Upload 3 files

app.py

@@ -0,0 +1,70 @@
+import streamlit as st
+import json
+
+from chat_ai import generate_response_for_pre_indexed_repo, generate_response_for_custom_repo
+
+st.set_page_config(page_title="Tune AI Git Issue Chat")
+
+if 'clicked' not in st.session_state:
+    st.session_state.clicked = False
+
+
+def click_button():
+    st.session_state.clicked = True
+
+
+with st.sidebar:
+    option = st.selectbox(
+        'What repo you are looking for?',
+        ('Pre-Indexed', 'Custom'),
+        index=None,
+        placeholder="please choose an option"
+    )
+    if option == 'Pre-Indexed':
+        option_of_repo = st.selectbox(
+            'Choose any one of the pre-index repo',
+            ('Tensorflow', 'Pytorch'),
+            index=None,
+            placeholder="please choose an option"
+        )
+        st.write("Select number of top issues you are looking for!")
+        number_of_issues = st.number_input('Insert a number')
+    elif option == 'Custom':
+        repo_link = st.text_area("Please enter your public repo link!")
+        st.write("Select number of top issues you are looking for!")
+        number_of_issues = st.number_input('Insert a number')
+
+    st.button("Ask Tune AI!", on_click=click_button)
+
+if st.session_state.clicked:
+    with st.spinner("Generating, It may take some minutes🫡..."):
+        if option == 'Pre-Indexed' and number_of_issues:
+            if option_of_repo == "Tensorflow":
+                repo_choice = "Tensorflow"
+            elif option_of_repo == "Pytorch":
+                repo_choice = "Pytorch"
+            gpt_response = generate_response_for_pre_indexed_repo(repo_choice, number_of_issues)
+            if gpt_response["success"]:
+                try:
+                    json_data = json.loads(gpt_response["data"])
+                    for issue in json_data['issues']:
+                        st.markdown(f"**{issue['issue_title']}**", unsafe_allow_html=True)
+                        st.write("Rating:", issue['rating']['type'])
+                        st.write("Description:", issue['rating']['description'])
+                except:
+                    st.json(gpt_response["data"])
+            else:
+                st.write("Sorry we encountered some issues!")
+        elif option == 'Custom' and number_of_issues and repo_link:
+            gpt_response = generate_response_for_custom_repo(number_of_issues, repo_link)
+            if gpt_response["success"]:
+                try:
+                    json_data = json.loads(gpt_response["data"])
+                    for issue in json_data['issues']:
+                        st.markdown(f"**{issue['issue_title']}**", unsafe_allow_html=True)
+                        st.write("Rating:", issue['rating']['type'])
+                        st.write("Description:", issue['rating']['description'])
+                except:
+                    st.json(gpt_response["data"])
+            else:
+                st.write("Sorry we encountered some issues!")

chat_ai.py

@@ -0,0 +1,102 @@
+import os
+import re
+import pandas as pd
+from constants import JSON_SCHEMA_FOR_GPT, REPO_NAME_EXTRACTION_PATTERN
+from utils import create_open_ai_query, get_issues_csv, convert_repo_url_to_git_api_url
+
+
+def generate_response_for_pre_indexed_repo(repo_choice, number_of_issues):
+    if repo_choice == "Tensorflow":
+        csv_key = "tf.csv"
+        repo_name = "Tensorflow"
+    else:
+        csv_key = "torch.csv"
+        repo_name = "Pytorch"
+    issues_df = pd.read_csv(csv_key)
+    context_of_issues = ""
+    for i, row in issues_df.iterrows():
+        issue_title = row["Issue Title"]
+        issue_description = row["Description"]
+        issue_creation_date = row["Created At"]
+        issue_comments = row["Comments"]
+
+        formulated_issue = f"""
+          Issue_title : {issue_title},
+          Issue_description : {issue_description},
+          Issue_creation_date: {issue_creation_date},
+          Issue_comments: {issue_comments}
+
+          """
+        context_of_issues += formulated_issue
+    schema_context = f"""Output JSON format : {JSON_SCHEMA_FOR_GPT}"""
+    additional_prompt = f"""You have to provide top {number_of_issues}"""
+    prompt = f"""Act as a Software Developer, you are provided with Github Issues details: {context_of_issues} for 
+    github repo of {repo_name}. User has asked you to list top {number_of_issues} issues for this repository. 
+    Let's break down your task of listing top issues step by step:
+    1. First take time to think and understand the github repo.
+    2. Take time to think and understand the Github Issues details provided. Understand the title, description, 
+    number of comments.
+    3. Try to understand what impact each issue will have on the repository if it is resolved.
+    4. Understand why the issues which are highly commented with aspect of its impact on github repo
+    5. calculate rating for issues and select top {number_of_issues} issues.
+    5. Finally provide a JSON response which will have selected top {number_of_issues} issues. 
+    Follow the mentioned format for the JSON.
+    """
+    final_prompt = (schema_context + additional_prompt + prompt)
+    response = create_open_ai_query(final_prompt)
+    if response["success"]:
+        return {"success": True, "data": response["data"]}
+    else:
+        return {"success": False, "error": response}
+
+
+def generate_response_for_custom_repo(number_of_issues, repo_url):
+    converted_url = convert_repo_url_to_git_api_url(repo_url)
+    print(converted_url)
+    match = re.match(REPO_NAME_EXTRACTION_PATTERN, repo_url)
+    print(match)
+    if match:
+        repo_name = match.group(2)
+        csv_file_name = f"{repo_name}.csv"
+    else:
+        repo_name = repo_url
+        csv_file_name = "test.csv"
+    issues_csv = get_issues_csv(converted_url, csv_file_name)
+    if issues_csv["success"]:
+        issues_df = pd.read_csv(csv_file_name)
+        context_of_issues = ""
+        for i, row in issues_df.iterrows():
+            issue_title = row["Issue Title"]
+            issue_description = row["Description"]
+            issue_creation_date = row["Created At"]
+            issue_comments = row["Comments"]
+            formulated_issue = f"""
+              Issue_title : {issue_title},
+              Issue_description : {issue_description},
+              Issue_creation_date: {issue_creation_date},
+              Issue_comments: {issue_comments}
+              """
+            context_of_issues += formulated_issue
+        schema_context = f"""Output JSON format : {JSON_SCHEMA_FOR_GPT}"""
+        additional_prompt = f"""You have to provide top {number_of_issues}"""
+        prompt = f"""Act as a Software Developer, you are provided with Github Issues details: {context_of_issues} for 
+            github repo of {repo_name}. User has asked you to list top {number_of_issues} issues for this repository. 
+            Let's break down your task of listing top issues step by step:
+            1. First take time to think and understand the github repo.
+            2. Take time to think and understand the Github Issues details provided. Understand the title, description, 
+            number of comments.
+            3. Try to understand what impact each issue will have on the repository if it is resolved.
+            4. Understand why the issues which are highly commented with aspect of its impact on github repo
+            5. calculate rating for issues and select top {number_of_issues} issues.
+            5. Finally provide a JSON response which will have selected top {number_of_issues} issues. 
+            Follow the mentioned format for the JSON.
+            """
+        final_prompt = (schema_context + additional_prompt + prompt)
+        response = create_open_ai_query(final_prompt)
+        os.remove(csv_file_name)
+        if response["success"]:
+            return {"success": True, "data": response["data"]}
+        else:
+            return {"success": False, "error": response}
+    else:
+        return {"success": False}

constants.py

@@ -0,0 +1,55 @@
+import os
+
+# OPENAI
+
+OPENAI_API_BASE_URL = "https://api.openai.com/v1"
+OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
+TEXT_MODEL_ENGINE = 'gpt-4-0125-preview'
+
+# GITHUB
+GITHUB_AUTH_KEY = os.getenv("GITHUB_AUTH_KEY")
+
+# REGEX
+REPO_NAME_EXTRACTION_PATTERN = r"https://github.com/([^/]+)/([^/]+)$"
+
+# SCHEMA
+JSON_SCHEMA_FOR_GPT = {
+  "type": "object",
+  "properties": {
+    "issues": {
+      "type": "array",
+      "items": [
+        {
+          "type": "object",
+          "properties": {
+            "issue_title": {
+              "type": "string"
+            },
+            "rating": {
+              "type": "object",
+              "properties": {
+                "type": {
+                  "type": "string"
+                },
+                "description": {
+                  "type": "string"
+                }
+              },
+              "required": [
+                "type",
+                "description"
+              ]
+            }
+          },
+          "required": [
+            "issue_title",
+            "rating"
+          ]
+        }
+      ]
+    }
+  },
+  "required": [
+    "issues"
+  ]
+}