deal with stop reason being length problem (#109)

api/apps/conversation_app.py

@@ -176,7 +176,7 @@ def chat(dialog, messages, **kwargs):
     if not llm:
         raise LookupError("LLM(%s) not found" % dialog.llm_id)
     llm = llm[0]
-    question = messages[-1]["content"]
+    questions = [m["content"] for m in messages if m["role"] == "user"]
     embd_mdl = LLMBundle(dialog.tenant_id, LLMType.EMBEDDING)
     chat_mdl = LLMBundle(dialog.tenant_id, LLMType.CHAT, dialog.llm_id)
 
@@ -184,7 +184,7 @@ def chat(dialog, messages, **kwargs):
     ## try to use sql if field mapping is good to go
     if field_map:
         stat_logger.info("Use SQL to retrieval.")
-        markdown_tbl, chunks = use_sql(question, field_map, dialog.tenant_id, chat_mdl)
+        markdown_tbl, chunks = use_sql("\n".join(questions), field_map, dialog.tenant_id, chat_mdl)
         if markdown_tbl:
             return {"answer": markdown_tbl, "retrieval": {"chunks": chunks}}
 
@@ -195,7 +195,9 @@ def chat(dialog, messages, **kwargs):
         if p["key"] not in kwargs:
             prompt_config["system"] = prompt_config["system"].replace("{%s}" % p["key"], " ")
 
-    kbinfos = retrievaler.retrieval(question, embd_mdl, dialog.tenant_id, dialog.kb_ids, 1, dialog.top_n,
+    for _ in range(len(questions)//2):
+        questions.append(questions[-1])
+    kbinfos = retrievaler.retrieval(" ".join(questions), embd_mdl, dialog.tenant_id, dialog.kb_ids, 1, dialog.top_n,
                                     dialog.similarity_threshold,
                                     dialog.vector_similarity_weight, top=1024, aggs=False)
     knowledges = [ck["content_with_weight"] for ck in kbinfos["chunks"]]
@@ -224,13 +226,14 @@ def chat(dialog, messages, **kwargs):
 
 
 def use_sql(question, field_map, tenant_id, chat_mdl):
-    sys_prompt = "你是一个DBA。你需要这对以下表的字段结构，根据我的问题写出sql。"
+    sys_prompt = "你是一个DBA。你需要这对以下表的字段结构，根据用户的问题列表，写出最后一个问题对应的SQL。"
     user_promt = """
 表名：{}；
 数据库表字段说明如下：
 {}
 
-问题：{}
+问题如下：
+{}
 请写出SQL，且只要SQL，不要有其他说明及文字。
 """.format(
         index_name(tenant_id),

api/apps/user_app.py

@@ -100,12 +100,14 @@ def github_callback():
             if len(users) > 1: raise Exception('Same E-mail exist!')
             user = users[0]
             login_user(user)
+            return redirect("/?auth=%s"%user.get_id())
         except Exception as e:
             rollback_user_registration(user_id)
             stat_logger.exception(e)
             return redirect("/?error=%s"%str(e))
-
-    return redirect("/?auth=%s"%user_id)
+    user = users[0]
+    login_user(user)
+    return redirect("/?auth=%s" % user.get_id())
 
 
 def user_info_from_github(access_token):

deepdoc/vision/t_recognizer.py

@@ -28,7 +28,7 @@ def main(args):
     images, outputs = init_in_out(args)
     if args.mode.lower() == "layout":
         labels = LayoutRecognizer.labels
-        detr = Recognizer(labels, "layout.paper", os.path.join(get_project_base_directory(), "rag/res/deepdoc/"))
+        detr = Recognizer(labels, "layout", os.path.join(get_project_base_directory(), "rag/res/deepdoc/"))
     if args.mode.lower() == "tsr":
         labels = TableStructureRecognizer.labels
         detr = TableStructureRecognizer()

rag/app/presentation.py

@@ -73,12 +73,13 @@ class Pdf(PdfParser):
         return res
 
 
-def chunk(filename, binary=None, from_page=0, to_page=100000, callback=None, **kwargs):
+def chunk(filename, binary=None, from_page=0, to_page=100000, lang="Chinese", callback=None, **kwargs):
     """
     The supported file formats are pdf, pptx.
     Every page will be treated as a chunk. And the thumbnail of every page will be stored.
     PPT file will be parsed by using this method automatically, setting-up for every PPT file is not necessary.
     """
+    eng = lang.lower() == "english"
     doc = {
         "docnm_kwd": filename,
         "title_tks": huqie.qie(re.sub(r"\.[a-zA-Z]+$", "", filename))
@@ -98,8 +99,10 @@ def chunk(filename, binary=None, from_page=0, to_page=100000, callback=None, **k
         for pn, (txt,img) in enumerate(pdf_parser(filename if not binary else binary, from_page=from_page, to_page=to_page, callback=callback)):
             d = copy.deepcopy(doc)
             d["image"] = img
-            d["page_num_obj"] = [pn+1]
-            tokenize(d, txt, pdf_parser.is_english)
+            d["page_num_int"] = [pn+1]
+            d["top_int"] = [0]
+            d["position_int"].append((pn + 1, 0, img.size[0], 0, img.size[1]))
+            tokenize(d, txt, eng)
             res.append(d)
         return res
 

rag/llm/chat_model.py

@@ -14,9 +14,13 @@
 #  limitations under the License.
 #
 from abc import ABC
+from copy import deepcopy
+
 from openai import OpenAI
 import openai
 
+from rag.nlp import is_english
+
 
 class Base(ABC):
     def __init__(self, key, model_name):
@@ -34,13 +38,17 @@ class GptTurbo(Base):
     def chat(self, system, history, gen_conf):
         if system: history.insert(0, {"role": "system", "content": system})
         try:
-            res = self.client.chat.completions.create(
+            response = self.client.chat.completions.create(
                 model=self.model_name,
                 messages=history,
                 **gen_conf)
-            return res.choices[0].message.content.strip(), res.usage.completion_tokens
+            ans = response.output.choices[0]['message']['content'].strip()
+            if response.output.choices[0].get("finish_reason", "") == "length":
+                ans += "...\nFor the content length reason, it stopped, continue?" if is_english(
+                    [ans]) else "······\n由于长度的原因，回答被截断了，要继续吗？"
+            return ans, response.usage.completion_tokens
         except openai.APIError as e:
-            return "ERROR: "+str(e), 0
+            return "**ERROR**: "+str(e), 0
 
 
 from dashscope import Generation
@@ -59,9 +67,16 @@ class QWenChat(Base):
             result_format='message',
             **gen_conf
         )
+        ans = ""
+        tk_count = 0
         if response.status_code == HTTPStatus.OK:
-            return response.output.choices[0]['message']['content'], response.usage.output_tokens
-        return "ERROR: " + response.message, 0
+            ans += response.output.choices[0]['message']['content']
+            tk_count += response.usage.output_tokens
+            if response.output.choices[0].get("finish_reason", "") == "length":
+                ans += "...\nFor the content length reason, it stopped, continue?" if is_english([ans]) else "······\n由于长度的原因，回答被截断了，要继续吗？"
+            return ans, tk_count
+
+        return "**ERROR**: " + response.message, tk_count
 
 
 from zhipuai import ZhipuAI
@@ -73,11 +88,16 @@ class ZhipuChat(Base):
     def chat(self, system, history, gen_conf):
         from http import HTTPStatus
         if system: history.insert(0, {"role": "system", "content": system})
-        response = self.client.chat.completions.create(
-            self.model_name,
-            messages=history,
-            **gen_conf
-        )
-        if response.status_code == HTTPStatus.OK:
-            return response.output.choices[0]['message']['content'], response.usage.completion_tokens
-        return "ERROR: " + response.message, 0
+        try:
+            response = self.client.chat.completions.create(
+                self.model_name,
+                messages=history,
+                **gen_conf
+            )
+            ans = response.output.choices[0]['message']['content'].strip()
+            if response.output.choices[0].get("finish_reason", "") == "length":
+                ans += "...\nFor the content length reason, it stopped, continue?" if is_english(
+                    [ans]) else "······\n由于长度的原因，回答被截断了，要继续吗？"
+            return ans, response.usage.completion_tokens
+        except Exception as e:
+            return "**ERROR**: " + str(e), 0

rag/nlp/search.py

@@ -224,12 +224,13 @@ class Dealer:
                                                             chunks_tks,
                                                             tkweight, vtweight)
             mx = np.max(sim) * 0.99
-            if mx < 0.35:
+            if mx < 0.66:
                 continue
             cites[idx[i]] = list(
                 set([str(ii) for ii in range(len(chunk_v)) if sim[ii] > mx]))[:4]
 
         res = ""
+        seted = set([])
         for i, p in enumerate(pieces):
             res += p
             if i not in idx:
@@ -237,7 +238,10 @@ class Dealer:
             if i not in cites:
                 continue
             for c in cites[i]: assert int(c) < len(chunk_v)
-            for c in cites[i]: res += f" ##{c}$$"
+            for c in cites[i]:
+                if c in seted:continue
+                res += f" ##{c}$$"
+                seted.add(c)
 
         return res
 
@@ -318,7 +322,7 @@ class Dealer:
             if dnm not in ranks["doc_aggs"]:
                 ranks["doc_aggs"][dnm] = {"doc_id": did, "count": 0}
             ranks["doc_aggs"][dnm]["count"] += 1
-        ranks["doc_aggs"] = [{"doc_name": k, "doc_id": v["doc_id"], "count": v["count"]} for k,v in sorted(ranks["doc_aggs"].items(), key=lambda x:x[1]["count"]*-1)]
+        ranks["doc_aggs"] = []#[{"doc_name": k, "doc_id": v["doc_id"], "count": v["count"]} for k,v in sorted(ranks["doc_aggs"].items(), key=lambda x:x[1]["count"]*-1)]
 
         return ranks