WIP standard / aggressive value matching modes

Author: thunderer

Makefile

@@ -7,7 +7,7 @@ composer-update:
 
 test:
 	docker-compose run --rm php-${PHP} php -v
-	docker-compose run --rm php-${PHP} php /app/vendor/bin/phpunit -c /app/phpunit.xml.dist
+	docker-compose run --rm php-${PHP} php /app/vendor/bin/phpunit -c /app/phpunit.xml.dist --filter "${TEST}"
 test-local:
 	php -v
 	php vendor/bin/phpunit

src/Parser/RegularParser.php

@@ -35,6 +35,12 @@ final class RegularParser implements ParserInterface
     const TOKEN_STRING = 6;
     const TOKEN_WS = 7;
 
+    const VALUE_REGULAR = 0x01;
+    const VALUE_AGGRESSIVE = 0x02;
+
+    /** @var int */
+    public $valueMode = self::VALUE_REGULAR;
+
     public function __construct(SyntaxInterface $syntax = null)
     {
         $this->lexerRegex = $this->prepareLexer($syntax ?: new CommonSyntax());
@@ -239,7 +245,16 @@ private function value()
         }
 
         if($this->lookahead(self::TOKEN_STRING) || $this->lookahead(self::TOKEN_MARKER)) {
-            while(false === ($this->lookahead(self::TOKEN_WS) || $this->lookahead(self::TOKEN_CLOSE) || $this->lookaheadN(array(self::TOKEN_MARKER, self::TOKEN_CLOSE)))) {
+            while(true) {
+                if($this->lookahead(self::TOKEN_WS) || $this->lookahead(self::TOKEN_CLOSE)) {
+                    break;
+                }
+                if($this->lookaheadN(array(self::TOKEN_MARKER, self::TOKEN_CLOSE))) {
+                    if($this->valueMode === self::VALUE_AGGRESSIVE) {
+                        $value .= $this->match(null, false);
+                    }
+                    break;
+                }
                 $value .= $this->match(null, false);
             }
 
@@ -301,6 +316,11 @@ private function lookahead($type)
         return $this->position < $this->tokensCount && $this->tokens[$this->position][0] === $type;
     }
 
+    /**
+     * @param int[] $types
+     *
+     * @return bool
+     */
     private function lookaheadN(array $types)
     {
         $count = count($types);
@@ -323,6 +343,12 @@ private function lookaheadN(array $types)
         return true;
     }
 
+    /**
+     * @param int|null $type
+     * @param bool $ws
+     *
+     * @return string
+     */
     private function match($type, $ws)
     {
         if($this->position >= $this->tokensCount) {

tests/ParserTest.php

@@ -252,7 +252,7 @@ public function provideShortcodes()
          *
          * Tests cases from array above with identifiers in the array below must be skipped.
          */
-        $wordpressSkip = array(3, 6, 16, 21, 22, 23, 25, 32, 33, 34, 46, 47, 49, 51);
+        $wordpressSkip = array(3, 6, 16, 21, 22, 23, 25, 32, 33, 34, 46, 47, 49, 51, 52);
         $result = array();
         foreach($tests as $key => $test) {
             $syntax = array_shift($test);
@@ -276,12 +276,35 @@ public function testIssue77()
             new ParsedShortcode(new Shortcode('x', array(), '', null), '[x][/x]', 3),
             new ParsedShortcode(new Shortcode('y', array(), 'x', null), '[y]x[/y]', 22),
         ));
-
         $this->assertShortcodes($parser->parse('[a k="v][x][/x]'), array(
             new ParsedShortcode(new Shortcode('x', array(), '', null), '[x][/x]', 8),
         ));
     }
 
+    public function testValueModeAggressive()
+    {
+        $parser = new RegularParser(new CommonSyntax());
+        $parser->valueMode = RegularParser::VALUE_AGGRESSIVE;
+        $parsed = $parser->parse('[x=/[/] [y a=/"//] [z=http://url/] [a=http://url ]');
+        $tested = array(
+            new ParsedShortcode(new Shortcode('x', array(), null, '/[/'), '[x=/[/]', 0),
+            new ParsedShortcode(new Shortcode('y', array('a' => '/"//'), null, null), '[y a=/"//]', 8),
+            new ParsedShortcode(new Shortcode('z', array(), null, 'http://url/'), '[z=http://url/]', 19),
+            new ParsedShortcode(new Shortcode('a', array(), null, 'http://url'), '[a=http://url ]', 35),
+        );
+
+        $count = count($tested);
+        static::assertCount($count, $parsed, 'counts');
+        for ($i = 0; $i < $count; $i++) {
+            static::assertSame($tested[$i]->getName(), $parsed[$i]->getName(), 'name');
+            static::assertSame($tested[$i]->getParameters(), $parsed[$i]->getParameters(), 'parameters');
+            static::assertSame($tested[$i]->getContent(), $parsed[$i]->getContent(), 'content');
+            static::assertSame($tested[$i]->getText(), $parsed[$i]->getText(), 'text');
+            static::assertSame($tested[$i]->getOffset(), $parsed[$i]->getOffset(), 'offset');
+            static::assertSame($tested[$i]->getBbCode(), $parsed[$i]->getBbCode(), 'bbCode');
+        }
+    }
+
     public function testWordPress()
     {
         $parser = new WordpressParser();