refactor/move-prompts-to-jinja-templates

app/backend/approaches/chatapproach.py

@@ -2,39 +2,27 @@
 import re
 from abc import ABC, abstractmethod
 from typing import Any, AsyncGenerator, Optional
+from jinja2 import Environment, FileSystemLoader
 
 from openai.types.chat import ChatCompletion, ChatCompletionMessageParam
 
 from approaches.approach import Approach
 
-
 class ChatApproach(Approach, ABC):
-    query_prompt_few_shots: list[ChatCompletionMessageParam] = [
-        {"role": "user", "content": "How did crypto do last year?"},
-        {"role": "assistant", "content": "Summarize Cryptocurrency Market Dynamics from last year"},
-        {"role": "user", "content": "What are my health plans?"},
-        {"role": "assistant", "content": "Show available health plans"},
-    ]
+
     NO_RESPONSE = "0"
 
-    follow_up_questions_prompt_content = """Generate 3 very brief follow-up questions that the user would likely ask next.
-    Enclose the follow-up questions in double angle brackets. Example:
-    <<Are there exclusions for prescriptions?>>
-    <<Which pharmacies can be ordered from?>>
-    <<What is the limit for over-the-counter medication?>>
-    Do no repeat questions that have already been asked.
-    Make sure the last question ends with ">>".
-    """
-
-    query_prompt_template = """Below is a history of the conversation so far, and a new question asked by the user that needs to be answered by searching in a knowledge base.
-    You have access to Azure AI Search index with 100's of documents.
-    Generate a search query based on the conversation and the new question.
-    Do not include cited source filenames and document names e.g info.txt or doc.pdf in the search query terms.
-    Do not include any text inside [] or <<>> in the search query terms.
-    Do not include any special characters like '+'.
-    If the question is not in English, translate the question to English before generating the search query.
-    If you cannot generate a search query, return just the number 0.
-    """
+    def __init__(self):
+        self._initialize_templates()
+
+    def _initialize_templates(self):
+        self.env = Environment(loader=FileSystemLoader('approaches/prompts/chat'))
+        json_content = self.env.loader.get_source(self.env, 'query_few_shots.json')[0]
+        self.query_prompt_few_shots: list[ChatCompletionMessageParam] = json.loads(json_content)
+        self.query_prompt_template = self.env.get_template('query_template.jinja').render()
+        self.follow_up_questions_prompt = self.env.get_template('follow_up_questions.jinja').render()
+        self.system_message_chat_conversation_template = self.env.get_template('system_message.jinja')
+        self.system_message_chat_conversation_vision_template = self.env.get_template('system_message_vision.jinja')
 
     @property
     @abstractmethod
@@ -47,12 +35,14 @@ async def run_until_final_call(self, messages, overrides, auth_claims, should_st
 
     def get_system_prompt(self, override_prompt: Optional[str], follow_up_questions_prompt: str) -> str:
         if override_prompt is None:
-            return self.system_message_chat_conversation.format(
-                injected_prompt="", follow_up_questions_prompt=follow_up_questions_prompt
+            return self.system_message_chat_conversation_template.render(
+                follow_up_questions_prompt=follow_up_questions_prompt,
+                injected_prompt=""
             )
         elif override_prompt.startswith(">>>"):
-            return self.system_message_chat_conversation.format(
-                injected_prompt=override_prompt[3:] + "\n", follow_up_questions_prompt=follow_up_questions_prompt
+            return self.system_message_chat_conversation_template.render(
+                follow_up_questions_prompt=follow_up_questions_prompt,
+                injected_prompt=override_prompt[3:] + "\n"
             )
         else:
             return override_prompt.format(follow_up_questions_prompt=follow_up_questions_prompt)

app/backend/approaches/chatreadretrieveread.py

@@ -39,6 +39,7 @@ def __init__(
         query_language: str,
         query_speller: str,
     ):
+        super().__init__()
         self.search_client = search_client
         self.openai_client = openai_client
         self.auth_helper = auth_helper
@@ -55,13 +56,10 @@ def __init__(
 
     @property
     def system_message_chat_conversation(self):
-        return """Assistant helps the company employees with their healthcare plan questions, and questions about the employee handbook. Be brief in your answers.
-        Answer ONLY with the facts listed in the list of sources below. If there isn't enough information below, say you don't know. Do not generate answers that don't use the sources below. If asking a clarifying question to the user would help, ask the question.
-        If the question is not in English, answer in the language used in the question.
-        Each source has a name followed by colon and the actual information, always include the source name for each fact you use in the response. Use square brackets to reference the source, for example [info1.txt]. Don't combine sources, list each source separately, for example [info1.txt][info2.pdf].
-        {follow_up_questions_prompt}
-        {injected_prompt}
-        """
+        return self.system_message_chat_conversation_template.render(
+            follow_up_questions_prompt="",
+            injected_prompt=""
+        )
 
     @overload
     async def run_until_final_call(
@@ -177,7 +175,7 @@ async def run_until_final_call(
         # Allow client to replace the entire prompt, or to inject into the exiting prompt using >>>
         system_message = self.get_system_prompt(
             overrides.get("prompt_template"),
-            self.follow_up_questions_prompt_content if overrides.get("suggest_followup_questions") else "",
+            self.follow_up_questions_prompt_content if overrides.get("suggest_followup_questions") else ""
         )
 
         response_token_limit = 1024

app/backend/approaches/chatreadretrievereadvision.py

@@ -46,6 +46,7 @@ def __init__(
         vision_endpoint: str,
         vision_token_provider: Callable[[], Awaitable[str]]
     ):
+        super().__init__()
         self.search_client = search_client
         self.blob_container_client = blob_container_client
         self.openai_client = openai_client
@@ -67,19 +68,10 @@ def __init__(
 
     @property
     def system_message_chat_conversation(self):
-        return """
-        You are an intelligent assistant helping analyze the Annual Financial Report of Contoso Ltd., The documents contain text, graphs, tables and images.
-        Each image source has the file name in the top left corner of the image with coordinates (10,10) pixels and is in the format SourceFileName:<file_name>
-        Each text source starts in a new line and has the file name followed by colon and the actual information
-        Always include the source name from the image or text for each fact you use in the response in the format: [filename]
-        Answer the following question using only the data provided in the sources below.
-        If asking a clarifying question to the user would help, ask the question.
-        Be brief in your answers.
-        The text and image source can be the same file name, don't use the image title when citing the image source, only use the file name as mentioned
-        If you cannot answer using the sources below, say you don't know. Return just the answer without any input texts.
-        {follow_up_questions_prompt}
-        {injected_prompt}
-        """
+        return self.system_message_chat_conversation_vision_template.render(
+            follow_up_questions_prompt="",
+            injected_prompt=""
+        )
 
     async def run_until_final_call(
         self,

app/backend/approaches/prompts/ask/few_shots.json

@@ -0,0 +1,10 @@
+{
+    "question": "What is the deductible for the employee plan for a visit to Overlake in Bellevue?",
+    "sources": {
+      "info1.txt": "deductibles depend on whether you are in-network or out-of-network. In-network deductibles are $500 for employee and $1000 for family. Out-of-network deductibles are $1000 for employee and $2000 for family.",
+      "info2.pdf": "Overlake is in-network for the employee plan.",
+      "info3.pdf": "Overlake is the name of the area that includes a park and ride near Bellevue.",
+      "info4.pdf": "In-network institutions include Overlake, Swedish and others in the region."
+    },
+    "answer": "In-network deductibles are $500 for employee and $1000 for family [info1.txt] and Overlake is in-network for the employee plan [info2.pdf][info4.pdf]."
+}

app/backend/approaches/prompts/ask/system_message.jinja

@@ -0,0 +1,5 @@
+You are an intelligent assistant helping Contoso Inc employees with their healthcare plan questions and employee handbook questions.
+Use 'you' to refer to the individual asking the questions even if they ask with 'I'. 
+Answer the following question using only the data provided in the sources below.
+Each source has a name followed by colon and the actual information, always include the source name for each fact you use in the response.
+If you cannot answer using the sources below, say you don't know. Use below example to answer

app/backend/approaches/prompts/ask/system_message_vision.jinja

@@ -0,0 +1,7 @@
+You are an intelligent assistant helping analyze the Annual Financial Report of Contoso Ltd., The documents contain text, graphs, tables and images.
+Each image source has the file name in the top left corner of the image with coordinates (10,10) pixels and is in the format SourceFileName:<file_name>.
+Each text source starts in a new line and has the file name followed by colon and the actual information.
+Always include the source name from the image or text for each fact you use in the response in the format: [filename].
+Answer the following question using only the data provided in the sources below.
+The text and image source can be the same file name, don't use the image title when citing the image source, only use the file name as mentioned.
+If you cannot answer using the sources below, say you don't know. Return just the answer without any input texts.

app/backend/approaches/prompts/chat/follow_up_questions.jinja

@@ -0,0 +1,7 @@
+Generate 3 very brief follow-up questions that the user would likely ask next.
+Enclose the follow-up questions in double angle brackets. Example:
+<<Are there exclusions for prescriptions?>>
+<<Which pharmacies can be ordered from?>>
+<<What is the limit for over-the-counter medication?>>
+Do not repeat questions that have already been asked.
+Make sure the last question ends with ">>".

app/backend/approaches/prompts/chat/query_few_shots.json

@@ -0,0 +1,18 @@
+[
+    {
+        "role": "user",
+        "content": "How did crypto do last year?"
+    },
+    {
+        "role": "assistant",
+        "content": "Summarize Cryptocurrency Market Dynamics from last year"
+    },
+    {
+        "role": "user",
+        "content": "What are my health plans?"
+    },
+    {
+        "role": "assistant",
+        "content": "Show available health plans"
+    }
+]

app/backend/approaches/prompts/chat/query_template.jinja

@@ -0,0 +1,8 @@
+Below is a history of the conversation so far, and a new question asked by the user that needs to be answered by searching in a knowledge base.
+You have access to Azure AI Search index with 100's of documents.
+Generate a search query based on the conversation and the new question.
+Do not include cited source filenames and document names e.g. info.txt or doc.pdf in the search query terms.
+Do not include any text inside [] or <<>> in the search query terms.
+Do not include any special characters like '+'.
+If the question is not in English, translate the question to English before generating the search query.
+If you cannot generate a search query, return just the number 0.

app/backend/approaches/prompts/chat/system_message.jinja

@@ -0,0 +1,6 @@
+Assistant helps the company employees with their healthcare plan questions, and questions about the employee handbook. Be brief in your answers.
+Answer ONLY with the facts listed in the list of sources below. If there isn't enough information below, say you don't know. Do not generate answers that don't use the sources below. If asking a clarifying question to the user would help, ask the question.
+If the question is not in English, answer in the language used in the question.
+Each source has a name followed by colon and the actual information, always include the source name for each fact you use in the response. Use square brackets to reference the source, for example [info1.txt]. Don't combine sources, list each source separately, for example [info1.txt][info2.pdf].
+{{ follow_up_questions_prompt }}
+{{ injected_prompt }}

app/backend/approaches/prompts/chat/system_message_vision.jinja

@@ -0,0 +1,11 @@
+You are an intelligent assistant helping analyze the Annual Financial Report of Contoso Ltd., The documents contain text, graphs, tables and images.
+Each image source has the file name in the top left corner of the image with coordinates (10,10) pixels and is in the format SourceFileName:<file_name>
+Each text source starts in a new line and has the file name followed by colon and the actual information
+Always include the source name from the image or text for each fact you use in the response in the format: [filename]
+Answer the following question using only the data provided in the sources below.
+If asking a clarifying question to the user would help, ask the question.
+Be brief in your answers.
+The text and image source can be the same file name, don't use the image title when citing the image source, only use the file name as mentioned
+If you cannot answer using the sources below, say you don't know. Return just the answer without any input texts.
+{follow_up_questions_prompt}
+{injected_prompt}

app/backend/approaches/retrievethenread.py

@@ -1,10 +1,14 @@
 from typing import Any, Optional
+import json
+import re
+import ast
 
 from azure.search.documents.aio import SearchClient
 from azure.search.documents.models import VectorQuery
 from openai import AsyncOpenAI
 from openai.types.chat import ChatCompletionMessageParam
 from openai_messages_token_helper import build_messages, get_token_limit
+from jinja2 import Environment, FileSystemLoader
 
 from approaches.approach import Approach, ThoughtStep
 from core.authentication import AuthenticationHelper
@@ -17,26 +21,6 @@ class RetrieveThenReadApproach(Approach):
     (answer) with that prompt.
     """
 
-    system_chat_template = (
-        "You are an intelligent assistant helping Contoso Inc employees with their healthcare plan questions and employee handbook questions. "
-        + "Use 'you' to refer to the individual asking the questions even if they ask with 'I'. "
-        + "Answer the following question using only the data provided in the sources below. "
-        + "Each source has a name followed by colon and the actual information, always include the source name for each fact you use in the response. "
-        + "If you cannot answer using the sources below, say you don't know. Use below example to answer"
-    )
-
-    # shots/sample conversation
-    question = """
-'What is the deductible for the employee plan for a visit to Overlake in Bellevue?'
-
-Sources:
-info1.txt: deductibles depend on whether you are in-network or out-of-network. In-network deductibles are $500 for employee and $1000 for family. Out-of-network deductibles are $1000 for employee and $2000 for family.
-info2.pdf: Overlake is in-network for the employee plan.
-info3.pdf: Overlake is the name of the area that includes a park and ride near Bellevue.
-info4.pdf: In-network institutions include Overlake, Swedish and others in the region
-"""
-    answer = "In-network deductibles are $500 for employee and $1000 for family [info1.txt] and Overlake is in-network for the employee plan [info2.pdf][info4.pdf]."
-
     def __init__(
         self,
         *,
@@ -68,6 +52,14 @@ def __init__(
         self.query_speller = query_speller
         self.chatgpt_token_limit = get_token_limit(chatgpt_model, self.ALLOW_NON_GPT_MODELS)
 
+        self._initialize_templates()
+
+    def _initialize_templates(self):
+        self.env = Environment(loader=FileSystemLoader('approaches/prompts/ask'))
+        self.system_chat_template = self.env.get_template('system_message.jinja').render()
+        json_content = self.env.loader.get_source(self.env, 'few_shots.json')[0]
+        self.few_shots = json.loads(json_content)
+
     async def run(
         self,
         messages: list[ChatCompletionMessageParam],
@@ -114,11 +106,22 @@ async def run(
         content = "\n".join(sources_content)
         user_content = q + "\n" + f"Sources:\n {content}"
 
+        few_shots = [
+            {
+                "role": "user",
+                "content": f"{self.few_shots['question']}\nSources:\n" + "\n".join([f"{k}: {v}" for k, v in self.few_shots['sources'].items()])
+            },
+            {
+                "role": "assistant",
+                "content": self.few_shots["answer"]
+            }
+        ]
+
         response_token_limit = 1024
         updated_messages = build_messages(
             model=self.chatgpt_model,
             system_prompt=overrides.get("prompt_template", self.system_chat_template),
-            few_shots=[{"role": "user", "content": self.question}, {"role": "assistant", "content": self.answer}],
+            few_shots=[{"role": "user", "content": self.few_shots["question"]}, {"role": "assistant", "content": self.few_shots["answer"]}],
             new_user_content=user_content,
             max_tokens=self.chatgpt_token_limit - response_token_limit,
             fallback_to_default=self.ALLOW_NON_GPT_MODELS,

app/backend/approaches/retrievethenreadvision.py

@@ -9,6 +9,7 @@
     ChatCompletionMessageParam,
 )
 from openai_messages_token_helper import build_messages, get_token_limit
+from jinja2 import Environment, FileSystemLoader
 
 from approaches.approach import Approach, ThoughtStep
 from core.authentication import AuthenticationHelper
@@ -22,16 +23,6 @@ class RetrieveThenReadVisionApproach(Approach):
     (answer) with that prompt.
     """
 
-    system_chat_template_gpt4v = (
-        "You are an intelligent assistant helping analyze the Annual Financial Report of Contoso Ltd., The documents contain text, graphs, tables and images. "
-        + "Each image source has the file name in the top left corner of the image with coordinates (10,10) pixels and is in the format SourceFileName:<file_name> "
-        + "Each text source starts in a new line and has the file name followed by colon and the actual information "
-        + "Always include the source name from the image or text for each fact you use in the response in the format: [filename] "
-        + "Answer the following question using only the data provided in the sources below. "
-        + "The text and image source can be the same file name, don't use the image title when citing the image source, only use the file name as mentioned "
-        + "If you cannot answer using the sources below, say you don't know. Return just the answer without any input texts "
-    )
-
     def __init__(
         self,
         *,
@@ -68,6 +59,10 @@ def __init__(
         self.vision_token_provider = vision_token_provider
         self.gpt4v_token_limit = get_token_limit(gpt4v_model, self.ALLOW_NON_GPT_MODELS)
 
+        self.env = Environment(loader=FileSystemLoader('approaches/prompts/ask'))
+        self.system_chat_template_gpt4v = self.env.get_template('system_message_vision.jinja').render()
+
+
     async def run(
         self,
         messages: list[ChatCompletionMessageParam],