app.py

import streamlit as st
import pandas as pd
from eventbrite_scrapper import Eventbrite
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity
import numpy as np
from datetime import datetime
from dataclasses import dataclass, field, replace
from typing import List, Any

# Dataclasses for event structure
@dataclass(frozen=True)
class EventAddress:
    latitude: float = None
    longitude: float = None
    region: str = None
    postal_code: str = None
    address_1: str = None

@dataclass(frozen=True)
class EventVenue:
    id: str = None
    name: str = None
    url: str = None
    address: EventAddress = field(default_factory=lambda: EventAddress())

@dataclass(frozen=True)
class EventImage:
    url: str = None

@dataclass(frozen=True)
class EventTag:
    text: str = None

@dataclass(frozen=True)
class Event:
    id: str = None
    name: str = None
    url: str = None
    is_online_event: bool = False
    short_description: str = None
    published_datetime: datetime = None
    start_datetime: datetime = None
    end_datetime: datetime = None
    timezone: str = None
    hide_start_date: bool = False
    hide_end_date: bool = False
    parent_event_url: str = None
    series_id: str = None
    primary_venue: EventVenue = field(default_factory=lambda: EventVenue())
    tickets_url: str = None
    checkout_flow: str = None
    language: str = None
    image: EventImage = field(default_factory=lambda: EventImage())
    tags_categories: tuple = field(default_factory=tuple)
    tags_formats: tuple = field(default_factory=tuple)
    tags_by_organizer: tuple = field(default_factory=tuple)

    def __hash__(self):
        return hash(self.id) if self.id else hash((self.name, self.is_online_event, self.start_datetime, self.primary_venue.name))

# Event Retrieval Pipeline
class EventbriteRAGPipeline:
    def __init__(self, events: List[Event], embedding_model: str = 'all-MiniLM-L6-v2'):
        self.events = [
            replace(
                event,
                tags_categories=tuple(event.tags_categories),
                tags_formats=tuple(event.tags_formats),
                tags_by_organizer=tuple(event.tags_by_organizer),
            )
            for event in events
        ]
        self.model = SentenceTransformer(embedding_model)
        self.event_embeddings = self._compute_embeddings()

    def _compute_embeddings(self) -> List[np.ndarray]:
        def event_to_text(event: Event) -> str:
            text_parts = [
                event.name or '',
                event.short_description or '',
                ' '.join(tag.text for tag in event.tags_categories),
                ' '.join(tag.text for tag in event.tags_formats),
                ' '.join(tag.text for tag in event.tags_by_organizer),
                event.primary_venue.name or '',
                event.primary_venue.address.region or '',
                event.language or ''
            ]
            return ' '.join(filter(bool, text_parts))

        return self.model.encode([event_to_text(event) for event in self.events])

    def query_events(self, query: str, top_k: int = 5) -> List[Event]:
        # query_embedding = self.model.encode(query).reshape(1, -1)
        # similarities = cosine_similarity(query_embedding, self.event_embeddings)[0]
        # top_indices = similarities.argsort()[-top_k:][::-1]
        # return [self.events[idx] for idx in top_indices]
        query_embedding = self.model.encode(query).reshape(1, -1)
        similarities = cosine_similarity(query_embedding, self.event_embeddings)[0]

        top_indices = similarities.argsort()[-(top_k * 2):][::-1]  # Get extra events to filter duplicates

        unique_events = {}
        for idx in top_indices:
            event = self.events[idx]
            if event.id not in unique_events:
                unique_events[event.id] = event
            if len(unique_events) == top_k:
                break

        return list(unique_events.values())

# Event Evaluator
class EventEvaluator:
    def __init__(self, pipeline):
        self.pipeline = pipeline

    def evaluate_query(self, query):
        """Evaluate a single query and return results."""
        # top_events = self.pipeline.query_events(query)
        # results = []
        # for event in top_events:
        #     result = {
        #         "Event Name": event.name,
        #         "Online Event": event.is_online_event,
        #         "Start Time": event.start_datetime,
        #         "Venue Address": event.primary_venue.address.address_1,
        #         "Venue Name": event.primary_venue.name,
        #         "Description": event.short_description,
        #         "Tickets URL": event.tickets_url,
        #         "Language": event.language,
        #         "Categories": [tag.text for tag in event.tags_categories],
        #     }
        #     results.append(result)
        top_events = self.pipeline.query_events(query)
    
        results = []
        seen = set()
    
        for event in top_events:
            if event.id not in seen:  # Ensure unique events
                seen.add(event.id)
                results.append({
                    "Event Name": event.name,
                    "Online Event": event.is_online_event,
                    "Start Time": event.start_datetime,
                    "Venue Address": event.primary_venue.address.address_1,
                    "Venue Name": event.primary_venue.name,
                    "Description": event.short_description,
                    "Tickets URL": event.tickets_url,
                    "Language": event.language,
                    "Categories": [tag.text for tag in event.tags_categories],
                })
        return results

# Fetch events from Eventbrite API
client = Eventbrite()
events = client.search_events.get_results(
    region="ca--los-angeles",
    dt_start="2025-02-26",
    dt_end="2025-02-28",
    max_pages=6,
)

# Initialize pipeline and evaluator
rag_pipeline = EventbriteRAGPipeline(events)
evaluator = EventEvaluator(rag_pipeline)

# Streamlit UI
st.title("🎟️ Event Search App")

st.write("Find events based on your interests!")

query = st.text_input("🔎 Enter your search query:")
# if query:
#     results = evaluator.evaluate_query(query)

#     if results:
#         df = pd.DataFrame(results)
#         st.dataframe(df)  # Display results as a formatted table
#     else:
#         st.warning("No results found.")
if query:
    print(f"🔍 Processing query: {query}")  # Debugging query input
    results = evaluator.evaluate_query(query)

    if results:
        df = pd.DataFrame(results)
        st.dataframe(df)  # Display results as a formatted table
    else:
        st.warning("No results found.")