From 976aa271514cbe8006b56c50fe86099db5ef047c Mon Sep 17 00:00:00 2001
From: Wesley Kerfoot <wjak56@gmail.com>
Date: Sun, 15 Sep 2013 22:05:25 -0400
Subject: [PATCH] worked on toknizer some more

---
 representation.js |  2 ++
 tokenize.js       | 22 +++++++------
 tools.js          | 82 +++++++++++++++++++++++++++++++++++++++++++----
 3 files changed, 91 insertions(+), 15 deletions(-)

diff --git a/representation.js b/representation.js
index d1067a1..8fcb323 100644
--- a/representation.js
+++ b/representation.js
@@ -161,7 +161,9 @@ OPInfo = {"+" : [3, "Left"],
       "++" : [3, "Left"],
       "==" : [2, "Left"],
       ">" : [2, "Left"],
+      ">=" : [2, "Left"],
       "<" : [2, "Left"],
+      "<=" : [2, "Left"],
       ":" : [2, "Left"],
       "$" : [1, "Left"]}
 
diff --git a/tokenize.js b/tokenize.js
index a860b12..0bcd5af 100755
--- a/tokenize.js
+++ b/tokenize.js
@@ -5,8 +5,6 @@ var rep = require("./representation.js");
 var tools = require("./tools.js");
 var operators = Object.keys(rep.OPInfo);
 
-hasOp = tools.hasOperator(operators);
-
 function isDigit(a) {
   if (!a)
     return false;
@@ -84,13 +82,17 @@ function tokenizeIdent(tokstream) {
     n++;
   }
   identifier = identifier.join('');
-  var op = hasOp(identifier);
+/*  var op = hasOp(identifier);
   if (op) {
+    if (identifier === op)
+      return [[tools.len(op), ["identifier", op]]];
     var splitted = identifier.split(op);
+    console.log(splitted);
     var newIdent = splitted[0];
     tokstream = splitted[1]+tokstream;
     return [[n-(tools.len(op)), ["identifier", newIdent]], [tools.len(op), ["identifier", op]]];
   }
+*/
   return [[n, ["identifier", identifier]]];
 }
 
@@ -278,10 +280,10 @@ function tokenize(tokstream) {
           break;
         }
         var result = tokenizeIdent(tokstream);
-        var i = result[0];
-        var ident = result[1];
-        tokens.push(ident);
-        tokstream = tokstream.substr(i);
+        result.map(function(x) {
+          tokens.push(x[1]);
+          tokstream = tokstream.substr(x[0]);
+        });
     }
   }
   return tokens;
@@ -290,8 +292,10 @@ function tokenize(tokstream) {
 module.exports = {tokenize : tokenize};
 
 //var tokstream = fs.readFileSync("/dev/stdin").toString();
-//console.log(isIdentifier(')'));
 //console.log(tokenize(tokstream));
+console.log(tools.buildTrie('', operators)[1][6]);
+//console.log(isIdentifier(')'));
 //console.log(tools.maxBy(tools.len, operators.filter(function (x) { return "#".indexOf(x) != -1;})));
-console.log(tokenizeIdent("abc%%3"));
+//console.log(tokenizeIdent("abc%%3"));
+
 
diff --git a/tools.js b/tools.js
index 303006a..1b38181 100644
--- a/tools.js
+++ b/tools.js
@@ -30,17 +30,87 @@ function len(xs) {
   return xs.length;
 }
 
-function hasOperator(operators) {
-  return function (str) {
-    if (len(operators) < 1)
-      return false;
-    return maxBy(len, operators.filter(function (x) { return str.indexOf(x) != -1; }));
+function takeWhile(f, xs) {
+  var result = [];
+  for (i = 0; i < xs.length; i++) {
+    if (f(xs[i]))
+      result.push(xs[i]);
+    else
+      break;
+  }
+  return result;
+}
+
+function dropWhile(f, xs) {
+  for (i = 0; i < xs.length; i++) {
+    if (!f(xs[i]))
+      break;
+  }
+  return xs.slice(i);
+}
+
+function span(f, xs) {
+  return [takeWhile(f, xs), dropWhile(f, xs)];
+}
+
+function eq(a) {
+  return function(b) {
+    return a[0] === b[0];
   };
 }
 
+function groupBy(eq, xs) {
+  var groups = [];
+  var spanned;
+  while (xs.length > 0) {
+    spanned = span(eq(xs[0]), xs.slice(1));
+    groups.push([xs[0]].concat(spanned[0]));
+    xs = spanned[1];
+  }
+  return groups;
+}
+
+function groupOps(ops) {
+  return groupBy(eq, ops.sort());
+}
+
+// See: https://en.wikipedia.org/wiki/Trie
+function buildTrie(prefix, operators) {
+  var grps = groupOps(operators);
+  return [prefix, grps.map(
+    function(ops) {
+      if (ops.length === 1)
+        return ops;
+      return buildTrie(ops[0][0],
+        ops.map(
+          function(op) {
+            return op.slice(1);
+          }).filter(function (x){ return x;}));
+    })];
+}
+
+function matchStr(n, str, trie) {
+  if (trie[0][0].length > 1 && str.length <= 1)
+    return false;
+  else if (trie[0].every(function(x){return x[0][0] !== str[0];})) {
+    return false;
+  }
+  else if (trie[0].length === 1) {
+    if (trie[0][0] !== str.slice(n, trie[0][0].length))
+      return false;
+    return trie[0][0].length + n;
+  }
+  else {
+    return matchStr(n+1, str.slice(1), trie[0][1])
+  }
+}
+
+console.log(matchStr(0, "**^*$4545", buildTrie("", ["**", "**^"])[1]));
+
 module.exports = {compose : compose,
 				  not	  : not,
 				  on	  : on,
           maxBy : maxBy,
           len   : len,
-          hasOperator : hasOperator}
+          groupOps : groupOps,
+          buildTrie : buildTrie}