#!/bin/sh
# txt2regexp.sh - Regular Expressions "wizard" 
#
# A T E N T I O N: only works in bash >= 2.04 
#  
# all REs for the S2_PROG arrays was taken from the PROG man page
# or missing it, from the 'mastering regular expressions' book
#
# versions tested:
# ed: GNU ed version 0.2
# egrep: egrep (GNU grep) 2.4.2
# find: GNU find version 4.1
# gawk: GNU Awk 3.0.4
# grep: grep (GNU grep) 2.4.2
# mawk: mawk 1.2
# sed: GNU sed version 3.02.80
# vim: VIM - Vi IMproved 5.7
#
# $STATUS:
#   0 begining of the regexp
#   1 defining regexp
#   12 defining type of letters
#   2 defining quantifier
#   3 end of the regexp
#
# 20001019 <aurelio@conectiva.com.br> ** 1st version
# 20001026 <aurelio@conectiva.com.br> ++ lots of changes and tests
# 20001028 <aurelio@conectiva.com.br> ++ improvements, public release
#
# TODO i18n
# TODO capture blanks on Get* (via menu)
# TODO create [] mixing letters/numbers/blanks
# TODO EscChar - delimiters too? like / (sed, *awk)
# TODO expr, oawk, nawk, MKS awk, flex, php
# TODO s/vi/n&/

# take out from here programs you don't want to know about
# or to minimize the lines printed on the screen
#progs=(awk ed egrep emacs expect find gawk grep lex lisp mawk perl python sed tcl vi vim)
progs=(egrep emacs gawk grep perl sed vim)


set -o noglob

# here's all the text and RegExps arrays
S0_txt=('comece a casar a partir' 'do começo da linha'\
        'de qualquer lugar da linha')
 S0_re=('' '^' '')

S1_txt=('seguido de' 'apenas números' 'apenas letras' 'letras e números'\
        'um caractere qualquer' 'um caractere específico'\
        'uma lista de caracteres permitidos'\
        'uma lista de caracteres proibidos'\
        'uma cadeia de caracteres literais' 'qualquer coisa')
S1_re=( '' '[0-9]' '' '0-9' '.' '' '' '' '' '.*')

S12_txt=('tipo das letras' 'apenas letras maiúsculas'\
        'apenas letras minúsculas' 'letras maiúsculas e minúsculas')
S12_re=( '' 'A-Z' 'a-z' 'A-Za-z')

S2_txt=('quantas vezes' 'uma' 'zero ou uma (opcional)' 'zero ou mais'\
        'uma ou mais' 'exatamente N' 'até N' 'no mínimo N')
S2_sed=(   '' '' '\?' '*' '\+' '\{@\}' '\{1,@\}' '\{@,\}')
S2_ed=(    '' '' '\?' '*' '\+' '\{@\}' '\{1,@\}' '\{@,\}')
S2_grep=(  '' '' '\?' '*' '\+' '\{@\}' '\{1,@\}' '\{@,\}')
S2_vim=(   '' '' '\=' '*' '\+' '\{@}'  '\{1,@}'  '\{@,}' )
S2_egrep=( '' ''  '?' '*'  '+'  '{@}'   '{1,@}'   '{@,}' )
S2_python=('' ''  '?' '*'  '+'  '{@}'   '{1,@}'   '{@,}' )
S2_lex=(   '' ''  '?' '*'  '+'  '{@}'   '{1,@}'   '{@,}' )
S2_perl=(  '' ''  '?' '*'  '+'  '{@}'   '{1,@}'   '{@,}' )
S2_gawk=(  '' ''  '?' '*'  '+'  '{@}'   '{1,@}'   '{@,}' )
S2_mawk=(  '' ''  '?' '*'  '+'  '!!'    '!!'      '!!'   )
S2_awk=(   '' ''  '?' '*'  '+'  '!!'    '!!'      '!!'   )
S2_find=(  '' ''  '?' '*'  '+'  '!!'    '!!'      '!!'   )
S2_emacs=( '' ''  '?' '*'  '+'  '!!'    '!!'      '!!'   )
S2_lisp=(  '' ''  '?' '*'  '+'  '!!'    '!!'      '!!'   )
S2_tcl=(   '' ''  '?' '*'  '+'  '!!'    '!!'      '!!'   )
S2_expect=('' ''  '?' '*'  '+'  '!!'    '!!'      '!!'   )
S2_vi=(    '' '' '\?' '*' '\+'  '__'    '__'      '__'   )
#63# cause on table 6-1 it seems that the vi part is wrong

# mastering regular expressions:
# egrep 29 1-3
# .* 182 6-1
# grep 183 6-2
# *awk 184 6-3
# tcl 189 6-4
# emacs 194 6-7
# perl 201 7-1


# tst: \/_$[]{}()|+?^_/p
# [gm]awk = egrep
ax_txt=('detalhes' 'agrupamento' 'alternância' 'escapar normal' 'escapar lista []')
#                              \.*[]{}()|+?^$   ,=tested  space=pending
ax_ed=(    ''  '\(,\)'   '\|' '\.*[,,,,,,,,,,' ',')
ax_vim=(   ''  '\(,\)'   '\|' '\.*[,,,,,,,,,,' '\')
ax_sed=(   ''  '\(,\)'   '\|' '\.*[,,,,,,,,,,' ',')
ax_grep=(  ''  '\(,\)'   '\|' '\.*[,,,,,,,,,,' ',')
ax_find=(  ''  '\(,\)'   '\|' '\.*[,,,,,,+?,,' ',')
ax_egrep=( ''   '(,)'     '|' '\.*[,{,()|+?^$' ',')
ax_python=(''   '(,)'     '|' '\.*[,{,()|+?  ' '\')
ax_lex=(   ''   '(,)'     '|' '\.*[ { ( |+?  ' ' ')
ax_perl=(  ''   '(,)'     '|' '\.*[ { ( |+?  ' '\')
ax_gawk=(  ''   '(,)'     '|' '\.*[,,,(,|+?^$' '\')
ax_mawk=(  ''   '(,)'     '|' '\.*[,,,()|+?^$' '\')
ax_awk=(   ''   '(,)'     '|' '\.*[   (,|+?  ' '\')
ax_emacs=( ''  '\(,\)'   '\|' '\.*[      +?  ' ',')
ax_lisp=(  '' '\\(,\\)' '\\|' '\.*[      +?  ' ',')
ax_tcl=(   ''   '(,)'     '|' '\.*[   ( |+?  ' ',')
ax_expect=(''   '(,)'     '|' '\.*[   ( |+?  ' ' ')
ax_vi=(    ''  '\(,\)'   '!!' '\.*[          ' ' ')
#194# emacs: a backslash ... it is completely unspecial
#189# tcl: withing a class, a backslash is completely unspecial

# screen size/positioning issues
x_regexp=1 ; y_regexp=3
x_hist=3   ; y_hist=$((y_regexp+${#progs[*]}+1))
x_prompt=3 ; y_prompt=$((y_regexp+${#progs[*]}+2))
x_menu=3   ; y_menu=$((y_prompt+2))
x_prompt2=15
y_max=$((y_menu+${#S1_txt[*]}))
screensize=`stty size`
[ "${screensize% *}" -lt "$y_max" ] && { echo -e "
sua tela tem ${screensize% *} linhas e deve ter no mínimo $y_max para que
este programa caiba nela. aumente o número de linhas ou
retire alguns programas da variável \$progs\n" ; exit 1
}


_eol=`echo -ne "\033[0K"`  # clear trash until EOL

# the cool functions
gotoxy(){ echo -ne "\033[$2;$1H" ;}
ClearEnd(){ echo -ne "\033[0J"; }

ColorOnOff(){
  if [ "$cN" ]
  then unset cN cR cY cW
  else cN=`echo -ne "\033[m"`      # normal
       cR=`echo -ne "\033[1;31m"`  # red
       cY=`echo -ne "\033[1;33m"`  # yellow
       cW=`echo -ne "\033[1;37m"`  # white
  fi
}

DoMenu(){
  eval Menui=(\"\${$1[@]}\"); menu_n=$((${#Menui[*]}-1))  # ini
  gotoxy $x_hist $y_hist ; echo "   .oO($REPLIES)$_eol"   # history
  gotoxy $x_menu $y_menu ; echo "$cY${Menui[0]}:$cN$_eol" # title
  for i in `seq $menu_n`                                  # itens
  do echo "  $cW$i$cN) ${Menui[$i]}$_eol"; i=$((i+1)); done
  ClearEnd                                                # prompt
  gotoxy $x_prompt $y_prompt ; echo -ne "$cR[1-$menu_n]:$cN $_eol"
  read -n 1
}

Menu(){
  DoMenu "$1"
  case "$REPLY" in
    [1-9])if [ "$REPLY" -gt "$menu_n" ]
          then Menu "$1" ; else REPLIES="$REPLIES$REPLY"; fi;;
     [qQ])STATUS=3 ;; [rR])STATUS=0 ;; [cC])ColorOnOff; Menu "$1";;
        *)Menu "$1";;
  esac
  [ "${STATUS/[03]/}" ] || continue         # 0,3: escape status
}

GetChar(){ gotoxy $x_prompt2 $y_prompt
  echo -ne "${cR}qual?$cN " ; read -n 1 -r USERINPUT ; uin="$USERINPUT"
  F_ESCCHAR=1
}


#TODO 1st of all, take out repeated chars
GetCharList(){ gotoxy $x_prompt2 $y_prompt
  echo -ne "${cR}quais?$cN " ; read -r USERINPUT ; uin="$USERINPUT"
  # putting not special chars in not special places: [][^-]
  [ "${uin/^//}" != "$uin" ] && uin="${uin/^/}^"
  [ "${uin/-//}" != "$uin" ] && uin="${uin/-/}-"
  [ "${uin/[//}" != "$uin" ] && uin="[${uin/[/}"
  [ "${uin/]//}" != "$uin" ] && uin="]${uin/]/}"
  [ "$1" ] && uin="^$uin"                   # if any $1, negated list
  uin="[$uin]"
  F_ESCCHARLIST=1
}

GetString(){ gotoxy $x_prompt2 $y_prompt
  echo -ne "${cR}txt:$cN " ; read -r USERINPUT ; uin="$USERINPUT"
  F_ESCCHAR=1
}

GetNumber(){ gotoxy $x_prompt2 $y_prompt
  echo -ne "${cR}N=$cN$_eol" ; read USERINPUT
  uin="${USERINPUT//[^0-9]/}"               # extracting !numbers
  [ "${uin/666/}" ] || { gotoxy 35 1 ; echo "$cR]:|$cN" ; } # ee
  [ "$uin" ] || GetNumber                   # there _must_ be a number
}

EscChar(){ # escape userinput chars as .,*,[ and friends
  local c x x2 z i ui esc
  esc='\'; [ "$1" == 'lisp' ] && esc='\\'   # double escape for lisp
  ui="$uin"
  eval x=\"\${ax_$1[3]}\" ; x="${x//[, ]/}" # list of escapable chars
  [ "${ui/[\\\\$x]/}" != "$ui" ] && {       # test for speed up
    for i in `seq 0 $((${#ui}-1))`          # for each user char
    do c="${ui:$i:1}"
       case "$c" in                         # special bash chars
         [?*#%])x2="${x/[$c]/}";;
           [/}])x2="${x/\\$c/}";;
           [\\])x2="${x/$c$c/}";;
              *)x2="${x/$c/}"  ;;
       esac
       [ "$x2" != "$x" ] && c="$esc$c"      # escaping
       z="$z$c"
    done
    uin="$z"                                # ah, the escaped string
  }
}

EscCharList(){
  local x esc='\' ; eval x=\"\${ax_$1[4]}\"
  [ "$x" == '\' ] && uin="${uin/\\\\/$esc$esc}" # escaping escape
}

Reset(){ gotoxy $x_regexp $y_regexp
  unset REPLIES HUMAN Regexp[*]
  for p in ${progs[*]}; do printf " RegExp %-6s: $_eol\n" "$p"; done
}

ShowRegExp(){ gotoxy $x_regexp $y_regexp
  local i save="$uin"
  for i in `seq 0 $((${#progs[*]}-1))`      # for each program
  do [ "$F_ESCCHAR"     == '1' ] && EscChar     ${progs[$i]}
     [ "$F_ESCCHARLIST" == '1' ] && EscCharList ${progs[$i]}
     case "$1" in                           # check status
       S2) eval Regexp[$i]="\${Regexp[$i]}\${S2_${progs[$i]}[$REPLY]/@/$uin}";;
       S0) Regexp[$i]="${Regexp[$i]}${S0_re[$REPLY]}";;
       S1) Regexp[$i]="${Regexp[$i]}${uin:-${S1_re[$REPLY]}}";;
      S12) Regexp[$i]="${Regexp[$i]}$TMP_RE";;
     esac
     printf " RegExp %-6s: %s\n" "${progs[$i]}" "${Regexp[$i]}"
     uin="$save"
  done
  unset uin USERINPUT F_ESCCHAR F_ESCCHARLIST
}



clear
ColorOnOff
echo 'Q)uit   R)eset   C)olor On/Off          __: unknown  !!: not supported'

while : ; do
case ${STATUS:=0} in
 0) Reset
    STATUS=1
    Menu S0_txt
    HUMAN="$S0_txt $cW${S0_txt[$REPLY]}$cN"
    ShowRegExp S0
    STATUS=1
    ;;
 1) Menu S1_txt
    TMP_RE="${S1_re[$REPLY]}"
    HUMAN="$HUMAN, $S1_txt $cW${S1_txt[$REPLY]}$cN"
    [ "$REPLY" -eq 2 ] && STATUS=12 && continue
    [ "$REPLY" -eq 3 ] && STATUS=12 && continue
    [ "$REPLY" -eq 5 ] && GetChar
    [ "$REPLY" -eq 6 ] && GetCharList
    [ "$REPLY" -eq 7 ] && GetCharList negated
    STATUS=2
    [ "$REPLY" -eq 8 ] && GetString && STATUS=1
    [ "$REPLY" -eq 9 ] && STATUS=1
    ShowRegExp S1
    ;;
12) Menu S12_txt
    TMP_RE="[$TMP_RE${S12_re[$REPLY]}]"
    ShowRegExp S12
    STATUS=2
    ;;
 2) Menu S2_txt
    [ "$REPLY" -ge 5 ] && GetNumber
    HUMAN="$HUMAN, $cW${S2_txt[$REPLY]}$cN vez(es)"
    ShowRegExp S2
    STATUS=1
    ;;
 3) echo -ne "\033[0G" ; ClearEnd
    echo -e "\n  ${HUMAN:-no RegExp}.\n"
    exit 0
    ;;
 *) echo "Error: STATUS = $STATUS"
    exit 1
    ;;
esac
done