RAG Business in progress

default-config.yaml

@@ -35,7 +35,7 @@ llm:
         - name     : "z-ai/glm-5"
 
 ragroleplay:
-  db_path      : "./.ragroleplay"
+  db_path      : "~/.config/hendrik/ragroleplay"
   vector_size  : 768 # used on table create only
   model_url    : "http://localhost:11434/api/embed"
   model_name   : "nomic-embed-text" # Need to download model from local ollama
@@ -64,4 +64,4 @@ delay: # Humanize Delay (anti-bot detection)
   typing_max    : 10.0 # max typing delay limit per second
 
 user:
-  id: "YOUR_USER_ID_HERE"
+  id: # Fill your user UUID here

interfaces/tui/agent.py

@@ -130,6 +130,30 @@ def _agent_loop(app):
         if response.warning:
             log(app, "system", f"  {response.warning}")
 
+        # Cek apakah response adalah error dari LLM client
+        content = response.content
+        is_error = isinstance(content, str) and (content.startswith("Error:") or content.startswith("HTTP Error:"))
+
+        if is_error:
+            # Hapus "Thinking..." log dan placeholder
+            for i in range(len(app.log) - 1, -1, -1):
+                if app.log[i].get('role') == 'system' and 'Thinking' in app.log[i].get('text', ''):
+                    app.log.pop(i)
+                    break
+            if placeholder_marker is not None:
+                for i in range(len(app.log) - 1, -1, -1):
+                    if app.log[i].get('role') == 'ai':
+                        text = app.log[i].get('text', '')
+                        if placeholder_marker in text or text == "" or text == "...":
+                            app.log.pop(i)
+                            break
+            log(app, "error", content)
+            _add_msg(app, "assistant", content)
+            log(app, "sep", "")
+            ntro.end(stamp)
+            app.agent_done.set()
+            return
+
         # Cek apakah ada tool_calls
         has_tool_calls = bool(response.tool_calls)
 

lib/ragbusiness.py

@@ -0,0 +1,84 @@
+import requests, gc, lancedb, pyarrow, uuid
+from datetime import datetime
+
+def _schema_domainlogic(vector_size):
+    return pyarrow.schema([
+        pyarrow.field('timestamp',          pyarrow.timestamp('ms'),                       metadata={'description': 'When record created'}),
+        pyarrow.field('content',            pyarrow.string(),                              metadata={'description': 'Content'}),
+        pyarrow.field('category',           pyarrow.string(),                              metadata={'description': 'Category. Comma separated.'}),
+        pyarrow.field('vector_context',     pyarrow.list_(pyarrow.float32(), vector_size), metadata={'description': 'Vector data of combined data'}),
+    ])
+
+_TABLE_SCHEMAS = {
+    'knowledge_domainlogic':  _schema_domainlogic,
+}
+
+def db_init(db_path, vector_size):
+    db        = lancedb.connect(db_path)
+    existing  = db.table_names()
+    for name, schema_fn in _TABLE_SCHEMAS.items():
+        if name not in existing:
+            db.create_table(name, schema=schema_fn(vector_size))
+            print(f"[ragroleplay] Created table: {name}", flush=True)
+    del db
+    gc.collect()
+    
+def table_ensure(db_path, table_name, vector_size):
+    db        = lancedb.connect(db_path)
+    existing  = db.table_names()
+    if table_name not in existing:
+        schema_fn = _TABLE_SCHEMAS.get(table_name)
+        if schema_fn:
+            db.create_table(table_name, schema=schema_fn(vector_size))
+    del db
+    gc.collect()
+
+def _uuid(val):
+    return uuid.UUID(val) if isinstance(val, str) else val
+
+def embed_text(url, model, text):
+    response  = requests.post(url=url, json={"model": model, "input": text} )
+    data      = response.json()
+    return data["embeddings"][0]
+
+def domainlogic_store(db_path, model_url, model_name, payload):
+    try:
+        db      = lancedb.connect(db_path)
+        table   = db.open_table("knowledge_domainlogic")
+        
+        vector  = embed_text(model_url, model_name, f'Category: {payload["category"]}. \n{payload["content"]}')
+        
+        record  = {
+            "timestamp"       : datetime.now(),
+            "content"         : payload["content"   ],
+            "category"        : payload["category"  ],
+            "vector_context"  : vector
+        }
+        
+        table.add([record])
+        del table
+        del db
+        gc.collect()
+        
+        return "success"
+    
+    except Exception as e:
+        return e
+
+def domainlogic_load(db_path, payload):
+    try:
+        db       = lancedb.connect(db_path)
+        table    = db.open_table("knowledge_domainlogic")
+        
+        results  = table.search().to_list()
+        
+        del table
+        del db
+        gc.collect()
+        
+        return results
+    
+    except Exception as e:
+        print(f"error: {e}")
+        return []
+

lib/ragroleplay.py

@@ -115,11 +115,8 @@ def ensure_table(db_path, table_name, vector_size): # table_ensure seharusnya
     del db
     gc.collect()
 
-
 def _uuid(val):
-    if isinstance(val, str):
-        return uuid.UUID(val)
-    return val
+    return uuid.UUID(val) if isinstance(val, str) else val
 
 def embed_text(url, model, text):
     response  = requests.post(url=url, json={"model": model, "input": text} )