display wordclouds along with summary

app.py

@@ -1,43 +1,94 @@
-import gradio as gr
-import requests
+from wordcloud import WordCloud
+import matplotlib.pyplot as plt
 import json
+import requests
+import gradio as gr
+from io import BytesIO
+import base64
 import os
+
 AUTH = os.environ["AUTH"]
 
+def generate_wordcloud(keywords, theme):
+    # Sort keywords in descending order based on their values
+    sorted_keywords = [entry['keyword'] for entry in sorted(keywords, key=lambda x: x['value'], reverse=True)]
+
+    keyword_dict = {entry['keyword']: entry['value'] for entry in keywords}
+    
+    colour_range =['#00B3CF', '#FF0079', '#8C72B4', '#99B561', '#C49C8E']
+    
+    # Create a color function mapping each word to a color from the range based on grouped index
+    def color_func(word, font_size, position, orientation, random_state=42, **kwargs):
+        index = sorted_keywords.index(word) // 2 % len(colour_range)
+        return colour_range[index]
+    
+    # Generate the word cloud
+    wordcloud = WordCloud(width=800, height=400,
+                          background_color='white',
+                          color_func=color_func,
+                          collocations=False,  # Avoid combining words
+                          random_state=42
+                         ).generate_from_frequencies(keyword_dict)
+    
+    # Save the word cloud image to a BytesIO object
+    image_stream = BytesIO()
+    wordcloud.to_image().save(image_stream, format='PNG')
+    image_stream.seek(0)
+
+    # Convert the BytesIO object to a base64-encoded image string
+    image_data = base64.b64encode(image_stream.read()).decode('utf-8')
+    image_src = f"data:image/png;base64,{image_data}"
+
+    return f'<img src="{image_src}" alt="{theme} Wordcloud" style="width:270px;height:150px;margin-right:10px;">'
+
+
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     with gr.Row():
-        answers = gr.TextArea(label="List of responses",placeholder='["this is cool", "this is very cool"]')
+        answers = gr.TextArea(label="List of responses", placeholder='["this is cool", "this is very cool"]')
         with gr.Column():
-            nr_topics = gr.Slider(minimum=5, maximum=15,step=1)
-            greet_btn = gr.Button("Generate themes")
+            nr_topics = gr.Slider(minimum=5, maximum=15, step=1)
+    greet_btn = gr.Button("Generate themes")
    
-    output = gr.HTML(label="Themes")
+    output = gr.HTML(label="Themes and Wordclouds")
     
 
     @greet_btn.click(inputs=[answers, nr_topics], outputs=output)
     def greet(answers, nr_topics):
         inputs = {"answers": json.loads(answers), "nr_topics": nr_topics}
         p = requests.post(
-            url = 'https://pulsifi-dev--thematic-analysis-app.modal.run/analysis',
+            url='https://pulsifi-dev--thematic-analysis-app.modal.run/analysis',
             headers={"Authorization": AUTH},
-            json = inputs
+            json=inputs
         )
 
         result = p.json()
+        #print(result)
         data = result.get("data")
         if data["thematic_results"]:
-            output = {x["theme"]: x["description"] for x in data["thematic_results"]}
-
             output_text = ""
-            for key, description in output.items():
-                output_text +=  f"""<h3 style="color:blue">{key}:</h3>
-                                    <p> {description.strip()}
-                                 """
+            for theme_data in data["thematic_results"]:
+                theme = theme_data["theme"]
+                description = theme_data["description"]
+                keywords_list = theme_data['keywords']
+
+                # Generate word cloud and get HTML content
+                wordcloud_html = generate_wordcloud(keywords_list, theme)
+
+                # Add theme description to the output text
+                output_text += f"""
+                    <div style="display:flex; align-items: center;">
+                        {wordcloud_html}
+                        <div>
+                            <h3 style="color:blue">{theme}:</h3>
+                            <p style="text-align: justify;">{description.strip()}</p>
+                        </div>
+                    </div>
+                """
+
         else:
             output_text = data["reason"]
 
         return output_text
 
-
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()

requirements.txt

@@ -1 +1,2 @@
 requests
+wordcloud