Update app.py

app.py

@@ -1,121 +1,124 @@
-import gradio as gr
-import requests
-import os
-import random
-import pandas as pd
-
-# Load instructions from local files
-def load_instruction(persona):
-    try:
-        with open(f"instructions/{persona.lower()}.txt", "r") as file:
-            return file.read()
-    except FileNotFoundError:
-        return ""
-
-# Call Cohere API
-def call_cohere_api(system_instruction, user_prompt):
-    headers = {
-        "Authorization": f"Bearer {os.getenv('COHERE_API_KEY')}",
-        "Content-Type": "application/json"
-    }
-    
-    # Append word limit instruction
-    user_prompt += "\n\nAnswer in 100 words or fewer."
-    payload = {
-        "model": "command-r-plus",
-        "message": user_prompt,
-        "preamble": system_instruction,
-        "max_tokens": 300
-    }
-    response = requests.post("https://api.cohere.ai/v1/chat", headers=headers, json=payload)
-    return response.json().get("text", "No response").strip()
-
-# Load questions from file
-def load_questions():
-    try:
-        with open("questions.txt", "r") as file:
-            return [line.strip() for line in file if line.strip()]
-    except FileNotFoundError:
-        return []
-
-questions_list = load_questions()
-
-# Generate random question
-def get_random_question():
-    return random.choice(questions_list) if questions_list else "No questions available."
-
-# Load counter-narratives CSV
-def load_counternarratives():
-    try:
-        df = pd.read_csv("counternarratives.csv")
-        return df
-    except FileNotFoundError:
-        print("counternarratives.csv not found.")
-        return pd.DataFrame(columns=["myth", "fact", "persona"])
-
-counternarratives = load_counternarratives()
-
-# Generate Random Myth or Fact
-def get_random_myth_or_fact():
-    if counternarratives.empty:
-        return "No myths or facts available.", "Fact-Checker", "", ""
-    
-    row = counternarratives.sample(1).iloc[0]
-    selected_column = random.choice(["myth", "fact"])
-    myth_or_fact = row[selected_column]
-    persona = row["persona"]
-
-    # Set Persona 1 to the one in the myth/fact and Persona 2 to "Fact-Checker"
-    if selected_column == "myth":
-        fact_check_response = f"🛑 **MYTH**\n\nThe fact is: {row['fact']}"
-    else:
-        fact_check_response = f"✅ **FACT**\n\nIndeed, {row['fact']}"
-    
-    # Return responses and personas for the UI
-    return myth_or_fact, persona, "Fact-Checker", fact_check_response
-
-# Dynamically load persona names from instructions folder
-personas = [os.path.splitext(f)[0].capitalize() for f in os.listdir("instructions") if f.endswith(".txt")]
-
-# Gradio Interface
-with gr.Blocks() as demo:
-    with gr.Row():
-        gr.Markdown("# 🌲 **Wild Voices** — *Listening to the More-than-Human World*")
-        gr.Markdown("Ask questions to rivers, trees, owls, and more. Generate myths and facts, and let nature's wisdom respond.")
-
-    with gr.Row():
-        persona1 = gr.Dropdown(personas, label="Choose First Persona", value="Tree")
-        persona2 = gr.Dropdown(personas, label="Choose Second Persona", value="Crow")
-
-    with gr.Row():
-        user_input = gr.Textbox(label="🌱 Your Question", placeholder="e.g., What do you think of humans?", lines=2)
-        random_button = gr.Button("🎲 Generate Random Question")
-        myth_fact_button = gr.Button("🔍 Generate Random Myth/Fact")
-
-    with gr.Row():
-        ask_button = gr.Button("🌎 Submit Question")
-
-    with gr.Row():
-        output1 = gr.Textbox(label=f"{persona1.value} Responds")
-        output2 = gr.Textbox(label=f"{persona2.value} Responds")
-
-    # Button events
-    random_button.click(fn=get_random_question, inputs=[], outputs=[user_input])
-    
-    # Myth/Fact button click event
-    myth_fact_button.click(
-        fn=get_random_myth_or_fact, 
-        inputs=[], 
-        outputs=[user_input, persona1, persona2, output2]
-    )
-
-    # Ask button for normal questions
-    ask_button.click(
-        fn=lambda p1, p2, q: (call_cohere_api(load_instruction(p1), q), 
-                              call_cohere_api(load_instruction(p2), q)), 
-        inputs=[persona1, persona2, user_input], 
-        outputs=[output1, output2]
-    )
-
-if __name__ == "__main__":
-    demo.launch()
+import gradio as gr
+import requests
+import os
+import random
+import pandas as pd
+
+# Load instructions from local files
+def load_instruction(persona):
+    try:
+        with open(f"instructions/{persona.lower()}.txt", "r") as file:
+            return file.read()
+    except FileNotFoundError:
+        return ""
+
+# Call Cohere API
+def call_cohere_api(system_instruction, user_prompt):
+    headers = {
+        "Authorization": f"Bearer {os.getenv('COHERE_API_KEY')}",
+        "Content-Type": "application/json"
+    }
+    
+    # Append word limit instruction
+    user_prompt += "\n\nAnswer in 100 words or fewer."
+    payload = {
+        "model": "command-r-plus",
+        "message": user_prompt,
+        "preamble": system_instruction,
+        "max_tokens": 300
+    }
+    response = requests.post("https://api.cohere.ai/v1/chat", headers=headers, json=payload)
+    return response.json().get("text", "No response").strip()
+
+# Load questions from file
+def load_questions():
+    try:
+        with open("questions.txt", "r") as file:
+            return [line.strip() for line in file if line.strip()]
+    except FileNotFoundError:
+        return []
+
+questions_list = load_questions()
+
+# Generate random question
+def get_random_question():
+    return random.choice(questions_list) if questions_list else "No questions available."
+
+# Load counter-narratives CSV
+def load_counternarratives():
+    try:
+        df = pd.read_csv("counternarratives.csv")
+        return df
+    except FileNotFoundError:
+        print("counternarratives.csv not found.")
+        return pd.DataFrame(columns=["myth", "fact", "persona"])
+
+# Generate Random Myth or Fact
+def get_random_myth_or_fact():
+    if counternarratives.empty:
+        return "No myths or facts available.", "Fact-Checker", "", "", ""
+
+    # Randomly select a row from the dataframe
+    row = counternarratives.sample(1).iloc[0]
+    selected_column = random.choice(["myth", "fact"])
+    myth_or_fact = row[selected_column]
+    persona = row["persona"]
+
+    # Call the Cohere API to get the persona's response
+    persona_instruction = load_instruction(persona)
+    persona_response = call_cohere_api(persona_instruction, myth_or_fact)
+
+    # Fact-checker response logic
+    if selected_column == "myth":
+        fact_check_response = f"🛑 **MYTH**\n\nThe fact is: {row['fact']}"
+    else:
+        fact_check_response = f"✅ **FACT**\n\nIndeed, {row['fact']}"
+
+    # Return the myth/fact, update the personas, and fill the responses
+    return myth_or_fact, persona, "Fact-Checker", persona_response, fact_check_response
+
+# Dynamically load persona names from instructions folder
+personas = [os.path.splitext(f)[0].capitalize() for f in os.listdir("instructions") if f.endswith(".txt")]
+
+# Gradio Interface
+with gr.Blocks() as demo:
+    with gr.Row():
+        gr.Markdown("# 🌲 **Wild Voices** — *Listening to the More-than-Human World*")
+        gr.Markdown("Ask questions to rivers, trees, owls, and more. Generate myths and facts, and let nature's wisdom respond.")
+
+    with gr.Row():
+        persona1 = gr.Dropdown(personas, label="Choose First Persona", value="Tree")
+        persona2 = gr.Dropdown(personas, label="Choose Second Persona", value="Crow")
+
+    with gr.Row():
+        user_input = gr.Textbox(label="🌱 Your Question", placeholder="e.g., What do you think of humans?", lines=2)
+        random_button = gr.Button("🎲 Generate Random Question")
+        myth_fact_button = gr.Button("🔍 Generate Random Myth/Fact")
+
+    with gr.Row():
+        ask_button = gr.Button("🌎 Submit Question")
+
+    with gr.Row():
+        output1 = gr.Textbox(label=f"{persona1.value} Responds")
+        output2 = gr.Textbox(label=f"{persona2.value} Responds")
+
+    # Button events
+    random_button.click(fn=get_random_question, inputs=[], outputs=[user_input])
+    
+    # Myth/Fact button click event
+    myth_fact_button.click(
+        fn=get_random_myth_or_fact, 
+        inputs=[], 
+        outputs=[user_input, persona1, persona2, output2]
+    )
+
+    # Ask button for normal questions
+    ask_button.click(
+        fn=lambda p1, p2, q: (call_cohere_api(load_instruction(p1), q), 
+                              call_cohere_api(load_instruction(p2), q)), 
+        inputs=[persona1, persona2, user_input], 
+        outputs=[output1, output2]
+    )
+
+if __name__ == "__main__":
+    demo.launch()