Update app.py

app.py

@@ -3,7 +3,6 @@ import os
 import subprocess
 import pandas as pd
 from PIL import Image
-import io
 import folium
 
 # --- Setup: Clone the repository for helper classes ---
@@ -13,7 +12,7 @@ if not os.path.exists(REPO_DIR):
     try:
         print(f"Cloning repository into ./{REPO_DIR}...")
         subprocess.run(
-            ["git", "clone", "https://github.com/ioscbasotcstw/TheAmateur.git"],
+            ["git", "clone", "https://github.com/OrlovVladislav/TheAmateur.git"],
             check=True,
             capture_output=True,
             text=True
@@ -21,15 +20,13 @@ if not os.path.exists(REPO_DIR):
         print("Repository cloned successfully.")
     except subprocess.CalledProcessError as e:
         print(f"Error cloning repository: {e.stderr}")
-        # Exit if cloning fails, as the app cannot run without it.
-        exit()
+        exit() # Exit if cloning fails, as the app cannot run without it.
 
 # Now that the repo is cloned, we can import the necessary modules
 try:
     from TheAmateur.src.rainbolt_parody import RainboltParody
     from TheAmateur.src.map_gen import MapGenerator
     from TheAmateur.src.geo_info import GeoInfo
-    from TheAmateur.src.utils import prepare_csv as original_prepare_csv
 except ImportError as e:
     print(f"Failed to import from the cloned repository: {e}")
     exit()
@@ -45,14 +42,10 @@ You will follow a strict, three-stage analytical process.
 
 ## STAGE 1: OBSERVATION & CLUE EXTRACTION
 Internally, create a structured list of all visual evidence. Do not output this list.
-- **Road & Infrastructure:**
-  - Road Lines, Bollards, Poles, Signage (Language, Script, Style)
-- **Environment & Nature:**
-  - Sun Position & Climate, Vegetation & Trees, Soil & Topography
-- **Human & Cultural Markers:**
-  - Architecture Style, Vehicle Models & License Plates
-- **Meta Clues:**
-  - Google Car Generation/Antenna, Image Quality/Season
+- **Road & Infrastructure:** Road Lines, Bollards, Poles, Signage (Language, Script, Style)
+- **Environment & Nature:** Sun Position & Climate, Vegetation & Trees, Soil & Topography
+- **Human & Cultural Markers:** Architecture Style, Vehicle Models & License Plates
+- **Meta Clues:** Google Car Generation/Antenna, Image Quality/Season
 
 ## STAGE 2: DEDUCTION & SYNTHESIS
 Internally, reason through the clues from Stage 1.
@@ -88,7 +81,8 @@ def handle_error(message, e=None):
 
 def pinpoint_location(api_key, uploaded_image):
     """
-    Takes an API key and an uploaded image, returns location data and a map.
+    Takes an API key and an uploaded image, returns location data.
+    The `gr.Image(type="pil")` component handles the upload, replacing the need for a custom upload function.
     """
     if not api_key:
         handle_error("Google AI API Key is required.")
@@ -97,11 +91,14 @@ def pinpoint_location(api_key, uploaded_image):
 
     try:
         rp = RainboltParody(title="Location Analysis", api_key=api_key)
-        # The 'get_info' method in the original code expects a PIL image
+        # 'uploaded_image' is already a PIL Image object, thanks to `type="pil"`.
         response, _, _ = rp.get_info(SYSTEM_PROMPT=SYSTEM_PROMPT, img=uploaded_image)
 
         # Prepare outputs for Gradio
-        info_df = pd.DataFrame([response])
+        # Convert the dictionary to a list of lists for the DataFrame
+        info_data = [[key, value] for key, value in response.items()]
+        info_df = pd.DataFrame(info_data, columns=["Attribute", "Value"])
+
         reasoning = response.get('reasoning', 'No reasoning provided.')
         coords_str = response.get('coordinates', '0,0')
         lat, lon = map(float, coords_str.split(','))
@@ -128,13 +125,11 @@ def calculate_error_distance(prediction_state, url, manual_coords_str):
         handle_error("Please provide either a Google Maps URL or manual coordinates for the actual location.")
 
     try:
-        # Determine the source of the actual coordinates
         if url:
             gi = GeoInfo(response=prediction_state, url=url)
         else:
             gi = GeoInfo(response=prediction_state, coordinates=manual_coords_str)
 
-        # Calculate error and bearing
         error_km, lat_actual, lon_actual, _, _ = gi.calculate_error()
         bearing = gi.calculate_bearing()
         direction = gi.get_direction(angle=bearing)
@@ -142,7 +137,6 @@ def calculate_error_distance(prediction_state, url, manual_coords_str):
         info_text = gi.combine_info(error=error_km, direction=direction)
         actual_coords = f"{lat_actual},{lon_actual}"
 
-        # Generate map with both points and a line
         mgk = MapGenerator()
         map_with_line = mgk.get_map(
             location1=prediction_state['coordinates'],
@@ -151,7 +145,6 @@ def calculate_error_distance(prediction_state, url, manual_coords_str):
             direction=direction
         )
         
-        # Convert folium map to HTML for display in Gradio
         map_html = map_with_line._repr_html_()
         
         return info_text, gr.HTML(value=map_html, visible=True), error_km, direction, url if url else ""
@@ -162,7 +155,8 @@ def calculate_error_distance(prediction_state, url, manual_coords_str):
 
 def save_and_download(prediction_state, error_km, direction, original_url, filename):
     """
-    Saves the combined results to a CSV file and provides a download link.
+    Saves the combined results to a CSV file and provides it for download.
+    This function replaces the Colab-specific `prepare_csv`.
     """
     if not prediction_state:
         handle_error("No prediction data to save.")
@@ -172,34 +166,30 @@ def save_and_download(prediction_state, error_km, direction, original_url, filen
         filename += '.csv'
 
     try:
-        # Re-using the logic from the original notebook's prepare_csv
-        # but adapting it for Gradio by removing the colab-specific download part.
-        
-        file_exists = os.path.exists(filename)
-        
+        # Prepare the new row of data
         data_to_add = prediction_state.copy()
-        data_to_add['error'] = error_km if error_km is not None else 'N/A'
+        data_to_add['error'] = f"{error_km:.2f}" if error_km is not None else 'N/A'
         data_to_add['direction'] = direction if direction else 'N/A'
         data_to_add['actual_location_url'] = original_url if original_url else 'N/A'
-        
         new_df = pd.DataFrame([data_to_add])
         
-        if file_exists:
+        # Check if file exists to append or create new
+        if os.path.exists(filename):
             existing_df = pd.read_csv(filename)
             final_df = pd.concat([existing_df, new_df], ignore_index=True)
         else:
             final_df = new_df
         
+        # Save the final dataframe to the CSV file
         final_df.to_csv(filename, index=False)
 
         # Return the updated DataFrame for display and the file path for download
-        return gr.DataFrame(value=final_df, visible=True), gr.File(value=filename, visible=True)
+        return gr.DataFrame(value=final_df, visible=True), gr.File(value=filename, label="Download CSV", visible=True)
 
     except Exception as e:
         handle_error("Failed to save data to CSV", e)
 
 # --- Gradio UI ---
-
 with gr.Blocks(theme=gr.themes.Soft(), title="GeoGuessr AI") as app:
     gr.Markdown("# GeoGuessr AI 🔎")
     gr.Markdown("Pinpoint the location of any Google Street View image.")
@@ -212,27 +202,18 @@ with gr.Blocks(theme=gr.themes.Soft(), title="GeoGuessr AI") as app:
 
     with gr.Row():
         with gr.Column(scale=1):
-            # --- STEP 1: Input and Prediction ---
             gr.Markdown("## 1. Pinpoint Location")
-            api_key_input = gr.Textbox(
-                label="Google AI API Key",
-                placeholder="Enter your API key here...",
-                type="password",
-                interactive=True
-            )
+            api_key_input = gr.Textbox(label="Google AI API Key", placeholder="Enter your API key here...", type="password")
             image_input = gr.Image(type="pil", label="Upload Street View Image")
             pinpoint_btn = gr.Button("Pinpoint Location", variant="primary")
-
-            # --- STEP 2: Prediction Output ---
+            
             gr.Markdown("### Prediction Results")
             info_output_df = gr.DataFrame(label="Location Information", headers=["Attribute", "Value"])
             reasoning_output = gr.Textbox(label="AI Reasoning", lines=5, interactive=False)
 
         with gr.Column(scale=2):
-            # --- Map Display ---
             gr.Markdown("### Map View")
             map_output_single = gr.Map(label="Predicted Location", interactive=False)
-            # This HTML component will be used for the two-point map later
             map_output_comparison = gr.HTML(visible=False)
 
     with gr.Accordion("2. Calculate Error (Optional)", open=False, visible=False) as error_accordion:
@@ -250,29 +231,24 @@ with gr.Blocks(theme=gr.themes.Soft(), title="GeoGuessr AI") as app:
             download_file = gr.File(label="Download CSV", visible=False)
 
     # --- Event Handlers ---
-
-    # Pinpoint button click event
     pinpoint_btn.click(
         fn=pinpoint_location,
         inputs=[api_key_input, image_input],
         outputs=[info_output_df, reasoning_output, map_output_single, prediction_state, error_accordion, save_accordion]
     )
-
-    # Calculate distance button click event
+    
     calculate_btn.click(
         fn=calculate_error_distance,
         inputs=[prediction_state, url_input, manual_coords_input],
         outputs=[distance_output, map_output_comparison, error_state, direction_state, url_state]
     )
-
-    # Save to CSV button click event
+    
     save_btn.click(
         fn=save_and_download,
         inputs=[prediction_state, error_state, direction_state, url_state, filename_input],
         outputs=[csv_preview_df, download_file]
     )
     
-    # # --- Example Images ---
     # gr.Examples(
     #     examples=[
     #         [os.path.join(REPO_DIR, "examples/example_1.png")],