Test APIs and fix bugs (#41)

api/apps/chunk_app.py

@@ -214,7 +214,7 @@ def retrieval_test():
     question = req["question"]
     kb_id = req["kb_id"]
     doc_ids = req.get("doc_ids", [])
-    similarity_threshold = float(req.get("similarity_threshold", 0.4))
+    similarity_threshold = float(req.get("similarity_threshold", 0.2))
     vector_similarity_weight = float(req.get("vector_similarity_weight", 0.3))
     top = int(req.get("top", 1024))
     try:

api/apps/conversation_app.py

@@ -170,7 +170,7 @@ def chat(dialog, messages, **kwargs):
         if p["key"] not in kwargs:
             prompt_config["system"] = prompt_config["system"].replace("{%s}"%p["key"], " ")
 
-    model_config = TenantLLMService.get_api_key(dialog.tenant_id, LLMType.CHAT.value, dialog.llm_id)
+    model_config = TenantLLMService.get_api_key(dialog.tenant_id, dialog.llm_id)
     if not model_config: raise LookupError("LLM({}) API key not found".format(dialog.llm_id))
 
     question = messages[-1]["content"]
@@ -186,10 +186,10 @@ def chat(dialog, messages, **kwargs):
     kwargs["knowledge"] = "\n".join(knowledges)
     gen_conf = dialog.llm_setting[dialog.llm_setting_type]
     msg = [{"role": m["role"], "content": m["content"]} for m in messages if m["role"] != "system"]
-    used_token_count = message_fit_in(msg, int(llm.max_tokens * 0.97))
+    used_token_count, msg = message_fit_in(msg, int(llm.max_tokens * 0.97))
     if "max_tokens" in gen_conf:
         gen_conf["max_tokens"] = min(gen_conf["max_tokens"], llm.max_tokens - used_token_count)
-    mdl = ChatModel[model_config.llm_factory](model_config["api_key"], dialog.llm_id)
+    mdl = ChatModel[model_config.llm_factory](model_config.api_key, dialog.llm_id)
     answer = mdl.chat(prompt_config["system"].format(**kwargs), msg, gen_conf)
 
     answer = retrievaler.insert_citations(answer,
@@ -198,4 +198,6 @@ def chat(dialog, messages, **kwargs):
                                  embd_mdl,
                                  tkweight=1-dialog.vector_similarity_weight,
                                  vtweight=dialog.vector_similarity_weight)
+    for c in kbinfos["chunks"]:
+        if c.get("vector"):del c["vector"]
     return {"answer": answer, "retrieval": kbinfos}

api/apps/document_app.py

@@ -11,7 +11,8 @@
 #  distributed under the License is distributed on an "AS IS" BASIS,
 #  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 #  See the License for the specific language governing permissions and
-#  limitations under the License.
+#  limitations under the License
+#
 #
 import base64
 import pathlib
@@ -65,7 +66,7 @@ def upload():
         while MINIO.obj_exist(kb_id, location):
             location += "_"
         blob = request.files['file'].read()
-        MINIO.put(kb_id, filename, blob)
+        MINIO.put(kb_id, location, blob)
         doc = DocumentService.insert({
             "id": get_uuid(),
             "kb_id": kb.id,
@@ -188,7 +189,10 @@ def rm():
         e, doc = DocumentService.get_by_id(req["doc_id"])
         if not e:
             return get_data_error_result(retmsg="Document not found!")
-        ELASTICSEARCH.deleteByQuery(Q("match", doc_id=doc.id), idxnm=search.index_name(doc.kb_id))
+        tenant_id = DocumentService.get_tenant_id(req["doc_id"])
+        if not tenant_id:
+            return get_data_error_result(retmsg="Tenant not found!")
+        ELASTICSEARCH.deleteByQuery(Q("match", doc_id=doc.id), idxnm=search.index_name(tenant_id))
 
         DocumentService.increment_chunk_num(doc.id, doc.kb_id, doc.token_num*-1, doc.chunk_num*-1, 0)
         if not DocumentService.delete_by_id(req["doc_id"]):

api/apps/llm_app.py

@@ -75,7 +75,7 @@ def list():
         llms = LLMService.get_all()
         llms = [m.to_dict() for m in llms if m.status == StatusEnum.VALID.value]
         for m in llms:
-            m["available"] = m.llm_name in mdlnms
+            m["available"] = m["llm_name"] in mdlnms
 
         res = {}
         for m in llms: