Update app.py

app.py

@@ -51,27 +51,42 @@ def talk_to_llm(user_prompt):
     }
 
     prompt_text = f'''
-    As StarGateVR, your role is specialized in customizing ADQL (Astronomical Data Query Language) queries for astronomers. Your focus is particularly on integrating specific 'WHERE' clauses into a standard query template. We will put your WHERE clause into the completed query template. In some cases, you may need to add additional parameters to the SELECT clause.
+    As StarGateVR, your role is specialized in customizing ADQL (Astronomical Data Query Language) 
+    queries for astronomers. Your focus is particularly on integrating specific 'WHERE' clauses into 
+    a standard query template. We will put your WHERE clause into the completed query template. 
 
-    The query includes essential fields like source_id, positional data (RA, DEC), motion data (pmra, pmdec), light parameters, and placeholders for additional parameters. 
+    The query includes essential SELECT fields like source_id, positional data (ra, dec), 
+    motion data (pmra, pmdec), and light parameters. Note that any
+    fields used in the WHERE clause must also be added to the SELECT clause.
 
-    Customizing 'WHERE' Clause: Your primary task is to adapt the 'WHERE' clause to fit the user's specific astronomical requirements. This often involves filtering stars based on various criteria such as distance, location in the sky, brightness, etc. Always include, at a minimum, the SELECT and FROM clauses exactly as given in the template. Note that the WHERE clause must reference variables by the field name and not the "AS" name. There is a special case for the part of the SELECT that is " 'Gaia DR3 ' || source_id as source_id", in the WHERE clause this field should always be referred to by "source_id".
-
-    Note that any parameter used in the WHERE clause should also be included in the SELECT clause. 
+    Customizing 'WHERE' Clause: Your primary task is to adapt the 'WHERE' clause to fit 
+    the user's specific astronomical requirements. This often involves filtering stars based on 
+    various criteria such as distance, location in the sky, brightness, etc. 
+    
+    Always include, at a minimum, the SELECT and FROM clauses as given in the template. Note that 
+    the WHERE clause must reference variables by the field name and not the "AS" name. 
+    There is a special case for the part of the SELECT that is " 'Gaia DR3 ' || source_id as source_id", 
+    in the WHERE clause this field should always be referred to by "source_id".
 
     Here is an example of the WHERE clause:
-
     ```
     WHERE (parallax >= 11.11 AND parallax_over_error>=20 AND
     astrometric_excess_noise<=2)
     ```
 
-    Bounds on Parallax: Always include bounds on parallax in the 'WHERE' clause. This is important as it helps in retrieving stars within a specified 3D region of space.
+    Bounds on Parallax: Always include bounds on parallax in the 'WHERE' clause. This is 
+    important as it helps in retrieving stars within a specified 3D region of space.
 
     The json structure to return is
     {{
-        "reasoning": "<Think through what the user is asking for, and what you know about the GAIA DB and astrsonomy to create their request.>",
-        "where_clause": "<a properly formatted ADQL query that will return the stars the user is asking for>"
+        "reasoning": "<Think through what the user is asking for, and what you know about the GAIA DB 
+        and astronomy to create their request. Because the WHERE clause you are generating 
+        is going to be concatenated into a larger SQL query, consider how to structure the 
+        query such that everything fits in a single WHERE clause.  Otherwise it will break 
+        the downstream logic.>",
+        "the_query": "<a properly formatted ADQL query that will return the stars the 
+        user is asking for>"
+        
     }}
 
     The users prompt is "{user_prompt}"
@@ -94,12 +109,12 @@ def talk_to_llm(user_prompt):
     output_json = json.loads(response_json['choices'][0]['message']['content'])
     
     try:
-        where_clause = output_json['where_clause']
+        the_query = output_json['the_query']
         reasoning = output_json['reasoning']
         print(f"dialog: {reasoning}")
-        print(f"query: {where_clause}")
+        print(f"query: {the_query}")
         # Update theme on the server
-        return reasoning, where_clause
+        return reasoning, the_query
     except KeyError as e:
         print(f"Key error: {e}")
         return "Failed to generate query."
@@ -121,7 +136,7 @@ def complete_query(partial_query):
     -- Measured Position
     ra,
     dec,
-    (1 / parallax)*1000 as dist,
+    parallax,
     -- Measured Motion
     pmra,
     pmdec,
@@ -140,7 +155,6 @@ def complete_query(partial_query):
     parallax_error,
 
     -- Additional parameters that appear in the WHERE clause should be added here
-    -- Example: MH_GSPPhot as 
 
     -- Note: No comma after this last SELECT item
     parallax_over_error
@@ -195,27 +209,29 @@ def process_queries():
     results = {}
     for prompt in user_prompt_examples:
         print (f"Processing prompt: {prompt}")
-        reasoning, where_clause = talk_to_llm(prompt)
-        query = complete_query(where_clause)
+        reasoning, the_query = talk_to_llm(prompt)
+        query = the_query #complete_query(the_query)
         download_url_from_query(query, prompt)
     return results
 
 with gr.Blocks() as demo:
     with gr.Row():
-        user_prompt = gr.Textbox(label="Enter your query for the LLM", value="Find me the stars in the hayedes cluster.")
+        user_prompt = gr.Textbox(label="Enter your query for the LLM", value="Find me the stars in 
+        the hayedes cluster.")
         submit_btn = gr.Button("Ask LLM")
     reasoning_output = gr.Textbox(label="Reasoning")
-    where_clause_output = gr.Textbox(label="Where Clause")
+    the_query_output = gr.Textbox(label="The Query")
 
     
     make_query_complete_bt  = gr.Button("Complete Query")
     complete_query_output = gr.Textbox(label="Complete Query")
     make_query_complete_bt.click(fn=complete_query, inputs=where_clause_output, outputs=[complete_query_output])
-    submit_btn.click(fn=talk_to_llm, inputs=user_prompt, outputs=[reasoning_output, where_clause_output])
+    submit_btn.click(fn=talk_to_llm, inputs=user_prompt, outputs=[reasoning_output, the_query_output])
 
     create_tap_url_bt = gr.Button("Create TAP URL")
     download_data_output = gr.Markdown()
-    create_tap_url_bt.click(fn=create_markdown_url_from_query, inputs=complete_query_output, outputs=[download_data_output])
+    create_tap_url_bt.click(fn=create_markdown_url_from_query, inputs=complete_query_output, 
+                            outputs=[download_data_output])
 
 #demo.launch(server_name="0.0.0.0", server_port=7861, share=True, debug=True)
 demo.launch()