Update core/sources/cdc.py

core/sources/cdc.py

@@ -1,9 +1,22 @@
-"""CDC / ATSDR ToxProfiles lookup (offline list)."""
+"""CDC/ATSDR ToxProfiles local index + search.
+
+Why local?
+- CDC ToxProfiles don't have a simple public search API.
+- Production code typically uses a prebuilt index (CAS/name -> URL).
+
+This file mirrors that approach: you maintain TOXPROFILES as a list of dicts.
+
+Each item MUST look like:
+    {"name": "Acetone", "cas": "67-64-1", "url": "https://wwwn.cdc.gov/TSP/ToxProfiles/ToxProfiles.aspx?id=5&tid=1"}
+
+Return shape is stable for Gradio rendering:
+    {"ok": True, "query": "...", "cas": "...", "matches": [...], "total": N}
+"""
 
 from __future__ import annotations
 
 import re
-from typing import Dict, List
+from typing import Any, Dict, List, Optional
 
 # Each entry: {"name": str, "cas": str, "url": str}
 TOXPROFILES: List[Dict[str, str]] = [
@@ -258,21 +271,67 @@ TOXPROFILES: List[Dict[str, str]] = [
 
 _CAS_RE = re.compile(r"^\d{2,7}-\d{2}-\d$")
 
-
+def _norm(s: str) -> str:
+    return re.sub(r"\s+", " ", (s or "").strip()).lower()
+    
 def is_cas(s: str) -> bool:
     return bool(_CAS_RE.match((s or "").strip()))
 
 
 def search(query: str, *, limit: int = 8) -> Dict[str, object]:
-    """Search toxprofiles by CAS (preferred) or by substring on name."""
+      """Search the local toxprofile index.
+
+    Behavior:
+    1) If `cas` is provided and CAS-like, prefer exact CAS matches.
+    2) Else if `query` is CAS-like, use exact CAS matches.
+    3) If no CAS matches, fall back to case-insensitive substring name match.
+
+    Args:
+        query: Original user query (CAS or name).
+        cas: Resolved CAS (preferred) if you have it.
+        limit: Max results.
+    """
     q = (query or "").strip()
-    if not q:
-        return {"ok": False, "error": "empty query", "matches": []}
+    cas_q = (cas or "").strip()
+
+    matches: List[Dict[str, str]] = []
+
+    # 1) CAS-first
+    cas_key = cas_q if is_cas(cas_q) else (q if is_cas(q) else "")
+    if cas_key:
+        for item in TOXPROFILES:
+            if (item.get("cas") or "").strip() == cas_key:
+                matches.append(
+                    {
+                        "name": item.get("name") or "ToxProfile",
+                        "cas": item.get("cas") or cas_key,
+                        "url": item.get("url") or "",
+                    }
+                )
+
+    # 2) Name fallback (when CAS isn't resolved OR CAS didn't match anything)
+    if not matches and q:
+        qn = _norm(q)
+        for item in TOXPROFILES:
+            if qn and qn in _norm(item.get("name") or ""):
+                matches.append(
+                    {
+                        "name": item.get("name") or "ToxProfile",
+                        "cas": item.get("cas") or "",
+                        "url": item.get("url") or "",
+                    }
+                )
 
-    if is_cas(q):
-        matches = [x for x in TOXPROFILES if x.get("cas") == q]
-    else:
-        qn = q.casefold()
-        matches = [x for x in TOXPROFILES if qn in (x.get("name") or "").casefold()]
+    # Deterministic order: exact name matches first, then substring
+    if matches and q:
+        qn = _norm(q)
+        matches.sort(key=lambda m: 0 if _norm(m.get("name", "")) == qn else 1)
 
-    return {"ok": True, "query": q, "matches": matches[: max(0, int(limit))], "total": len(matches)}
+    total = len(matches)
+    return {
+        "ok": True,
+        "query": q,
+        "cas": cas_key or cas_q or (q if is_cas(q) else ""),
+        "total": total,
+        "matches": matches[: max(0, int(limit))],
+    }