adipro

adipro

One import for the whole toolbox.

import adipro as ap

df   = ap.pd.read_csv("rows.csv")
soup = ap.BeautifulSoup(ap.requests.get(url).text, "lxml")
doc  = ap.Document()
ap.load_dotenv()

No import block at the top of every script, and one pip install on a new machine instead of forty.

Install

pip install adipro

That installs everything — all 50 packages, every group below. On a machine where you only need a slice of it:

pip install adipro[scraping]   # bs4, lxml, cloudscraper, curl_cffi, ddgs, ...
pip install adipro[llm]        # openai, groq, anthropic, langgraph, tiktoken, ...
pip install adipro[docs]       # python-docx, python-pptx, openpyxl, pdf
pip install adipro[browser]    # playwright, selenium
pip install adipro[media]      # gTTS, edge-tts, pillow, lameenc
pip install adipro[data]       # gspread, pyodbc
pip install adipro[cloud]      # boto3
pip install adipro[app]        # fastapi, streamlit, gradio, pydantic, ...

Development install:

git clone <this repo> && cd adipro
pip install -e ".[dev]"
pytest

Nothing loads until you touch it

import adipro takes about 4 ms whether you have 5 packages installed or 50. Aliases resolve on first attribute access and are cached afterwards, so ap.pd imports pandas once and every later ap.pd is a dict lookup.

Two things follow from that:

• A missing package only breaks the line that needs it. If playwright failed to build on this machine, ap.pd still works fine.

• You get told how to fix it, not a bare traceback:

  MissingPackage: adipro.sync_playwright needs 'playwright', which is not installed.
  
      pip install adipro[browser]   # the whole browser group
      pip install playwright        # just this one
  

Finding things

ap.search("pdf")     # {'PdfReader': 'pypdf:PdfReader', 'open_pdf': ..., ...}
ap.doctor()          # what is installed, what is missing, how to fix it
ap.loaded()          # aliases resolved so far this process
dir(ap)              # every alias — tab-completion works in Jupyter

From the shell:

python -m adipro          # health report
python -m adipro pdf      # search aliases

Import styles

from adipro import * is deliberately limited to the stdlib and core aliases — the ones a base install guarantees are present. Star-importing the heavy groups would have to load all of them eagerly and would die if any single one were missing, so __all__ leaves them out. Reach for them through ap..

What is in it

Run python -m adipro for the live list with install status.

Adding a package

Two edits, in this order:

1. Add it to the right group in src/adipro/_registry.py — its import root and pip name under packages, and the aliases you want under aliases.
2. Add the pip name to that extra and to dependencies and all in pyproject.toml.

tests/test_adipro.py::test_registry_matches_pyproject fails if you do one and forget the other, and test_every_installed_alias_resolves fails if an alias points at a module path or attribute that does not exist.

Notes

• Versions are loose on purpose. Only a few floors are pinned. Hard pins here would fight the pins in projects you install alongside this, so installing adipro will not drag your other venvs sideways.
• google-generativeai is end-of-life upstream. It still works and is still in the llm group because it is what is installed, but Google has stopped updating it in favour of google-genai (from google import genai). Worth migrating when convenient; the alias is ap.genai.
• Two aliases are renamed to avoid a clash. readability-lxml's Document is ap.ReadabilityDoc, because python-docx already owns ap.Document.

Licence

MIT