README - Llama Index

IMDB MOVIE REVIEWS LOADER

bash

pip install llama-index-readers-imdb-review

This loader fetches all the reviews of a movie or a TV-series from IMDB official site. This loader is working on Windows machine and it requires further debug on Linux. Fixes are on the way

Install the required dependencies

pip install -r requirements.txt

The IMDB downloader takes in two attributes

movie_name_year: The name of the movie or series and year
webdriver_engine: To use edge, google or gecko (mozilla) webdriver
generate_csv: Whether to generate csv file
multithreading: whether to use multithreading or not

Usage

python

from llama_index.readers.imdb_review import IMDBReviews

loader = IMDBReviews(
    movie_name_year="The Social Network 2010", webdriver_engine="edge"
)
docs = loader.load_data()

The metadata has the following information

date of the review (date)
title of the review (title)
rating of the review (rating)
link of the review (link)
whether the review is spoiler or not (spoiler)
number of people found the review helpful (found_helpful)
total number of votes (total)

It will download the files inside the folder movie_reviews with the filename as the movie name

EXAMPLES

This loader can be used with both Langchain and LlamaIndex.

LlamaIndex

python

from llama_index.core import VectorStoreIndex, download_loader
from llama_index.core import VectorStoreIndex

from llama_index.readers.imdb_review import IMDBReviews

loader = IMDBReviewsloader(
    movie_name_year="The Social Network 2010",
    webdriver_engine="edge",
    generate_csv=False,
    multithreading=False,
)
docs = loader.load_data()

index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()

response = query_engine.query(
    "What did the movie say about Mark Zuckerberg?",
)
print(response)

Langchain

python

from langchain.llms import OpenAI
from langchain.agents.agent_toolkits.pandas import (
    create_pandas_dataframe_agent,
)
from langchain.agents import Tool
from langchain.agents import initialize_agent
from langchain.chat_models import ChatOpenAI

from llama_index.readers.imdb_review import IMDBReviews

loader = IMDBReviewsloader(
    movie_name_year="The Social Network 2010",
    webdriver_engine="edge",
    generate_csv=False,
    multithreading=False,
)
docs = loader.load_data()
tools = [
    Tool(
        name="LlamaIndex",
        func=lambda q: str(index.as_query_engine().query(q)),
        description="useful for when you want to answer questions about the movies and their reviews. The input to this tool should be a complete english sentence.",
        return_direct=True,
    ),
]
llm = ChatOpenAI(temperature=0)
agent = initialize_agent(tools, llm, agent="conversational-react-description")
agent.run("What did the movie say about Mark Zuckerberg?")