Skip to content

Latest commit

 

History

History
161 lines (109 loc) · 5.07 KB

README.md

File metadata and controls

161 lines (109 loc) · 5.07 KB

geopolitex-analyzer

PyPI version License: MIT Downloads LinkedIn

geopolitex-analyzer is a lightweight Python package that turns short news headlines or text snippets about geopolitics, defense, and international competition into structured, LLM‑generated summaries. The output always follows a predefined pattern (subject, comparative context, implications) and is validated with robust regex matching to guarantee consistency and accuracy.

Features

  • One‑line summarisation for geopolitical and defense‑related texts.
  • Pattern‑based validation using llmatch to ensure the LLM output conforms to the expected format.
  • Pluggable LLM backend – defaults to ChatLLM7 but works with any LangChain‑compatible chat model (OpenAI, Anthropic, Google, etc.).
  • Simple API: just pass the raw text and get a list of structured strings back.

Installation

pip install geopolitex_analyzer

Quick Start

from geopolitex_analyzer import geopolitex_analyzer

# Simple usage with the default ChatLLM7 (requires an API key)
summary = geopolitex_analyzer(
    user_input="China launches a new hypersonic missile system, threatening regional stability."
)

print(summary)
# -> ['Subject: China...', 'Comparative Context: ...', 'Implications: ...']

Using a custom LangChain LLM

You can provide any LangChain chat model that implements BaseChatModel. This is useful if you prefer OpenAI, Anthropic, Google Gemini, etc.

OpenAI example

from langchain_openai import ChatOpenAI
from geopolitex_analyzer import geopolitex_analyzer

my_llm = ChatOpenAI(model="gpt-4o-mini")
summary = geopolitex_analyzer(
    user_input="Russia expands its naval presence in the Arctic.",
    llm=my_llm,
)

print(summary)

Anthropic example

from langchain_anthropic import ChatAnthropic
from geopolitex_analyzer import geopolitex_analyzer

my_llm = ChatAnthropic(model="claude-3-haiku-20240307")
summary = geopolitex_analyzer(
    user_input="India announces a new defense budget focusing on AI‑driven weaponry.",
    llm=my_llm,
)

print(summary)

Google Gemini example

from langchain_google_genai import ChatGoogleGenerativeAI
from geopolitex_analyzer import geopolitex_analyzer

my_llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash")
summary = geopolitex_analyzer(
    user_input="EU sanctions target new Russian aerospace firms.",
    llm=my_llm,
)

print(summary)

Supplying an API key for the default LLM7 backend

If you rely on the built‑in ChatLLM7 model, provide an API key either via the LLM7_API_KEY environment variable or directly:

from geopolitex_analyzer import geopolitex_analyzer

summary = geopolitex_analyzer(
    user_input="Turkey acquires advanced drone technology from South Korea.",
    api_key="your_llm7_api_key_here",
)

print(summary)

A free API key can be obtained by registering at https://token.llm7.io/. The free tier’s rate limits are sufficient for most typical use cases.

Function Reference

geopolitex_analyzer(
    user_input: str,
    api_key: Optional[str] = None,
    llm: Optional[BaseChatModel] = None,
) -> List[str]
Parameter Type Description
user_input str Raw headline or short text describing a geopolitical/defense topic.
api_key Optional[str] API key for the default ChatLLM7. If omitted, the function looks for the LLM7_API_KEY env var.
llm Optional[BaseChatModel] Any LangChain chat model. If provided, it overrides the default ChatLLM7.

Return value: A list of strings that match the defined output pattern (e.g., subject, comparative context, implications). If the LLM response does not satisfy the pattern, a RuntimeError is raised.

Dependencies

  • llmatch_messages – pattern‑matching helper used to enforce output format.
  • langchain-core – core interfaces for LLMs (BaseChatModel, message types).
  • langchain-llm7 – wrapper for the LLM7 service (automatically installed).
  • geopolitex-analyzer’s own prompts.py (contains the system / human prompts and the regex pattern).

Contributing & Support

Author

Eugene Evstafev
Email: hi@euegne.plus
GitHub: chigwell

License

MIT License – see the LICENSE file in the repository for details.

Happy analysing!