FMN_bot/src/imdb_datasets_worker.py

import sqlite3
import gzip

conn = sqlite3.connect("imdb_titles.sqlite")
c = conn.cursor()

c.execute(
    '''CREATE TABLE IF NOT EXISTS titles(tt_id INTEGER UNIQUE, type VARCHAR (50), original_name VARCHAR (500) DEFAULT NULL, ru_name VARCHAR (500) DEFAULT NULL, year INTEGER DEFAULT NULL)''')
c.execute("PRAGMA synchronous = OFF")
c.execute("PRAGMA optimize")
conn.commit()


def convert_tsv_to_db(title_basics_tsv):
    '''Конвертирование основного датасета в sqlite базу, выполняется долго (~5 минут)'''
    with gzip.open(title_basics_tsv, mode='rt') as file:
        write_dataset = []
        counter = 0
        for line in file:
            line = line.split("\t")
            try:
                tt_id = int(line[0].split("tt")[1])
                tt_type = line[1]
                original_name = line[3]
                ru_name = None
                year = line[5]

                if tt_type not in ("movie", "video"):
                    original_name = None
                    year = "\\N"
                else:
                    print(tt_id, tt_type, original_name, ru_name, year)

                if year == "\\N":
                    year = None
                else:
                    year = int(year)
                write_dataset.append((tt_id, tt_type, original_name, ru_name, year))
                counter += 1
                if counter >= 1000:
                    c.executemany("INSERT OR REPLACE INTO titles(tt_id, type, original_name, ru_name, year) VALUES (?, ?, ?, ?, ?)", write_dataset)
                    write_dataset = []
                    counter = 0
            except Exception as E:
                print(E)
                pass
        conn.commit()

def extract_ru_locale_from_tsv(title_akas_tsv):
    '''Конвертирование датасета с локализованными названиями и последующее добавление в базу'''
    with gzip.open(title_akas_tsv, mode='rt') as file:
        ru_name_writer = []
        for line in file:
            line = line.split("\t")
            try:
                tt_region = line[3]
                if tt_region != "RU": 
                    continue

                tt_id = int(line[0].split("tt")[1])
                tt_type = c.execute(f"SELECT type FROM titles WHERE tt_id={tt_id}").fetchone()[0]
                if tt_type not in ("movie", "video"):
                    continue
                ru_name = line[2]
                print(ru_name, tt_type)
                ru_name_writer.append((ru_name, tt_id))

            except Exception as E:
                print(E)
                pass
        
        c.executemany("UPDATE titles SET ru_name = ? WHERE tt_id = ?", ru_name_writer)
        conn.commit()

def convert_datasets_to_db():
    print("Converting tsv dataset to sqlite...")
    convert_tsv_to_db("title.basics.tsv.gz")
    print("Unpack ru locale...")
    extract_ru_locale_from_tsv("title.akas.tsv.gz")


def get_title_by_id(films_ids=list):
    tt_list = []
    for i in films_ids:
        tt_film = c.execute(f"SELECT * FROM titles WHERE tt_id={i}").fetchone()
        tt_list.append(tt_film)
    return tt_list