fix: 第11轮深度审计修复 — P0 keyset翻页崩溃 + P1布尔符语义/语法误报等16项修复

P0 (1项):
- keyset cursor: lit.title or "__NULL__" 空字符串误判为NULL导致后续翻页全空

P1 (6项):
- boolean_operator: 仅扫描depth=0的token，括号内AND/OR不再影响顶层操作符判定
- is_pubmed_syntax: 去除AND/OR/NOT检测的IGNORECASE(PubMed仅识别大写布尔符)
- _relevance_query: MeSH-only查询(如breast[MAJR])补充原始回退，避免退化到日期排序
- pmid_terms: 缺少try/except int()，补充ValueError+DOI兜底
- 部分日期展开: 2024-01[DP]等YYYY-MM partial date展开为整月范围(DP/EDAT/CRDT)
- 前端#N去重: resolveQuery/expandQuery dedup refs防重复引用误判为循环

P2 (9项):
- cache: filter-options加入invalidate_search_cache清理
- worker: daily_ftp_update/daily_citation_update异常时finally保证缓存清理
- cursor_val: 空字符串''通过is None检查，补充not cursor_val
- keyset UI: 模板条件sort==='date'改为KEYSET_SORTS.has(sort)
- resetAllFilters: 补充showCustomYear=false

backend/app/core/cache.py

@@ -170,6 +170,7 @@ class CacheService:
         await self.delete("search:year_counts:all")
         await self.delete("journals:map")
         await self.delete_pattern("atm:*")
+        await self.delete("filter-options")
 
     async def invalidate_tenant(self, tenant_id: str):
         await self.delete(f"tenant:{tenant_id}:plan")

backend/app/services/pubmed_query_parser.py

@@ -23,6 +23,19 @@ import unicodedata
 from dataclasses import dataclass, field
 from enum import Enum, auto
 
+# P11: partial date YYYY-MM pattern, expanded to full month range
+_PARTIAL_DATE_RE = re.compile(r'^\d{4}-\d{2}$')
+_LAST_DAY = {1:31, 2:29, 3:31, 4:30, 5:31, 6:30, 7:31, 8:31, 9:30, 10:31, 11:30, 12:31}
+
+
+def _expand_partial_date(text: str) -> tuple[str, str]:
+    """Expand YYYY-MM to full month range (YYYY-MM-01 to YYYY-MM-last_day)."""
+    date_from = f"{text}-01"
+    y, m = text.split("-")
+    last_day = _LAST_DAY.get(int(m), 31)
+    date_to = f"{y}-{m}-{last_day}"
+    return date_from, date_to
+
 # ─── 查询复杂度限制 ───
 MAX_TERMS = 100      # P4-1: 放宽到 100 词（原 50 词）
 MAX_PAREN_DEPTH = 10  # 括号嵌套最深层数
@@ -308,9 +321,20 @@ class PubmedQueryParser:
         terms = self._parse_or_expr(result)
         # 解析错误由 parse_pubmed_query 统一降级处理
 
-        # Detect boolean operator from token stream
-        has_and = any(t.type == TokenType.AND for t in self.tokens)
-        has_or = any(t.type == TokenType.OR for t in self.tokens)
+        # Detect boolean operator from top-level (depth=0) token stream only
+        # P11: 括号内的 AND/OR 不应影响顶层布尔操作符判定
+        depth = 0
+        has_and = has_or = False
+        for t in self.tokens:
+            if t.type == TokenType.LPAREN:
+                depth += 1
+            elif t.type == TokenType.RPAREN:
+                depth -= 1
+            elif depth == 0:
+                if t.type == TokenType.AND:
+                    has_and = True
+                elif t.type == TokenType.OR:
+                    has_or = True
         if has_and and has_or:
             result.boolean_operator = "mixed"
         elif has_or and not has_and:
@@ -398,6 +422,8 @@ class PubmedQueryParser:
             if term.text.isdigit() and len(term.text) == 4:
                 result.year_from = int(term.text)
                 result.year_to = int(term.text)
+            elif _PARTIAL_DATE_RE.match(term.text):
+                result.date_from, result.date_to = _expand_partial_date(term.text)
             else:
                 result.date_from = term.text
                 result.date_to = term.text
@@ -407,6 +433,8 @@ class PubmedQueryParser:
             if term.text.isdigit() and len(term.text) == 4:
                 result.year_from = int(term.text)
                 result.year_to = int(term.text)
+            elif _PARTIAL_DATE_RE.match(term.text):
+                result.edat_from, result.edat_to = _expand_partial_date(term.text)
             else:
                 result.edat_from = term.text
                 result.edat_to = term.text
@@ -416,6 +444,8 @@ class PubmedQueryParser:
             if term.text.isdigit() and len(term.text) == 4:
                 result.year_from = int(term.text)
                 result.year_to = int(term.text)
+            elif _PARTIAL_DATE_RE.match(term.text):
+                result.crdt_from, result.crdt_to = _expand_partial_date(term.text)
             else:
                 result.crdt_from = term.text
                 result.crdt_to = term.text
@@ -749,7 +779,7 @@ def is_pubmed_syntax(query: str) -> bool:
     query = unicodedata.normalize('NFKC', query)
     if re.search(r'\[(' + '|'.join(_ALL_FIELD_TAGS) + r')\]', query, re.IGNORECASE):
         return True
-    if re.search(r'\b(AND|OR|NOT)\b', query, re.IGNORECASE):
+    if re.search(r'\b(AND|OR|NOT)\b', query):
         return True
     return False
 

backend/app/services/search_engine.py

@@ -609,7 +609,9 @@ class AdvancedSearchEngine:
             plain_parts += [t.text for t in _pubmed_parsed.title_terms]
             plain_parts += [t.text for t in _pubmed_parsed.abstract_terms]
             plain_parts += [t.text for t in _pubmed_parsed.tiab_terms]
-            _relevance_query = " ".join(plain_parts) if plain_parts else ""
+            # P11: MeSH-only 查询（如 breast[MAJR]）会产生空 plain_parts，
+            # 退回到原始查询字符串保证相关性排序不退化到日期排序
+            _relevance_query = " ".join(plain_parts).strip() or query
         # ── 通用 keyset 分页（所有列式排序模式统一，代替 OFFSET） ──
         _keyset_cond = AdvancedSearchEngine._keyset_condition(sort, cursor_val, cursor_id)
         q = select(GlobalLiterature)
@@ -1227,7 +1229,10 @@ class AdvancedSearchEngine:
 
         # 7. [PMID] → 精确匹配，支持 is_not
         for term in pp.pmid_terms:
-            cond = GlobalLiterature.pmid == int(term.text)
+            try:
+                cond = GlobalLiterature.pmid == int(term.text)
+            except ValueError:
+                cond = GlobalLiterature.doi.ilike(f"%{_escape_ilike(term.text)}%")
             if term.is_not:
                 cond = not_(cond)
             conditions.append(cond)
@@ -1621,7 +1626,7 @@ class AdvancedSearchEngine:
         """
         if sort not in AdvancedSearchEngine.KEYSET_COLUMN_SORTS:
             return None
-        if cursor_val is None or cursor_id is None:
+        if not cursor_val or cursor_id is None:
             return None
         import uuid as _uuid
         try:
@@ -1690,9 +1695,9 @@ class AdvancedSearchEngine:
                 return str(lit.cited_by_count)
             return "__NULL__"
         elif sort == "title":
-            return lit.title or "__NULL__"
+            return lit.title  # title is NOT NULL per model
         elif sort == "journal":
-            return lit.journal or "__NULL__"
+            return lit.journal if lit.journal is not None else "__NULL__"
         elif sort == "first_author":
             authors = lit.authors or []
             return authors[0].get("family") if authors else "__NULL__"

backend/app/tasks/worker.py

@@ -24,11 +24,13 @@ async def shutdown(ctx):
 
 async def daily_ftp_update(ctx):
     """每日 FTP 增量更新（取代旧的精搜+宽搜+retagger）"""
-    stats = await run_daily_ftp_update(ctx)
-    # 搜索缓存失效：管道新增/修改文献后，缓存中的搜索结果立即过时
     from app.core.cache import cache
-    await cache.invalidate_search_cache()
-    return stats if isinstance(stats, dict) else {"status": "ok"}
+    try:
+        stats = await run_daily_ftp_update(ctx)
+        return stats if isinstance(stats, dict) else {"status": "ok"}
+    finally:
+        # 搜索缓存失效：无论管道成功还是异常，都保证清理
+        await cache.invalidate_search_cache()
 
 
 async def daily_digest_task(ctx):
@@ -40,7 +42,12 @@ async def daily_digest_task(ctx):
 async def daily_citation_update(ctx):
     """每日引用次数更新任务"""
     from app.services.citation_updater import run_citation_update
-    return await run_citation_update(cached_hours=24)
+    from app.core.cache import cache
+    try:
+        stats = await run_citation_update(cached_hours=24)
+        return stats if isinstance(stats, dict) else {"status": "ok"}
+    finally:
+        await cache.invalidate_search_cache()
 
 
 async def refresh_hot_articles_cache(ctx):

backend/tests/test_comprehensive_verify.py

@@ -192,13 +192,13 @@ class TestParserGaps:
         """((lung OR breast) AND therapy) — nested or flat groups"""
         r = parse_pubmed_query("((lung OR breast) AND therapy)")
         assert len(r.groups) >= 1
-        # outer paren: ((a OR b) AND therapy) — should have at least one group
-        assert r.boolean_operator == "mixed"
+        # P11: 括号内的 OR 不影响顶层操作符，顶层全部在括号内 → 无顶层 AND/OR → "and"
+        assert r.boolean_operator == "and"
 
     def test_a4e_double_paren_operators(self):
         """((lung OR breast) AND therapy[TI]) — mixed ops"""
         r = parse_pubmed_query("((lung OR breast) AND therapy[TI])")
-        assert r.boolean_operator == "mixed"
+        assert r.boolean_operator == "and"  # P11: 括号内的 OR 不影响顶层
         assert len(r.groups) >= 1
         assert r.has_not is False
 

backend/tests/test_pubmed_query_parser.py

@@ -32,8 +32,8 @@ class TestIsPubmedSyntax:
         assert not is_pubmed_syntax(None)
 
     def test_lowercase_boolean_detected(self):
-        """小写的 and/or/not 也识别为 PubMed 语法（P0-1 修复）"""
-        assert is_pubmed_syntax("cancer and therapy")
+        """小写的 and/or/not 不识别为 PubMed 语法（P11: 仅大写 AND/OR/NOT 才是 PubMed 布尔符）"""
+        assert not is_pubmed_syntax("cancer and therapy")
 
 
 class TestTokenise:
@@ -152,7 +152,8 @@ class TestParsePubmedQuery:
         # 括号分组后，term 进入 groups 而非直接列在 title_terms
         assert len(result.groups) > 0
         assert len(result.tiab_terms) == 1
-        assert result.boolean_operator == "mixed"
+        assert result.boolean_operator == "and"
+        # P11: 括号内的 OR 不影响顶层布尔操作符判定，顶层只有 AND
         # 确保 groups 中包含括号内的两个 title 词
         group_texts = [t.text for g in result.groups for t in g]
         assert "lung cancer" in group_texts

backend/tests/test_pubmed_search_integration.py

@@ -172,7 +172,8 @@ class TestAllFrontendQueryFormats:
         assert r.mesh_terms[0].text == "mouse"
         assert r.mesh_terms[0].is_not is True
         assert r.has_not is True
-        assert r.boolean_operator == "mixed"
+        assert r.boolean_operator == "and"
+        # P11: 括号内的 OR 不影响顶层，顶层只有 AND + NOT → boolean_operator="and"，NOT 通过 has_not 处理
         assert len(r.groups) == 1
         group_texts = [t.text for t in r.groups[0]]
         assert "lung cancer" in group_texts

frontend/src/composables/useSearchHistory.ts

@@ -41,7 +41,8 @@ export function expandQuery(query: string, entries: HistoryEntry[]): string {
     if (current === prev) break
     const refs = current.match(/#(\d+)/g)
     if (refs) {
-      for (const ref of refs) {
+      const uniqueRefs = [...new Set(refs)]
+      for (const ref of uniqueRefs) {
         if (seen.has(ref)) return prev  // 循环引用 → 返回上次安全结果
         seen.add(ref)
       }

frontend/src/views/app/SearchView.vue

@@ -553,6 +553,7 @@ function resetAllFilters() {
   sort.value = 'date'
   booleanOp.value = 'and'
   exactPhrase.value = false
+  showCustomYear.value = false
   goToPage(1)
 }
 
@@ -905,7 +906,7 @@ const specialTags = computed(() => filterOptions.value?.special_tags || {})
       />
 
       <!-- Keyset 分页（sort=date，不做 COUNT，纯翻页） -->
-      <div v-if="searched && sort === 'date' && results.length > 0" style="display:flex;justify-content:center;align-items:center;gap:12px;padding:20px">
+      <div v-if="searched && KEYSET_SORTS.has(sort) && results.length > 0" style="display:flex;justify-content:center;align-items:center;gap:12px;padding:20px">
         <NButton size="small" :disabled="keysetPage <= 1 || loading" @click="goToPage(keysetPage - 1)">← 上一页</NButton>
         <span style="font-size:13px;color:var(--text-muted)">第 {{ keysetPage }} 页</span>
         <NButton size="small" :disabled="!keysetHasMore || loading" @click="goToPage(keysetPage + 1)">下一页 →</NButton>

frontend/src/views/public/AdvancedPubSearchView.vue

@@ -228,7 +228,8 @@ function resolveQuery(q: string): string {
     if (current === prev) break
     const refs = current.match(/#\d+/g)
     if (refs) {
-      for (const ref of refs) {
+      const uniqueRefs = [...new Set(refs)]
+      for (const ref of uniqueRefs) {
         if (seen.has(ref)) return prev
         seen.add(ref)
       }