Headline Parser

A package to process news headlines or short text snippets, extracting key entities, topics, and sentiment, and outputting structured summaries in a consistent format.

Installation

pip install headline_parser

Overview

This package uses a large language model (LLM) to process input text and generate structured summaries of current events. It's suitable for:

• News aggregation
• Content moderation analysis
• Tracking public discourse on free speech and regulatory issues
• Applications not requiring raw media files

Example usage

from headline_parser import headline_parser

response = headline_parser(
    user_input="Us Democrats pass a health care reform bill.",
)
print(response)

The headline_parser function can take in optional arguments:

• llm: an instance of langchain_core.language_models.BaseChatModel, which defaults to langchain_llm7.ChatLLM7
• api_key: an optional API key for LLM7, which defaults to the LLM7_API_KEY environment variable

You can pass your own LLM instance from other providers:

# Using OpenAI
from langchain_openai import ChatOpenAI
from headline_parser import headline_parser

llm = ChatOpenAI()
response = headline_parser("Your news headline", llm=llm)
print(response)

# Using Anthropic
from langchain_anthropic import ChatAnthropic
from headline_parser import headline_parser

llm = ChatAnthropic()
response = headline_parser("Your news headline", llm=llm)
print(response)

# Using Google
from langchain_google_genai import ChatGoogleGenerativeAI
from headline_parser import headline_parser

llm = ChatGoogleGenerativeAI()
response = headline_parser("Your news headline", llm=llm)
print(response)

For higher rate limits with LLM7, you can obtain a free API key by registering at https://token.llm7.io/ and pass it either as an environment variable or directly to the function:

from headline_parser import headline_parser

# Via environment variable
os.environ["LLM7_API_KEY"] = "your_api_key"

# Or directly
response = headline_parser("Your news headline", api_key="your_api_key")
print(response)

Contributing

Found an issue or have a feature request? Please report them on GitHub.

Author

Eugene Evstafev (hi@eugene.plus)