(.*?)<', html) if ttm: title = ttm.group(1).strip() date = "" for pm in re.finditer(r'PubDate[^>]*content="([^"]*)"', html): d = pm.group(1).strip() cm = re.match(r'(\d{4})-(\d{1,2})-(\d{1,2})', d) if cm: date = f"{cm.group(1)}-{cm.group(2).zfill(2)}-{cm.group(3).zfill(2)}" break if not date: dm = re.search(r'(\d{4}-\d{2}-\d{2})', html) if dm: date = dm.group(1) return title, date def fetch_url(url): for retry in range(3): try: r = requests.get(url, headers=HEADERS, timeout=30) if r.status_code == 200: r.encoding = "utf-8" return r.text except: if retry < 2: time.sleep(2) return None def insert_article(url, title, date, content): db = sqlite3.connect(SEARCH_DB, timeout=60) db.execute("PRAGMA journal_mode=WAL") try: db.execute( "INSERT OR IGNORE INTO gov_raw(site_name,source_url,page_url,title,publish_date,content,summary,status,category) VALUES(?,?,?,?,?,?,?,?,?)", (SITE_NAME, url, url, title, date, content, title[:200], "active", "环境影响评价")) affected = db.total_changes db.commit() if affected: db.execute( "INSERT INTO gov_search(rowid,title,site_name,summary) SELECT r.id,r.title,r.site_name,r.summary FROM gov_raw r WHERE r.page_url=? AND r.id NOT IN (SELECT rowid FROM gov_search)", (url,)) db.commit() db.close() return affected except Exception as e: db.close() print(f" DB error: {e}") return 0 def run(max_pages=50): print(f"\n{'='*50}") print(f"🚀 {SITE_NAME}") print(f"{'='*50}") items, total = fetch_list(1) if not items: print("⚠ API无返回，尝试fallback参数...") items, total = fetch_list(1) if not items: print("❌ API仍然无返回") return print(f" 共{total}条") total_pages = min(max_pages, (total + 14) // 15) all_items = list(items) for pg in range(2, min(total_pages, _MAX_PG or total_pages) + 1): its, _ = fetch_list(pg) if not its: print(f" 第{pg}页: 空 -> 结束") break all_items.extend(its) dates = [it["date"] for it in its if it["date"]] if dates: print(f" 第{pg}页: {len(its)} 条 ({min(dates)} ~ {max(dates)})") if max(dates) < CUTOFF: print(f" 全部早于{CUTOFF}，停止翻页") break else: print(f" 第{pg}页: {len(its)} 条") time.sleep(0.15) print(f"\n📊 API共 {len(all_items)} 条") new = skip = no_body = err = 0 for i, item in enumerate(all_items, 1): if item["date"] and item["date"] < CUTOFF: skip += 1 continue html = fetch_url(item["url"]) if not html: err += 1 continue pt, pd = extract_meta(html) real_title = pt if pt else item["title"] real_date = pd if pd else item["date"] if real_date and real_date < CUTOFF: skip += 1 continue content = extract_content(html) text_len = len(re.sub(r'<[^>]+>', '', content).strip()) if content else 0 if text_len < 10: no_body += 1 continue if insert_article(item["url"], real_title, real_date, content): new += 1 else: skip += 1 if i % 5 == 0: print(f" ...{i}/{len(all_items)}") time.sleep(0.15) print(f"\n📊 FTS同步...") db = sqlite3.connect(SEARCH_DB, timeout=60) db.execute("DELETE FROM gov_search WHERE site_name=?", (SITE_NAME,)) db.execute("INSERT INTO gov_search(rowid,title,site_name,summary) SELECT rowid,title,site_name,summary FROM gov_raw WHERE site_name=?", (SITE_NAME,)) db.commit() c = db.execute("SELECT COUNT(*) FROM gov_search WHERE site_name=?", (SITE_NAME,)) cnt = c.fetchone()[0] db.close() print(f" FTS: {cnt} 条") print(f"\n✅ {SITE_NAME}: 新增{new}, 跳过{skip}, 空正文{no_body}, 错误{err}, FTS共{cnt}条") if __name__ == "__main__": mp = int(sys.argv[1]) if len(sys.argv)

内的实际文章内容""" # Step 1: Find the article body container # Look for: class="w100 text" div which contains the actual article pats = ['class="w100 text"', 'class=\'w100 text\'', 'w100 text'] text_div_start = -1 for pat in pats: text_div_start = html.find(pat) if text_div_start >= 0: break if text_div_start < 0: # Fallback: try barrierfree container for pat, tag in [ ("id='barrierfree_container'", 'div'), ('id="barrierfree_container', 'div'), ('id="zoom"', 'div'), ('class="nr_content"', 'div'), ]: i = html.find(pat) if i < 0: continue ds = html.rfind(f"<{tag}", 0, i) if ds < 0: continue s = html[ds:] d = 0 for j in range(len(s)): if s[j:j+4] == f"<{tag}" and (j+4 >= len(s) or s[j+4] in " >\n\r\t"): d += 1 elif s[j:j+3+len(tag)] == f"": d -= 1 if d == 0: gt = s.find(">", 0, j) c = s[gt+1:j] if gt > 0 else s[7:j] c = re.sub(r'<(script|style)[^>]*>.*?', '', c, flags=re.DOTALL|re.I) text = re.sub(r'<[^>]+>', '', c).strip() if len(text) > 50: return c.strip() return "" # Find the opening

= len(s) or s[j+4] in " >\n\r\t"): d += 1 elif s[j:j+6] == "

": d -= 1 if d == 0: gt = s.find(">", 0, j) c = s[gt+1:j] if gt > 0 else s[7:j] # Clean up c = re.sub(r'<(script|style)[^>]*>.*?', '', c, flags=re.DOTALL|re.I) # Remove mso-* and other MS Office cruft from style attributes c = re.sub(r'style="[^"]*mso-[^"]*"', '', c, flags=re.I) # Clean up empty/unnecessary span styles that are just MS Word artifacts c = re.sub(r']*mso-spacerun[^>]*>\s*', '', c, flags=re.DOTALL|re.I) # Remove hardcoded width=1200px, width=1000px, width=888px etc. c = re.sub(r'width\s*[:=]\s*["\']?\d+px["\']?', 'max-width:100%', c, flags=re.I) # Remove the "关闭窗口/返回顶部/打印文章" footer c = re.sub(r']*main_5124[^>]*>.*?