[refactor] Struggling to get self-evaluating components working.

Makefile

@@ -24,7 +24,7 @@ test: clean node_modules
 		--reporter mocha-jenkins-reporter --compilers coffee:coffee-script/register || true
 
 ltest: node_modules
-	@node_modules/.bin/mocha --compilers coffee:coffee-script/register
+	node_modules/.bin/mocha --compilers coffee:coffee-script/register
 
 watch:
 	while inotifywait $(SOURCES) ; do make test; done

chapter1/interpreter.coffee

@@ -2,19 +2,15 @@
  cdar, cadr, caadr, cadar, caddr, nilp, nil, setcdr, metacadr} = require "cons-lists/lists"
 readline = require "readline"
 {inspect} = require "util"
+{Symbol} = require "./reader_types"
 
 class LispInterpreterError extends Error
   name: 'LispInterpreterError'
-  constructor: (@message) ->
+  constructor: (@message, position = null) ->
-
 
 env_init = nil
 env_global = env_init
 
-defpredicate = (name, nativ, arity) ->
-  defprimitive name, ((a, b) -> if nativ.call(null, a, b) then true else the_false_value), arity
-
-
 definitial = (name, value = nil) ->
   env_global = (cons (cons name, value), env_global)
   name
@@ -27,6 +23,9 @@ defprimitive = (name, nativ, arity) ->
     else
       throw (new LispInterpreterError "Incorrect arity"))
 
+defpredicate = (name, nativ, arity) ->
+  defprimitive name, ((a, b) -> if nativ.call(null, a, b) then true else the_false_value), arity
+
 definitial "#t", true
 definitial "#f", the_false_value
 definitial "nil", nil
@@ -47,32 +46,7 @@ defpredicate "eq?", ((a, b) -> a == b), 2
 
 the_false_value = (cons "false", "boolean")
 
-cadddr = metacadr('cadddr')
+extend = (env, variables, values) ->
-
-metadata_evaluation =
-  listp:   (node) -> (car node) == 'list'
-  symbolp: (node) -> (car node) == 'symbol'
-  numberp: (node) -> (car node) == 'number'
-  stringp: (node) -> (car node) == 'string'
-  nvalu:   (node) -> cadr node
-
-straight_evaluation = 
-  listp:    (node) -> node.__type == 'list'
-  symbolp:  (node) -> node instanceOf Symbol
-  commentp: (node) -> node instanceOf Comment
-  numberp:  (node) -> typeof node == 'number'
-  stringp:  (node) -> typeof node == 'string'
-  boolp:    (node) -> typeof node == 'boolean'
-  nullp:    (node) -> node == null
-  vectorp:  (node) -> (not listp node) and toString.call(node) == '[object Array]'
-  recordp:  (node) -> (not x._prototype?) and toSTring.call(node) == '[object Object]')
-  objectp:  (node) -> (x._prototype?) and toString.call(node) == '[object Object]')
-  nilp:     (node) -> node == nilp
-  nvalu:    (node) -> node
-
-makeEvaluator = (ix = straight_evaluation) ->
-  (exp, env) ->
-    extend = (env, variables, values) ->
   if (pairp variables)
     if (pairp values)
       (cons (cons (car variables), (car values)),
@@ -87,18 +61,18 @@ makeEvaluator = (ix = straight_evaluation) ->
     else
       nil
 
-    make_function = (variables, body, env) ->
+make_function = (variables, body, env) ->
   (values) -> eprogn body, (extend env, variables, values)
 
-    invoke = (fn, args) ->
+invoke = (fn, args) ->
   (fn args)
 
-    # Takes a list of nodes and calls evaluate on each one, returning the
+# Takes a list of nodes and calls evaluate on each one, returning the
-    # last one as the value of the total expression.  In this example, we
+# last one as the value of the total expression.  In this example, we
-    # are hard-coding what ought to be a macro, namely the threading
+# are hard-coding what ought to be a macro, namely the threading
-    # macros, "->"
+# macros, "->"
 
-    eprogn = (exps, env) ->
+eprogn = (exps, env) ->
   if (pairp exps)
     if pairp (cdr exps)
       evaluate (car exps), env
@@ -108,13 +82,13 @@ makeEvaluator = (ix = straight_evaluation) ->
   else
     nil
 
-    evlis = (exps, env) ->
+evlis = (exps, env) ->
   if (pairp exps)
     (cons (evaluate (car exps), env), (evlis (cdr exps), env))
   else
     nil
     
-    lookup = (id, env) ->
+lookup = (id, env) ->
   if (pairp env)
     if (caar env) == id
       cdar env
@@ -123,7 +97,7 @@ makeEvaluator = (ix = straight_evaluation) ->
   else
     nil
 
-    update = (id, env, value) ->
+update = (id, env, value) ->
   if (pairp env)
     if (caar env) == id
       setcdr value, (car env)
@@ -133,11 +107,8 @@ makeEvaluator = (ix = straight_evaluation) ->
   else
     nil
 
-    # This really ought to be the only place where the AST meets the
+astSymbolsToLispSymbols = (node) ->
-    # interpreter core.  I can't help but think that this design precludes
+  nvalu = (node) -> cadr node
-    #  pluggable interpreter core.
-    
-    astSymbolsToLispSymbols = (node) ->
   return nil if nilp node
   throw (new LispInterpreterError "Not a list of variable names") if not (ntype(node) is 'list')
   handler = (node) ->
@@ -145,18 +116,49 @@ makeEvaluator = (ix = straight_evaluation) ->
     cons (nvalu car node), (handler cdr node)
   handler(nvalu node)
 
+
+cadddr = metacadr('cadddr')
+
+# This is really the only thing that changes behavior between "reader
+# nodes" (nodes loaded with debugging metadata) and a standard cons
+# object.  TODO: astSymbolsToLispSymbols should be deprecated in
+# favor of normalizeForm (s?) and Symbol extraction
+
+metadata_evaluation =
+  listp:   (node) -> (car node) == 'list'
+  symbolp: (node) -> (car node) == 'symbol'
+  numberp: (node) -> (car node) == 'number'
+  stringp: (node) -> (car node) == 'string'
+  nvalu:   (node) -> cadr node
+
+straight_evaluation = 
+  listp:    (node) -> node.__type == 'list'
+  symbolp:  (node) -> node instanceof Symbol
+  commentp: (node) -> node instanceof Comment
+  numberp:  (node) -> typeof node == 'number'
+  stringp:  (node) -> typeof node == 'string'
+  boolp:    (node) -> typeof node == 'boolean'
+  nullp:    (node) -> node == null
+  vectorp:  (node) -> (not straight_evaluation.listp node) and toString.call(node) == '[object Array]'
+  recordp:  (node) -> (not node._prototype?) and toSTring.call(node) == '[object Object]'
+  objectp:  (node) -> (node._prototype?) and toString.call(node) == '[object Object]'
+  nilp:     (node) -> nilp(node)
+  nvalu:    (node) -> node
+
+makeEvaluator = (ix = straight_evaluation, ty="straight") ->
+  (exp, env) ->
+    console.log("Type:", ty)
     # Takes an AST node and evaluates it and its contents.  A node may be
     # ("list" (... contents ...)) or ("number" 42) or ("symbol" x), etc.
   
-    [type, exp] = [(ntype e), (nvalu e)]
+    if ix.symbolp(exp)
-    if type == "symbol"
+      return lookup (ix.nvalu exp).v, env
-      return lookup exp, env
+    else if ([ix.numberp, ix.stringp].filter (i) -> i(exp)).length > 0
-    else if type in ["number", "string", "boolean", "vector"]
       return exp
-    else if type == "list"
+    else if ix.listp(exp)
-      head = car exp
+      head = car ix.nvalu exp
-      if (ntype head) == 'symbol'
+      if ix.symbolp(head)
-        switch (nvalu head)
+        switch (ix.nvalu head).v
           when "quote" then cdr exp
           when "if"
             unless (evaluate (cadr exp), env) == the_false_value
@@ -164,7 +166,7 @@ makeEvaluator = (ix = straight_evaluation) ->
             else
               evaluate (cadddr exp), env
           when "begin" then eprogn (cdr exp), env
-          when "set!" then update (nvalu cadr exp), env, (evaluate (caddr exp), env)
+          when "set!" then update (ix.nvalu cadr exp).v, env, (evaluate (caddr exp), env)
           when "lambda" then make_function (astSymbolsToLispSymbols cadr exp), (cddr exp), env
           else invoke (evaluate (car exp), env), (evlis (cdr exp), env)
       else
@@ -172,4 +174,10 @@ makeEvaluator = (ix = straight_evaluation) ->
     else
       throw new LispInterpreterError "Can't handle a #{type}"
 
+nodeEval = makeEvaluator(metadata_evaluation, "node")
+lispEval = makeEvaluator(straight_evaluation, "lisp")
+
+evaluate = (exp, env) ->
+  (if exp? and exp.__node then nodeEval else lispEval)(exp, env)
+
 module.exports = (c) -> evaluate c, env_global