Upload folder using huggingface_hub

app/agent.py

@@ -1,3 +1,4 @@
+from utils import preprocess_logs
 from langchain_openai import ChatOpenAI
 from config import examples, example_template, template_v2
 from langchain_core.prompts import PromptTemplate, FewShotPromptTemplate
@@ -25,13 +26,13 @@ class Agent:
         self.logs = None
         self.response = None
 
-    def process_request(self, question, logs):
+    def summarize(self, question, logs):
         self.question = question
-        self.logs = logs
+        self.logs = preprocess_logs(logs)
 
         prompt_formatted = self.prompt.format(
             question=question,
-            logs=logs
+            logs=self.logs
         )
 
         self.response = self.llm.predict(prompt_formatted)

app/app.py

@@ -1,8 +1,7 @@
-import time
 from agent import Agent
 from config import MODEL
 from flask import Flask, jsonify, request
-from utils import read_documents, preprocess_logs
+from utils import read_documents, validate_request_logs
 from models import GetQuestionAndFactsResponse, SubmitQuestionAndDocumentsResponse, SubmitQuestionAndDocumentRequest
 
 app = Flask(__name__)
@@ -15,16 +14,16 @@ submitted_data = None
 def get_response():
     global submitted_data, processing, agent
 
-    print(submitted_data)
-
+    # If no data found
     if not submitted_data:
         response = GetQuestionAndFactsResponse(
             question='',
             facts=[],
-            status='No data found.'
+            status='No data found, please submit data'
         )
         return jsonify(response.dict()), 200
 
+    # If still processing request
     if processing:
         response = GetQuestionAndFactsResponse(
             question=submitted_data.question,
@@ -33,6 +32,7 @@ def get_response():
         )
         return jsonify(response.dict()), 200
 
+    # Request processed, create response with Agent summarization
     response = GetQuestionAndFactsResponse(
         question=submitted_data.question,
         facts=agent.get_response_list(),
@@ -47,18 +47,43 @@ def submit_question():
     global submitted_data, processing, agent
     processing = True
     request_content = request.get_json()
-    submitted_data = SubmitQuestionAndDocumentRequest(**request_content)
 
-    logs = read_documents(submitted_data.urls)
-    processed_logs = preprocess_logs(logs)
+    # Submit payload read and validation
+    try:
+        submitted_data = SubmitQuestionAndDocumentRequest(**request_content)
+    except ValueError as e:
+        response = SubmitQuestionAndDocumentsResponse(status=f'Request payload does not match expected schema: {str(e)}')
+        return jsonify(response.dict()), 200
+
+    # Validate request URLS formats
+    try:
+        validate_request_logs(submitted_data.documents)
+    except ValueError as e:
+        # Respond with URL validation failed error
+        response = SubmitQuestionAndDocumentsResponse(status=f'URL validation failed: {e}')
+        return jsonify(response.dict()), 200
+
+    # Try loading documents
+    try:
+        logs = read_documents(submitted_data.documents)
+    except Exception as e:
+        # Respond with URL read fail if URL read error
+        response = SubmitQuestionAndDocumentsResponse(status=f'URL read failed: {e}')
+        return jsonify(response.dict()), 200
+
+    # If no data found
+    if len(logs) == 0:
+        response = SubmitQuestionAndDocumentsResponse(status=f'No data found in the URLs')
+        return jsonify(response.dict()), 200
 
-    agent.process_request(
+    # Call agent to summarize logs
+    agent.summarize(
         question=submitted_data.question,
-        logs=processed_logs
+        logs=logs
     )
 
     processing = False
-    response = SubmitQuestionAndDocumentsResponse()
+    response = SubmitQuestionAndDocumentsResponse(status='success')
     return jsonify(response.dict()), 200
 
 

app/models.py

@@ -9,9 +9,9 @@ class GetQuestionAndFactsResponse(BaseModel):
 
 
 class SubmitQuestionAndDocumentsResponse(BaseModel):
-    pass
+    status: str
 
 
 class SubmitQuestionAndDocumentRequest(BaseModel):
     question: str
-    urls: List[str]
+    documents: List[str]

app/ui.py

@@ -2,36 +2,40 @@ import time
 import requests
 import gradio as gr
 from utils import get_url_list
-from models import SubmitQuestionAndDocumentRequest, GetQuestionAndFactsResponse
+from models import SubmitQuestionAndDocumentRequest, GetQuestionAndFactsResponse, SubmitQuestionAndDocumentsResponse
 
-base_url = 'https://cleric-agent-ad76f992e8d8.herokuapp.com/'
+base_url = 'https://cleric-agent-api-untxx3isja-uc.a.run.app'
+# base_url = 'http://localhost:8000'
 
 
 def fetch_facts(question, call_log_urls):
+    if len(call_log_urls) == 0:
+        raise gr.Error('Please input call log.')
+
+    if len(question) == 0:
+        raise gr.Error('Please input question.')
+
     urls = get_url_list(call_log_urls)
     payload = SubmitQuestionAndDocumentRequest(
         question=question,
-        urls=urls
+        documents=urls
     ).dict()
+
     response = requests.post(f"{base_url}/submit_question_and_documents", json=payload)
+    response = SubmitQuestionAndDocumentsResponse(**response.json())
+    if response.status != 'success':
+        raise gr.Error('Input error: ' + response.status)
 
     start_time = time.time()
     while True:
         response = requests.get(f"{base_url}/get_question_and_facts")
         if response.status_code != 200:
-            # st.error(f"Unexpected status code when getting question and facts: {response.status_code}")
-            return None
-        try:
-            data = GetQuestionAndFactsResponse(**response.json())
-        except ValueError as e:
-            # st.error(f"The response data does not match the expected schema: {str(e)}")
-            # st.write(response.json())  # Print the invalid data for debugging
-            return None
+            raise gr.Error('Server response error.')
 
+        data = GetQuestionAndFactsResponse(**response.json())
         if data.status == "done":
             break
-        elif time.time() - start_time > 300:  # 5 minutes timeout
-            # st.error("Timeout: Facts not ready after 5 minutes")
+        elif time.time() - start_time > 300:
             return None
         time.sleep(1)
 
@@ -42,29 +46,24 @@ with gr.Blocks() as demo:
     gr.Markdown("""
     # Cleric Call Logs Summarize Agent
     
-    Please place the URLs in the Call Logs URLs text box, separated by new line.
-    Place your question to this call logs, then submit! 
+    ### Instructions:
+    1. Enter the URLs in the "Call Log URLs" text box, separating each URL with a new line.
+    2. Add your question related to these call logs.
+    3. Click the "Submit" button to proceed.
     """)
     error_box = gr.Textbox(label="Error", visible=False)
     with gr.Row(equal_height=True):
-        call_logs_box = gr.Textbox(label='Call Logs URLs', scale=2)
-        facts_box = gr.Textbox(label='Extracted Facts', scale=2)
+        call_logs_box = gr.Textbox(label='Call Log URLs', lines=10)
+        facts_box = gr.Textbox(label='Extracted Facts', lines=10)
 
     question_box = gr.Textbox(label='Question')
     submit_btn = gr.Button("Submit")
 
     submit_btn.click(
         fetch_facts,
-        inputs=[call_logs_box, question_box],
+        inputs=[question_box, call_logs_box],
         outputs=facts_box
     )
 
-# iface = gr.Interface(
-#     fn=fetch_facts,
-#     inputs=["text", "text"],
-#     outputs="text",
-#     allow_flagging="never",
-#     title="Cleric Call Logs Summarize Agent"
-# )
 demo.launch()
 

app/utils.py

@@ -1,4 +1,5 @@
 import requests
+import validators
 from typing import List
 
 
@@ -19,10 +20,17 @@ def read_documents(documents: List[str]) -> List[str]:
     logs = []
     for url in documents:
         response = requests.get(url)
+        response.raise_for_status()
         logs.append(response.text)
 
     return logs
 
 
-def preprocess_logs(logs: List[str]):
+def preprocess_logs(logs: List[str]) -> str:
     return '\n'.join(logs)
+
+
+def validate_request_logs(urls: List[str]):
+    for url in urls:
+        if not validators.url(url):
+            raise ValueError(f'The following URL is invalid: {url}')

requirements.txt

@@ -4,4 +4,5 @@ pydantic
 langchain
 langchain-openai
 pydantic
-gunicorn
+gunicorn
+validators