优化生成流程

backend/app/services/library.py

@@ -9,10 +9,28 @@ from app.settings import Settings
 
 
 TOKEN_PATTERN = re.compile(r"[a-zA-Z0-9_]+")
+CN_TOKEN_PATTERN = re.compile(r"[\u4e00-\u9fff]")
+
+TERM_ALIASES = {
+    "法兰": {"flange", "mounting_plate"},
+    "套筒": {"sleeve", "tube", "cylinder"},
+    "孔": {"hole", "circle", "circles"},
+    "沉头": {"countersunk", "counterbore"},
+    "沉孔": {"counterbore", "counterbored"},
+    "槽": {"slot", "obround"},
+    "底板": {"plate", "rectangle", "mounting_plate"},
+    "轴": {"shaft", "cylinder"},
+}
 
 
 def tokens(value: str) -> set[str]:
-    return {token.lower() for token in TOKEN_PATTERN.findall(value)}
+    text = str(value or "").lower()
+    result = {token.lower() for token in TOKEN_PATTERN.findall(text)}
+    result.update(CN_TOKEN_PATTERN.findall(text))
+    for term, aliases in TERM_ALIASES.items():
+        if term in text:
+            result.update(aliases)
+    return result
 
 
 class CdslLibrary:
@@ -38,6 +56,8 @@ class CdslLibrary:
                     "profiles": record.get("profiles", []),
                     "features": record.get("features", []),
                     "summary": record.get("summary", ""),
+                    "family": record.get("family", ""),
+                    "atomics": record.get("features", []),
                 }
                 for record in self._records()[:limit]
             ]
@@ -46,6 +66,8 @@ class CdslLibrary:
             corpus = " ".join([
                 record.get("part_id", ""),
                 record.get("source_name", ""),
+                record.get("summary", ""),
+                record.get("family", ""),
                 " ".join(record.get("profiles", [])),
                 " ".join(record.get("features", [])),
                 " ".join(record.get("parameters", [])),
@@ -60,6 +82,8 @@ class CdslLibrary:
                 "profiles": record.get("profiles", []),
                 "features": record.get("features", []),
                 "summary": record.get("summary", ""),
+                "family": record.get("family", ""),
+                "atomics": record.get("features", []),
             }
             for _, record in scored[:limit]
         ]