Convert Markdown tables to CSV

mdtable2csv.sh

#!/usr/bin/env bash
set -euo pipefail

# mdtable2csv: Convert Markdown tables to CSV.
# Handles alignment rows, trims cells, escapes CSV safely, and respects escaped pipes (\|).

mdtable2csv() {
  local in out

  if [[ $# -lt 1 || $# -gt 2 ]]; then
    echo "Usage: mdtable2csv INPUT.md [OUTPUT.csv]" >&2
    exit 2
  fi

  in="$1"
  out="${2:-${in%.*}.csv}"

  if [[ ! -f "$in" ]]; then
    echo "ERROR: input file not found: $in" >&2
    exit 1
  fi

  awk '
  function trim(s) {
    sub(/^[[:space:]]+/, "", s)
    sub(/[[:space:]]+$/, "", s)
    return s
  }

  function csv_escape(s,   t) {
    t = s
    gsub(/"/, "\"\"", t)
    if (t ~ /[,\n"]/)
      return "\"" t "\""
    return t
  }

  function is_separator_row(raw,   s) {
    s = raw
    s = trim(s)
    sub(/^\|/, "", s)
    sub(/\|$/, "", s)
    s = trim(s)

    if (s !~ /-/) return 0
    gsub(/[[:space:]\|\-:]/, "", s)
    return (s == "")
  }

  BEGIN {
    ESC = "\034"
  }

  # Optional stricter filter (recommended): uncomment to only process lines that start with a pipe
  # $0 !~ /^[[:space:]]*\|/ { next }

  /\|/ {
    raw = $0
    if (is_separator_row(raw)) next

    line = raw
    gsub(/\\\|/, ESC, line)
    if (line !~ /\|/) next

    line = trim(line)
    sub(/^\|/, "", line)
    sub(/\|$/, "", line)

    n = split(line, a, /\|/)
    row = ""

    for (i = 1; i <= n; i++) {
      field = a[i]
      gsub(ESC, "|", field)
      field = trim(field)
      field = csv_escape(field)
      row = row (i == 1 ? "" : ",") field
    }

    print row
  }
  ' "$in" > "$out"

  echo "Wrote: $out"
}

mdtable2csv "$@"