Fix using < or > in a reference name

These characters are valid and only the last `<...>` must be treated as the embeded URL. This stops using the inline lexer inside references, reSt doesn't allow nested inline nodes and the only thing required is getting the embeded URL (if there is any).
phpDocumentor · Dec 16, 2023 · cc2e28c · cc2e28c
1 parent b42e3e1
commit cc2e28c
Show file tree

Hide file tree

Showing 12 changed files with 29 additions and 127 deletions.
diff --git a/packages/guides-restructured-text/src/RestructuredText/Parser/InlineLexer.php b/packages/guides-restructured-text/src/RestructuredText/Parser/InlineLexer.php
@@ -12,6 +12,7 @@
 use function array_flip;
 use function parse_url;
 use function preg_match;
+use function strlen;
 
 use const PHP_URL_SCHEME;
 
@@ -23,8 +24,7 @@ final class InlineLexer extends AbstractLexer
     public const ANONYMOUS_END = 3;
     public const LITERAL = 5;
     public const BACKTICK = 6;
-    public const EMBEDED_URL_START = 9;
-    public const EMBEDED_URL_END = 10;
+    public const EMBEDED_URL = 9;
     public const NAMED_REFERENCE = 11;
     public const ANONYMOUSE_REFERENCE = 12;
     public const COLON = 13;
@@ -60,10 +60,9 @@ protected function getCatchablePatterns(): array
             '(?<=^|\s)[a-z0-9-]+_{2}', //Inline href.
             '(?<=^|\s)[a-z0-9-]+_{1}(?=[\s\.+]|$)', //Inline href.
             '``.+?``(?!`)',
+            '<[^<]+?>(?=`)',
             '_{2}',
             '_',
-            '<',
-            '>',
             '`',
             ':',
             '|',
@@ -133,13 +132,15 @@ protected function getType(string &$value)
             return self::WHITESPACE;
         }
 
+        if ($value[0] === '<' && strlen($value) > 1) {
+            return self::EMBEDED_URL;
+        }
+
         return match ($value) {
             '`' => self::BACKTICK,
             '**' => self::STRONG_DELIMITER,
             '*' => self::EMPHASIS_DELIMITER,
             '|' => self::VARIABLE_DELIMITER,
-            '<' => self::EMBEDED_URL_START,
-            '>' => self::EMBEDED_URL_END,
             '_' => self::UNDERSCORE,
             '__' => self::ANONYMOUS_END,
             ':' => self::COLON,

diff --git a/...ructured-text/src/RestructuredText/Parser/Productions/InlineRules/AnonymousPhraseRule.php b/...ructured-text/src/RestructuredText/Parser/Productions/InlineRules/AnonymousPhraseRule.php
@@ -8,6 +8,8 @@
 use phpDocumentor\Guides\RestructuredText\Parser\BlockContext;
 use phpDocumentor\Guides\RestructuredText\Parser\InlineLexer;
 
+use function substr;
+
 /**
  * Rule to parse for anonymous references
  *
@@ -45,11 +47,8 @@ public function apply(BlockContext $blockContext, InlineLexer $lexer): AbstractL
 
                     return $this->createAnonymousReference($blockContext, $text, $embeddedUrl);
 
-                case InlineLexer::EMBEDED_URL_START:
-                    $embeddedUrl = $this->parseEmbeddedUrl($lexer);
-                    if ($embeddedUrl === null) {
-                        $text .= '<';
-                    }
+                case InlineLexer::EMBEDED_URL:
+                    $embeddedUrl = substr($lexer->token->value, 1, -1);
 
                     break;
                 default:

diff --git a/...restructured-text/src/RestructuredText/Parser/Productions/InlineRules/NamedPhraseRule.php b/...restructured-text/src/RestructuredText/Parser/Productions/InlineRules/NamedPhraseRule.php
@@ -8,6 +8,8 @@
 use phpDocumentor\Guides\RestructuredText\Parser\BlockContext;
 use phpDocumentor\Guides\RestructuredText\Parser\InlineLexer;
 
+use function substr;
+
 /**
  * Rule to parse for named references
  *
@@ -48,11 +50,8 @@ public function apply(BlockContext $blockContext, InlineLexer $lexer): InlineNod
 
                     return $this->createReference($blockContext, $text, $embeddedUrl);
 
-                case InlineLexer::EMBEDED_URL_START:
-                    $embeddedUrl = $this->parseEmbeddedUrl($lexer);
-                    if ($embeddedUrl === null) {
-                        $text .= '<';
-                    }
+                case InlineLexer::EMBEDED_URL:
+                    $embeddedUrl = substr($lexer->token->value, 1, -1);
 
                     break;
                 case InlineLexer::WHITESPACE:

diff --git a/...s-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/ReferenceRule.php b/...s-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/ReferenceRule.php
@@ -8,7 +8,6 @@
 use phpDocumentor\Guides\Nodes\Inline\DocReferenceNode;
 use phpDocumentor\Guides\Nodes\Inline\HyperLinkNode;
 use phpDocumentor\Guides\RestructuredText\Parser\BlockContext;
-use phpDocumentor\Guides\RestructuredText\Parser\InlineLexer;
 
 use function filter_var;
 use function preg_replace;
@@ -41,35 +40,4 @@ protected function createReference(BlockContext $blockContext, string $link, str
 
         return new HyperLinkNode($link, $targetLink);
     }
-
-    protected function parseEmbeddedUrl(InlineLexer $lexer): string|null
-    {
-        if ($lexer->token === null) {
-            return null;
-        }
-
-        $startPosition = $lexer->token->position;
-        $text = '';
-
-        while ($lexer->moveNext()) {
-            $token = $lexer->token;
-            switch ($token->type) {
-                case InlineLexer::BACKTICK:
-                    //We did not find the expected SpanLexer::EMBEDED_URL_END
-                    $this->rollback($lexer, $startPosition);
-
-                    return null;
-
-                case InlineLexer::EMBEDED_URL_END:
-                    return $text;
-
-                default:
-                    $text .= $token->value;
-            }
-        }
-
-        $this->rollback($lexer, $startPosition);
-
-        return null;
-    }
 }
diff --git a/...ges/guides-restructured-text/src/RestructuredText/TextRoles/AbstractReferenceTextRole.php b/...ges/guides-restructured-text/src/RestructuredText/TextRoles/AbstractReferenceTextRole.php
@@ -6,25 +6,14 @@
 
 use phpDocumentor\Guides\Nodes\Inline\AbstractLinkInlineNode;
 use phpDocumentor\Guides\RestructuredText\Parser\DocumentParserContext;
-use phpDocumentor\Guides\RestructuredText\Parser\InlineLexer;
-use Psr\Log\LoggerInterface;
 
-use function sprintf;
-use function trim;
+use function preg_match;
 
 /** @see https://docutils.sourceforge.io/docs/ref/rst/restructuredtext.html#embedded-uris-and-aliases */
 abstract class AbstractReferenceTextRole implements TextRole
 {
     /** @see https://regex101.com/r/htMn5p/1 */
     public const INTERLINK_REGEX = '/^([a-zA-Z0-9-_]+):(.*$)/';
-    private readonly InlineLexer $lexer;
-
-    public function __construct(
-        private readonly LoggerInterface $logger,
-    ) {
-        // Do not inject the $lexer. It contains a state.
-        $this->lexer = new InlineLexer();
-    }
 
     public function processNode(
         DocumentParserContext $documentParserContext,
@@ -35,54 +24,15 @@ public function processNode(
         $referenceTarget = null;
         $value = null;
 
-        $part = '';
-        $this->lexer->setInput($content);
-        $this->lexer->moveNext();
-        $this->lexer->moveNext();
-        while ($this->lexer->token !== null) {
-            $token = $this->lexer->token;
-            switch ($token->type) {
-                case InlineLexer::EMBEDED_URL_START:
-                    $value = trim($part);
-                    $part = '';
-
-                    break;
-                case InlineLexer::EMBEDED_URL_END:
-                    if ($value === null) {
-                        // not inside the embedded URL
-                        $part .= $token->value;
-                        break;
-                    }
-
-                    if ($this->lexer->peek() !== null) {
-                        $this->logger->debug(
-                            sprintf(
-                                'Reference contains unexpected content after closing `>`, treating it as text like sphinx does: "%s"',
-                                $rawContent,
-                            ),
-                            $documentParserContext->getLoggerInformation(),
-                        );
-                        $part = $value . '<' . $part . '>';
-                        $value = null;
-                        break;
-                    }
-
-                    $referenceTarget = $part;
-                    $part = '';
-
-                    break 2;
-                default:
-                    $part .= $token->value;
-            }
-
-            $this->lexer->moveNext();
-        }
+        preg_match('/^(.+?)(?:\s*<(.+)>)?$/s', $content, $matches);
 
-        $value .= trim($part);
+        $value = $matches[1];
 
-        if ($referenceTarget === null) {
+        if (!isset($matches[2])) {
             $referenceTarget = $value;
             $value = null;
+        } else {
+            $referenceTarget = $matches[2];
         }
 
         return $this->createNode($referenceTarget, $value, $role);

diff --git a/packages/guides-restructured-text/src/RestructuredText/TextRoles/DocReferenceTextRole.php b/packages/guides-restructured-text/src/RestructuredText/TextRoles/DocReferenceTextRole.php
@@ -6,18 +6,11 @@
 
 use phpDocumentor\Guides\Nodes\Inline\AbstractLinkInlineNode;
 use phpDocumentor\Guides\Nodes\Inline\DocReferenceNode;
-use Psr\Log\LoggerInterface;
 
 use function preg_match;
 
 class DocReferenceTextRole extends AbstractReferenceTextRole
 {
-    public function __construct(
-        protected readonly LoggerInterface $logger,
-    ) {
-        parent::__construct($this->logger);
-    }
-
     final public const NAME = 'doc';
 
     public function getName(): string

diff --git a/...ages/guides-restructured-text/src/RestructuredText/TextRoles/GenericReferenceTextRole.php b/...ages/guides-restructured-text/src/RestructuredText/TextRoles/GenericReferenceTextRole.php
@@ -7,19 +7,16 @@
 use phpDocumentor\Guides\Nodes\Inline\AbstractLinkInlineNode;
 use phpDocumentor\Guides\Nodes\Inline\ReferenceNode;
 use phpDocumentor\Guides\ReferenceResolvers\AnchorReducer;
-use Psr\Log\LoggerInterface;
 
 use function array_keys;
 use function preg_match;
 
 class GenericReferenceTextRole extends AbstractReferenceTextRole
 {
     public function __construct(
-        protected readonly LoggerInterface $logger,
         private readonly GenericLinkProvider $genericLinkProvider,
         private readonly AnchorReducer $anchorReducer,
     ) {
-        parent::__construct($this->logger);
     }
 
     public function getName(): string

diff --git a/packages/guides-restructured-text/tests/unit/Parser/InlineTokenParserTest.php b/packages/guides-restructured-text/tests/unit/Parser/InlineTokenParserTest.php
@@ -4,7 +4,6 @@
 
 namespace phpDocumentor\Guides\RestructuredText\Parser;
 
-use Monolog\Logger;
 use phpDocumentor\Guides\Nodes\Inline\CitationInlineNode;
 use phpDocumentor\Guides\Nodes\Inline\DocReferenceNode;
 use phpDocumentor\Guides\Nodes\Inline\EmphasisInlineNode;
@@ -43,20 +42,18 @@
 
 final class InlineTokenParserTest extends TestCase
 {
-    public Logger $logger;
     private DocumentParserContext $documentParserContext;
     private InlineParser $inlineTokenParser;
     private DefaultTextRoleFactory $textRoleFactory;
 
     public function setUp(): void
     {
-        $this->logger = new Logger('test');
         $this->textRoleFactory = new DefaultTextRoleFactory(
             new GenericTextRole(),
             new LiteralTextRole(),
             [
-                new ReferenceTextRole($this->logger),
-                new DocReferenceTextRole($this->logger),
+                new ReferenceTextRole(),
+                new DocReferenceTextRole(),
             ],
         );
         $this->documentParserContext = new DocumentParserContext(

diff --git a/packages/guides-restructured-text/tests/unit/TextRoles/DocReferenceTextRoleTest.php b/packages/guides-restructured-text/tests/unit/TextRoles/DocReferenceTextRoleTest.php
@@ -4,7 +4,6 @@
 
 namespace phpDocumentor\Guides\RestructuredText\TextRoles;
 
-use Monolog\Logger;
 use phpDocumentor\Guides\Nodes\Inline\DocReferenceNode;
 use phpDocumentor\Guides\RestructuredText\Parser\DocumentParserContext;
 use PHPUnit\Framework\Attributes\DataProvider;
@@ -13,15 +12,13 @@
 
 class DocReferenceTextRoleTest extends TestCase
 {
-    private Logger $logger;
     private DocReferenceTextRole $docReferenceTextRole;
     private DocumentParserContext&MockObject $documentParserContext;
 
     public function setUp(): void
     {
-        $this->logger = new Logger('test');
         $this->documentParserContext = $this->createMock(DocumentParserContext::class);
-        $this->docReferenceTextRole = new DocReferenceTextRole($this->logger);
+        $this->docReferenceTextRole = new DocReferenceTextRole();
     }
 
     #[DataProvider('docReferenceProvider')]

diff --git a/packages/guides-restructured-text/tests/unit/TextRoles/ReferenceTextRoleTest.php b/packages/guides-restructured-text/tests/unit/TextRoles/ReferenceTextRoleTest.php
@@ -4,7 +4,6 @@
 
 namespace phpDocumentor\Guides\RestructuredText\TextRoles;
 
-use Monolog\Logger;
 use phpDocumentor\Guides\Nodes\Inline\ReferenceNode;
 use phpDocumentor\Guides\RestructuredText\Parser\DocumentParserContext;
 use PHPUnit\Framework\Attributes\DataProvider;
@@ -13,15 +12,13 @@
 
 class ReferenceTextRoleTest extends TestCase
 {
-    private Logger $logger;
     private ReferenceTextRole $referenceTextRole;
     private DocumentParserContext&MockObject $documentParserContext;
 
     public function setUp(): void
     {
-        $this->logger = new Logger('test');
         $this->documentParserContext = $this->createMock(DocumentParserContext::class);
-        $this->referenceTextRole = new ReferenceTextRole($this->logger);
+        $this->referenceTextRole = new ReferenceTextRole();
     }
 
     #[DataProvider('referenceProvider')]

diff --git a/tests/Functional/tests/links/links.html b/tests/Functional/tests/links/links.html
@@ -10,3 +10,4 @@
 <a href="http://www.ics.uci.edu/pub/ietf/uri/historical.html#WARNING">http://www.ics.uci.edu/pub/ietf/uri/historical.html#WARNING</a>.</p>
 <p>You can read more on the features of Embeddables objects <a href="http://docs.doctrine-project.org/en/latest/tutorials/embeddables.html">in the documentation</a>.</p>
 <p>Doctrine has a few links that are between brackets (<a href="https://www.doctrine-project.org/projects/doctrine-orm/en/2.16/reference/dql-doctrine-query-language.html#query-hints">such as here</a>)</p>
+<p>Links may use reserved characters as name like <a href="https://html.spec.whatwg.org/#the-head-element">this one linking to the &lt;head&gt; element</a>.</p>
diff --git a/tests/Functional/tests/links/links.rst b/tests/Functional/tests/links/links.rst
@@ -25,5 +25,8 @@ You can read more on the features of Embeddables objects `in the documentation
 Doctrine has a few links that are between brackets (`such as here
 <https://www.doctrine-project.org/projects/doctrine-orm/en/2.16/reference/dql-doctrine-query-language.html#query-hints>`_)
 
+Links may use reserved characters as name like `this one linking to the <head> element`_.
+
 .. _`xkcd`: http://xkcd.com/
 .. _something: http://something.com/
+.. _`this one linking to the <head> element`: https://html.spec.whatwg.org/#the-head-element