text-summarizer-gi 0.1.0

LLM based context aware text summarizer

azure-llm-summarizer

A lightweight Python library for summarizing text using Azure OpenAI, with built-in token counting for both input and output — no extra dependencies beyond openai.

Installation

pip install azure-llm-summarizer

Quick Start

from llm_summarizer import AzureSummarizer

summarizer = AzureSummarizer(
    api_key="<your-azure-api-key>",
    azure_endpoint="https://<your-resource>.openai.azure.com",
    deployment_name="gpt-4o-mini",   # your deployed model name
)

result = summarizer.summarize(
    text="Your long document or passage goes here...",
    summary_type="medium",    # "short" | "medium" | "detailed"
    tone="neutral",           # "neutral" | "formal" | "casual"
    focus_area="general",     # "general" | "technical insights" | "financial" ...
    output_format="text",     # "text" | "bullets" | "json"
)

print(result.summary)          # the summary
print(result.input_tokens)     # token count of the original passage
print(result.output_tokens)    # token count of the summary
print(result)                  # summary + token counts in one print

Example Output

Generative AI (GenAI) refers to AI systems that create new content by learning
from large datasets, using architectures like LLMs and diffusion models...

[Tokens — Input: 312 | Summary: 47]

API Reference

AzureSummarizer(azure_endpoint, api_key, api_version, deployment_name)

Parameter	Type	Default	Description
azure_endpoint	str	env AZURE_OPENAI_ENDPOINT	Azure resource endpoint
api_key	str	env AZURE_OPENAI_API_KEY	Azure API key
api_version	str	"2024-02-15-preview"	API version
deployment_name	str	"gi-local-gpt-5-mini"	Deployed model name

.summarize(text, summary_type, tone, focus_area, output_format) → SummaryResult

Parameter	Options	Default
summary_type	"short", "medium", "detailed"	"medium"
tone	"neutral", "formal", "casual"	"neutral"
focus_area	any string	"general"
output_format	"text", "bullets", "json"	"text"

SummaryResult

Attribute	Type	Description
summary	str	The generated summary
input_tokens	int	Estimated tokens in the original text
output_tokens	int	Estimated tokens in the summary

Standalone token counter

from llm_summarizer import count_tokens

count_tokens("Hello, world!")  # → 4

Environment Variables

You can skip passing credentials directly and use env vars instead:

export AZURE_OPENAI_ENDPOINT="https://<resource>.openai.azure.com"
export AZURE_OPENAI_API_KEY="<your-key>"

Publishing to PyPI

pip install build twine
python -m build
twine upload dist/*

License

MIT