Refactor message handling to replace viral messages with categorized …

…message display; add MessageCategory model and update analysis prompt for improved insights.
zoharbabin · Dec 26, 2024 · 8aadfe8 · 8aadfe8
1 parent 33aab66
commit 8aadfe8
Show file tree

Hide file tree

Showing 3 changed files with 118 additions and 46 deletions.
diff --git a/gh_static_front/app.js b/gh_static_front/app.js
@@ -231,7 +231,7 @@ function displayResults(data) {
     displayMemorableMoments(data);
     displayHappiestDays(data);
     displaySaddestDays(data);
-    displayViralMessages(data);
+    displayMessageCategories(data);
     displayMediaStats(data);
     displaySharedLinks(data);
 
@@ -406,39 +406,6 @@ function displaySaddestDays(data) {
     document.getElementById('saddest-days').innerHTML = saddestHtml;
 }
 
-function displayViralMessages(data) {
-    if (!data.viral_messages) {
-        console.log('No viral messages found');
-        return;
-    }
-
-    const viralHtml = data.viral_messages
-        .map(msg => {
-            return `
-                <div class="p-2 bg-purple-50 rounded hover:bg-purple-100 transition-all duration-200 cursor-pointer"
-                    onclick="createConfetti(event.clientX, event.clientY)">
-                    <div class="flex items-start gap-2">
-                        <div class="flex-grow">
-                            <p class="font-medium text-purple-900 mb-1">${msg.message}</p>
-                            <div class="text-xs text-purple-600 flex items-center gap-2 mb-1">
-                                <span class="inline-flex items-center">💬 ${msg.replies}</span>
-                                <span class="inline-flex items-center">❤️ ${msg.reactions}</span>
-                            </div>
-                        </div>
-                    </div>
-                    <div class="text-sm text-gray-600 border-l-2 border-purple-200 pl-2 mt-1">
-                        ${msg.thread.map(reply => `
-                            <p class="text-sm py-0.5 hover:bg-purple-50 rounded">${reply}</p>
-                        `).join('')}
-                    </div>
-                </div>
-            `;
-        })
-        .join('');
-
-    document.getElementById('viral-messages').innerHTML = viralHtml;
-}
-
 function displayMediaStats(data) {
     if (!data.media_stats) {
         console.log('No media stats found');
@@ -532,4 +499,72 @@ function displaySharedLinks(data) {
         .join('');
 
     document.getElementById('shared-links').innerHTML = linksHtml;
+}
+
+function displayMessageCategories(data) {
+    if (!data.message_categories || data.message_categories.length === 0) {
+        console.log('No message categories found');
+        return;
+    }
+
+    const categoryColors = {
+        'celebration': 'from-yellow-50 to-yellow-100',
+        'business': 'from-blue-50 to-blue-100',
+        'team': 'from-green-50 to-green-100',
+        'strategic': 'from-purple-50 to-purple-100',
+        'knowledge': 'from-red-50 to-red-100'
+    };
+
+    const getGradient = (category) => {
+        const baseCategory = Object.keys(categoryColors).find(key =>
+            category.toLowerCase().includes(key.toLowerCase())
+        );
+        return categoryColors[baseCategory] || 'from-gray-50 to-gray-100';
+    };
+
+    const categoriesHtml = data.message_categories
+        .map(category => `
+            <div class="mb-6">
+                <div class="bg-gradient-to-r ${getGradient(category.category)} p-6 rounded-lg shadow-md
+                    hover:shadow-lg transition-all duration-300 cursor-pointer"
+                    onclick="createConfetti(event.clientX, event.clientY)">
+                    <div class="flex justify-between items-start mb-4">
+                        <div>
+                            <h3 class="text-xl font-bold text-gray-800">${category.category}</h3>
+                            <p class="text-sm text-gray-600">${category.subcategory}</p>
+                        </div>
+                        <span class="text-sm font-semibold bg-white px-3 py-1 rounded-full shadow">
+                            Impact: ${(category.impact_score * 100).toFixed(0)}%
+                        </span>
+                    </div>
+                    
+                    <div class="space-y-2">
+                        ${category.messages.map(msg => `
+                            <div class="bg-white bg-opacity-60 p-3 rounded">
+                                <p class="text-gray-800">${msg}</p>
+                            </div>
+                        `).join('')}
+                    </div>
+                    
+                    <div class="mt-4 flex flex-wrap gap-2">
+                        ${category.participants.map(participant => `
+                            <span class="text-sm bg-white bg-opacity-70 px-2 py-1 rounded">
+                                ${participant}
+                            </span>
+                        `).join('')}
+                    </div>
+                    
+                    <div class="mt-4 text-sm text-gray-600 italic">
+                        ${category.context}
+                    </div>
+                    
+                    <div class="mt-2 text-xs text-gray-500">
+                        ${category.timestamp}
+                    </div>
+                </div>
+            </div>
+        `)
+        .join('');
+
+    document.getElementById('message-categories').innerHTML = categoriesHtml;
 }
diff --git a/gh_static_front/index.html b/gh_static_front/index.html
@@ -120,10 +120,12 @@ <h2 class="text-2xl font-bold mb-4 text-blue-600">Most Reflective Days</h2>
                 </div>
             </div>
 
-            <!-- Viral Messages -->
+            <!-- Message Categories -->
             <div class="bg-white rounded-lg shadow-md p-6 col-span-1 md:col-span-2">
-                <h2 class="text-2xl font-bold mb-4">Epic Chat Moments</h2>
-                <div id="viral-messages" class="space-y-2"></div>
+                <h2 class="text-2xl font-bold mb-4">Message Categories</h2>
+                <div id="message-categories" class="grid gap-6">
+                    <!-- Categories will be dynamically inserted here -->
+                </div>
             </div>
 
             <!-- Media Statistics -->
@@ -162,6 +164,7 @@ <h2 class="text-2xl font-bold mb-4">Holiday Greeting</h2>
                 <h2 class="text-2xl font-bold mb-4">Your Group's Holiday Poem</h2>
                 <div id="chat-poem" class="text-center text-lg font-serif p-6 bg-gradient-to-r from-red-50 to-green-50 rounded-lg whitespace-pre-line"></div>
             </div>
+
         </div>
     </main>
 

diff --git a/main.py b/main.py
@@ -160,6 +160,15 @@ class MediaStats(BaseModel):
     top_media_sharers: List[UserActivity]
     most_reacted_media: List[MediaItem]
 
+class MessageCategory(BaseModel):
+    category: str
+    subcategory: str
+    messages: List[str]
+    context: str
+    participants: List[str]
+    impact_score: float
+    timestamp: str
+
 class ChatSummary(BaseModel):
     most_active_users: List[UserActivity]
     popular_topics: List[str]
@@ -175,6 +184,7 @@ class ChatSummary(BaseModel):
     shared_links: List[SharedLink]
     chat_poem: str
     media_stats: MediaStats
+    message_categories: List[MessageCategory]
 
 def calculate_md5(content: bytes) -> str:
     """Calculate MD5 hash of file content."""
@@ -585,13 +595,18 @@ async def process_date_group(dates):
             batch_results = []
 
             for date in dates:
-                day_messages = daily_messages[date]
-                if len(day_messages) > 5:
-                    indices = [0, len(day_messages)//4, len(day_messages)//2,
-                             (3*len(day_messages))//4, len(day_messages)-1]
-                    group_messages.extend([day_messages[i] for i in indices])
+                # Filter out media messages first
+                filtered_messages = [
+                    msg for msg in daily_messages[date]
+                    if not MEDIA_PATTERN.search(str(msg))
+                ]
+
+                if len(filtered_messages) > 5:
+                    indices = [0, len(filtered_messages)//4, len(filtered_messages)//2,
+                             (3*len(filtered_messages))//4, len(filtered_messages)-1]
+                    group_messages.extend([filtered_messages[i] for i in indices])
                 else:
-                    group_messages.extend(day_messages)
+                    group_messages.extend(filtered_messages)
 
                 if len(group_messages) >= 15:
                     sentiment = await analyze_sentiment_batch(group_messages)
@@ -695,11 +710,29 @@ async def analyze_chat(file: UploadFile = File(...)):
             remaining = df[~df['message'].isin(samples)].sample(n=min(sample_size - len(samples), len(df)))
             samples.extend(remaining['message'].tolist())
 
-        prompt = f"""Analyze this WhatsApp chat and provide insights in the following format:
+        prompt = f"""Analyze this WhatsApp chat and provide comprehensive insights with the following structure:
+
         1. Key topics discussed (max 5)
         2. Three most memorable moments
         3. A festive holiday greeting based on the chat context
         4. Create a comedic rhyming poem (at least 8 lines) that tells a story about the group's memorable moments and inside jokes. Make it festive and entertaining!
+        5. Categorize messages into meaningful groups by analyzing:
+           - Type of interaction (celebration, milestone, discussion, etc.)
+           - Context and significance
+           - Participant dynamics
+           - Impact on team/organization
+           - Cultural significance
+           
+           For each identified category, provide:
+           - Category name and subcategory
+           - Representative messages
+           - Context and significance
+           - Involved participants
+           - Impact score (0.0 to 1.0)
+           - Timestamp
+
+        Don't use predetermined categories - identify natural patterns and groupings that emerge from the content.
+        Consider message context, participant engagement, long-term significance, and cultural dynamics.
 
         Chat sample: {' '.join(samples)}"""
 
@@ -879,7 +912,7 @@ def process_message_threads(messages_df: pd.DataFrame) -> List[ViralMessage]:
         word_counts_converted = {k: int(v) for k, v in word_counts.items()}
         activity_converted = {k: int(v) for k, v in activity.items()}
 
-        # Create summary with properly structured data
+        # Create summary with properly structured data including message categories
         summary = ChatSummary(
             most_active_users=[UserActivity(name=k, count=v) for k, v in most_active_converted.items()],
             popular_topics=response.popular_topics,
@@ -894,7 +927,8 @@ def process_message_threads(messages_df: pd.DataFrame) -> List[ViralMessage]:
             viral_messages=viral_messages,
             shared_links=shared_links,
             chat_poem=response.chat_poem,
-            media_stats=media_stats
+            media_stats=media_stats,
+            message_categories=response.message_categories if hasattr(response, 'message_categories') else []
         )
 
         analysis_time = time.time() - analysis_start