Новый раг

app/modules/rag/retrieval/query_terms.py

@@ -0,0 +1,45 @@
+from __future__ import annotations
+
+import re
+
+
+def extract_query_terms(query_text: str) -> list[str]:
+    raw_terms = re.findall(r"[A-Za-z_][A-Za-z0-9_]{2,}", query_text or "")
+    normalized: list[str] = []
+    for term in raw_terms:
+        for variant in _identifier_variants(term):
+            if variant not in normalized:
+                normalized.append(variant)
+    for variant in _intent_variants(query_text):
+        if variant not in normalized:
+            normalized.append(variant)
+    return normalized[:6]
+
+
+def _identifier_variants(term: str) -> list[str]:
+    lowered = term.lower()
+    variants = [lowered]
+    snake = _camel_to_snake(term)
+    if snake and snake not in variants:
+        variants.append(snake)
+    if lowered.endswith("manager") and len(lowered) > len("manager"):
+        manager_split = lowered[: -len("manager")] + "_manager"
+        if manager_split not in variants:
+            variants.append(manager_split)
+    compact = snake.replace("_", "") if snake else ""
+    if compact and compact not in variants:
+        variants.append(compact)
+    return variants
+
+
+def _camel_to_snake(term: str) -> str:
+    first = re.sub(r"(.)([A-Z][a-z]+)", r"\1_\2", term)
+    return re.sub(r"([a-z0-9])([A-Z])", r"\1_\2", first).lower()
+
+
+def _intent_variants(query_text: str) -> list[str]:
+    lowered = (query_text or "").lower()
+    variants: list[str] = []
+    if any(token in lowered for token in ("управ", "control", "manage", "management")):
+        variants.extend(["control", "management", "start", "stop", "status"])
+    return variants