fix: 第七轮深度搜索审计修复 — 20项

- HIGH: isdecimal() → re.match() for PMID (non-ASCII digit crash)
- HIGH: boolean_operator 应用到字段分组和无标签词（AND/OR 错误）
- HIGH: Cursor 分页 pub_date 优先（article_date 导致数据错位）
- Parser: parse 异常处理器清理字段标签/布尔符/引号
- Parser: _parse_range date:year 反向交换（第4分支）
- Parser: MAX_TERMS 保护、TITLE/ABSTRACT 大写键补充
- Engine: best_match 排序剥离字段标签
- Engine: year_from/year_to is not None（年份=0 被 falsy 跳过）
- Engine: _single_term_condition SB 字段全量分发
- Engine: journal_tiers/nlm_subsets 空条件预警（0结果保障）
- Engine: ATM 展开和 _expand_mesh_tag_ids pass → logger.exception
- Engine: 降级 regex [\w/: -] 兼容 [MH:noexp] 冒号
- Frontend: syncSearchToUrl 移到 finally 块
- Frontend: 年份滑块/resetAllFilters 清除 URL 日期
- Frontend: Null cursor 日期安全守卫
- Docs: 新增 P7-1~P7-20 记录，测试数更新至 276

backend/app/services/pubmed_query_parser.py

@@ -621,6 +621,8 @@ class PubmedQueryParser:
                 start_val, end_val = end_val, start_val
             elif _start_is_digit and not _end_is_digit and int(start_val) > int(end_val[:4]):
                 start_val, end_val = end_val, start_val
+            elif not _start_is_digit and _end_is_digit and int(start_val[:4]) > int(end_val):
+                start_val, end_val = end_val, start_val
             # 确定两端是否是 4 位年份
             _start_is_year = start_val.isdigit() and len(start_val) == 4
             _end_is_year = end_val.isdigit() and len(end_val) == 4
@@ -717,10 +719,15 @@ def parse_pubmed_query(query: str) -> ParsedPubmedQuery:
         parser = PubmedQueryParser(tokens)
         return parser.parse()
     except (ParseError, IndexError, ValueError):
-        # P0-1: 降级时返回原始查询作为 plain_terms，不丢失用户输入
+        # P0-1: 降级时清理查询中的 [field] 标签、布尔符、引号和括号
         degraded = ParsedPubmedQuery()
-        for t in query.strip().split():
-            degraded.plain_terms.append(Term(text=t))
+        import re as _degrade_re
+        _clean = _degrade_re.sub(r'\[[\w/: -]+\]', '', query)
+        _clean = _degrade_re.sub(r'\b(AND|OR|NOT)\b', '', _clean)
+        _clean = _clean.replace('"', '').replace('(', '').replace(')', '')
+        for t in _clean.split():
+            if t.strip():
+                degraded.plain_terms.append(Term(text=t.strip()))
         return degraded
 
 

backend/app/services/query_expansion.py

@@ -99,7 +99,7 @@ async def _find_mesh_tags(db: AsyncSession, query: str) -> list[UUID]:
         stmt = select(GlobalTag.id).where(
             GlobalTag.source.in_(["mesh", "manual"]),
             GlobalTag.name_zh.ilike(like_pattern),
-        )
+        ).limit(100)
         rows = await db.execute(stmt)
         for (tid,) in rows:
             if tid not in seen:

backend/app/services/search_engine.py

@@ -219,7 +219,7 @@ class AdvancedSearchEngine:
                      or _pubmed_parsed.year_from or _pubmed_parsed.year_to)
             ]):
                 # 解析失败但检测到 PubMed 语法 — 擦除 [field] 标签、布尔符、引号
-                query = re.sub(r'\[[\w/-]+\]', '', query)  # P5: [\w/-] 覆盖 [Title/Abstract]
+                query = re.sub(r'\[[\w/: -]+\]', '', query)  # P5: [\w/: -] 覆盖 [Title/Abstract] 和 [MH:noexp]
                 query = re.sub(r'\b(AND|OR|NOT)\b', '', query)
                 query = query.replace('"', '').replace('(', '').replace(')', '')
                 query = ' '.join(query.split())
@@ -245,8 +245,8 @@ class AdvancedSearchEngine:
                 terms = [p for p in _phrases if p.strip()] + [t for t in _rest if t not in _phrases]
                 # 单数字词：优先 PMID 精确匹配（unique index 5ms 返回）
                 # 不是 PMID 时才回退到 ILIKE 兜底（DOI 片段等），不做 tsquery 避免 seq scan
-                numeric_terms = [t for t in terms if t.isdecimal() and len(t) <= 15]
-                text_terms = [t for t in terms if not (t.isdecimal() and len(t) <= 15)]
+                numeric_terms = [t for t in terms if re.match(r'^\d{1,15}$', t)]
+                text_terms = [t for t in terms if not re.match(r'^\d{1,15}$', t)]
                 if numeric_terms:
                     num_conds = []
                     if exact_phrase:
@@ -287,6 +287,7 @@ class AdvancedSearchEngine:
                         try:
                             _atm_cond = await _expand_atm(db, _atm_query)
                         except Exception:
+                            logger.exception("ATM expansion failed (flat text): %s", _atm_query[:100])
                             _atm_cond = None
 
                     _cond_before = len(conditions)
@@ -335,8 +336,9 @@ class AdvancedSearchEngine:
             subq = select(GlobalJournal.issn).where(GlobalJournal.tier.in_(journal_tiers))
             result = await db.execute(subq)
             issns = [r for (r,) in result.all()]
-            if issns:
-                conditions.append(GlobalLiterature.journal_issn.in_(issns))
+            if not issns:
+                logger.warning("journal_tiers filter matched zero journals: %s", journal_tiers)
+            conditions.append(GlobalLiterature.journal_issn.in_(issns))
 
         # 标签筛选（含子标签递归）
         if tag_ids:
@@ -386,8 +388,9 @@ class AdvancedSearchEngine:
             subq = select(GlobalJournal.issn).where(GlobalJournal.nlm_subsets.overlap(nlm_subsets))
             result = await db.execute(subq)
             issns = [r for (r,) in result.all()]
-            if issns:
-                conditions.append(GlobalLiterature.journal_issn.in_(issns))
+            if not issns:
+                logger.warning("nlm_subsets filter matched zero journals: %s", nlm_subsets)
+            conditions.append(GlobalLiterature.journal_issn.in_(issns))
 
         # ── PubMed 筛选器 ──
 
@@ -494,7 +497,7 @@ class AdvancedSearchEngine:
 
         # 排序（PubMed 查询时跳过 ts_rank，避免语法标签噪音）
         _relevance_query = query
-        if _pubmed_parsed and sort == "relevance":
+        if _pubmed_parsed and sort in ("relevance", "best_match"):
             # 用纯文本词做相关性排序，去掉 [field] 标签
             plain_parts = [t.text for t in _pubmed_parsed.plain_terms]
             plain_parts += [t.text for t in _pubmed_parsed.title_terms]
@@ -609,6 +612,7 @@ class AdvancedSearchEngine:
         term_conditions: list = []
 
         # 1. 字段级搜索 [TI] [AB] [TIAB] [AU] [TA] [LA] [VI] [IP] [PG] [LID]
+        field_combine = or_ if pp.boolean_operator == "or" else and_
         field_map = {
             "title": pp.title_terms,
             "abstract": pp.abstract_terms,
@@ -631,7 +635,7 @@ class AdvancedSearchEngine:
                 if term.is_not:
                     cond = not_(cond)
                 field_conds.append(cond)
-            term_conditions.append(and_(*field_conds) if len(field_conds) > 1 else field_conds[0])
+            term_conditions.append(field_combine(*field_conds) if len(field_conds) > 1 else field_conds[0])
 
         # 2. 纯文本词（无字段标签）— P0-2: 对无标签词补充 ATM MeSH 展开
         if pp.plain_terms:
@@ -648,14 +652,15 @@ class AdvancedSearchEngine:
                     try:
                         atm_cond = await _expand_atm_inline(db, combined)
                     except Exception:
+                        logger.exception("ATM expansion failed (pubmed plain_terms): %s", combined[:100])
                         atm_cond = None
                     if atm_cond is not None:
-                        text_cond = and_(*plain_conds) if len(plain_conds) > 1 else plain_conds[0]
+                        text_cond = field_combine(*plain_conds) if len(plain_conds) > 1 else plain_conds[0]
                         term_conditions.append(or_(atm_cond, text_cond))
                     else:
-                        term_conditions.append(and_(*plain_conds) if len(plain_conds) > 1 else plain_conds[0])
+                        term_conditions.append(field_combine(*plain_conds) if len(plain_conds) > 1 else plain_conds[0])
                 else:
-                    term_conditions.append(and_(*plain_conds) if len(plain_conds) > 1 else plain_conds[0])
+                    term_conditions.append(field_combine(*plain_conds) if len(plain_conds) > 1 else plain_conds[0])
 
         # 3. [MH] → tree_number 展开，支持 is_not 和 _noexp
         if pp.mesh_terms:
@@ -964,9 +969,9 @@ class AdvancedSearchEngine:
         # 6. [DP] → 年份/日期范围
         dp_negated = "DP" in getattr(pp, 'negated_date_ranges', set())
         dp_conds = []
-        if pp.year_from:
+        if pp.year_from is not None:
             dp_conds.append(GlobalLiterature.pub_year >= pp.year_from)
-        if pp.year_to:
+        if pp.year_to is not None:
             dp_conds.append(GlobalLiterature.pub_year <= pp.year_to)
         if pp.date_from:
             from datetime import date as _dt_date
@@ -1095,10 +1100,18 @@ class AdvancedSearchEngine:
         if field == "TT":
             return GlobalLiterature.vernacular_title.ilike(f"%{_escape_ilike(term.text)}%")
         if field == "SB":
-            subq = select(GlobalJournal.issn).where(
-                GlobalJournal.nlm_subsets.overlap([term.text.upper()])
-            )
-            return GlobalLiterature.journal_issn.in_(subq)
+            val = term.text.upper()
+            if val == "PUBMED":
+                return text("TRUE")  # no-op: 所有记录都是 PubMed
+            elif val == "MEDLINE":
+                return GlobalLiterature.citation_status == "medline"
+            elif len(val) == 1 and val.isalpha():
+                subq = select(GlobalJournal.issn).where(
+                    GlobalJournal.nlm_subsets.overlap([val])
+                )
+                return GlobalLiterature.journal_issn.in_(subq)
+            else:
+                return GlobalLiterature.citation_status == val.lower()
         if field == "STAT":
             return GlobalLiterature.citation_status == term.text.lower()
         if field == "UID":
@@ -1239,7 +1252,7 @@ class AdvancedSearchEngine:
                 for (tid,) in rows:
                     mesh_tag_ids.add(tid)
         except Exception:
-            pass
+            logger.exception("MeSH tag lookup failed for mesh_names=%s", mesh_names[:5])
 
         if not mesh_tag_ids:
             return None
@@ -1263,7 +1276,7 @@ class AdvancedSearchEngine:
                     )).scalars().all()
                     mesh_tag_ids.update(children)
             except Exception:
-                pass
+                logger.exception("Tree number expansion failed for mesh_names=%s", mesh_names[:5])
 
         uids = list(mesh_tag_ids)
         if major_only: