par.nim

import npeg

let parser = peg "COMMANDS":
  COMMANDS <- COMMAND * *( SEPERATOR_SYM * COMMAND )
  COMMAND  <- +(
    SPACES        |
    REDIRECT_SYM  |
    COMMAND_SUB   |
    VARIABLE_SUB  |
    DQ_STRING_LIT |
    WORD_LIT
  )

  # Words
  WORD_LIT   <- >+( WORD_CHAR - SUB_MARKER_SYM )                        : echo "WORD_LIT: ", $1
  WORD_CHAR  <- Alpha | Digit
  UTF8       <- {0..244}


  # Double Quoted Strings
  DQ_STRING_LIT   <-
    DQ_STRING_START *
    DQ_STRING_DATA  *
    DQ_STRING_END

  DQ_STRING_START <- DQ_STRING_START_SYM                               : echo "DQ_STRING_START"
  DQ_STRING_END   <- DQ_STRING_END_SYM | E_MISSING_QUOTE               : echo "DQ_STRING_END"

  DQ_STRING_DATA  <- *(
    COMMAND_SUB      |
    VARIABLE_SUB     |
    DQ_STRING_ESCAPE |
    DQ_STRING_CHARS
  )
  DQ_STRING_CHARS  <- >+DQ_STRING_CHAR                                  : echo "DQ_STRING_CHARS: ", $1
  DQ_STRING_CHAR   <- >(UTF8 - SUB_MARKER_SYM - DQ_STRING_ESCAPE - '"' )
  DQ_STRING_ESCAPE <- '\\' * ( SUB_ESCAPE | UTF8_ESCAPE | CHAR_ESCAPE | '"')


  # Single Quoted Strings
  SQ_STRING_LIT   <-
    SQ_STRING_START *
    SQ_STRING_DATA  *
    SQ_STRING_END

  SQ_STRING_START <- SQ_STRING_START_SYM                               : echo "SQ_STRING_START"
  SQ_STRING_END   <- SQ_STRING_END_SYM | E_MISSING_QUOTE               : echo "SQ_STRING_END"

  SQ_STRING_DATA  <- *(
    COMMAND_SUB      |
    VARIABLE_SUB     |
    SQ_STRING_ESCAPE |
    SQ_STRING_CHARS
  )
  SQ_STRING_CHARS  <- >+SQ_STRING_CHAR                                  : echo "SQ_STRING_CHARS: ", $1
  SQ_STRING_CHAR   <- >(UTF8 - SUB_MARKER_SYM - SQ_STRING_ESCAPE - '\'' )
  SQ_STRING_ESCAPE <- '\\' * ( SUB_ESCAPE | UTF8_ESCAPE | CHAR_ESCAPE | '\'')

  # Escapes
  UTF8_ESCAPE  <- >('u' * Xdigit[4])                                     : echo "UTF8_ESCAPE: ", $1
  SUB_ESCAPE   <- >SUB_MARKER_SYM                                        : echo "SUB_ESCAPE: ", $1
  CHAR_ESCAPE <- >{ '\\', 'b', 'f', 'n', 'r', 't' }                      : echo "CHAR_ESCAPE: ", $1



  # Command Substitutions
  COMMAND_SUB <-
    COMMAND_SUB_START *
    COMMAND_SUB_DATA  *
    COMMAND_SUB_END

  COMMAND_SUB_START <- "$("                                             : echo "COMMAND_SUB_START"
  COMMAND_SUB_END   <- ")" | E_MISSING_CBRACE                           : echo "COMMAND_SUB_END"
  COMMAND_SUB_DATA  <- COMMAND | E_INVALID_SUB


  # Command Substitutions
  VARIABLE_SUB <-
    VARIABLE_SUB_START *
    VARIABLE_SUB_DATA  *
    VARIABLE_SUB_END

  VARIABLE_SUB_START <- "${"                                            : echo "VARIABLE_SUB_START"
  VARIABLE_SUB_END   <- "}" | E_MISSING_PAREN                           : echo "VARIABLE_SUB_END"
  VARIABLE_SUB_DATA  <- WORD_LIT | E_INVALID_SUB


  # Symbols
  SUB_MARKER_SYM <- (
    COMMAND_SUB_START_SYM  |
    COMMAND_SUB_END_SYM    |
    VARIABLE_SUB_START_SYM |
    VARIABLE_SUB_END_SYM
  )

  DQ_STRING_START_SYM <- '"'
  DQ_STRING_END_SYM   <- '"'
  SQ_STRING_START_SYM <- '\''
  SQ_STRING_END_SYM   <- '\''

  COMMAND_SUB_START_SYM  <- "$("
  COMMAND_SUB_END_SYM    <- ")"
  VARIABLE_SUB_START_SYM <- "${"
  VARIABLE_SUB_END_SYM   <- "}"

  SEPERATOR_SYM <- ( "&&" | "||" | ";")
  REDIRECT_SYM  <- ( ">>" | "!>" | "!>>" | ">" | "<")

  SPACES        <- +Space


  # Errors
  E_MISSING_QUOTE  <- E"Umatched quotation"
  E_MISSING_CBRACE <- E"Missing curly brace"
  E_MISSING_PAREN  <- E"Missing parenthesis"
  E_INVALID_SUB    <- E"Invalid substitution"


echo parser.match("""echo hello"jj$($("world"))\" >""").captures