Improved OS.parse()

GunioRobot · Mar 3, 2010 · 106140a · 106140a
1 parent 0ea83b4
commit 106140a
Show file tree

Hide file tree

Showing 2 changed files with 84 additions and 63 deletions.
diff --git a/lib/os.js b/lib/os.js
@@ -89,73 +89,89 @@ exports.enquote = function (word) {
  * This utility is used by Narwhal to pare arguments from
  * system.env.NARWHAL_OPT.
  */
-exports.parse = function (args) {
-    var startsWith = function(self, str) {
-        return (self.match("^"+str.replace('\\', '\\\\'))==str)
-    }
-    var endsWith = function(self, str) {
-        return (self.match(str.replace('\\', '\\\\')+"$")==str)
-    }
-    var results = [], quoteType = null, quotedElement = null;
-    args.split(' ').forEach(function(element) {
-        if (
-            quoteType || endsWith(element, '\\') ||
-            startsWith(element, '\'') || startsWith(element, '"')
-        ) {
-            if (quoteType) {
-                if (endsWith(element, quoteType)) {
-                    results.push(
-                        quotedElement + " " +
-                        element.substring(0, element.length - 1)
-                    );
-                quotedElement = null;
-                quoteType = null;
-                return;
+
+var STATE_NORMAL    = 0; // waiting for non whitespace/quote
+var STATE_ARG       = 1; // nextArg is an argument, even if empty
+var STATE_IN_QUOTE  = 2; // within a ' or " quote
+
+exports.parse = function (argString) {
+    var args = [];
+
+    var nextArg = "";
+    var state = STATE_NORMAL;
+    var escapeNext = false;
+    var delimiter;
+
+    var tokens = argString.split("");
+    while (tokens.length > 0) {
+        var token = tokens.shift();
+
+        if (state === STATE_NORMAL || state === STATE_ARG) {
+            if (!escapeNext && token === "\\") {
+                escapeNext = true;
             }
-            else { quotedElement += " " + element; return; }
-        }
-        if (!quoteType && startsWith(element, '\'')) {
-            quoteType = '\'';
-            quotedElement = element.substring(1, element.length);
-            return;
-        }
-        if (!quoteType && startsWith(element, '"')) {
-            quoteType = '"';
-            quotedElement = element.substring(1, element.length);
-                return;
+            else if (escapeNext) {
+                state = STATE_ARG;
+                escapeNext = false;
+                nextArg += token;
             }
-            if (!quoteType) {
-                if (endsWith(element, '\\')) {
-                    if (quotedElement) {
-                        quotedElement += " " + element.substring(
-                            0,
-                            element.length - 1
-                        );
-                    }
-                    else {
-                        quotedElement = element.substring(
-                            0,
-                            element.length - 1
-                        );
-                    }
-                }
-                else {
-                    results.push(quotedElement + ' ' + element);
-                    quotedElement = null;
-                    return;
+            else if (token === "'" || token === '"') {
+                delimiter = token;
+                state = STATE_IN_QUOTE;
+            }
+            else if (token === " ") {
+                if (state === STATE_ARG) {
+                    args.push(nextArg);
+                    nextArg = "";
                 }
+                state = STATE_NORMAL;
+            }
+            else {
+                nextArg += token;
+                state = STATE_ARG;
             }
         }
-        else {
-            if (quotedElement) {
-                results.push(quotedElement + ' ' + element);
-                quotedElement = null;
+        else if (state === STATE_IN_QUOTE) {
+            if (!escapeNext && token === "\\") {
+                escapeNext = true;
+            }
+            else if (delimiter === token) {
+                if (escapeNext) {
+                    nextArg += token;
+                    escapeNext = false;
+                } else {
+                    state = STATE_ARG;
+                }
             }
             else {
-                results.push(element);
+                if (escapeNext) {
+                    // if not a quote (above) or other special character that needs to be escaped then include the backslash
+                    if (token !== "\\")
+                        nextArg += "\\";
+                    nextArg += token;
+                    escapeNext = false;
+                } else {
+                    nextArg += token;
+                }
             }
         }
-    });
-    return results;
+        else {
+            throw "wtf " + state;
+        }
+    }
+
+    if (state === STATE_IN_QUOTE) {
+        if (token === delimiter) {
+            args.push(nextArg.slice(0,-1) + "\\");
+        }
+        else {
+            // throw "Invalid or not yet implemented case"
+        }
+    }
+    else if (state === STATE_ARG) {
+        args.push(nextArg);
+    }
+
+    return args;
 };
 
diff --git a/tests/os/parse-fuzzer.js b/tests/os/parse-fuzzer.js
@@ -3,12 +3,17 @@ var ASSERT = require("assert");
 var UTIL = require("UTIL");
 var stream = require("term").stream;
 
+// config
 var random = false;
-var tokens = ["x", "'", '"', " "];
+var maxLength = 10;
+var tokens = ["x", "'", '"', " ", "\\"];
 
 function shellParse(argString) {
+    // print(argString)
     var p = OS.popen(["sh", "-c", 'for i in "$@"; do echo "$i"; done', "--"].map(OS.enquote).join(" ") + " " + argString);
-    var result = p.stdout.read().split("\n").slice(0,-1);
+    var x = p.stdout.read();
+    // print("["+x+"]")
+    var result = x.split("\n").slice(0,-1);
     if (p.wait())
         throw result;
     return result;
@@ -19,7 +24,7 @@ function buildTestString() {
 
     if (random) {
         var components = [];
-        var n = Math.round(Math.random()*10);
+        var n = Math.round(Math.random()*maxLength);
         while (components.length < n)
             components.push(tokens[Math.floor(Math.random()*tokens.length)]);
         return components.join("");
@@ -50,5 +55,5 @@ while (true) {
     var pass = UTIL.eq(expectedArgs, actualArgs);
 
     stream.print("[" + testIndex + "] " + (pass ? "\0green(PASS\0)" : "\0red(FAIL\0)") +
-        ": string=["+argString+"] expected=["+expectedArgs+"] actual=["+actualArgs+"]");
+        ": string=["+argString+"] expected=["+expectedArgs+"]("+expectedArgs.length+") actual=["+actualArgs+"]("+actualArgs.length+")");
 }