Merge pull request #6 from clue-labs/unquoted-escapes

Support escape sequences in unquoted strings

README.md

@@ -98,8 +98,9 @@ way more sophisticated. It also supports parsing the following:
 * Single quoted strings (`'hello world'`) which preserve any whitespace characters and
   only accept escape sequences `\\` and `\'`, e.g. `'let\'s go'`.
 * Double quoted strings (`"hello world"`) which preserve any whitespace characters and
-  support common escape sequences such as `\t\r\n` etc., e.g. `"hello\nworld!"`.
-* Unquoted strings are terminated at the next whitespace character.
+  support common escape sequences such as `\t\r\n` etc., e.g. `"hi there\nworld!"`.
+* Unquoted strings are terminated at the next (unescaped) whitespace character and
+  support common escape sequences such as `\t\r\n` etc., e.g. `hi\ there\nworld!`.
 * Ignores excessive whitespace around arguments, such as trailing whitespace or
   multiple spaces between arguments.
 * Makes no assumptions about your input encoding, so this works with binary data

src/functions.php

@@ -46,8 +46,19 @@ function split($command)
         for (; isset($command[$i]); ++$i) {
             $c = $command[$i];
 
-            if ($inQuote === '"') {
-                // we're within a "double quoted" string
+            if ($inQuote === "'") {
+                // we're within a 'single quoted' string
+                if ($c === '\\' && isset($command[$i + 1]) && ($command[$i + 1] === "'" || $command[$i + 1] === '\\')) {
+                    // escaped single quote or backslash ends up as char in argument
+                    $argument .= $command[++$i];
+                    continue;
+                } elseif ($c === "'") {
+                    // single quote ends
+                    $inQuote = null;
+                    continue;
+                }
+            } else {
+                // we're not within any quotes or within a "double quoted" string
                 if ($c === '\\' && isset($command[$i + 1])) {
                     // any escaped character will be processed
                     $c = $command[++$i];
@@ -59,29 +70,15 @@ function split($command)
                         $argument .= $c;
                     }
                     continue;
-                } else if ($c === '"') {
+                } elseif ($inQuote === '"' && $c === '"') {
                     // double quote ends
                     $inQuote = null;
                     continue;
-                }
-            } elseif ($inQuote === "'") {
-                // we're within a 'single quoted' string
-                if ($c === '\\' && isset($command[$i + 1]) && ($command[$i + 1] === "'" || $command[$i + 1] === '\\')) {
-                    // escaped single quote or backslash ends up as char in argument
-                    $argument .= $command[++$i];
-                    continue;
-                } elseif ($c === "'") {
-                    // single quote ends
-                    $inQuote = null;
-                    continue;
-                }
-            } else {
-                // we're not within any quotes
-                if ($c === '"' || $c === "'") {
+                } elseif ($inQuote === null && ($c === '"' || $c === "'")) {
                     // start of quotes found
                     $inQuote = $c;
                     continue;
-                }elseif (in_array($c, $ws)) {
+                } elseif ($inQuote === null && in_array($c, $ws)) {
                     // whitespace character terminates unquoted argument
                     break;
                 }

tests/SplitTest.php

@@ -32,13 +32,27 @@ public function testSingleStringWithWhitespace()
         $this->assertEquals(array('hello'), $args);
     }
 
+    public function testSingleStringWithEscapedWhitespace()
+    {
+        $args = Arguments\split('hello\\ world');
+
+        $this->assertEquals(array('hello world'), $args);
+    }
+
     public function testSimpleStringWithArgument()
     {
         $args = Arguments\split('hello world');
 
         $this->assertEquals(array('hello', 'world'), $args);
     }
 
+    public function testSimpleStringWithArgumentWithInterpretedEscape()
+    {
+        $args = Arguments\split('hello world\\nthere!');
+
+        $this->assertEquals(array('hello', "world\nthere!"), $args);
+    }
+
     public function testSimpleStringWithArgumentsAndWhitespace()
     {
         $args = Arguments\split('  hello  world  ');