Update app.py

app.py

@@ -4,6 +4,36 @@ import os
 
 openai.api_key = os.environ["OpenAPI_Key"]
 
+manual = "uitleg"
+
+# Function to make API call
+def api_call(messages, temperature=0.5, model="gpt-3.5-turbo"):
+    return openai.ChatCompletion.create( 
+        messages=messages,
+        temperature=temperature,
+        model=model
+    ).choices[0].message.content
+
+def chatbot(input):
+    # Check if input is empty
+    if not input:
+        return manual
+    # Step 1: first API call ~5secs
+        stepOne = [
+            {"role": "system", "content": "You distill a takeaway out of an exercise's question and answer. Mark the answer part in triple quotes '''. Don't do anything else."},
+            {"role": "user", "content": UserPrompt1},
+            {"role": "assistant", "content": AssistantPrompt1},
+            {"role": "user", "content": UserPrompt2},
+            {"role": "assistant", "content": AssistantPrompt2},
+            {"role": "user", "content": UserPrompt3},
+            {"role": "assistant", "content": AssistantPrompt3},
+            {"role": "user", "content": input}
+        ]
+        Takeaway = api_call(stepOne, 0.4)
+        strippedTakeaway = Takeaway.replace("'''", '')
+
+
+
 def chunk_HTMLs(text, delimiter="HTML (ID:"):
     """ Splits the text into chunks based on a delimiter, extracts the ID, and stores chunks in a dictionary. """
     HTMLs = text.split(delimiter)
@@ -18,16 +48,28 @@ def chunk_HTMLs(text, delimiter="HTML (ID:"):
             html_dict[html_id] = delimiter + "HTML (ID:" + html_id + ") " + html_content  # Store in dictionary with full label
     
     return html_dict
-
     
 def process_course(input_text):
-    """ Processes input text, extracts HTML chunks, and returns only the HTML IDs. """
+    """ Processes input text, extracts HTML chunks, analyzes each using an OpenAI model, and returns the analysis results. """
     stored_HTMLs = chunk_HTMLs(input_text)
-    return '\n\n'.join(stored_HTMLs.keys())  # Return the keys of the dictionary
+
+    results = []
+    for html_id, html_content in stored_HTMLs.items():
+        # Prepare the message list for the API call
+        messages = [
+            {"role": "system", "content": "Analyseer het perspectief van deze tekst, bedoeld voor zorgprofessionals. Adresseert de tekst altijd de zorgverlener? Of wordt soms ook de zorgvrager aangesproken? Geef je antwoord in een van de volgende zinnen: \nOordeel: het perspectief is altijd zorgverlener. \nOordeel: het perspectief is deels zorgvrager."},
+            {"role": "user", "content": html_content}
+        ]
+        # Make API call
+        oordeel = api_call(messages)
+        # Store results
+        results.append((html_id, oordeel))
+    
+    return results
 
     
 inputs = gr.Textbox(lines=7, label="Input Text")
-outputs = gr.Textbox(label="Processed HTML Chunks")
+outputs = gr.Textbox(label="Result")
 
 # Create the Gradio interface with HTML-formatted output
 iface = gr.Interface(