first commit

scripts/map_downloads.py

@@ -0,0 +1,171 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Part0: 把 downloads 表里 node_key 为空的导出文件回填到对应功能页, 实现导出↔页面一一对应。
+匹配策略:
+  1) 普通导出: 文件目录(如 downloads/02_人事/员工关系) 对应功能页 screenshot 目录(模块/分组);
+     在该分组下的候选页中, 选文件名里能命中的页面名(最长命中优先)。
+  2) _任务中心 收割文件: 无目录线索, 用文件名关键词对全站页面名/路径做全局匹配。
+仅回填 node_key 为空者; 打印每条匹配结果与未匹配项供人工核对。
+"""
+import os, sqlite3, re, collections
+
+BASE = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+DB = os.path.join(BASE, "db", "research.db")
+
+
+def db():
+    c = sqlite3.connect(DB); c.row_factory = sqlite3.Row; return c
+
+
+def strip_num(s):
+    return re.sub(r"^\d+_", "", s or "").strip()
+
+
+def load_pages():
+    c = db()
+    rows = [dict(r) for r in c.execute(
+        "SELECT node_key,name,full_path,screenshot_path FROM features "
+        "WHERE screenshot_path IS NOT NULL AND screenshot_path<>''")]
+    c.close()
+    for r in rows:
+        # screenshots/02_人事/员工管理/01_员工花名册.jpg -> 目录 02_人事/员工管理
+        d = os.path.dirname(r["screenshot_path"]).replace("screenshots/", "", 1)
+        r["shot_dir"] = d
+        r["name_core"] = strip_num(r["name"]).replace("员工", "").strip()
+    return rows
+
+
+def best_page_in_group(fname, candidates):
+    """在同分组候选页里, 选页面名能在文件名中命中且命中最长的。"""
+    best, blen = None, 0
+    for p in candidates:
+        for token in (p["name"], strip_num(p["name"]), p["name_core"]):
+            token = (token or "").strip()
+            if len(token) >= 2 and token in fname and len(token) > blen:
+                best, blen = p, len(token)
+    return best
+
+
+def global_match(fname, full_path_hint, pages):
+    """任务中心文件: 按文件名关键词对全站页面做匹配。"""
+    key = fname
+    best, blen = None, 0
+    for p in pages:
+        hay = (p["full_path"] or "") + " " + (p["name"] or "")
+        for token in (strip_num(p["name"]), p["name_core"]):
+            token = (token or "").strip()
+            if len(token) >= 2 and token in key and len(token) > blen:
+                best, blen = p, len(token)
+    return best
+
+
+# 任务中心文件名 -> 目标页面关键词(人工先验, 提高命中)
+TC_HINTS = {
+    "参保": ["参保档案", "社保"],
+    "薪酬档案": ["薪酬档案", "薪资档案"],
+    "已签署": ["已签署"],
+    "电子合同": ["已签署", "合同"],
+}
+
+# 显式关键词->页面 node_key (第一遍模糊未命中时的人工规则, 按顺序取首个命中)
+EXPLICIT = [
+    ("异动", "人事::/personnel/relation/unusual-action"),
+    ("谈话", "人事::/personnel/relation/talk"),
+    ("实习生", "人事::/personnel/relation/internship"),
+    ("已签订合同", "人事::/personnel/relation/contract"),
+    ("合同人员", "人事::/personnel/relation/contract"),
+    ("待入职", "人事::/personnel/relation/enrollment"),
+    ("待离职", "人事::/personnel/relation/departure"),
+    ("试用期", "人事::/personnel/relation/probation"),
+    ("考勤明细", "考勤假期::/attendance/stat/records"),
+    ("打卡记录", "考勤假期::/attendance/stat/records"),
+    ("工资档案", "薪酬::/payroll/salary/profiles"),
+]
+# 目录兜底: 该目录剩余未匹配文件统一归属(员工管理各类统计报表 -> 人事报表)
+DIR_FALLBACK = {
+    "02_人事/员工管理": "人事::/personnel/employee/personnel-report",
+    "02_人事/员工关系": "人事::/personnel/relation/office",
+}
+
+
+def map_explicit():
+    """第二遍: 对仍为空的行用显式关键词规则 + 目录兜底回填。"""
+    c = db()
+    fp = {r["node_key"]: r["full_path"] for r in c.execute("SELECT node_key,full_path FROM features")}
+    rows = [dict(r) for r in c.execute(
+        "SELECT id,file_path FROM downloads WHERE node_key IS NULL OR node_key=''")]
+    done, left = 0, 0
+    for r in rows:
+        rel = r["file_path"].replace("downloads/", "", 1)
+        parts = rel.split("/"); fname = parts[-1]; dir_key = "/".join(parts[:-1])
+        nk = None
+        for sub, node in EXPLICIT:
+            if sub in fname:
+                nk = node; break
+        if not nk:
+            nk = DIR_FALLBACK.get(dir_key)
+        if nk and nk in fp:
+            c.execute("UPDATE downloads SET node_key=?, full_path=? WHERE id=?", (nk, fp[nk], r["id"]))
+            done += 1
+            print("  [OK2] %-40s -> %s" % (fname[:40], fp[nk]))
+        else:
+            left += 1
+            print("  [??2] %-40s (dir=%s)" % (fname[:40], dir_key))
+    c.execute("PRAGMA journal_mode=DELETE")
+    c.commit(); c.close()
+    print("显式回填: 命中 %d, 仍未匹配 %d" % (done, left))
+
+
+
+def main():
+    pages = load_pages()
+    by_dir = collections.defaultdict(list)
+    for p in pages:
+        by_dir[p["shot_dir"]].append(p)
+
+    c = db()
+    rows = [dict(r) for r in c.execute(
+        "SELECT id,file_path FROM downloads WHERE node_key IS NULL OR node_key=''")]
+    print("待回填导出文件: %d" % len(rows))
+    matched, unmatched = 0, []
+    for r in rows:
+        rel = r["file_path"].replace("downloads/", "", 1)
+        parts = rel.split("/")
+        fname = parts[-1]
+        dir_key = "/".join(parts[:-1])
+        page = None
+        if dir_key.startswith("_任务") or dir_key.startswith("_"):
+            # 任务中心: 关键词优先 + 全局匹配
+            hint_pages = None
+            for kw, targets in TC_HINTS.items():
+                if kw in fname:
+                    hint_pages = [p for p in pages if any(t in (p["full_path"] or "") + (p["name"] or "") for t in targets)]
+                    break
+            if hint_pages:
+                page = best_page_in_group(fname, hint_pages) or hint_pages[0]
+            if not page:
+                page = global_match(fname, dir_key, pages)
+        else:
+            cands = by_dir.get(dir_key, [])
+            page = best_page_in_group(fname, cands)
+            if not page and len(cands) == 1:
+                page = cands[0]
+        if page:
+            c.execute("UPDATE downloads SET node_key=?, full_path=? WHERE id=?",
+                      (page["node_key"], page["full_path"], r["id"]))
+            matched += 1
+            print("  [OK] %-46s -> %s" % (fname[:46], page["full_path"]))
+        else:
+            unmatched.append((fname, dir_key))
+            print("  [??] %-46s (dir=%s) 未匹配" % (fname[:46], dir_key))
+    c.execute("PRAGMA journal_mode=DELETE")
+    c.commit(); c.close()
+    print("\n回填完成: 命中 %d, 未匹配 %d" % (matched, len(unmatched)))
+    if unmatched:
+        print("--- 第二遍: 显式关键词 + 目录兜底 ---")
+        map_explicit()
+
+
+if __name__ == "__main__":
+    main()