dsh-biocore 0.1.5 → 0.1.6

package/lib/modules/core-turn.js

@@ -94,6 +94,17 @@ export function diagnosticsFromState(state, userText) {
   }
 }
 
+/** Final assistant text: core clean_text wins, JS sanitizer is the fallback.
+ * sanitizeEnabled=false keeps the original (user disabled the switch). */
+export function chooseCleanReply(original, coreCleanText, options) {
+  const opts = options || {}
+  const source = String(original == null ? '' : original)
+  if (opts.sanitizeEnabled === false) return source
+  if (typeof coreCleanText === 'string') return coreCleanText
+  const sanitize = typeof opts.sanitize === 'function' ? opts.sanitize : sanitizeReply
+  return String(sanitize(source))
+}
+
 export function turnStartPayload(options) {
   const opts = options || {}
   const expression = opts.expression && typeof opts.expression === 'object'
@@ -218,6 +229,14 @@ export function makeCoreSessionListener(options) {
           })
           const result = await runAction('turn_end', payload)
           if (!result || result.ok !== true) return
+          const reply = result.reply && typeof result.reply === 'object' ? result.reply : {}
+          const coreCleanText = typeof reply.clean_text === 'string'
+            ? reply.clean_text
+            : (typeof result.clean_text === 'string' ? result.clean_text : null)
+          const finalText = chooseCleanReply(original, coreCleanText, {
+            sanitizeEnabled,
+            sanitize: sanitizeReply,
+          })
           const gate = result.gate && typeof result.gate === 'object' ? result.gate : {}
           const action = stringId(gate.action || 'pass')
           if ((action === 'replace' || action === 'truncate')
@@ -228,9 +247,9 @@ export function makeCoreSessionListener(options) {
               + ' blocked=' + (gate.blocked === true ? '1' : '0')
               + ' rule=' + stringId(gate.rule || '')
               + ' chars=' + original.length + '->' + gate.text.length)
-          } else if (cleaned !== original) {
-            // No gate rewrite: still write the sanitized text back.
-            applyGateText(event, cleaned)
+          } else if (finalText !== original) {
+            // No gate rewrite: write the core-cleaned / JS-fallback text back.
+            applyGateText(event, finalText)
           }
           if (result.trace_file) {
             logLine(log, CORE_TRACE_PREFIX + ' session=' + sessionId

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-biocore",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "type": "module",
   "description": "BioCore DSH UI: per-session state tab, standalone instrument panel, /api/biocore/state, session/event one_turn updates, /bio sync.",
   "main": "lib/index.js",

package/python/biocore_runtime/pipeline.py

@@ -96,6 +96,11 @@ try:
 except Exception:  # pragma: no cover
     _rerank_mod = None
 
+try:
+    from biocore_model import sanitize as _sanitize
+except Exception:  # pragma: no cover
+    _sanitize = None
+
 try:  # self-layer modules (all pure; optional so a stripped checkout still runs)
     from biocore_model import x_directives as _x_directives
     from biocore_model import self_intent as _self_intent
@@ -1794,10 +1799,35 @@ def dispatch(action, payload=None):
                 return bad("missing session_id")
             if "assistant_text" not in data:
                 return bad("missing assistant_text")
-            assistant_text = clean_text(data.get("assistant_text"))
+            raw_assistant = clean_text(data.get("assistant_text"))
+            do_sanitize = _as_bool(data.get("sanitize_reply"), True)
+            cleaned = raw_assistant
+            sanitized = False
+            removed_tags = 0
+            if do_sanitize and _sanitize is not None:
+                stats = getattr(_sanitize, "STATS", {}) or {}
+                before = (int(stats.get("tag_stripped", 0) or 0),
+                          int(stats.get("json_stripped", 0) or 0),
+                          int(stats.get("fenced_stripped", 0) or 0))
+                try:
+                    cleaned = _sanitize.clean(raw_assistant)
+                except Exception:
+                    cleaned = raw_assistant
+                after = (int(stats.get("tag_stripped", 0) or 0),
+                         int(stats.get("json_stripped", 0) or 0),
+                         int(stats.get("fenced_stripped", 0) or 0))
+                removed_tags = max(0, sum(after) - sum(before))
+                sanitized = cleaned != raw_assistant or removed_tags > 0
+            assistant_text = cleaned
             switches = normalize_switches(data.get("switches"))
             gate = gate_reply(assistant_text, {
                 "switches": switches, "user_text": data.get("user_text", "")})
+            if do_sanitize and _sanitize is not None:
+                try:
+                    gate = dict(gate)
+                    gate["text"] = _sanitize.clean(gate.get("text") or "")
+                except Exception:
+                    pass
             captured = capture(session_id, "assistant", assistant_text, store, {
                 "enabled": _as_bool(data.get("memoryEnabled",
                                              switches.get("memoryEnabled", True)),
@@ -1818,6 +1848,9 @@ def dispatch(action, payload=None):
                 "captured": {"captured": bool(decision.get("capture")),
                              "entry_id": str(entry.get("id") or ""),
                              "reason": str(decision.get("reason") or "")},
+                "clean_text": cleaned,
+                "sanitized": bool(sanitized),
+                "removed_tags": int(removed_tags),
                 "trace": trace,
                 "trace_file": trace_file,
             }