Current state of mal for Clojure West lighting talk.

1 files changed, 153 insertions, 0 deletions

diff --git a/bash/reader.sh b/bash/reader.sh
new file mode 100644
index 0000000..bc32fa7
--- /dev/null
+++ b/bash/reader.sh
@@ -0,0 +1,153 @@
+#
+# mal (Make Lisp) Parser/Reader
+#
+
+source $(dirname $0)/types.sh
+
+READ_ATOM () {
+    local token=${__reader_tokens[${__reader_idx}]}
+    __reader_idx=$(( __reader_idx + 1 ))
+    case "${token}" in
+        [0-9]*) number "${token}" ;;
+        \"*)    token="${token:1:-1}"
+                token="${token//\\\"/\"}"
+                string "${token}" ;;
+        nil)    r="${__nil}" ;;
+        true)   r="${__true}" ;;
+        false)  r="${__false}" ;;
+        *)      symbol "${token}" ;;
+    esac
+}
+
+# Return seqence of tokens into r.
+#   ${1}: Type of r (vector, list)
+#   ${2}: starting symbol
+#   ${3}: ending symbol
+READ_SEQ () {
+    local start="${1}"
+    local end="${2}"
+    local items=""
+    local token=${__reader_tokens[${__reader_idx}]}
+    __reader_idx=$(( __reader_idx + 1 ))
+    if [[ "${token}" != "${start}" ]]; then
+        r=
+        _error "expected '${start}'"
+        return
+    fi
+    token=${__reader_tokens[${__reader_idx}]}
+    while [[ "${token}" != "${end}" ]]; do
+        if [[ ! "${token}" ]]; then
+            r=
+            _error "exepected '${end}', got EOF"
+            return
+        fi
+        READ_FORM
+        items="${items} ${r}"
+        token=${__reader_tokens[${__reader_idx}]}
+    done
+    __reader_idx=$(( __reader_idx + 1 ))
+    r="${items:1}"
+}
+
+# Return form in r
+READ_FORM () {
+    local token=${__reader_tokens[${__reader_idx}]}
+    case "${token}" in
+        \')     __reader_idx=$(( __reader_idx + 1 ))
+                symbol quote; local q="${r}"
+                READ_FORM; local f="${r}"
+                list "${q}" "${f}" ;;
+        \`)     __reader_idx=$(( __reader_idx + 1 ))
+                symbol quasiquote; local q="${r}"
+                READ_FORM; local f="${r}"
+                list "${q}" "${f}" ;;
+        \~)     __reader_idx=$(( __reader_idx + 1 ))
+                symbol unquote; local q="${r}"
+                READ_FORM; local f="${r}"
+                list "${q}" "${f}" ;;
+        \~\@)   __reader_idx=$(( __reader_idx + 1 ))
+                symbol splice-unquote; local q="${r}"
+                READ_FORM; local f="${r}"
+                list "${q}" "${f}" ;;
+        ^)      __reader_idx=$(( __reader_idx + 1 ))
+                symbol with-meta; local wm="${r}"
+                READ_FORM; local meta="${r}"
+                READ_FORM; local obj="${r}"
+                list "${wm}" "${obj}" "${meta}" ;;
+        @)     __reader_idx=$(( __reader_idx + 1 ))
+                symbol deref; local d="${r}"
+                READ_FORM; local f="${r}"
+                list "${d}" "${f}" ;;
+        \))     _error "unexpected ')'" ;; 
+        \()     READ_SEQ "(" ")"
+                list ${r} ;;
+        \])     _error "unexpected ']'" ;; 
+        \[)     READ_SEQ "[" "]"
+                vector ${r} ;;
+        \})     _error "unexpected '}'" ;; 
+        \{)     READ_SEQ "{" "}"
+                hash_map ${r} ;;
+        *)      READ_ATOM
+    esac
+}
+
+# Returns __reader_tokens as an indexed array of tokens
+TOKENIZE () {
+    local data="${*}"
+    local datalen=${#data}
+    local idx=0
+    local chunk=0
+    local chunksz=500
+    local match=
+    local token=
+    local str=
+
+    __reader_idx=0
+    __reader_tokens=
+    while true; do
+        if (( ${#str} < ( chunksz / 2) )) && (( chunk < datalen )); then
+            str="${str}${data:${chunk}:${chunksz}}"
+            chunk=$(( chunk + ${chunksz} ))
+        fi
+        (( ${#str} == 0 )) && break
+        [[ "${str}" =~ ^^([][{}\(\)^@])|^(~@)|(\"(\\.|[^\\\"])*\")|^(;[^$'\n']*)|^([~\'\`])|^([^][ ~\`\'\";{}\(\)^@]+)|^[,]|^[[:space:]]+ ]]
+        match=${BASH_REMATCH[0]}
+        str="${str:${#match}}"
+        token="${match//$'\n'/}"
+        #echo "MATCH: '${token}' / [${str}]"
+        if ! [[ "${token}" =~ (^[,]$|^[[:space:]]*;.*$|^[[:space:]]*$) ]]; then 
+            __reader_tokens[${idx}]="${token}"
+            idx=$(( idx + 1 ))
+        fi
+        if [ -z "${match}" ]; then
+            echo >&2 "Tokenizing error at: ${str:0:50}"
+            _error "Tokenizing error at: ${str:0:50}"
+            break
+        fi
+    done
+}
+
+# read-str from a raw "string" or from a string object. Retruns object
+# read in r.
+READ_STR () {
+    declare -a __reader_tokens
+    TOKENIZE "${*}"  # sets __reader_tokens
+    #set | grep ^__reader_tokens
+    if [ -z "${__reader_tokens[k]}" ]; then
+        r=
+        return 1  # No tokens
+    fi
+    READ_FORM
+    #echo "Token: ${r}: <${ANON["${r}"]}>"
+    return
+}
+
+# Call readline and save the history. Returns the string read in r.
+READLINE_EOF=
+READLINE_HISTORY_FILE=${HOME}/.mal-history
+READLINE () {
+    history -r "${READLINE_HISTORY_FILE}"
+    read -r -e -p "${1}" r || return "$?"
+    history -s -- "${r}"
+    history -a "${READLINE_HISTORY_FILE}"
+}