Tidy up

app.py

@@ -1,9 +1,18 @@
+import datetime
 import gradio as gr
+import os
 from find_similar_issues import get_similar_issues
 import requests
 
 from defaults import OWNER, REPO
 
+from build_saved_issues import get_issues
+from update_stored_issues import update_issues
+import build_issue_dict
+import build_embeddings
+import shutil
+
+
 
 def get_query_issue_information(issue_no, token):
     headers = {
@@ -35,24 +44,47 @@ def run_find_similar_issues(token, n_issues, issue_no, query):
     return issues_html
 
 
+def update_issues():
+    # Archive the stored issues
+    if os.path.exists("issues.json"):
+        date_time = datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
+        shutil.copy("issues.json", f"{date_time}_issues.json")
+
+    # Retrieve new issues
+    get_issues(overwrite=False, update=True, output_filename="issues.json")
+    # Update any issues that have been updated since the last update
+    update_issues()
+    # Update the dictionary of issues
+    build_issue_dict.build_json_file("issues.json", "issues_dict.json")
+    # Update the embeddings
+    build_embeddings.embed_issues(
+        input_filename="issues_dict.json",
+        issue_type="issue",
+        model_id="all-mpnet-base-v2",
+        update=True
+    )
+
+
 with gr.Blocks(title="Github Bot") as demo:
     with gr.Tab("Find similar issues"):
         with gr.Row():
+            # with gr.Column():
             with gr.Column():
-                with gr.Row():
-                    issue_no = gr.Textbox(label="Github Issue", placeholder="Github issue you want to find similar issues to")
-                    query = gr.Textbox(label="Query", placeholder="Search for issues")
-                with gr.Row():
-                    token = gr.Textbox(label="Github Token", placeholder="Your github token for authentication. This is not stored anywhere.")
-                    n_issues = gr.Slider(1, 50, value=5, step=1, label="Number of similar issues", info="Choose between 1 and 50")
+                gr.Markdown("Find similar issues to a given issue or query")
+                issue_no = gr.Textbox(label="Github Issue", placeholder="Github issue you want to find similar issues to")
+                query = gr.Textbox(label="Query", placeholder="Search for issues")
+            with gr.Column():
+                token = gr.Textbox(label="Github Token", placeholder="Your github token for authentication. This is not stored anywhere.")
+                n_issues = gr.Slider(1, 50, value=5, step=1, label="Number of similar issues", info="Choose between 1 and 50")
+                update_button = gr.Button(value="Update issues")
+                update_button.click(update_issues)
 
         with gr.Row():
-            submit_button = gr.Button(value="Submit")
+                submit_button = gr.Button(value="Submit")
 
         with gr.Row():
             with gr.Row():
                 issues_html = gr.HTML(label="Issue text", elem_id="issue_html")
-            with gr.Row():
                 submit_button.click(run_find_similar_issues, outputs=[issues_html], inputs=[token, n_issues, issue_no, query])
 
     with gr.Tab("Find maintainers to ping"):

get_topic.py

@@ -44,7 +44,7 @@ model = LlamaForCausalLM.from_pretrained("meta-llama/Llama-2-7b-hf")
 tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-hf")
 
 # prompt = f"Which of the following topics {list(topic_maintainers_map.keys())} is this issue about:\n{issue['body']}"
-prompt = f"What is the provided issue about? Pick up to 3 topics from the following list: {list(topic_maintainers_map.keys())} \nIssue:\n{issue['body']}"
+prompt = f"QUESTION: What is the provided issue about? Pick up to 3 topics from the following list: {list(topic_maintainers_map.keys())} \nISSUE START:\n{issue['body']} \n ISSUE END. \n ANSWER:"
 inputs = tokenizer(prompt, return_tensors="pt")
 
 prefix_len = inputs.input_ids.shape[1]

update_stored_issues.py

@@ -35,7 +35,7 @@ TOKEN = os.environ.get("GITHUB_TOKEN")
 JSON_FILE = f"issues.json"
 
 
-def get_issues(
+def update_issues(
     input_filename=JSON_FILE,
     output_filename=JSON_FILE,
     github_api_version=GITHUB_API_VERSION,
@@ -151,4 +151,4 @@ if __name__ == "__main__":
     parser.add_argument("--token", type=str, default=TOKEN)
     parser.add_argument("--n_pages", type=int, default=-1)
     args = parser.parse_args()
-    get_issues(**vars(args))
+    update_issues(**vars(args))