fix: chatglm not working in doc qa, meta qa and plugin

pilot/model/llm_out/chatglm_llm.py

@@ -1,5 +1,8 @@
 #!/usr/bin/env python3
 # -*- coding:utf-8 -*-
+
+from typing import List
+import re
 import copy
 
 import torch
@@ -33,34 +36,36 @@ def chatglm_generate_stream(
     messages = prompt.split(stop)
     #
     # # Add history conversation
-    hist = []
+    hist = [HistoryEntry()]
-    once_conversation = []
+    system_messages = []
     for message in messages[:-2]:
         if len(message) <= 0:
             continue
-
         if "human:" in message:
-            once_conversation.append(message.split("human:")[1])
+            hist[-1].add_question(message.split("human:")[1])
-        # elif "system:" in message:
+        elif "system:" in message:
-        #     once_conversation.append(f"""###system:{message.split("system:")[1]} """)
+            msg = message.split("system:")[1]
+            hist[-1].add_question(msg)
+            system_messages.append(msg)
         elif "ai:" in message:
-            once_conversation.append(message.split("ai:")[1])
+            hist[-1].add_answer(message.split("ai:")[1])
-            last_conversation = copy.deepcopy(once_conversation)
+            hist.append(HistoryEntry())
-            hist.append(last_conversation)
+        else:
-            once_conversation = []
+            # TODO
-        # else:
+            # hist[-1].add_question(message.split("system:")[1])
-        #     once_conversation.append(f"""###system:{message} """)
+            # once_conversation.append(f"""###system:{message} """)
+            pass
 
     try:
         query = messages[-2].split("human:")[1]
     except IndexError:
-        # fix doc qa: https://github.com/csunny/DB-GPT/issues/274
+        query = messages[-3].split("human:")[1]
-        doc_qa_message = messages[-2]
+    hist = build_history(hist)
-        if "system:" in doc_qa_message:
+    if not hist:
-            query = doc_qa_message.split("system:")[1]
+        # No history conversation, but has system messages, merge to user`s query
-        else:
+        query = prompt_adaptation(system_messages, query)
-            query = messages[-3].split("human:")[1]
     print("Query Message: ", query)
+    print("hist: ", hist)
     # output = ""
     # i = 0
 
@@ -75,3 +80,43 @@ def chatglm_generate_stream(
         yield output
 
     yield output
+
+
+class HistoryEntry:
+    def __init__(self, question: str = "", answer: str = ""):
+        self.question = question
+        self.answer = answer
+
+    def add_question(self, question: str):
+        self.question += question
+
+    def add_answer(self, answer: str):
+        self.answer += answer
+
+    def to_list(self):
+        if self.question == "" or self.answer == "":
+            return None
+        return [self.question, self.answer]
+
+
+def build_history(hist: List[HistoryEntry]) -> List[List[str]]:
+    return list(filter(lambda hl: hl is not None, map(lambda h: h.to_list(), hist)))
+
+
+def prompt_adaptation(system_messages: List[str], human_message: str) -> str:
+    if not system_messages:
+        return human_message
+    system_messages_str = " ".join(system_messages)
+    adaptation_rules = [
+        r"Question:\s*{}\s*",  # chat_db scene
+        r"Goals:\s*{}\s*",  # chat_execution
+        r"问题:\s*{}\s*",  # chat_knowledge zh
+        r"question:\s*{}\s*",  # chat_knowledge en
+    ]
+    # system message has include human question
+    for rule in adaptation_rules:
+        pattern = re.compile(rule.format(re.escape(human_message)))
+        if re.search(pattern, system_messages_str):
+            return system_messages_str
+    # https://huggingface.co/THUDM/chatglm2-6b/blob/e186c891cf64310ac66ef10a87e6635fa6c2a579/modeling_chatglm.py#L926
+    return f"{system_messages_str}\n\n问：{human_message}\n\n答："