初始化提交

backend/app/modules/lorebook/matching.py

@@ -0,0 +1,80 @@
+"""世谱匹配引擎：关键词 / 正则 / selective / constant / 递归扫描。
+
+纯逻辑，无 IO。扫描文本由调用方（集成层）从对话历史截取后传入。
+"""
+
+from __future__ import annotations
+
+import re
+
+from app.modules.lorebook.types import LorebookBook, LorebookEntry, MatchedEntry
+
+_RECURSION_MAX_ITERATIONS = 5
+
+
+def match_entries(book: LorebookBook, scan_text: str) -> list[MatchedEntry]:
+    """返回命中的条目（含 constant 与关键词命中）；递归扫描按 book 配置。"""
+    matched: list[MatchedEntry] = []
+    for entry in book.entries:
+        if not entry.enabled:
+            continue
+        if entry.constant:
+            matched.append(MatchedEntry(entry=entry, reason="constant"))
+            continue
+        if _entry_matches(entry, scan_text):
+            matched.append(MatchedEntry(entry=entry, reason="key"))
+
+    if book.recursive_scanning:
+        matched = _expand_recursive(matched, book, scan_text)
+    return matched
+
+
+def _entry_matches(entry: LorebookEntry, scan_text: str) -> bool:
+    if entry.selective and entry.secondary_keys:
+        return _any_key_hit(entry.keys, entry, scan_text) and _any_key_hit(
+            entry.secondary_keys, entry, scan_text
+        )
+    return _any_key_hit(entry.keys, entry, scan_text)
+
+
+def _any_key_hit(keys: list[str], entry: LorebookEntry, scan_text: str) -> bool:
+    for key in keys:
+        if not key:
+            continue
+        if entry.use_regex:
+            flags = 0 if entry.case_sensitive else re.IGNORECASE
+            try:
+                if re.search(key, scan_text, flags):
+                    return True
+            except re.error:
+                # 非法正则降级为字面量包含
+                pass
+        else:
+            hay = scan_text if entry.case_sensitive else scan_text.casefold()
+            needle = key if entry.case_sensitive else key.casefold()
+            if needle in hay:
+                return True
+    return False
+
+
+def _expand_recursive(
+    matched: list[MatchedEntry], book: LorebookBook, original_text: str
+) -> list[MatchedEntry]:
+    """已命中条目的 content 追加进扫描文本，再扫未命中条目；限迭代防振荡。"""
+    hit_ids = {m.entry.id for m in matched}
+    pool = list(matched)
+    extra_text = original_text + "\n" + "\n".join(m.entry.content for m in matched)
+
+    for _ in range(_RECURSION_MAX_ITERATIONS):
+        grew = False
+        for entry in book.entries:
+            if entry.id in hit_ids or not entry.enabled or entry.constant:
+                continue
+            if _entry_matches(entry, extra_text):
+                pool.append(MatchedEntry(entry=entry, reason="recursive"))
+                hit_ids.add(entry.id)
+                extra_text += "\n" + entry.content
+                grew = True
+        if not grew:
+            break
+    return pool