Rewrite parser and fix escaped quotes in multiline values (#322)

vlucas · Jan 26, 2019 · 21c540a · 21c540a
1 parent 2ecc2da
commit 21c540a
Show file tree

Hide file tree

Showing 9 changed files with 145 additions and 99 deletions.
diff --git a/composer.json b/composer.json
@@ -12,7 +12,8 @@
     ],
     "require": {
         "php": "^5.4 || ^7.0",
-        "phpoption/phpoption": "^1.5"
+        "phpoption/phpoption": "^1.5",
+        "symfony/polyfill-ctype": "^1.9"
     },
     "require-dev": {
         "phpunit/phpunit": "^4.8.35 || ^5.0 || ^6.0"

diff --git a/src/Lines.php b/src/Lines.php
@@ -22,7 +22,7 @@ public static function process(array $lines)
         foreach ($lines as $line) {
             list($multiline, $line, $multilineBuffer) = self::multilineProcess($multiline, $line, $multilineBuffer);
 
-            if (!$multiline && !self::isComment($line) && self::looksLikeSetter($line)) {
+            if (!$multiline && !self::isComment($line) && self::isSetter($line)) {
                 $output[] = $line;
             }
         }
@@ -68,7 +68,11 @@ private static function multilineProcess($multiline, $line, array $buffer)
      */
     private static function looksLikeMultilineStart($line)
     {
-        return strpos($line, '="') !== false && substr_count($line, '"') === 1;
+        if (strpos($line, '="') === false) {
+            return false;
+        }
+
+        return self::looksLikeMultilineStop($line) === false;
     }
 
     /**
@@ -80,7 +84,31 @@ private static function looksLikeMultilineStart($line)
      */
     private static function looksLikeMultilineStop($line)
     {
-        return strpos($line, '"') !== false && substr_count($line, '="') === 0;
+        if ($line === '"') {
+            return true;
+        }
+
+        foreach (self::getCharPairs(str_replace('\\\\', '', $line)) as $pair) {
+            if ($pair[0] !== '\\' && $pair[0] !== '=' && $pair[1] === '"') {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    /**
+     * Get all pairs of adjacent characters within the line.
+     *
+     * @param string $line
+     *
+     * @return bool
+     */
+    private static function getCharPairs($line)
+    {
+        $chars = str_split($line);
+
+        return array_map(null, $chars, array_slice($chars, 1));
     }
 
     /**
@@ -104,7 +132,7 @@ private static function isComment($line)
      *
      * @return bool
      */
-    private static function looksLikeSetter($line)
+    private static function isSetter($line)
     {
         return strpos($line, '=') !== false;
     }

diff --git a/src/Parser.php b/src/Parser.php
@@ -3,18 +3,19 @@
 namespace Dotenv;
 
 use Dotenv\Exception\InvalidFileException;
-use Dotenv\Regex\Regex;
 
 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 environment variable entry into a name and value.
      *
-     * Takes value as passed in by developer and:
-     * - breaks up the line into a name and value,
-     * - cleaning the value of quotes,
-     * - cleaning the name of quotes.
-     *
      * @param string $entry
      *
      * @throws \Dotenv\Exception\InvalidFileException
@@ -25,14 +26,12 @@ public static function parse($entry)
     {
         list($name, $value) = self::splitStringIntoParts($entry);
 
-        return [self::sanitiseName($name), self::sanitiseValue($value)];
+        return [self::parseName($name), self::parseValue($value)];
     }
 
     /**
      * Split the compound string into parts.
      *
-     * If the `$line` contains an `=` sign, then we split it into 2 parts.
-     *
      * @param string $line
      *
      * @throws \Dotenv\Exception\InvalidFileException
@@ -66,7 +65,7 @@ private static function splitStringIntoParts($line)
      *
      * @return string
      */
-    private static function sanitiseName($name)
+    private static function parseName($name)
     {
         $name = trim(str_replace(['export ', '\'', '"'], '', $name));
 
@@ -100,71 +99,60 @@ private static function isValidName($name)
      *
      * @return string|null
      */
-    private static function sanitiseValue($value)
+    private static function parseValue($value)
     {
         if ($value === null || trim($value) === '') {
             return $value;
         }
 
-        if (self::beginsWithAQuote($value)) {
-            return self::processQuotedValue($value);
-        }
-
-        // Strip comments from the left
-        $value = explode(' #', $value, 2)[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(
-                    self::getErrorMessage('an unexpected space', $value)
-                );
+        return array_reduce(str_split($value), function ($data, $char) use ($value) {
+            switch ($data[1]) {
+                case self::INITIAL_STATE:
+                    if ($char === '"') {
+                        return [$data[0], self::QUOTED_STATE];
+                    } elseif ($char === '#') {
+                        return [$data[0], self::COMMENT_STATE];
+                    } else {
+                        return [$data[0].$char, self::UNQUOTED_STATE];
+                    }
+                case self::UNQUOTED_STATE:
+                    if ($char === '#') {
+                        return [$data[0], self::COMMENT_STATE];
+                    } elseif (ctype_space($char)) {
+                        return [$data[0], self::WHITESPACE_STATE];
+                    } else {
+                        return [$data[0].$char, self::UNQUOTED_STATE];
+                    }
+                case self::QUOTED_STATE:
+                    if ($char === '"') {
+                        return [$data[0], self::WHITESPACE_STATE];
+                    } elseif ($char === '\\') {
+                        return [$data[0], self::ESCAPE_STATE];
+                    } else {
+                        return [$data[0].$char, self::QUOTED_STATE];
+                    }
+                case self::ESCAPE_STATE:
+                    if ($char === '"' || $char === '\\') {
+                        return [$data[0].$char, self::QUOTED_STATE];
+                    } else {
+                        throw new InvalidFileException(
+                            self::getErrorMessage('an unexpected escape sequence', $value)
+                        );
+                    }
+                case self::WHITESPACE_STATE:
+                    if ($char === '#') {
+                        return [$data[0], self::COMMENT_STATE];
+                    } elseif (!ctype_space($char)) {
+                        throw new InvalidFileException(
+                            self::getErrorMessage('unexpected whitespace', $value)
+                        );
+                    } else {
+                        return [$data[0], self::WHITESPACE_STATE];
+                    }
+                case self::COMMENT_STATE:
+                    return [$data[0], self::COMMENT_STATE];
             }
-        }
-
-        return $value;
-    }
-
-    /**
-     * Strips quotes from the environment variable value.
-     *
-     * @param string $value
-     *
-     * @return string
-     */
-    private static function processQuotedValue($value)
-    {
-        $quote = $value[0];
-
-        $pattern = sprintf(
-            '/^
-            %1$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',
-            $quote
-        );
-
-        return Regex::replace($pattern, '$1', $value)
-            ->mapSuccess(function ($str) use ($quote) {
-                return str_replace('\\\\', '\\', str_replace("\\$quote", $quote, $str));
-            })
-            ->mapError(function ($err) use ($value) {
-                throw new InvalidFileException(
-                    self::getErrorMessage(sprintf('a quote parsing error (%s)', $err), $value)
-                );
-            })
-            ->getSuccess();
+        }, ['', self::INITIAL_STATE])[0];
     }
 
     /**
@@ -183,16 +171,4 @@ private static function getErrorMessage($cause, $subject)
             strtok($subject, "\n")
         );
     }
-
-    /**
-     * Determine if the given string begins with a quote.
-     *
-     * @param string $value
-     *
-     * @return bool
-     */
-    private static function beginsWithAQuote($value)
-    {
-        return isset($value[0]) && ($value[0] === '"' || $value[0] === '\'');
-    }
 }
diff --git a/tests/Dotenv/DotenvTest.php b/tests/Dotenv/DotenvTest.php
@@ -72,6 +72,8 @@ public function testCommentedDotenvLoadsEnvironmentVars()
         $this->assertSame('a value with a # character & a quote " character inside quotes', getenv('CQUOTESWITHQUOTE'));
         $this->assertEmpty(getenv('CNULL'));
         $this->assertEmpty(getenv('EMPTY'));
+        $this->assertEmpty(getenv('EMPTY2'));
+        $this->assertSame('foo', getenv('FOOO'));
     }
 
     public function testQuotedDotenvLoadsEnvironmentVars()
@@ -266,6 +268,13 @@ public function testDotenvAllowsSpecialCharacters()
         $this->assertSame('test some escaped characters like a quote " or maybe a backslash \\', getenv('SPVAR5'));
     }
 
+    public function testMutlilineLoading()
+    {
+        $dotenv = Dotenv::create($this->fixturesFolder, 'multiline.env');
+        $dotenv->load();
+        $this->assertSame("test\n     test\"test\"\n     test", getenv('TEST'));
+    }
+
     public function testDotenvAssertions()
     {
         $dotenv = Dotenv::create($this->fixturesFolder, 'assertions.env');

diff --git a/tests/Dotenv/LinesTest.php b/tests/Dotenv/LinesTest.php
@@ -5,7 +5,7 @@
 
 class LinesTest extends TestCase
 {
-    public function testProcess()
+    public function testProcessBasic()
     {
         $content = file_get_contents(dirname(__DIR__).'/fixtures/env/assertions.env');
 
@@ -14,13 +14,37 @@ public function testProcess()
             'ASSERTVAR2=""',
             'ASSERTVAR3="val3   "',
             'ASSERTVAR4="0" # empty looking value',
-            'ASSERTVAR5=#foo',
+            'ASSERTVAR5="#foo"',
             "ASSERTVAR6=\"val1\nval2\"",
             "ASSERTVAR7=\"\nval3\" #",
             "ASSERTVAR8=\"val3\n\"",
-            "ASSERTVAR9=\"\n\"",
+            "ASSERTVAR9=\"\n\n\"",
         ];
 
         $this->assertSame($expected, Lines::process(preg_split("/(\r\n|\n|\r)/", $content)));
     }
+
+    public function testProcessQuotes()
+    {
+        $content = file_get_contents(dirname(__DIR__).'/fixtures/env/multiline.env');
+
+        $expected = [
+            "TEST=\"test\n     test\\\"test\\\"\n     test\"",
+        ];
+
+        $this->assertSame($expected, Lines::process(preg_split("/(\r\n|\n|\r)/", $content)));
+    }
+
+    public function testProcessClosingSlash()
+    {
+        $lines = [
+            'SPVAR5="test some escaped characters like a quote \" or maybe a backslash \\" # not escaped',
+        ];
+
+        $expected = [
+            'SPVAR5="test some escaped characters like a quote \" or maybe a backslash \\" # not escaped',
+        ];
+
+        $this->assertSame($expected, $lines);
+    }
 }
diff --git a/tests/Dotenv/ParserTest.php b/tests/Dotenv/ParserTest.php
@@ -25,9 +25,17 @@ public function testExportParse()
         $this->assertSame(['FOO', 'bar baz'], Parser::parse('export FOO="bar baz"'));
     }
 
+    public function testClosingSlashParse()
+    {
+        $content = 'SPVAR5="test some escaped characters like a quote \\" or maybe a backslash \\\\" # not escaped';
+        $expected = ['SPVAR5', 'test some escaped characters like a quote " or maybe a backslash \\'];
+
+        $this->assertSame($expected, Parser::parse($content));
+    }
+
     /**
      * @expectedException \Dotenv\Exception\InvalidFileException
-     * @expectedExceptionMessage Failed to parse dotenv file due to an unexpected space. Failed at [bar baz].
+     * @expectedExceptionMessage Failed to parse dotenv file due to unexpected whitespace. Failed at [bar baz].
      */
     public function testParseInvalidSpaces()
     {
@@ -54,16 +62,10 @@ public function testParseInvalidName()
 
     /**
      * @expectedException \Dotenv\Exception\InvalidFileException
-     * @expectedExceptionMessage Failed to parse dotenv file due to a quote parsing error (PREG_
+     * @expectedExceptionMessage Failed to parse dotenv file due to an unexpected escape sequence. Failed at ["iiiiviiiixiiiiviiii\n"].
      */
-    public function testParserFailsWithException()
+    public function testParserEscaping()
     {
-        $limit = (int) ini_get('pcre.backtrack_limit');
-
-        if ($limit > 1000000) {
-            $this->markTestSkipped('System pcre.backtrack_limit too large.');
-        }
-
         Parser::parse('FOO_BAD="iiiiviiiixiiiiviiii\\n"');
     }
 }
diff --git a/tests/fixtures/env/assertions.env b/tests/fixtures/env/assertions.env
@@ -4,7 +4,7 @@ ASSERTVAR2=""
 
 ASSERTVAR3="val3   "
 ASSERTVAR4="0" # empty looking value
-ASSERTVAR5=#foo
+ASSERTVAR5="#foo"
 ASSERTVAR6="val1
 val2"
 ASSERTVAR7="
@@ -14,4 +14,5 @@ val3" #
 ASSERTVAR8="val3
 "
 ASSERTVAR9="
+
 "
diff --git a/tests/fixtures/env/commented.env b/tests/fixtures/env/commented.env
@@ -6,6 +6,8 @@ CSPACED="with spaces" # this is a comment
 CQUOTES="a value with a # character" # this is a comment
 CQUOTESWITHQUOTE="a value with a # character & a quote \" character inside quotes" # " this is a comment
 EMPTY= # comment with empty variable
+EMPTY2=# comment with empty variable
+FOOO=foo# comment with no space
 BOOLEAN=yes # (yes, no)
 
 CNULL=

diff --git a/tests/fixtures/env/multiline.env b/tests/fixtures/env/multiline.env
@@ -0,0 +1,3 @@
+TEST="test
+     test\"test\"
+     test"