bookmark-analyzer

bookmark-analyzer is a lightweight Python package that turns unstructured bookmark titles or user‑written descriptions into structured tags, categories, or summaries. It leverages a LLM (by default ChatLLM7) to understand the meaning of a bookmark and return machine‑readable metadata, making it far easier to search, filter, and rediscover saved content.

🚀 Installation

pip install bookmark_analyzer

📖 Quick Start

from bookmark_analyzer import bookmark_analyzer

# Simple usage – the default ChatLLM7 will be used
tags = bookmark_analyzer(
    user_input="How to bake a perfect sourdough loaf – an in‑depth guide with photos"
)

print(tags)   # e.g. ['baking', 'sourdough', 'cooking', 'guide']

Using Your Own LLM

If you prefer another LangChain‑compatible model (OpenAI, Anthropic, Google, …) just pass the instance:

OpenAI

from langchain_openai import ChatOpenAI
from bookmark_analyzer import bookmark_analyzer

llm = ChatOpenAI(model="gpt-4o-mini")
tags = bookmark_analyzer(
    user_input="Why the 2024 election matters for climate policy",
    llm=llm,
)
print(tags)

Anthropic

from langchain_anthropic import ChatAnthropic
from bookmark_analyzer import bookmark_analyzer

llm = ChatAnthropic(model="claude-3-haiku-20240307")
tags = bookmark_analyzer(
    user_input="Best practices for micro‑frontends in a large SaaS product",
    llm=llm,
)
print(tags)

Google Gemini

from langchain_google_genai import ChatGoogleGenerativeAI
from bookmark_analyzer import bookmark_analyzer

llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash")
tags = bookmark_analyzer(
    user_input="Top 10 must‑read sci‑fi novels of the 21st century",
    llm=llm,
)
print(tags)

🛠️ Parameters

🔑 API Key for the Default Model

• ChatLLM7 is the out‑of‑the‑box model used by bookmark-analyzer.
• Free tier rate limits are sufficient for typical personal use.
• To obtain a free key, register at https://token.llm7.io/.
• You can provide the key either:
  • Via the environment variable LLM7_API_KEY, or
  • Directly when calling the function:

tags = bookmark_analyzer(
    user_input="My favorite podcasts about AI",
    api_key="sk-xxxxxx"
)

📚 How It Works

bookmark_analyzer builds a prompt (see the internal prompts module) and sends it to the LLM together with a regular‑expression pattern that describes the expected output format. The response is validated against that pattern and the extracted tags are returned as a list of strings.

🐞 Issues & Contributions

If you encounter any bugs or have feature requests, please open an issue:

https://github.com/chigwell/bookmark-analyzer/issues

Feel free to fork the repository, submit pull requests, or improve documentation.

✍️ Author

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

Happy bookmarking! 🎉