# ====================== 导入模块区 ====================== from flask import Flask, request, render_template_string, session, redirect, url_for import openai import io import re import zipfile import xml.etree.ElementTree as ET import traceback from functools import wraps import subprocess import shutil # 文件解析依赖兼容判断 # docx try: from docx import Document HAS_DOCX = True except ImportError: HAS_DOCX = False print("[WARN] python-docx未安装，docx解析不可用") # PyPDF2 try: from PyPDF2 import PdfReader HAS_PYPDF2 = True except ImportError: HAS_PYPDF2 = False print("[WARN] PyPDF2未安装，pdf解析不可用") # pdfplumber try: import pdfplumber HAS_PDFPLUMBER = True except ImportError: HAS_PDFPLUMBER = False print("[WARN] pdfplumber未安装，高质量pdf解析不可用") # ====================== Flask 基础配置 ====================== app = Flask(__name__) app.secret_key = "luowenfanwenji547b4f7882714c2994cd9155671c76e7.M1Lx5hRGdBlpq8Ty" app.config["MAX_CONTENT_LENGTH"] = 16 * 1024 * 1024 # 16MB 文件限制 # 大模型GLM4-Flash配置 openai.api_key = "sk-替换成你的智谱API密钥" openai.api_base = "https://open.bigmodel.cn/api/paas/v4" MODEL_NAME = "glm-4-flash" # 登录用户 USERS = { "admin": "123456" } # 支持上传文件格式 ALLOWED_EXTENSIONS = { 'txt', 'md', 'csv', 'log', 'json', 'xml', 'docx', 'doc', 'pdf', 'rtf', 'odt', 'html', 'htm', } FORMAT_LABELS = { 'txt': 'TXT纯文本', 'md': 'Markdown', 'csv': 'CSV表格', 'log': '日志文件', 'json': 'JSON', 'xml': 'XML', 'docx': 'Word文档', 'doc': '旧版Word', 'pdf': 'PDF文档', 'rtf': 'RTF富文本', 'odt': 'OpenDocument', 'html': 'HTML网页', 'htm': 'HTML网页', } # ====================== 工具函数 ====================== def allowed_file(filename): """校验文件后缀是否合法""" return '.' in filename and filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS # ---------------------- 文件解析器 子函数 ---------------------- def _extract_txt(raw_bytes): """纯文本多编码兼容解析""" enc_list = ['utf-8', 'utf-8-sig', 'gbk', 'gb2312', 'gb18030', 'big5', 'euc-jp', 'shift_jis', 'latin-1'] for enc in enc_list: try: text = raw_bytes.decode(enc) if text.strip(): return text except (UnicodeDecodeError, LookupError): continue # 兜底忽略错误解码 return raw_bytes.decode('utf-8', errors='ignore') def _extract_csv(raw_bytes): """CSV解析，清理空行""" text = _extract_txt(raw_bytes) lines = [line.strip() for line in text.split('\n') if line.strip()] return '\n'.join(lines) def _extract_docx_xml(raw_bytes): """docx基础XML解析：正文+页眉页脚""" W_NS = '{http://schemas.openxmlformats.org/wordprocessingml/2006/main}' try: with zipfile.ZipFile(io.BytesIO(raw_bytes), 'r') as z: target_files = [ n for n in z.namelist() if n == 'word/document.xml' or n.startswith('word/header') or n.startswith('word/footer') ] all_lines = [] for fname in target_files: try: root = ET.fromstring(z.read(fname)) for p_node in root.iter(f'{W_NS}p'): txt_list = [t.text for t in p_node.iter(f'{W_NS}t') if t.text] line = ''.join(txt_list).strip() if line: all_lines.append(line) except Exception: continue # 去重相邻重复行 dedup = [] for line in all_lines: if not dedup or line != dedup[-1]: dedup.append(line) print(f"[docx-XML]提取{len(dedup)}行") return '\n'.join(dedup) if dedup else "" except Exception as e: print(f"[docx-XML]解析失败: {e}") return "" def _extract_docx_full_scan(raw_bytes): """docx全量XML扫描：文本框、形状、绘图文字兜底""" W_NS = '{http://schemas.openxmlformats.org/wordprocessingml/2006/main}' A_NS = '{http://schemas.openxmlformats.org/drawingml/2006/main}' try: with zipfile.ZipFile(io.BytesIO(raw_bytes), 'r') as z: xml_files = [f for f in z.namelist() if f.endswith('.xml')] all_lines = [] for fname in xml_files: try: root = ET.fromstring(z.read(fname)) # Word正文文本 for p_node in root.iter(f'{W_NS}p'): txts = [t.text for t in p_node.iter(f'{W_NS}t') if t.text] line = ''.join(txts).strip() if line: all_lines.append(line) # 绘图/形状文本 for sp_node in root.iter(f'{A_NS}sp'): txts = [t.text for t in sp_node.iter(f'{A_NS}t') if t.text] line = ''.join(txts).strip() if line and line not in all_lines: all_lines.append(line) except Exception: continue dedup = [] for line in all_lines: if not dedup or line != dedup[-1]: dedup.append(line) print(f"[docx-全量扫描]解析{len(xml_files)}个XML，提取{len(dedup)}行") return '\n'.join(dedup) if dedup else "" except Exception as e: print(f"[docx-全量扫描]失败: {e}") return "" def _extract_docx(raw_bytes): """docx主解析逻辑：优先python-docx，失败走XML兜底""" parts = [] # 方案1 python-docx if HAS_DOCX: try: doc = Document(io.BytesIO(raw_bytes)) # 段落 for p in doc.paragraphs: t = p.text.strip() if t: parts.append(t) # 表格 for table in doc.tables: for row in table.rows: cells = [cell.text.strip() for cell in row.cells if cell.text.strip()] if cells: parts.append(' | '.join(cells)) if parts: print(f"[docx]常规提取{len(parts)}段") return '\n'.join(parts) except Exception as e: print(f"[docx]python-docx解析失败: {e}") # 方案2 页眉页脚XML xml_res = _extract_docx_xml(raw_bytes) if xml_res: return xml_res # 方案3 全量扫描兜底 return _extract_docx_full_scan(raw_bytes) def _extract_doc(raw_bytes): """旧版 .doc 文件多重解析策略""" # 策略1 antiword命令行 if shutil.which("antiword"): try: proc = subprocess.run( ["antiword", "-"], input=raw_bytes, capture_output=True, timeout=15 ) if proc.returncode == 0 and proc.stdout: text = proc.stdout.decode("utf-8", errors="ignore") if text.strip(): print("[doc] antiword提取成功") return text except Exception: pass # 策略2 catdoc if shutil.which("catdoc"): try: proc = subprocess.run( ["catdoc", "-"], input=raw_bytes, capture_output=True, timeout=15 ) if proc.returncode == 0 and proc.stdout: text = proc.stdout.decode("utf-8", errors="ignore") if text.strip(): print("[doc] catdoc提取成功") return text except Exception: pass # 策略3 伪装docx兼容 if HAS_DOCX: try: doc = Document(io.BytesIO(raw_bytes)) parts = [p.text.strip() for p in doc.paragraphs if p.text.strip()] if parts: print("[doc] python-docx兼容提取成功") return '\n'.join(parts) except Exception: pass # 策略4 UTF16LE二进制提取 try: text = raw_bytes.decode("utf-16-le", errors="ignore") text = re.sub(r'[\x00-\x08\x0b\x0c\x0e-\x1f]', "", text) text = re.sub(r'\n{3,}', '\n\n', text) matches = re.findall(r'[\u4e00-\u9fff\w\s,.:;!?，。、：；！？\-()（）/]+', text) res = '\n'.join(chunk.strip() for chunk in matches if len(chunk.strip()) > 1) if res and len(res) > 30: print(f"[doc]二进制提取成功，字符数：{len(res)}") return res except Exception: pass # 策略5 GBK兜底 try: text = raw_bytes.decode("gbk", errors="ignore") matches = re.findall(r'[\u4e00-\u9fff\w\s,.:;!?，。、：；！？\-()（）/]+', text) res = '\n'.join(chunk.strip() for chunk in matches if len(chunk.strip()) > 1) if res and len(res) > 30: return res except Exception: pass return "" def _extract_pdf(raw_bytes): """PDF解析：优先pdfplumber，兜底PyPDF2""" # 方案1 pdfplumber高质量 if HAS_PDFPLUMBER: try: with pdfplumber.open(io.BytesIO(raw_bytes)) as pdf: parts = [] for page in pdf.pages: page_txt = page.extract_text() if page_txt and page_txt.strip(): parts.append(page_txt.strip()) if parts: print(f"[pdf] pdfplumber提取{len(parts)}页") return '\n'.join(parts) except Exception as e: print(f"[pdf] pdfplumber失败: {e}") # 方案2 PyPDF2 if HAS_PYPDF2: try: reader = PdfReader(io.BytesIO(raw_bytes)) parts = [] for page in reader.pages: page_txt = page.extract_text() if page_txt and page_txt.strip(): parts.append(page_txt.strip()) if parts: print(f"[pdf] PyPDF2提取{len(parts)}页") return '\n'.join(parts) except Exception as e: print(f"[pdf] PyPDF2失败: {e}") return "" def _extract_rtf(raw_bytes): """RTF富文本清洗提取文字""" text = _extract_txt(raw_bytes) if not text.strip().startswith(r'{\rtf'): return text # 移除RTF资源组 text = re.sub(r'\{\\fonttbl[^}]*\}', "", text) text = re.sub(r'\{\\colortbl[^}]*\}', "", text) text = re.sub(r'\{\\stylesheet[^}]*\}', "", text) text = re.sub(r'\{\\\*\\[^}]*\}', "", text) # Unicode转义 text = re.sub(r"\\u(\d+)\??", lambda m: chr(int(m.group(1))), text) # 移除控制符 text = re.sub(r'\\[a-zA-Z]+\d*\s?', ' ', text) text = re.sub(r"\\'[0-9a-fA-F]{2}", "", text) text = re.sub(r'[{}]', "", text) # 空白清理 text = re.sub(r' +', ' ', text) text = re.sub(r'\n{3,}', '\n\n', text) return text.strip() def _extract_odt(raw_bytes): """ODT OpenDocument文本提取""" try: with zipfile.ZipFile(io.BytesIO(raw_bytes), 'r') as z: if "content.xml" in z.namelist(): xml_content = z.read("content.xml").decode("utf-8") text = re.sub(r'<[^>]+>', '\n', xml_content) text = re.sub(r'\n{2,}', '\n', text) text = re.sub(r'[ \t]+', ' ', text) lines = [l.strip() for l in text.split('\n') if l.strip()] return '\n'.join(lines) except Exception: pass return "" def _extract_html(raw_bytes): """HTML网页纯文本提取，过滤标签、脚本、注释""" text = _extract_txt(raw_bytes) # 移除脚本、样式、注释 text = re.sub(r']*>.*?', "", text, flags=re.DOTALL | re.IGNORECASE) text = re.sub(r']*>.*?', "", text, flags=re.DOTALL | re.IGNORECASE) text = re.sub(r'', "", text, flags=re.DOTALL) # 换行标签替换 text = re.sub(r'', '\n', text, flags=re.IGNORECASE) block_tags = r'' text = re.sub(block_tags, '\n', text, flags=re.IGNORECASE) # 删除所有标签 text = re.sub(r'<[^>]+>', "", text) # HTML实体替换 entity_map = { ' ': ' ', '&': '&', '<': '<', '>': '>', '"': '"', ''': "'", ' ': ' ', ' ': ' ' } for ent, char in entity_map.items(): text = text.replace(ent, char) text = re.sub(r'&#(\d+);', lambda m: chr(int(m.group(1))), text) # 空白压缩 text = re.sub(r'\n{3,}', '\n\n', text) text = re.sub(r'[ \t]+', ' ', text) return text.strip() def parse_file(filename, raw_bytes): """文件统一分发解析入口，返回 (文本内容, 错误信息)""" ext = filename.rsplit('.', 1)[1].lower() extract_map = { 'txt': _extract_txt, 'md': _extract_txt, 'csv': _extract_csv, 'log': _extract_txt, 'json': _extract_txt, 'xml': _extract_txt, 'docx': _extract_docx, 'doc': _extract_doc, 'pdf': _extract_pdf, 'rtf': _extract_rtf, 'odt': _extract_odt, 'html': _extract_html, 'htm': _extract_html, } extract_func = extract_map.get(ext, _extract_txt) try: content = extract_func(raw_bytes) if content and content.strip(): return content.strip(), None else: return None, f"文件内容为空（{ext}格式无可用文字）" except Exception as e: err_log = traceback.format_exc() print(f"[文件解析异常] {ext}:\n{err_log}") return None, f"解析失败：{str(e)}" # ====================== 登录鉴权装饰器 ====================== def login_required(f): @wraps(f) def decorated(*args, **kwargs): if "username" not in session: return redirect(url_for("login")) return f(*args, **kwargs) return decorated # ====================== 前端HTML模板字符串 ====================== LOGIN_HTML = """ Login · Resume Fairy
✦✦✦✦✦
✨

Resume Fairy

你的简历魔法优化师
{% if error %}
⚠ {{ error }}
{% endif %}
👤
🔒
♥
html2.link ·粘贴 HTML，一键生成链接
""" MAIN_HTML = """ Resume Fairy