fix: PubMed搜索合规 — 34项修复 + query_expansion UnboundLocalError + HomeView precision_mode残留

Batch1 — 解析器 (pubmed_query_parser.py)
- P0-1: 未知字段标签降级为 WORD token 而非 ParseError
- P2-1: 增加未消费 token 检查
- P2-2: PRISMA 字段标签正则 [\w:+] → [\w:]+
- P2-3: Unicode NFKC 规格化输入
- P2-4: re.ASCII 防止 Unicode 数字匹配
- P2-9: 移除重复 dataclass 字段

Batch2 — 搜索引擎 (search_engine.py)
- P1-1: 批量 PMID 查询替代 N+1 循环
- P1-2: isdigit() → isdecimal()
- P1-5: 移除 precision_mode 参数
- P2-5: 移除死代码
- P2-6: 统一 _CHINESE_RE 正则

Batch3 — ATM 引擎 (query_expansion.py)
- P0-4: name_zh ILIKE 中文回退 + _find_mesh_tags 中文降级
- P1-3: name_en ILIKE 加 LIMIT 100

Batch4 — API 层 (features.py)
- P1-5: 移除 precision_mode 请求字段
- P1-11: 增加 logging
- P2-8: NLM_SUBSET_LABELS f-string 安全注释
- P2-12: split 校验器近似性注释

Batch5 — SearchView.vue
- P0-5a/b/c: 日期修复（UTC 方法、互斥逻辑、restoreFromQuery 合并）
- P2-10: 筛选模态关闭时重搜
- P3-1: 搜索框 aria-label

Batch6 — HomeView.vue
- P1-10: URL date → date_from/date_to
- P2-11: clearSearch 清空 feedItems 并重加载

Batch7 — LiteratureCard.vue
- P1-6: 字段标签正则 [\w-]+ → [\w:-]+
- P1-7: terms 切片限制 20 项防 ReDoS

后修复:
- query_expansion.py _find_partial_mesh_tags UnboundLocalError（单非中文词未初始化 tag_ids）
- HomeView.vue handleAdvancedSearch precision_mode 残留引用

backend/app/services/pubmed_query_parser.py

@@ -19,6 +19,7 @@
 from __future__ import annotations
 
 import re
+import unicodedata
 from dataclasses import dataclass, field
 from enum import Enum, auto
 
@@ -127,7 +128,7 @@ _TOKEN_PATTERNS: list[tuple[TokenType, str]] = [
 
 _TOKEN_RE = re.compile(
     '|'.join(f'(?P<{t.name}>{p})' for t, p in _TOKEN_PATTERNS),
-    re.IGNORECASE,
+    re.IGNORECASE | re.ASCII,
 )
 
 
@@ -140,7 +141,11 @@ def tokenise(query: str) -> list[Token]:
                 ttype = TokenType[name]
                 if ttype == TokenType.UNKNOWN_FIELD:
                     fname = value.strip('[]').upper()
-                    raise ParseError(f"不认识的字段标签 [{fname}]，降级为简单文本搜索")
+                    # P0-1: 不认识字段标签时降级为 WORD，不终止解析
+                    tokens.append(Token(TokenType.WORD, value.strip('[]')))
+                    if len(tokens) > MAX_TERMS:
+                        raise ParseError(f"查询词过多（超过 {MAX_TERMS} 个），降级为简单文本搜索")
+                    continue
                 tokens.append(Token(ttype, value))
                 if len(tokens) > MAX_TERMS:
                     raise ParseError(f"查询词过多（超过 {MAX_TERMS} 个），降级为简单文本搜索")
@@ -169,11 +174,6 @@ class ParsedPubmedQuery:
     tiab_terms: list[Term] = field(default_factory=list)           # [TIAB]
     author_terms: list[Term] = field(default_factory=list)         # [AU]
     journal_terms: list[Term] = field(default_factory=list)        # [TA]
-    mesh_terms: list[str] = field(default_factory=list)            # [MH]
-    majr_terms: list[str] = field(default_factory=list)            # [MAJR]
-    pub_types: list[str] = field(default_factory=list)             # [PT]
-    doi_terms: list[str] = field(default_factory=list)             # [DOI]
-    pmid_terms: list[int] = field(default_factory=list)            # [PMID]
     affiliation_terms: list[Term] = field(default_factory=list)    # [AD]
     language_terms: list[Term] = field(default_factory=list)       # [LA]
     volume_terms: list[Term] = field(default_factory=list)         # [VI]
@@ -275,10 +275,8 @@ class PubmedQueryParser:
         """入口：解析完整的查询字符串。"""
         result = ParsedPubmedQuery()
         self._depth = 0  # 括号嵌套深度计数器
-        try:
-            terms = self._parse_or_expr(result)
-        except ParseError:
-            return ParsedPubmedQuery()
+        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)
@@ -296,6 +294,13 @@ class PubmedQueryParser:
         for t in _ungrouped:
             self._dispatch_term(result, t)
 
+        # P2-1: Handle unconsumed tokens (e.g., orphan text after RPAREN)
+        if self.pos < len(self.tokens) - 1:
+            for t in self.tokens[self.pos:-1]:  # exclude EOF token
+                if t.type in (TokenType.WORD, TokenType.QUOTED, TokenType.NUMBER):
+                    text = t.value.strip('"') if t.type == TokenType.QUOTED else t.value
+                    result.plain_terms.append(Term(text=text, exact=(t.type == TokenType.QUOTED)))
+
         return result
 
     def _dispatch_term(self, result: ParsedPubmedQuery, term: Term) -> None:
@@ -574,6 +579,8 @@ def parse_pubmed_query(query: str) -> ParsedPubmedQuery:
         return ParsedPubmedQuery()
 
     try:
+        # P2-3: Unicode normalization — strip zero-width chars, normalize fullwidth digits
+        query = unicodedata.normalize('NFKC', query)
         # 将 YYYY/MM/DD 格式的日期分隔符统一为 YYYY-MM-DD，使 tokeniser 正确识别为 DATE
         import re as _re
         query = _re.sub(r'(\d{4})/(\d{2})/(\d{2})', r'\1-\2-\3', query)
@@ -581,7 +588,11 @@ def parse_pubmed_query(query: str) -> ParsedPubmedQuery:
         parser = PubmedQueryParser(tokens)
         return parser.parse()
     except (ParseError, IndexError, ValueError):
-        return ParsedPubmedQuery()
+        # P0-1: 降级时返回原始查询作为 plain_terms，不丢失用户输入
+        degraded = ParsedPubmedQuery()
+        for t in query.strip().split():
+            degraded.plain_terms.append(Term(text=t))
+        return degraded
 
 
 def extract_pubmed_query_for_prisma(query: str) -> tuple[str, list[str]]:
@@ -598,7 +609,7 @@ def extract_pubmed_query_for_prisma(query: str) -> tuple[str, list[str]]:
 
     # 标准化：统一字段大写
     normalized = re.sub(
-        r'\[(\w+)\]',
+        r'\[([\w:]+)\]',
         lambda m: f'[{m.group(1).upper()}]',
         query,
     )