Feature/submit yt issue (#3)

* WIP

* yt sumission is working

* bot provides yt issue link now

* fix prompt and function call escription

slack_bot/db.py

@@ -0,0 +1,33 @@
+from __future__ import annotations
+
+import contextlib
+from pathlib import Path
+from typing import Any
+
+
+# This class represents a simple database that stores its data as files in a directory.
+class DB:
+    def __init__(self, path: Path) -> None:
+        """Initialize a DB instance."""
+        self.path = path
+
+    def get(self, key: str) -> str | None:
+        """Get a value from the DB."""
+        try:
+            with (self.path / key).open(mode="rb") as f:
+                return f.read().decode("utf-8")
+        except FileNotFoundError:
+            return None
+
+    def __getitem__(self, key: str) -> str | None:
+        return self.get(key)
+
+    def write(self, key: str, value: Any) -> None:
+        """Set a value in the DB."""
+        with (self.path / key).open(mode="wb") as f:
+            f.write(value)
+
+    def delete(self, key: str) -> None:
+        """Delete a value from the DB."""
+        with contextlib.suppress(FileNotFoundError):
+            Path.unlink(self.path / key)

slack_bot/functions/create_issue

@@ -0,0 +1,18 @@
+{
+    "name": "create_issue",
+    "description": "Create a YouTrack issue",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "summary": {
+                "type": "string",
+                "description": "Issue summary (come up with an issue title here)"
+            },
+            "description": {
+                "type": "string",
+                "description": "Issue description (pass filled ${template} here)"
+            }
+        },
+        "required": ["summary", "description"]
+    }
+}

slack_bot/functions/run_query

@@ -0,0 +1,13 @@
+{
+    "name": "run_query",
+    "description": "Runs SQL query against a db and returns result",
+    "parameters": {
+        "type": "object",
+        "properties": {
+            "sql_query": {
+                "type": "string",
+                "description": "SQL query to run"
+            }
+        }
+    }
+}

slack_bot/prompts/clarification

@@ -0,0 +1,24 @@
+Act as a ClarifAI🧙🏾‍♂️, an expert at technical specifications clarification.
+You are going to assist user with ${technical specification}, which will be assigned to Analytics team.
+The ${user} is not an analyst, so you are helping ${user} with adjustment of ${technical specification} to map their ${goal}.
+
+Your initialization: ${emoji}: My name is ${name}.
+My task ends when technical specification will be adjusted.
+${first step, question}."
+
+Follow these steps:
+1. ${Initialization}; Ask ${user} to describe the ${goal} in a few words.
+2. Learn ${Product knowledge}.
+3. Fill ${template}, based on the user's answers:
+{template}
+4. 🧙🏾‍♂️ make sure that ${user} is agreed with the final technical specification.
+5. When the ${final technical specification} is confirmed, end the task with function call: `create_issue`, fill `description` with the final technical specification.
+
+Rules:
+-End every output with a multiple choice question with several options (not more than 4) to chose from
+-Don't ask a colleague about data structure, data sources or any information about the tables and fields available in database
+-Don't work with SQL, work only with ${technical specification} as an YouTrack issue
+
+Product knowledge:
+At our product users can make submissions. Submission is a user answer to a step.
+Step is a educational unit that contains a question or some task to do.

slack_bot/prompts/developing

@@ -0,0 +1,12 @@
+Act as a QueryMaster🧙🏾‍♂️, an expert SQL developer, who is using "ClickHouse" dialect.
+Your job is to map technical specification to SQL query.
+Develop a SQL query.
+
+Rules:
+-There are several available tables for you to work with:
+  `hyperskill.content` – events mart, contains information about user actions
+  `hyperskill_private.users` – users mart, contains user properties
+-Use CTEs (common table expressions) to break down complex queries into smaller steps
+-Don't run queries against `hyperskill.content` without `date` column filter or it will take consume all server resources
+-Don't use `CASE`, use if() or multiIf() instead
+-Don't use `COUNT(DISTINCT {column})` use `uniqExact({column})` instead

slack_bot/prompts/reviewing

@@ -0,0 +1,14 @@
+Act as a QueryReviewer🧙🏾‍♂️, an expert SQL developer, who is reviewing quires in "ClickHouse" dialect.
+Your job is to review a SQL query:
+{sql_query}
+
+Follow the next steps:
+1. Check if {query} makes sense in context of the technical specification:
+    {tech_spec}
+2. Check if {query} is valid according to docs:
+    {docs}
+3. Check some other aspects of the {query}
+4. Provide a step by step review of the {query} as a list of comments
+
+Useful reminders:
+-Don't run queries against `hyperskill.content` without `date` column filter or it will take consume all server resources

slack_bot/prompts/testing

@@ -0,0 +1,5 @@
+Act as a QueryTester🧙🏾‍♂️, an expert SQL developer, who is testing queries in ClickHouse dialect.
+Your job is to test this SQL query:
+{sql_query}
+
+Run {query} against a db, if errors occur, you will be informed, so you can fix them.

slack_bot/run-tasks.sh

@@ -1,7 +1,5 @@
 #!/bin/bash
 
-cd slack_bot || exit
-
 echo "Formatting code..."
 black .
 

slack_bot/templates/yt_issue

@@ -0,0 +1,10 @@
+### Motivation (Why?)
+[comment]: <> (Please, describe the motivation for this task. Why do you need it? What is the problem you are trying to solve?)
+
+### TODO:
+[comment]: <> (What do you expect to see as a result of this task? What should the results of this task contain?)
+
+### Priority & Urgency
+[comment]: <> (Please, fill the following checklist:)
+- is important [comment]: <> (Ask user about issue priority: Critical, Major, Normal, Minor.)
+- is urgent [comment]: <> (Ask user about issue urgency: yes, no, may be there is a due date.)

slack_bot/utils.py

@@ -1,14 +1,20 @@
 from __future__ import annotations
 
+import json
 import os
 import re
+import traceback
+from pathlib import Path
 from typing import Any, TYPE_CHECKING
 
 import tiktoken
 from dotenv import load_dotenv
 from trafilatura import extract, fetch_url
 from trafilatura.settings import use_config
 
+from slack_bot.db import DB
+from youtrack import YouTrack
+
 load_dotenv()
 
 
@@ -22,14 +28,17 @@
 SLACK_BOT_TOKEN = os.environ.get("SLACK_BOT_TOKEN")
 SLACK_APP_TOKEN = os.environ.get("SLACK_APP_TOKEN")
 OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
-
-SYSTEM_PROMPT = """
-You are an AI assistant.
-You will answer the question as truthfully as possible.
-If you're unsure of the answer, say Sorry, I don't know.
-"""
+YT_BASE_URL = os.environ.get("YT_BASE_URL")
+YT_API_TOKEN = os.environ.get("YT_API_TOKEN")
+
+prompts = DB(Path(__file__).parent / "prompts")
+templates = DB(Path(__file__).parent / "templates")
+functions = DB(Path(__file__).parent / "functions")
+AN_COMMAND = "an"
+YT_COMMAND = "yt"
 WAIT_MESSAGE = "Got your request. Please wait."
 MAX_TOKENS = 8192
+MODEL = "gpt-4"
 
 
 def extract_url_list(text: str) -> list[str] | None:
@@ -76,7 +85,7 @@ def num_tokens_from_messages(
 
     elif model == "gpt-4":  # noqa: RET505
         print(  # noqa: T201
-            "Warning: gpt-4 may change over time."
+            "Warning: gpt-4 may change over time. "
             "Returning num tokens assuming gpt-4-0314."
         )
 
@@ -136,7 +145,18 @@ def process_message(message: dict[str, str], bot_user_id: str, role: str) -> str
 def process_conversation_history(
     conversation_history: SlackResponse, bot_user_id: str
 ) -> list[dict[str, str]]:
-    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
+    messages = []
+
+    cleaned_message = (
+        conversation_history["messages"][0]["text"]
+        .replace(f"<@{bot_user_id}>", "")
+        .strip()
+    )
+
+    if cleaned_message == AN_COMMAND:
+        conversation_history["messages"].pop(0)
+        system = prompts["clarification"].replace("{template}", templates["yt_issue"])
+        messages.append({"role": "system", "content": system})
 
     for message in conversation_history["messages"][:-1]:
         role = "assistant" if message["user"] == bot_user_id else "user"
@@ -156,6 +176,38 @@ def get_conversation_history(
     )
 
 
+def submit_issue(messages: list[dict[str, str]], openai: Any) -> str:
+    funcs = [json.loads(functions["create_issue"])]
+    openai_response = openai.ChatCompletion.create(
+        model=MODEL,
+        messages=messages,
+        functions=funcs,
+        function_call={"name": "create_issue"},
+    )
+
+    arguments = json.loads(
+        openai_response.choices[0]
+        .message.get("function_call", {})
+        .get("arguments", {}),
+        strict=False,
+    )
+
+    yt = YouTrack(
+        base_url=YT_BASE_URL,
+        token=YT_API_TOKEN,
+    )
+
+    response_text = yt.create_issue(
+        summary=arguments["summary"],
+        description=arguments["description"],
+    )
+
+    if isinstance(response_text, Exception):
+        raise response_text
+
+    return f"{YT_BASE_URL}/issue/{response_text['id']}"
+
+
 def make_ai_response(
     app: App, body: dict[str, dict[str, str]], context: dict[str, str], openai: Any
 ) -> None:
@@ -171,13 +223,21 @@ def make_ai_response(
 
         conversation_history = get_conversation_history(app, channel_id, thread_ts)
         messages = process_conversation_history(conversation_history, bot_user_id)
+
         num_tokens = num_tokens_from_messages(messages)
         print(f"Number of tokens: {num_tokens}")  # noqa: T201
 
-        openai_response = openai.ChatCompletion.create(
-            model="gpt-3.5-turbo", messages=messages
-        )
-        response_text = openai_response.choices[0].message["content"]
+        last_msg = messages[-1]
+
+        if (last_msg["role"] == "user") & (last_msg["content"] == YT_COMMAND):
+            messages.pop()
+            response_text = submit_issue(messages=messages, openai=openai)
+        else:
+            openai_response = openai.ChatCompletion.create(
+                model=MODEL, messages=messages
+            )
+            response_text = openai_response.choices[0].message["content"]
+
         app.client.chat_update(
             channel=channel_id, ts=reply_message_ts, text=response_text
         )
@@ -188,3 +248,4 @@ def make_ai_response(
             thread_ts=thread_ts,
             text=f"I can't provide a response. Encountered an error:\n`\n{e}\n`",
         )
+        traceback.print_exc()

slack_bot/youtrack.py

@@ -0,0 +1,48 @@
+from __future__ import annotations
+
+from typing import Any
+
+import requests
+
+
+class IssueCreationError(Exception):
+    """Exception raised for errors in the issue creation process."""
+
+    def __init__(self, status_code: int, error_message: str) -> None:
+        self.status_code = status_code
+        self.error_message = error_message
+        super().__init__(
+            f"Failed to create issue. Status code: {status_code}.\n"
+            f"Error: {error_message}"
+        )
+
+
+class YouTrack:
+    def __init__(self, base_url: str | None, token: str | None) -> None:
+        if base_url or token:
+            self.base_url = base_url
+            self.token = token
+        else:
+            raise ValueError("YouTrack base URL and API token are required.")
+
+    def create_issue(
+        self, summary: str, description: str, project: str = "43-46"
+    ) -> dict[str, Any] | IssueCreationError:
+        """Create an issue in YouTrack."""
+        url = f"{self.base_url}/api/issues"
+        headers = {
+            "Content-Type": "application/json",
+            "Authorization": f"Bearer {self.token}",
+        }
+        payload = {
+            "project": {"id": project},
+            "summary": summary + " [created by AI Data Assistant]",
+            "description": description,
+        }
+
+        response = requests.post(url, headers=headers, json=payload, timeout=30)
+
+        if response.status_code != 200:  # noqa: PLR2004
+            return IssueCreationError(response.status_code, response.text)
+
+        return response.json()