bash/reader.sh

   1 #
   2 # mal (Make Lisp) Parser/Reader
   3 #
   4
   5 if [ -z "${__mal_readerr_included__}" ]; then
   6 __mal_readerr_included=true
   7
   8 source $(dirname $0)/types.sh
   9
  10 READ_ATOM () {
  11     local token=${__reader_tokens[${__reader_idx}]}
  12     __reader_idx=$(( __reader_idx + 1 ))
  13     case "${token}" in
  14         [0-9]*)  _number "${token}" ;;
  15         -[0-9]*) _number "${token}" ;;
  16         \"*)    token="${token:1:-1}"
  17                 token="${token//\\\\/${__keyw}}"
  18                 token="${token//\\\"/\"}"
  19                 token="${token//\\n/$'\n'}"
  20                 token="${token//${__keyw}/\\}"
  21                 _string "${token}" ;;
  22         :*)     _keyword "${token:1}" ;;
  23         nil)    r="${__nil}" ;;
  24         true)   r="${__true}" ;;
  25         false)  r="${__false}" ;;
  26         *)      _symbol "${token}" ;;
  27     esac
  28 }
  29
  30 # Return seqence of tokens into r.
  31 #   ${1}: Type of r (vector, list)
  32 #   ${2}: starting symbol
  33 #   ${3}: ending symbol
  34 READ_SEQ () {
  35     local start="${1}"
  36     local end="${2}"
  37     local items=""
  38     local token=${__reader_tokens[${__reader_idx}]}
  39     __reader_idx=$(( __reader_idx + 1 ))
  40     if [[ "${token}" != "${start}" ]]; then
  41         r=
  42         _error "expected '${start}'"
  43         return
  44     fi
  45     token=${__reader_tokens[${__reader_idx}]}
  46     while [[ "${token}" != "${end}" ]]; do
  47         if [[ ! "${token}" ]]; then
  48             r=
  49             _error "exepected '${end}', got EOF"
  50             return
  51         fi
  52         READ_FORM
  53         items="${items} ${r}"
  54         token=${__reader_tokens[${__reader_idx}]}
  55     done
  56     __reader_idx=$(( __reader_idx + 1 ))
  57     r="${items:1}"
  58 }
  59
  60 # Return form in r
  61 READ_FORM () {
  62     local token=${__reader_tokens[${__reader_idx}]}
  63     case "${token}" in
  64         \')     __reader_idx=$(( __reader_idx + 1 ))
  65                 _symbol quote; local q="${r}"
  66                 READ_FORM; local f="${r}"
  67                 _list "${q}" "${f}" ;;
  68         \`)     __reader_idx=$(( __reader_idx + 1 ))
  69                 _symbol quasiquote; local q="${r}"
  70                 READ_FORM; local f="${r}"
  71                 _list "${q}" "${f}" ;;
  72         \~)     __reader_idx=$(( __reader_idx + 1 ))
  73                 _symbol unquote; local q="${r}"
  74                 READ_FORM; local f="${r}"
  75                 _list "${q}" "${f}" ;;
  76         \~\@)   __reader_idx=$(( __reader_idx + 1 ))
  77                 _symbol splice-unquote; local q="${r}"
  78                 READ_FORM; local f="${r}"
  79                 _list "${q}" "${f}" ;;
  80         ^)      __reader_idx=$(( __reader_idx + 1 ))
  81                 _symbol with-meta; local wm="${r}"
  82                 READ_FORM; local meta="${r}"
  83                 READ_FORM; local obj="${r}"
  84                 _list "${wm}" "${obj}" "${meta}" ;;
  85         @)     __reader_idx=$(( __reader_idx + 1 ))
  86                 _symbol deref; local d="${r}"
  87                 READ_FORM; local f="${r}"
  88                 _list "${d}" "${f}" ;;
  89         \))     _error "unexpected ')'" ;;
  90         \()     READ_SEQ "(" ")"
  91                 _list ${r} ;;
  92         \])     _error "unexpected ']'" ;;
  93         \[)     READ_SEQ "[" "]"
  94                 _vector ${r} ;;
  95         \})     _error "unexpected '}'" ;;
  96         \{)     READ_SEQ "{" "}"
  97                 _hash_map ${r} ;;
  98         *)      READ_ATOM
  99     esac
 100 }
 101
 102 # Returns __reader_tokens as an indexed array of tokens
 103 TOKENIZE () {
 104     local data="${*}"
 105     local datalen=${#data}
 106     local idx=0
 107     local chunk=0
 108     local chunksz=500
 109     local match=
 110     local token=
 111     local str=
 112
 113     __reader_idx=0
 114     __reader_tokens=
 115     while true; do
 116         if (( ${#str} < ( chunksz / 2) )) && (( chunk < datalen )); then
 117             str="${str}${data:${chunk}:${chunksz}}"
 118             chunk=$(( chunk + ${chunksz} ))
 119         fi
 120         (( ${#str} == 0 )) && break
 121         [[ "${str}" =~ ^^([][{}\(\)^@])|^(~@)|(\"(\\.|[^\\\"])*\")|^(;[^$'\n']*)|^([~\'\`])|^([^][ ~\`\'\";{}\(\)^@\,]+)|^[,]|^[[:space:]]+ ]]
 122         match=${BASH_REMATCH[0]}
 123         str="${str:${#match}}"
 124         token="${match//$'\n'/}"
 125         #echo "MATCH: '${token}' / [${str}]"
 126         if ! [[ "${token}" =~ (^[,]$|^[[:space:]]*;.*$|^[[:space:]]*$) ]]; then
 127             __reader_tokens[${idx}]="${token}"
 128             idx=$(( idx + 1 ))
 129         fi
 130         if [ -z "${match}" ]; then
 131             _error "Tokenizing error at: ${str:0:50}"
 132             return 1
 133         fi
 134     done
 135 }
 136
 137 # read-str from a raw "string" or from a string object. Retruns object
 138 # read in r.
 139 READ_STR () {
 140     declare -a __reader_tokens
 141     TOKENIZE "${*}" || return 1  # sets __reader_tokens
 142     #set | grep ^__reader_tokens
 143     if [ -z "${__reader_tokens[0]}" ]; then
 144         r=
 145         return 1  # No tokens
 146     fi
 147     READ_FORM
 148     #echo "Token: ${r}: <${ANON["${r}"]}>"
 149     return
 150 }
 151
 152 # Call readline and save the history. Returns the string read in r.
 153 READLINE_EOF=
 154 READLINE_HISTORY_FILE=${HOME}/.mal-history
 155 READLINE () {
 156     history -r "${READLINE_HISTORY_FILE}" 2>/dev/null || true
 157     read -r -e -p "${1}" r || return "$?"
 158     history -s -- "${r}"
 159     history -a "${READLINE_HISTORY_FILE}" 2>/dev/null || true
 160 }
 161
 162 fi