Update Chat API, Logs, Interfaces to store, use references as list

- Remove the need to split by magic string in emacs and chat interfaces
- Move compiling references into string as context for GPT to GPT layer
- Update setup in tests to use new style of setting references
- Name first argument to converse as more appropriate "references"

src/interface/emacs/khoj.el

@@ -443,9 +443,8 @@ RECEIVE-DATE is the message receive date."
   (let* ((message (cdr (or (assoc 'response json-response) (assoc 'message json-response))))
          (sender (cdr (assoc 'by json-response)))
          (receive-date (cdr (assoc 'created json-response)))
-         (context (or (cdr (assoc 'context json-response)) ""))
+         (references (or (cdr (assoc 'context json-response)) '()))
-         (reference-source-texts (split-string context "\n\n# " t))
+         (footnotes (mapcar #'khoj--generate-reference references))
-         (footnotes (mapcar #'khoj--generate-reference reference-source-texts))
          (footnote-links (mapcar #'car footnotes))
          (footnote-defs (mapcar #'cdr footnotes)))
     (thread-first

src/khoj/interface/web/chat.html

@@ -39,7 +39,6 @@
             let references = '';
             if (context) {
                 references = context
-                    .split("\n\n# ")
                     .map((reference, index) => generateReference(reference, index))
                     .join("<sup>,</sup>");
             }

src/khoj/processor/conversation/gpt.py

@@ -223,13 +223,14 @@ A:{ "search-type": "notes" }"""
     return json.loads(story.strip(empty_escape_sequences))
 
 
-def converse(text, user_query, conversation_log={}, api_key=None, temperature=0.2):
+def converse(references, user_query, conversation_log={}, api_key=None, temperature=0.2):
     """
     Converse with user using OpenAI's ChatGPT
     """
     # Initialize Variables
     model = "gpt-3.5-turbo"
     openai.api_key = api_key or os.getenv("OPENAI_API_KEY")
+    compiled_references = "\n\n".join({f"# {item}" for item in references})
 
     personality_primer = "You are Khoj, a friendly, smart and helpful personal assistant."
     conversation_primer = f"""
@@ -237,7 +238,7 @@ Using the notes and our past conversations as context, answer the following ques
 Current Date: {datetime.now().strftime("%Y-%m-%d")}
 
 Notes:
-{text}
+{compiled_references}
 
 Question: {user_query}"""
 

src/khoj/routers/api.py

@@ -215,11 +215,11 @@ def chat(q: Optional[str] = None):
         result_list = []
         for query in inferred_queries:
             result_list.extend(search(query, n=5, r=True, score_threshold=-5.0, dedupe=False))
-        collated_result = "\n\n".join({f"# {item.additional['compiled']}" for item in result_list})
+        compiled_references = [item.additional["compiled"] for item in result_list]
 
     try:
         with timer("Generating chat response took", logger):
-            gpt_response = converse(collated_result, q, meta_log, api_key=api_key)
+            gpt_response = converse(compiled_references, q, meta_log, api_key=api_key)
         status = "ok"
     except Exception as e:
         gpt_response = str(e)
@@ -231,8 +231,8 @@ def chat(q: Optional[str] = None):
         q,
         gpt_response,
         user_message_metadata={"created": user_message_time},
-        khoj_message_metadata={"context": collated_result, "intent": {"inferred-queries": inferred_queries}},
+        khoj_message_metadata={"context": compiled_references, "intent": {"inferred-queries": inferred_queries}},
         conversation_log=meta_log.get("chat", []),
     )
 
-    return {"status": status, "response": gpt_response, "context": collated_result}
+    return {"status": status, "response": gpt_response, "context": compiled_references}

tests/test_chat_actors.py

@@ -111,7 +111,7 @@ def test_extract_multiple_implicit_questions_from_message():
 def test_generate_search_query_using_question_from_chat_history():
     # Arrange
     message_list = [
-        ("What is the name of Mr. Vaders daughter?", "Princess Leia", ""),
+        ("What is the name of Mr. Vader's daughter?", "Princess Leia", []),
     ]
 
     # Act
@@ -127,7 +127,7 @@ def test_generate_search_query_using_question_from_chat_history():
 def test_generate_search_query_using_answer_from_chat_history():
     # Arrange
     message_list = [
-        ("What is the name of Mr. Vaders daughter?", "Princess Leia", ""),
+        ("What is the name of Mr. Vader's daughter?", "Princess Leia", []),
     ]
 
     # Act
@@ -143,7 +143,7 @@ def test_generate_search_query_using_answer_from_chat_history():
 def test_generate_search_query_using_question_and_answer_from_chat_history():
     # Arrange
     message_list = [
-        ("Does Luke Skywalker have any Siblings?", "Yes, Princess Leia", ""),
+        ("Does Luke Skywalker have any Siblings?", "Yes, Princess Leia", []),
     ]
 
     # Act
@@ -159,7 +159,7 @@ def test_generate_search_query_using_question_and_answer_from_chat_history():
 def test_generate_search_query_with_date_and_context_from_chat_history():
     # Arrange
     message_list = [
-        ("When did I visit Masai Mara?", "You visited Masai Mara in April 2000", ""),
+        ("When did I visit Masai Mara?", "You visited Masai Mara in April 2000", []),
     ]
 
     # Act
@@ -184,7 +184,7 @@ def test_generate_search_query_with_date_and_context_from_chat_history():
 def test_chat_with_no_chat_history_or_retrieved_content():
     # Act
     response = converse(
-        text="",  # Assume no context retrieved from notes for the user_query
+        references=[],  # Assume no context retrieved from notes for the user_query
         user_query="Hello, my name is Testatron. Who are you?",
         api_key=api_key,
     )
@@ -202,13 +202,13 @@ def test_chat_with_no_chat_history_or_retrieved_content():
 def test_answer_from_chat_history_and_no_content():
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
     ]
 
     # Act
     response = converse(
-        text="",  # Assume no context retrieved from notes for the user_query
+        references=[],  # Assume no context retrieved from notes for the user_query
         user_query="What is my name?",
         conversation_log=populate_chat_history(message_list),
         api_key=api_key,
@@ -228,13 +228,17 @@ def test_answer_from_chat_history_and_previously_retrieved_content():
     "Chat actor needs to use context in previous notes and chat history to answer question"
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", "Testatron was born on 1st April 1984 in Testville."),
+        (
+            "When was I born?",
+            "You were born on 1st April 1984.",
+            ["Testatron was born on 1st April 1984 in Testville."],
+        ),
     ]
 
     # Act
     response = converse(
-        text="",  # Assume no context retrieved from notes for the user_query
+        references=[],  # Assume no context retrieved from notes for the user_query
         user_query="Where was I born?",
         conversation_log=populate_chat_history(message_list),
         api_key=api_key,
@@ -252,13 +256,15 @@ def test_answer_from_chat_history_and_currently_retrieved_content():
     "Chat actor needs to use context across currently retrieved notes and chat history to answer question"
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
     ]
 
     # Act
     response = converse(
-        text="Testatron was born on 1st April 1984 in Testville.",  # Assume context retrieved from notes for the user_query
+        references=[
+            "Testatron was born on 1st April 1984 in Testville."
+        ],  # Assume context retrieved from notes for the user_query
         user_query="Where was I born?",
         conversation_log=populate_chat_history(message_list),
         api_key=api_key,
@@ -275,13 +281,13 @@ def test_no_answer_in_chat_history_or_retrieved_content():
     "Chat actor should say don't know as not enough contexts in chat history or retrieved to answer question"
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
     ]
 
     # Act
     response = converse(
-        text="",  # Assume no context retrieved from notes for the user_query
+        references=[],  # Assume no context retrieved from notes for the user_query
         user_query="Where was I born?",
         conversation_log=populate_chat_history(message_list),
         api_key=api_key,
@@ -300,23 +306,20 @@ def test_no_answer_in_chat_history_or_retrieved_content():
 def test_answer_requires_current_date_awareness():
     "Chat actor should be able to answer questions relative to current date using provided notes"
     # Arrange
-    context = f"""
+    context = [
-    # {datetime.now().strftime("%Y-%m-%d")} "Naco Taco" "Tacos for Dinner"
+        f"""{datetime.now().strftime("%Y-%m-%d")} "Naco Taco" "Tacos for Dinner"
-      Expenses:Food:Dining  10.00 USD
+Expenses:Food:Dining  10.00 USD""",
-
+        f"""{datetime.now().strftime("%Y-%m-%d")} "Sagar Ratna" "Dosa for Lunch"
-    # {datetime.now().strftime("%Y-%m-%d")} "Sagar Ratna" "Dosa for Lunch"
+Expenses:Food:Dining  10.00 USD""",
-      Expenses:Food:Dining  10.00 USD
+        f"""2020-04-01 "SuperMercado" "Bananas"
-
+Expenses:Food:Groceries  10.00 USD""",
-    # 2020-04-01 "SuperMercado" "Bananas"
+        f"""2020-01-01 "Naco Taco" "Burittos for Dinner"
-      Expenses:Food:Groceries  10.00 USD
+Expenses:Food:Dining  10.00 USD""",
-
+    ]
-    # 2020-01-01 "Naco Taco" "Burittos for Dinner"
-      Expenses:Food:Dining  10.00 USD
-    """
 
     # Act
     response = converse(
-        text=context,  # Assume context retrieved from notes for the user_query
+        references=context,  # Assume context retrieved from notes for the user_query
         user_query="What did I have for Dinner today?",
         api_key=api_key,
     )
@@ -334,23 +337,20 @@ def test_answer_requires_current_date_awareness():
 def test_answer_requires_date_aware_aggregation_across_provided_notes():
     "Chat actor should be able to answer questions that require date aware aggregation across multiple notes"
     # Arrange
-    context = f"""
+    context = [
-    # {datetime.now().strftime("%Y-%m-%d")} "Naco Taco" "Tacos for Dinner"
+        f"""# {datetime.now().strftime("%Y-%m-%d")} "Naco Taco" "Tacos for Dinner"
-      Expenses:Food:Dining  10.00 USD
+Expenses:Food:Dining  10.00 USD""",
-
+        f"""{datetime.now().strftime("%Y-%m-%d")} "Sagar Ratna" "Dosa for Lunch"
-    # {datetime.now().strftime("%Y-%m-%d")} "Sagar Ratna" "Dosa for Lunch"
+Expenses:Food:Dining  10.00 USD""",
-      Expenses:Food:Dining  10.00 USD
+        f"""2020-04-01 "SuperMercado" "Bananas"
-
+Expenses:Food:Groceries  10.00 USD""",
-    # 2020-04-01 "SuperMercado" "Bananas"
+        f"""2020-01-01 "Naco Taco" "Burittos for Dinner"
-      Expenses:Food:Groceries  10.00 USD
+Expenses:Food:Dining  10.00 USD""",
-
+    ]
-    # 2020-01-01 "Naco Taco" "Burittos for Dinner"
-      Expenses:Food:Dining  10.00 USD
-    """
 
     # Act
     response = converse(
-        text=context,  # Assume context retrieved from notes for the user_query
+        references=context,  # Assume context retrieved from notes for the user_query
         user_query="How much did I spend on dining this year?",
         api_key=api_key,
     )
@@ -366,14 +366,14 @@ def test_answer_general_question_not_in_chat_history_or_retrieved_content():
     "Chat actor should be able to answer general questions not requiring looking at chat history or notes"
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
-        ("Where was I born?", "You were born Testville.", ""),
+        ("Where was I born?", "You were born Testville.", []),
     ]
 
     # Act
     response = converse(
-        text="",  # Assume no context retrieved from notes for the user_query
+        references=[],  # Assume no context retrieved from notes for the user_query
         user_query="Write a haiku about unit testing in 3 lines",
         conversation_log=populate_chat_history(message_list),
         api_key=api_key,
@@ -393,20 +393,18 @@ def test_answer_general_question_not_in_chat_history_or_retrieved_content():
 def test_ask_for_clarification_if_not_enough_context_in_question():
     "Chat actor should ask for clarification if question cannot be answered unambiguously with the provided context"
     # Arrange
-    context = f"""
+    context = [
-    # Ramya
+        f"""# Ramya
-    My sister, Ramya, is married to Kali Devi. They have 2 kids, Ravi and Rani.
+My sister, Ramya, is married to Kali Devi. They have 2 kids, Ravi and Rani.""",
-
+        f"""# Fang
-    # Fang
+My sister, Fang Liu is married to Xi Li. They have 1 kid, Xiao Li.""",
-    My sister, Fang Liu is married to Xi Li. They have 1 kid, Xiao Li.
+        f"""# Aiyla
-
+My sister, Aiyla is married to Tolga. They have 3 kids, Yildiz, Ali and Ahmet.""",
-    # Aiyla
+    ]
-    My sister, Aiyla is married to Tolga. They have 3 kids, Yildiz, Ali and Ahmet.
-     """
 
     # Act
     response = converse(
-        text=context,  # Assume context retrieved from notes for the user_query
+        references=context,  # Assume context retrieved from notes for the user_query
         user_query="How many kids does my older sister have?",
         api_key=api_key,
     )

tests/test_chat_director.py

@@ -56,8 +56,8 @@ def test_chat_with_no_chat_history_or_retrieved_content(chat_client):
 def test_answer_from_chat_history(chat_client):
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
     ]
     populate_chat_history(message_list)
 
@@ -78,8 +78,12 @@ def test_answer_from_chat_history(chat_client):
 def test_answer_from_currently_retrieved_content(chat_client):
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", "Testatron was born on 1st April 1984 in Testville."),
+        (
+            "When was I born?",
+            "You were born on 1st April 1984.",
+            ["Testatron was born on 1st April 1984 in Testville."],
+        ),
     ]
     populate_chat_history(message_list)
 
@@ -97,8 +101,12 @@ def test_answer_from_currently_retrieved_content(chat_client):
 def test_answer_from_chat_history_and_previously_retrieved_content(chat_client):
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", "Testatron was born on 1st April 1984 in Testville."),
+        (
+            "When was I born?",
+            "You were born on 1st April 1984.",
+            ["Testatron was born on 1st April 1984 in Testville."],
+        ),
     ]
     populate_chat_history(message_list)
 
@@ -119,8 +127,8 @@ def test_answer_from_chat_history_and_previously_retrieved_content(chat_client):
 def test_answer_from_chat_history_and_currently_retrieved_content(chat_client):
     # Arrange
     message_list = [
-        ("Hello, my name is Xi Li. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Xi Li. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
     ]
     populate_chat_history(message_list)
 
@@ -143,8 +151,8 @@ def test_no_answer_in_chat_history_or_retrieved_content(chat_client):
     "Chat director should say don't know as not enough contexts in chat history or retrieved to answer question"
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
     ]
     populate_chat_history(message_list)
 
@@ -197,9 +205,9 @@ def test_answer_requires_date_aware_aggregation_across_provided_notes(chat_clien
 def test_answer_general_question_not_in_chat_history_or_retrieved_content(chat_client):
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
-        ("Where was I born?", "You were born Testville.", ""),
+        ("Where was I born?", "You were born Testville.", []),
     ]
     populate_chat_history(message_list)
 
@@ -243,9 +251,9 @@ def test_ask_for_clarification_if_not_enough_context_in_question(chat_client):
 def test_answer_in_chat_history_beyond_lookback_window(chat_client):
     # Arrange
     message_list = [
-        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", ""),
+        ("Hello, my name is Testatron. Who are you?", "Hi, I am Khoj, a personal assistant. How can I help?", []),
-        ("When was I born?", "You were born on 1st April 1984.", ""),
+        ("When was I born?", "You were born on 1st April 1984.", []),
-        ("Where was I born?", "You were born Testville.", ""),
+        ("Where was I born?", "You were born Testville.", []),
     ]
     populate_chat_history(message_list)