duckeye

duckeye for the terminal

…read anything that quacks like a doc, however you want: webpages, data, docs, code & more.

Read documents in your terminal: Markdown, HTML, PDF, DOCX, EPUB, LaTeX, Jupyter notebooks, source code and an offline Wikipedia. Each format is a DuckDB extension. duckeye is one script that needs only DuckDB, with pandoc for a few gaps.

Install

curl -fsSL https://raw.githubusercontent.com/teaguesterling/duckeye/main/install.sh | bash

The script puts duckeye in ~/.local/bin, adds the short names de, dep (paged) and der (raw table), and installs the DuckDB extensions. duckeye --update keeps everything current. Source and full docs: github.com/teaguesterling/duckeye (MIT licence).

Read what you can't cat

duckeye proposal.docx           # a Word document, in the terminal
duckeye -T book.epub            # the outline of a book
duckeye -S 'Results' paper.tex  # one section of a LaTeX paper
duckeye -P 1-5 manual.pdf       # pages 1 to 5 of a PDF
duckeye analysis.ipynb          # notebook prose and code, no Jupyter

Code, by its structure

duckeye -T server.go               # outline classes, functions and methods
duckeye -S handle_request server.go # one function, with its body
duckeye -Q .fn#foo test.py          # select code by CSS selector (experimental)

Code is read by sitting_duck as a syntax tree, in more than 20 languages.

Data files are tables

duckeye events.parquet
duckeye -w "level = 'ERROR'" events.parquet   # filter rows
duckeye spreadsheet.xlsx

Parquet, CSV, TSV, JSON, YAML, TOML, XLSX, zip archives and a git history open as tables.

And more

The same engine as the app

duckeye for the terminal and duckeye for Android read with the same DuckDB extensions. The terminal tool reads more today, such as PDF, ZIM archives and code syntax trees. Read about the approach.