rwxrob-dot/scripts/fmttable

#!/usr/bin/env bash

# Originally written by https://github.com/deyloop/dot

# Formats Markdown tables so that they are readable in text form
# Parses the table in order to print the formatted form.
# 
# Supposed to be used as an UNIX filter, where the unformatted table
# should be sent to `stdin` and the formatted output will be sent to
# `stdout`. 
# If a parsing error occures, prints the unformatted table unchanged,
# and exits with code 42
#
# In order to debug, look at `__fail`

# Does all the parsing in pure bash, in a single process. Printing uses
# a few subshells as that was convenient.

# TODO: support escaping pipe characters
# TODO: skipping to just tables when whole markdown document is passed
# TODO: allow filename to be passed as argument


#-------------------------------------------------------------------------
# The Parsing Grammar used:
#
# table       -> newline headingrow delimrow tablebody
#
# headingrow  -> pipe? headings pipe? newline
# headings    -> content pipe headingsB
# headingsB   -> content / content pipe headingsB
# content     -> [^pipe]*
#
# delimrow    -> pipe? delims pipe? newline
# delims      -> delim pipe delimsB
# delimsB     -> delim / delim pipe delimsB
# delim       -> noalign / leftalign / centeralign / rightalign
# noalign     -> dash dash dashes
# leftalign   -> colon dash dashes
# centeralign -> colon dashes colon
# rightalign  -> dash dashes colon
#
# tablebody   -> nothing / tablerow tablebody
# tablerow    -> pipe? body pipe? newline
# body        -> content pipe bodyB
# bodyB       -> content / content pipe bodyB
#
# dashes      -> dash / dash dashes
# pipe        -> '|'
# dash        -> '-'
# colon       -> ':'
#-------------------------------------------------------------------------


fmttable() {
  local in=/dev/stdin

  local buf
  buf="$(<"$in")"
  buf+=$'\n'

  local i=0
  local c

  local -a contents
  local col=0
  local row=0
  local totalcols=0
  local -a colwidths
  local -a colaligns

  # parsing
  _parse_table

  # print the headings
  printf "|"
  for (( x=0; x<totalcols; x++ )); do
    local w="${colwidths[$x]}"
    printf "%-${w}s|" "${contents[$x]}"
  done
  printf "\n"

  # print the delims
  printf "|"
  for (( x=0; x<${totalcols}; x++ )); do
    local con
    case ${colaligns[$x]} in
      left) 
        con=":$(for ((i=0; i<$((${colwidths[$x]}-1)); i++)); do 
          printf "-"; 
        done)"
        ;;
      right)
        con="$(for ((i=0; i<$((${colwidths[$x]}-1)); i++)); do 
          printf "-"; 
        done):"
        ;;
      center)
        con=":$(for ((i=0; i<$((${colwidths[$x]}-2)); i++)); do 
          printf "-"; 
        done):"
        ;;
      none) 
        con="$(for ((i=0; i<${colwidths[$x]}; i++)); do 
          printf "-"; 
        done)"
        ;;
    esac
    local w="${colwidths[$x]}"
    printf "%-${w}s|" "$con"
  done
  printf "\n"

  # print the body
  for (( y=2; y<row; y++ )); do
    printf "|"
    for (( x=0; x<totalcols; x++ )); do
      local w="${colwidths[$x]}"
      local a=""
      [[ ${colaligns[$x]} != right ]] && a="-" 
      printf "%${a}${w}s|" "${contents[$y*$totalcols+$x]}"
    done
    printf "\n"
  done
}

__fail(){
  printf "%s" "$buf"
  echo
  echo "$row:$col - $1"
  exit 42
}

_parse_table() {
  _parse_headingrow
  _parse_delimrow
  _parse_tablebody
}


_parse_headingrow(){
  _nextchar
  [[ $c != '|' ]] && _putback

  _parse_headings

  _nextchar
  [[ $c != '|' ]] && _putback

  _parse_newline

  totalcols=$col
  row=$((row+1))
}

_advance_col() {
  shopt -s extglob

  local trimed="${contents[$row*$totalcols+$col]}"
  trimed="${trimed##+([[:space:]])}"
  trimed="${trimed%%+([[:space:]])}"
  contents[$row*$totalcols+$col]="$trimed"

  local colw="${#trimed}"
  local w="${colwidths[$col]}"
  if [[ $colw -gt $w ]]; then colwidths[$col]="$colw"; fi

  col=$((col+1))
}

_parse_headings() {

  _parse_content
  _advance_col

  _parse_pipe
  _parse_headingsB

}

_parse_content() {
  _nextchar
  if [[ $c = '|' || $c = $'\n' ]]; then
    _putback
  else 
    contents[$row*$totalcols+$col]+="$c"

    _parse_content
  fi
}

_parse_pipe() {
  _nextchar
  [[ $c != '|' ]] && __fail "expected pipe, found '$c'"
}

_parse_headingsB() {
  _parse_content
  _advance_col

  _nextchar
  if [[ $c = '|' ]]; then
    _nextchar
    if [[ $c = $'\n' ]]; then
      _putback
      _putback
    else
      _putback
      _putback
      _parse_pipe
      _parse_headingsB
    fi
  else
    if [[ $c = $'\n' ]]; then
      _putback
    fi
  fi
   
}

_parse_newline() {
  _nextchar
  [[ $c != $'\n' ]] && __fail "expected newline, found '$c'"
}

_parse_delimrow(){
  col=0
  _nextchar
  [[ $c != '|' ]] && _putback

  _parse_delims

  _nextchar
  [[ $c != '|' ]] && _putback

  _parse_newline

  row=$((row+1))
}

_parse_delims() {
  _parse_delim
  col=$((col+1))

  _parse_pipe

  _parse_delimsB
}

_parse_delimsB() {
  _parse_delim
  col=$((col+1))

  _nextchar
  if [[ $c = '|' ]]; then
    _nextchar
    if [[ $c = $'\n' ]]; then
      _putback
      _putback
    else
      _putback
      _putback
      _parse_pipe
      _parse_delimsB
    fi
  else
    if [[ $c = $'\n' ]]; then
      _putback
    fi
  fi
}

_parse_delim() {
  local remainingbuf=${buf:$i}
  local lookahead=${remainingbuf%%\|*}

  if [[ $lookahead =~ ^[[:space:]]*:-+:[[:space:]]*$ ]]; then
    colaligns[$col]="center"
    _parse_centeralign
    return
  fi
  if [[ $lookahead =~ ^[[:space:]]*:--+[[:space:]]*$ ]]; then
    colaligns[$col]="left"
    _parse_leftalign
    return
  fi
  if [[ $lookahead =~ ^[[:space:]]*---+[[:space:]]*$ ]]; then
    colaligns[$col]="none"
    _parse_noalign
    return
  fi
  if [[ $lookahead =~ ^[[:space:]]*--+:[[:space:]]*$ ]]; then
    colaligns[$col]="right"
    _parse_rightalign
    return
  fi
}

_parse_skipspaces() {
  _nextchar
  if [[ $c = ' ' ]]; then
    _parse_skipspaces 
  else
    _putback
  fi
}

_parse_noalign() {
  _parse_skipspaces
  _parse_dash
  _parse_dash
  _parse_dashes
  _parse_skipspaces
}

_parse_leftalign() {
  _parse_skipspaces
  _parse_colon
  _parse_dash
  _parse_dashes
  _parse_skipspaces
}

_parse_centeralign(){
  _parse_skipspaces
  _parse_colon
  _parse_dashes
  _parse_colon
  _parse_skipspaces
}

_parse_rightalign() {
  _parse_skipspaces
  _parse_dash
  _parse_dashes
  _parse_colon
  _parse_skipspaces
}

_parse_dash() {
  _nextchar
  [[ $c != '-' ]] && __fail "expected dash, found '$c'"
}

_parse_dashes() {
  _parse_dash

  _nextchar
  if [[ $c = '-' ]]; then
    _putback
    _parse_dashes
  else
    _putback
  fi
}

_parse_colon() {
  _nextchar
  [[ $c != ':' ]] && __fail "expected colon, fount '$c'"
}

_parse_tablebody(){
  local bi=$((${#buf}-1))
  if [[ $i -lt $bi ]]; then
    _parse_tablerow
    _parse_tablebody
  else
    return
  fi
}

_parse_tablerow() {
  col=0
  _nextchar
  [[ $c != '|' ]] && _putback

  _parse_body

  _nextchar
  [[ $c != '|' ]] && _putback

  row=$((row+1))
  _parse_newline
}

_parse_body() {
  _parse_content
  _advance_col

  _parse_pipe
  _parse_bodyB
}

_parse_bodyB() {
  _parse_content
  _advance_col

  _nextchar
  if [[ $c = '|' ]]; then
    _nextchar
    if [[ $c = $'\n' ]]; then
      _putback
      _putback
    else
      _putback
      _putback
      _parse_pipe
      _parse_bodyB
    fi
  else
    if [[ $c = $'\n' ]]; then
      _putback
    fi
  fi
}

_nextchar() {
  if [[ $i -lt "${#buf}" ]]; then
    c="${buf:$i:1}"
    i=$((i+1))
  else
    __fail "Unexpectedly reached EOF"
  fi
}

_putback() {
  i=$((i-1))
}

fmttable