Backport new parser design, keeping BC

Author: GrahamCampbell

composer.json

@@ -11,7 +11,8 @@
         }
     ],
     "require": {
-        "php": ">=5.3.9"
+        "php": ">=5.3.9",
+        "symfony/polyfill-ctype": "^1.9"
     },
     "require-dev": {
         "phpunit/phpunit": "^4.8.35 || ^5.0"

src/Loader.php

@@ -2,7 +2,6 @@
 
 namespace Dotenv;
 
-use Dotenv\Exception\InvalidFileException;
 use Dotenv\Exception\InvalidPathException;
 
 /**
@@ -229,42 +228,7 @@ protected function sanitiseVariableValue($name, $value)
             return array($name, $value);
         }
 
-        if ($this->beginsWithAQuote($value)) { // value starts with a quote
-            $quote = $value[0];
-            $regexPattern = sprintf(
-                '/^
-                %1$s           # match a quote at the start of the value
-                (              # capturing sub-pattern used
-                 (?:           # we do not need to capture this
-                  [^%1$s\\\\]* # any character other than a quote or backslash
-                  |\\\\\\\\    # or two backslashes together
-                  |\\\\%1$s    # or an escaped quote e.g \"
-                 )*            # as many characters that match the previous rules
-                )              # end of the capturing sub-pattern
-                %1$s           # and the closing quote
-                .*$            # and discard any string after the closing quote
-                /mx',
-                $quote
-            );
-            $value = preg_replace($regexPattern, '$1', $value);
-            $value = str_replace("\\$quote", $quote, $value);
-            $value = str_replace('\\\\', '\\', $value);
-        } else {
-            $parts = explode(' #', $value, 2);
-            $value = trim($parts[0]);
-
-            // Unquoted values cannot contain whitespace
-            if (preg_match('/\s+/', $value) > 0) {
-                // Check if value is a comment (usually triggered when empty value with comment)
-                if (preg_match('/^#/', $value) > 0) {
-                    $value = '';
-                } else {
-                    throw new InvalidFileException('Dotenv values containing spaces must be surrounded by quotes.');
-                }
-            }
-        }
-
-        return array($name, trim($value));
+        return array($name, Parser::parseValue($value));
     }
 
     /**
@@ -308,21 +272,7 @@ function ($matchedPatterns) use ($loader) {
      */
     protected function sanitiseVariableName($name, $value)
     {
-        $name = trim(str_replace(array('export ', '\'', '"'), '', $name));
-
-        return array($name, $value);
-    }
-
-    /**
-     * Determine if the given string begins with a quote.
-     *
-     * @param string $value
-     *
-     * @return bool
-     */
-    protected function beginsWithAQuote($value)
-    {
-        return isset($value[0]) && ($value[0] === '"' || $value[0] === '\'');
+        return array(Parser::parseName($name), $value);
     }
 
     /**

src/Parser.php

@@ -0,0 +1,88 @@
+<?php
+
+namespace Dotenv;
+
+use Dotenv\Exception\InvalidFileException;
+
+class Parser
+{
+    const INITIAL_STATE = 0;
+    const UNQUOTED_STATE = 1;
+    const QUOTED_STATE = 2;
+    const ESCAPE_STATE = 3;
+    const WHITESPACE_STATE = 4;
+    const COMMENT_STATE = 5;
+
+    /**
+     * Parse the given variable name.
+     *
+     * @param string $name
+     *
+     * @return string
+     */
+    public static function parseName($name)
+    {
+        return trim(str_replace(array('export ', '\'', '"'), '', $name));
+    }
+
+    /**
+     * Parse the given variable value.
+     *
+     * @param string $value
+      *
+     * @throws \Dotenv\Exception\InvalidFileException
+     *
+     * @return string
+     */
+    public static function parseValue($value)
+    {
+        $data = array_reduce(str_split($value), function ($data, $char) use ($value) {
+            switch ($data[1]) {
+                case self::INITIAL_STATE:
+                    if ($char === '"') {
+                        return array($data[0], self::QUOTED_STATE);
+                    } else {
+                        return array($data[0].$char, self::UNQUOTED_STATE);
+                    }
+                case self::UNQUOTED_STATE:
+                    if ($char === '#') {
+                        return array($data[0], self::COMMENT_STATE);
+                    } elseif (ctype_space($char)) {
+                        return array($data[0], self::WHITESPACE_STATE);
+                    } else {
+                        return array($data[0].$char, self::UNQUOTED_STATE);
+                    }
+                case self::QUOTED_STATE:
+                    if ($char === '"') {
+                        return array($data[0], self::WHITESPACE_STATE);
+                    } elseif ($char === '\\') {
+                        return array($data[0], self::ESCAPE_STATE);
+                    } else {
+                        return array($data[0].$char, self::QUOTED_STATE);
+                    }
+                case self::ESCAPE_STATE:
+                    if ($char === '"' || $char === '\\') {
+                        return array($data[0].$char, self::QUOTED_STATE);
+                    } else {
+                        return array($data[0].'\\'.$char, self::QUOTED_STATE);
+                    }
+                case self::WHITESPACE_STATE:
+                    if ($char === '#') {
+                        return array($data[0], self::COMMENT_STATE);
+                    } elseif (!ctype_space($char)) {
+                        if ($data[0] !== '' && $data[0][0] === '#') {
+                            return array('', self::COMMENT_STATE);
+                        } else {
+                            throw new InvalidFileException('Dotenv values containing spaces must be surrounded by quotes.');
+                        }
+                    } else {
+                        return array($data[0], self::WHITESPACE_STATE);
+                    }
+                case self::COMMENT_STATE:
+                    return array($data[0], self::COMMENT_STATE);
+            }
+        }, array('', self::INITIAL_STATE));
+
+        return trim($data[0]);
+    }
+}