|||
|
|
提取 .docx 中的表格结构数据
|
|||
|
|
"""
|
|||
|
|
doc = Document(docx_path)
|
|||
|
|
results = defaultdict(dict)
|
|||
|
|
# 分析每个表格
|
|||
|
|
for table_idx, table in enumerate(doc.tables):
|
|||
|
|
print(f"\n处理表格 {table_idx + 1} ({len(table.rows)}行 × {len(table.columns)}列)")
|
|||
|
|
|
|||
|
|
# 获取表格结构
|
|||
|
|
table_structure = self._analyze_table_structure(table)
|
|||
|
|
# 提取键值对
|
|||
|
|
kv_pairs = self._extract_from_table_structure(table, table_structure)
|
|||
|
|
# 分类存储
|
|||
|
|
for key, value in kv_pairs:
|
|||
|
|
category = self._categorize_field(key)
|
|||
|
|
results[category][key] = value
|
|||
|
|
# 提取段落中的信息
|
|||
|
|
paragraph_info = self._extract_from_paragraphs(doc.paragraphs)
|
|||
|
|
for key, value in paragraph_info:
|
|||
|
|
category = self._categorize_field(key)
|
|||
|
|
results[category][key] = value
|
|||
|
|
|
|||
|
|
return dict(results)
|
|||
|
|
|
|||
|
|
def _analyze_table_structure(self, table) -> List[List[Dict]]:
|
|||
|
|
"""
|
|||
|
|
分析表格结构，返回每个单元格的元信息
|
|||
|
|
"""
|
|||
|
|
structure = []
|
|||
|
|
|
|||
|
|
for row_idx, row in enumerate(table.rows):
|
|||
|
|
row_info = []
|
|||
|
|
for col_idx, cell in enumerate(row.cells):
|
|||
|
|
cell_text = cell.text.strip()
|
|||
|
|
# 分析单元格属性
|
|||
|
|
cell_info = {
|
|||
|
|
'text': cell_text,
|
|||
|
|
'row': row_idx,
|
|||
|
|
'col': col_idx,
|
|||
|
|
'rowspan': 1,
|
|||
|
|
'colspan': 1,
|
|||
|
|
'is_key': self._is_likely_key(cell_text),
|
|||
|
|
'is_value': self._is_likely_value(cell_text),
|
|||
|
|
}
|
|||
|
|
row_info.append(cell_info)
|
|||
|
|
structure.append(row_info)
|
|||
|
|
|
|||
|
|
return structure
|
|||
|
|
|
|||
|
|
def _extract_from_table_structure(self, table, structure) -> List[Tuple[str, str]]:
|
|||
|
|
"""
|
|||
|
|
从表格结构中提取键值对
|
|||
|
|
"""
|
|||
|
|
kv_pairs = []
|
|||
|
|
visited = set()
|
|||
|
|
key_recode = []
|
|||
|
|
for row_idx, row in enumerate(structure):
|
|||
|
|
for col_idx, cell in enumerate(row):
|
|||
|
|
print(f"visited is {visited} ")
|
|||
|
|
print(f'row {row_idx} col {col_idx} all cell is {cell}')
|
|||
|
|
if (row_idx, col_idx) in visited:
|
|||
|
|
print(f'---{row_idx}, {col_idx} ')
|
|||
|
|
print(f'cell is {cell}')
|
|||
|
|
continue
|
|||
|
|
|
|||
|
|
if cell['is_key']:
|
|||
|
|
next_cell = structure[row_idx][col_idx+1]
|
|||
|
|
# 寻找对应的值
|
|||
|
|
print(f"cell2 is {cell} row {row_idx} col {col_idx}")
|
|||
|
|
value = self._find_value_for_key(table, structure, row_idx, col_idx, visited, kv_pairs)
|
|||
|
|
if value:
|
|||
|
|
key = self._normalize_key(cell['text'])
|
|||
|
|
kv_pairs.append((key, value))
|
|||
|
|
|
|||
|
|
else:
|
|||
|
|
print("不是key")
|
|||
|
|
return kv_pairs
|
|||
|
|
|
|||
|
|
def _find_value_for_key(self, table, structure, key_row, key_col, visited, kv_pairs) -> str:
|
|||
|
|
"""
|
|||
|
|
为键找到对应的值
|
|||
|
|
"""
|
|||
|
|
# 尝试右侧单元格
|
|||
|
|
if key_col + 1 < len(structure[key_row]):
|
|||
|
|
value_cell = structure[key_row][key_col + 1]
|
|||
|
|
if value_cell['is_key']:
|
|||
|
|
return None
|
|||
|
|
if value_cell['text'] and (key_row, key_col + 1) not in visited:
|
|||
|
|
# 检查这个值是否与前一个键提取的值相同（可能是合并单元格）
|
|||
|
|
if not self._is_key_duplicate_merged_cell(structure[key_row][key_col]['text'], kv_pairs):
|
|||
|
|
print("前一个不重复")
|
|||
|
|
print(f"visited add {key_row} {key_col + 1}")
|
|||
|
|
visited.add((key_row, key_col + 1))
|
|||
|
|
return value_cell['text']
|
|||
|
|
|
|||
|
|
# 尝试下方单元格
|
|||
|
|
if key_row + 1 < len(structure):
|
|||
|
|
value_cell = structure[key_row + 1][key_col]
|
|||
|
|
if value_cell['text'] and (key_row + 1, key_col) not in visited:
|
|||
|
|
# 检查这个值是否与前一个键提取的值相同（可能是合并单元格）
|
|||
|
|
if not self._is_key_duplicate_merged_cell(structure[key_row][key_col]['text'], kv_pairs):
|
|||
|
|
print("下一个不重复")
|
|||
|
|
print(f"visited add {key_row} {key_col + 1}")
|
|||
|
|
visited.add((key_row + 1, key_col))
|
|||
|
|
return value_cell['text']
|
|||
|
|
|
|||
|
|
# 尝试合并单元格的情况
|
|||
|
|
for row_idx in range(len(structure)):
|
|||
|
|
for col_idx in range(len(structure[row_idx])):
|
|||
|
|
cell = structure[row_idx][col_idx]
|
|||
|
|
if (row_idx, col_idx) not in visited and cell['text']:
|
|||
|
|
# 检查是否在键的附近
|
|||
|
|
if abs(row_idx - key_row) <= 2 and abs(col_idx - key_col) <= 2:
|
|||
|
|
# 检查这个值是否与前一个键提取的值相同
|
|||
|
|
if not self._is_key_duplicate_merged_cell(structure[key_row][key_col]['text'], kv_pairs):
|
|||
|
|
print("合并不重复")
|
|||
|
|
print(f"visited add {key_row} {key_col + 1}")
|
|||
|
|
visited.add((row_idx, col_idx))
|
|||
|
|
return cell['text']
|
|||
|
|
return None
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _is_key_duplicate_merged_cell(self, text, kv_pairs) -> bool:
|
|||
|
|
"""
|
|||
|
|
检查当前文本value是否可能和已收录的kv集合里的key值重复
|
|||
|
|
如下例：1行0列 ，2行0列 都是毕业院校
|
|||
|
|
第一次 1行0列:1行2列组成key:value
|
|||
|
|
第二次到2行0列，检测到 毕业院校已出现在kv_pairs中，不再组合2行0列:2行1列
|
|||
|
|
| 硕士学位/研究生学历：中国科学院计算技术研究所计算机技术专业
|
|||
|
|
毕业院校 |——————————————————————————————————————————————————
|
|||
|
|
|
|
|||
|
|
|————————————————————————————————————————————————————
|
|||
|
|
"""
|
|||
|
|
|
|||
|
|
for k, v in kv_pairs:
|
|||
|
|
if text == k:
|
|||
|
|
return True
|
|||
|
|
|
|||
|
|
return False
|
|||
|
|
|
|||
|
|
def _is_likely_key(self, text: str) -> bool:
|
|||
|
|
"""判断文本是否可能是键"""
|
|||
|
|
if not text or len(text) > 20:
|
|||
|
|
return False
|
|||
|
|
|
|||
|
|
# 检查是否包含常见字段词
|
|||
|
|
key_indicators = ['籍贯','籍 贯','政治面貌','政治\n面貌','姓名','性别','姓 名', '性 别', '出生年月', '民族','民 族', '单位', '部门','联系地址','主要学习经历',
|
|||
|
|
'职务','职 务','职\n务', '职称','职 称', '电话', '地址', '学历', '学位','现任职务','职业资格','奖惩情况'
|
|||
|
|
'专业', '岗位', '经历', '时间', '资格','现任职单位及部门','身份证号','婚姻状况','健康状况','应聘岗位','应聘部门/岗位','毕业院校系及专业']
|
|||
|
|
|
|||
|
|
for indicator in key_indicators:
|
|||
|
|
translation_table = str.maketrans('', '', ' \t\n\r\f\v')
|
|||
|
|
indicator = indicator.translate(translation_table)
|
|||
|
|
text = text.translate(translation_table)
|
|||
|
|
if indicator in text:
|
|||
|
|
return True
|
|||
|
|
|
|||
|
|
# 检查是否有冒号（中文文档常用）
|
|||
|
|
if '：' in text or ':' in text:
|
|||
|
|
key_part = text.split('：')[0].split(':')[0]
|
|||
|
|
if any(indicator in key_part for indicator in key_indicators):
|
|||
|
|
return True
|
|||
|
|
|
|||
|
|
return False
|
|||
|
|
|
|||
|
|
def _is_likely_value(self, text: str) -> bool:
|
|||
|
|
"""判断文本是否可能是值"""
|
|||
|
|
if not text:
|
|||
|
|
return False
|
|||
|
|
|
|||
|
|
# 值通常不是常见的字段名
|
|||
|
|
if self._is_likely_key(text):
|
|||
|
|
return False
|
|||
|
|
|
|||
|
|
# 值可能包含特定内容
|
|||
|
|
if re.match(r'^\d{11}$', text): # 手机号
|
|||
|
|
return True
|
|||
|
|
if re.match(r'^\d{4}年', text): # 日期
|
|||
|
|
return True
|
|||
|
|
if len(text) > 10: # 长文本可能是值
|
|||
|
|
return True
|
|||
|
|
|
|||
|
|
return True
|
|||
|
|
|
|||
|
|
def _normalize_key(self, key_text: str) -> str:
|
|||
|
|
"""标准化键名"""
|
|||
|
|
# 移除冒号和空格
|
|||
|
|
key_text = re.sub(r'[：:\s]+$', '', key_text)
|
|||
|
|
|
|||
|
|
# 映射到标准键名
|
|||
|
|
for std_key, variants in self.field_variants.items():
|
|||
|
|
for variant in variants:
|
|||
|
|
if variant == key_text or key_text in variant:
|
|||
|
|
return std_key
|
|||
|
|
|
|||
|
|
return key_text
|
|||
|
|
|
|||
|
|
def _categorize_field(self, key: str) -> str:
|
|||
|
|
"""将字段分类"""
|
|||
|
|
categories = {
|
|||
|
|
'基本信息': ['姓名', '性别', '出生年月', '民族', '政治面貌','学历学位','毕业院校系及专业'
|
|||
|
|
'婚姻状况', '健康状况', '籍贯', '身份证号','联系电话','婚姻状况','健康状况','身份证号','联系电话（手机）','毕业院校系及专业','联系地址','主要学习经历','奖惩情况'],
|
|||
|
|
'工作信息': ['现任职单位及部门', '现任职务', '职称', '职业资格',
|
|||
|
|
'参加工作时间', '职称取得时间','应聘部门/岗位','是否接受调剂职级/岗位'],
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
for category, fields in categories.items():
|
|||
|
|
translation_table = str.maketrans('', '', ' \t\n\r\f\v')
|
|||
|
|
key = key.translate(translation_table)
|
|||
|
|
if key in fields:
|
|||
|
|
# print(f"filed is {fields} key is {key} ")
|
|||
|
|
return category
|
|||
|
|
|
|||
|
|
return '其他信息'
|
|||
|
|
|
|||
|
|
def _extract_from_paragraphs(self, paragraphs) -> List[Tuple[str, str]]:
|
|||
|
|
"""从段落中提取信息"""
|
|||
|
|
kv_pairs = []
|
|||
|
|
|
|||
|
|
for para in paragraphs:
|
|||
|
|
text = para.text.strip()
|
|||
|
|
if not text:
|
|||
|
|
continue
|
|||
|
|
|
|||
|
|
# 尝试提取冒号分隔的键值对
|
|||
|
|
if '：' in text or ':' in text:
|
|||
|
|
separator = '：' if '：' in text else ':'
|
|||
|
|
parts = text.split(separator, 1)
|
|||
|
|
|
|||
|
|
if len(parts) == 2:
|
|||
|
|
key = parts[0].strip()
|
|||
|
|
value = parts[1].strip()
|
|||
|
|
|
|||
|
|
if self._is_likely_key(key) and value:
|
|||
|
|
normalized_key = self._normalize_key(key)
|
|||
|
|
kv_pairs.append((normalized_key, value))
|
|||
|
|
|
|||
|
|
return kv_pairs
|
|||
|
|
|
|||
|
|
|
|||
|
|
# 快速使用示例
|
|||
|
|
def quick_extract(docx_path: str):
|
|||
|
|
"""快速提取并显示结果"""
|
|||
|
|
extractor = EnhancedDocxExtractor()
|
|||
|
|
|
|||
|
|
try:
|
|||
|
|
result = extractor.extract_with_table_structure(docx_path)
|
|||
|
|
print("\n提取结果 (键值对格式):")
|
|||
|
|
print("=" * 60)
|
|||
|
|
|
|||
|
|
for category, fields in result.items():
|
|||
|
|
if fields:
|
|||
|
|
print(f"\n{category}:")
|
|||
|
|
for key, value in fields.items():
|
|||
|
|
print(f" {key}: {value}")
|
|||
|
|
return result
|
|||
|
|
|
|||
|
|
except Exception as e:
|
|||
|
|
print(f"提取失败: {e}")
|
|||
|
|
|
|||
|
|
|
|||
|
|
base_map = ['姓名','性别','籍贯','政治面貌','出生年月','身份证号','现居住地','民族','学历','学位','学历学位','特长','联系电话','联系电话（手机）',
|
|||
|
|
'婚姻状况','健康状况','毕业院校系及专业','主要学习经历','联系地址','入党/团时间']
|
|||
|
|
work_map = ['参加工作时间','现任职单位及部门','职务','现任职务','职称','奖惩','工作经历','主要工作经历','职称取得时间','职业资格','应聘部门/岗位']
|
|||
|
|
other_map = ['奖惩','工作经历','主要工作经历','职称取得时间','职业资格','应聘部门/岗位','是否接受调剂职级/岗位']
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
def fetch_info(data):
|
|||
|
|
map_word = base_map + work_map + other_map
|
|||
|
|
print("data is {0}".format(data))
|
|||
|
|
print("map_word is {0}".format(map_word))
|
|||
|
|
|
|||
|
|
for key, value in data.items():
|
|||
|
|
translation_table = str.maketrans('', '', ' \t\n\r\f\v')
|
|||
|
|
clean_key = key.translate(translation_table)
|
|||
|
|
print(f"key is {clean_key} ")
|
|||
|
|
if clean_key in map_word:
|
|||
|
|
clean_value = value.translate(translation_table)
|
|||
|
|
final_res.append({clean_key:clean_value})
|
|||
|
|
|
|||
|
|
return final_res
|
|||
|
|
|
|||
|
|
|
|||
|
|
if __name__ == "__main__":
|
|||
|
|
# 使用方法
|
|||
|
|
docx_file = "./1.报名登记表.docx" # 替换为你的文件
|
|||
|
|
result = quick_extract(docx_file)
|
|||
|
|
print(result)
|
|||
|
|
final_res = []
|
|||
|
|
base_data = result['基本信息']
|
|||
|
|
work_data = result['工作信息']
|
|||
|
|
other_data = result['其他信息']
|
|||
|
|
final_result = []
|
|||
|
|
data = {}
|
|||
|
|
data.update(base_data)
|
|||
|
|
data.update(work_data)
|
|||
|
|
data.update(other_data)
|
|||
|
|
res = fetch_info(data)
|
|||
|
|
print(res)
|
|||
|
|
|