from course_project import data_import
from course_project import embedding
from course_project import data_persistence
from course_project import gemma_llm
from course_project import constants
import pandas as pd
import movieposters as mp

DATA_FOLDER_LOCATION = "course_project/data/"
OLLAMA_EMBEDDING_BK_LOCATION = DATA_FOLDER_LOCATION + "ollama_embedding_bk_location.pkl"

def get_movie_info(id):
    movies_data = data_persistence.get_transformed_db()
    movie = movies_data.iloc[id, :]
    movie = (movie[["title", "director", "genres", "cast", "overview"]])
    movie_info = f'Title: {movie["title"]}. Director(s): {movie["director"]}. Genres: {movie["genres"]}. Cast: {movie["cast"]}. Overview: {movie["overview"]}'
    return movie_info

def get_data(overwrite_db=False):
    movies_data = data_persistence.get_transformed_db()
    if not isinstance(movies_data, pd.DataFrame) or overwrite_db:
        movies_data = data_import.read_or_generate_movies_data()
        movies_data["vector"] = \
            "Title: " + movies_data["title"] \
            + ". Director(s): " + movies_data["director"].apply(data_import.format_list_of_strings) \
            + ". Genres: " + movies_data["genres"].apply(data_import.format_list_of_strings) \
            + ". Cast: " + movies_data["cast"].apply(data_import.format_list_of_strings) \
            + ". Overview: " + movies_data["overview"]

        movies_data["vector"] = embedding.encode_text_series_ollama_method(movies_data["vector"], file_backup_location=OLLAMA_EMBEDDING_BK_LOCATION, batch_size=500)
        print(movies_data["vector"])

        data_persistence.save_transformed_db(movies_data)
    return movies_data

def replace_indexes_with_titles(suggestions, movies_data):
    suggestions_titles = suggestions.copy()
    for user, movie_id in suggestions_titles.items():
        movie_title = movies_data.at[movie_id, 'title']
        suggestions_titles[user] = movie_title
    return suggestions_titles

def get_recommendations_df(suggestions, n_results, similarity=constants.COSINE_SIMILARITY, rating_weight=0.15):
    movies_data = get_data()
    recommendations_raw = embedding.find_recommendations(suggestions, movies_data["vector"], n_results=n_results, similarity=similarity, rating_weight=rating_weight)
    recommendations_indexes = [recommendation["movie_index"] for recommendation in recommendations_raw]
    columns_of_interest = ["id", "title", "director", "genres", "cast", "overview", "rating", "poster_path", "imdb_id"]
    recommendations = (movies_data.loc[recommendations_indexes])[columns_of_interest].copy()
    suggestions_considered = [reco["suggestions_considered"] for reco in recommendations_raw]
    recommendations["based_on_index"] = pd.Series(suggestions_considered, index=recommendations.index)
    suggestions_considered_titles = [replace_indexes_with_titles(suggestion, movies_data) for suggestion in suggestions_considered]
    recommendations["based_on"] = pd.Series(suggestions_considered_titles, index=recommendations.index)
    return recommendations

def get_connections(row):
    connections = {}
    for person, request in row["based_on"].items():
        connections[person] = request
    return connections

def get_movie_poster(imdb_id):
    link = mp.get_poster(id=imdb_id)
    return link

def get_recommendations(suggestions, n_results=10, similarity=constants.COSINE_SIMILARITY, rating_weight=0.15):
    recommendations = get_recommendations_df(suggestions, n_results, similarity=similarity, rating_weight=rating_weight)
    recommendations["poster_path"] = recommendations["imdb_id"].apply(get_movie_poster)
    return recommendations

def get_explanation_for_reco(suggestion_1_id, suggestion_2_id, recommendation_id):
    suggestion_1_info = get_movie_info(suggestion_1_id)
    suggestion_2_info = get_movie_info(suggestion_2_id)
    recommendation_info = get_movie_info(recommendation_id)
    return gemma_llm.find_reason_for_recommendation(suggestion_1_info, suggestion_2_info, recommendation_info)