Parses a markdown table into a CSV file

extract_table.py

#!/usr/bin/env -S uv run --script
# Extracts data from the first markdown table found in the input file,
# converting it into CSV
# /// script
# dependencies = [
#   "marko~=2.1"
# ]
# ///
import csv
import sys

from marko.inline import CodeSpan, RawText
from marko.ext.gfm import gfm
from marko.ext.gfm.elements import TableCell, TableRow, Table


def extract_text(e) -> str:
    if isinstance(e, TableCell):
        assert len(e.children) == 1, e
        return extract_text(e.children[0])
    elif isinstance(e, (CodeSpan, RawText)):
        res = e.children
        assert isinstance(res, str), (type(res), e)
        return res
    else:
        raise TypeError(f"Unexpected element: {e!r}")

def extract_row_text(row: TableRow) -> list[str]:
    assert isinstance(row, TableRow)
    return [extract_text(child) for child in row.children]

def convert_table(t: Table):
    assert t.children[0] == t.head
    for child in t.children:
        yield extract_row_text(child)

def main(args):
    assert len(args) == 2, "Bad argcount"
    input_path = args[0]
    output_path = args[1]
    with open(input_path, 'rt') as f:
        text = f.read()
    tree = gfm.parse(text)
    table = None
    for element in tree.children:
        if isinstance(element, Table):
            table = element
            break
    assert table is not None, "Could not find a table!"
    assert table.head == table.children[0]
    # NOTE: Intentionally fails if output already exists
    with open(output_path, 'xt') as raw_output_file:
        writer = csv.writer(raw_output_file)
        for row in convert_table(table):
            writer.writerow(row)




if __name__ == "__main__":
    main(sys.argv[1:])