import gradio as gr
import torch
import pandas as pd
import numpy as np
from torch_geometric.data import Data
from torch_geometric.nn import GATConv
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity

# Define the GATConv model architecture
class ModeratelySimplifiedGATConvModel(torch.nn.Module):
    def __init__(self, in_channels, hidden_channels, out_channels):
        super().__init__()
        self.conv1 = GATConv(in_channels, hidden_channels, heads=2)
        self.dropout1 = torch.nn.Dropout(0.45)
        self.conv2 = GATConv(hidden_channels * 2, out_channels, heads=1)

    def forward(self, x, edge_index, edge_attr=None):
        x = self.conv1(x, edge_index, edge_attr)
        x = torch.relu(x)
        x = dropout1(x)
        x = self.conv2(x, edge_index, edge_attr)
        return x

# Load the dataset and the GATConv model
data = torch.load("graph_data.pt", map_location=torch.device("cpu"))

# Load the BERT-based sentence transformer model
model_bert = SentenceTransformer("all-mpnet-base-v2")

# Ensure the DataFrame is loaded properly
try:
    df = pd.read_json("combined_data.json.gz", orient='records', lines=True, compression='gzip')
except Exception as e:
    print(f"Error reading JSON file: {e}")

# Generate GNN-based embeddings
with torch.no_grad():
    all_video_embeddings = gatconv_model(data.x, data.edge_index, data.edge_attr).cpu()

# Function to find the most similar video and recommend the top 10 based on GNN embeddings
def get_similar_and_recommend(input_text):
    # Find the most similar video based on input text
    embeddings_matrix = np.array(df["embeddings"].tolist())
    input_embedding = model_bert.encode([input_text])[0]
    similarities = cosine_similarity([input_embedding], embeddings_matrix)[0]

    # Modify the similarity scores based on user input
    user_keywords = input_text.split()  # Create a list of keywords from user input
    weight = 1.0  # Initial weight factor

    for keyword in user_keywords:
        if keyword.lower() in df["title"].str.lower().tolist():  # Check if the keyword is in any title
            weight += 0.1  # Increase weight for matching keyword

    weighted_similarities = similarities * weight  # Apply the weight to the similarity score

    most_similar_index = np.argmax(weighted_similarities)  # Use weighted scores to find most similar

    # Get all features of the most similar video
    most_similar_video_features = df.iloc[most_similar_index].to_dict()

    # Recommend the top 10 videos based on GNN embeddings and dot product
    def recommend_next_10_videos(given_video_index, all_video_embeddings):
        dot_products = [
            torch.dot(all_video_embeddings[given_video_index], all_video_embeddings[i])
            for i in range(all_video_embeddings.shape[0])
        ]
        dot_products[given_video_index] = -float("inf")

        top_10_indices = np.argsort(dot_products)[::-1][:10]
        return [df.iloc[idx].to_dict() for idx in top_10_indices]

    top_10_recommended_videos_features = recommend_next_10_videos(most_similar_index, all_video_embeddings)

    # Exclude unwanted features for recommended videos
    for recommended_video in top_10_recommended_videos_features:
        if "text_for_embedding" in recommended_video:
            del recommended_video["text_for_embedding"]
        if "embeddings" in recommended_video:
            del recommended_video["embeddings"]

    # Create the output JSON with all features and the search context
    output = {
        "search_context": {
            "input_text": input_text,
            "weight": weight,  # The applied weight based on user input
        },
        "most_similar_video": most_similar_video_features,
        "top_10_recommended_videos": top_10_recommended_videos_features,
    }

    return output

# Update the Gradio interface to output JSON with weighted recommendations
interface = gr.Interface(
    fn=get_similar_and_recommend,
    inputs=gr.Textbox(label="Enter Text to Find Most Similar Video"),
    outputs=gr.JSON(),
    title="Video Recommendation System with GNN-based Recommendations",
    description="Enter text to find the most similar video and get top 10 recommended videos with search context and user-influenced weight factor.",
)

interface.launch()