Merge pull request #14 from tempestphp/terminated_underscore

fix: terminated underscores and asterisks

Author: brendt

mago.toml

@@ -46,6 +46,7 @@ interface-name = { psr = false }
 trait-name = { psr = false }
 class-name = { psr = false }
 literal-named-argument = { enabled = false } # todo
+no-insecure-comparison = { enabled = false }
 no-error-control-operator = { enabled = false }
 too-many-methods = { enabled = false }
 kan-defect = { enabled = false }

src/Rules/BoldRule.php

@@ -12,19 +12,31 @@
 final class BoldRule implements Rule, ProvidesFirstChar, ProvidesStopChar
 {
     private(set) string $firstChar = '*_';
-    private(set) string $stopChar = '_*';
+    private(set) string $stopChar = '*_';
 
     public function shouldParse(Parser $parser): bool
     {
-        if ($parser->comesNext('**', length: 2)) {
-            return ! $parser->comesNext('*', length: 1, offset: 2);
+        $stopToken = $parser->lookaheadUntil('*_')[0] ?? null;
+
+        if (! $stopToken) {
+            return false;
+        }
+
+        $lookahead = $parser->lookaheadUntil($stopToken, $stopToken, $stopToken, $stopToken);
+
+        if (count($lookahead) !== 4) {
+            return false;
         }
 
-        if ($parser->comesNext('__', length: 2)) {
-            return ! $parser->comesNext('_', length: 1, offset: 2);
+        $content = $lookahead[2];
+
+        $lastChar = substr($content, strlen($content) - 1, 1);
+
+        if ($lastChar !== $stopToken) {
+            return false;
         }
 
-        return false;
+        return true;
     }
 
     public function parse(Parser $parser): Token

src/Rules/ItalicRule.php

@@ -16,15 +16,32 @@ final class ItalicRule implements Rule, ProvidesFirstChar, ProvidesStopChar
 
     public function shouldParse(Parser $parser): bool
     {
-        if ($parser->comesNext('_', length: 1)) {
-            return ! $parser->comesNext('_', length: 1, offset: 1);
+        $stopToken = $parser->lookaheadUntil('*_')[0] ?? null;
+
+        if (! $stopToken) {
+            return false;
+        }
+
+        $lookahead = $parser->lookaheadUntil($stopToken, $stopToken);
+
+        if (count($lookahead) !== 2) {
+            return false;
+        }
+
+        $end = $lookahead[1];
+
+        $firstChar = substr($end, 0, 1);
+        $lastChar = substr($end, strlen($end) - 1, 1);
+
+        if ($firstChar === $stopToken) {
+            return false;
         }
 
-        if ($parser->comesNext('*', length: 1)) {
-            return ! $parser->comesNext('*', length: 1, offset: 1);
+        if ($lastChar !== $stopToken) {
+            return false;
         }
 
-        return false;
+        return true;
     }
 
     public function parse(Parser $parser): Token

tests/Rules/BoldAndItalicRuleTest.php

@@ -27,6 +27,38 @@ public function test_triple_underscore_bold_and_italic(): void
         $this->assertSame('<strong><em>text</em></strong>', $html);
     }
 
+    #[Test]
+    public function test_single_underscore_double_asterisk(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new BoldAndItalicRule(), new BoldRule(), new ItalicRule()])->parse('_**text**_');
+
+        $this->assertSame('<em><strong>text</strong></em>', $html);
+    }
+
+    #[Test]
+    public function test_single_asterisk_double_underscore(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new BoldAndItalicRule(), new BoldRule(), new ItalicRule()])->parse('*__text__*');
+
+        $this->assertSame('<em><strong>text</strong></em>', $html);
+    }
+
+    #[Test]
+    public function test_double_underscore_single_asterisk(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new BoldAndItalicRule(), new BoldRule(), new ItalicRule()])->parse('__*text*__');
+
+        $this->assertSame('<strong><em>text</em></strong>', $html);
+    }
+
+    #[Test]
+    public function test_double_asterisk_single_underscore(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new BoldAndItalicRule(), new BoldRule(), new ItalicRule()])->parse('**_text_**');
+
+        $this->assertSame('<strong><em>text</em></strong>', $html);
+    }
+
     #[Test]
     public function test_does_not_lex_double_asterisk(): void
     {

tests/Rules/BoldRuleTest.php

@@ -5,6 +5,9 @@
 use PHPUnit\Framework\Attributes\Test;
 use Tempest\Markdown\Parser;
 use Tempest\Markdown\Rules\BoldRule;
+use Tempest\Markdown\Rules\ItalicRule;
+use Tempest\Markdown\Rules\NewLineRule;
+use Tempest\Markdown\Rules\ParagraphRule;
 use Tempest\Markdown\Tests\ParserTestCase;
 
 class BoldRuleTest extends ParserTestCase
@@ -17,10 +20,18 @@ public function test_lex_double_asterisk(): void
         $this->assertSame('<strong>bold</strong>', $html);
     }
 
+    #[Test]
+    public function test_double_asterisk_must_be_terminated(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new NewLineRule(), new BoldRule(), new ParagraphRule()])->parse("Hello**world\n\nHi");
+
+        $this->assertSame("<p>Hello**world</p>\n\n<p>Hi</p>", $html);
+    }
+
     #[Test]
     public function test_lex_asterisk_with_underscore(): void
     {
-        $html = (string) new Parser(highlighter: null, rules: [new BoldRule()])->parse('**_bold_**');
+        $html = (string) new Parser(highlighter: null, rules: [new BoldRule(), new ItalicRule()])->parse('**_bold_**');
 
         $this->assertSame('<strong><em>bold</em></strong>', $html);
     }
@@ -41,6 +52,14 @@ public function test_lex_double_underscore(): void
         $this->assertSame('<strong>bold</strong>', $html);
     }
 
+    #[Test]
+    public function test_double_underscore_must_be_terminated(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new NewLineRule(), new BoldRule(), new ParagraphRule()])->parse("Hello__world\n\nHi");
+
+        $this->assertSame("<p>Hello__world</p>\n\n<p>Hi</p>", $html);
+    }
+
     #[Test]
     public function test_does_not_lex_single_underscore(): void
     {

tests/Rules/ItalicRuleTest.php

@@ -5,6 +5,8 @@
 use PHPUnit\Framework\Attributes\Test;
 use Tempest\Markdown\Parser;
 use Tempest\Markdown\Rules\ItalicRule;
+use Tempest\Markdown\Rules\NewLineRule;
+use Tempest\Markdown\Rules\ParagraphRule;
 use Tempest\Markdown\Tests\ParserTestCase;
 
 class ItalicRuleTest extends ParserTestCase
@@ -17,11 +19,27 @@ public function test_lex_with_underscore(): void
         $this->assertSame('<em>italic</em>', $html);
     }
 
+    #[Test]
+    public function test_underscore_must_be_terminated(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new NewLineRule(), new ItalicRule(), new ParagraphRule()])->parse("Hello_world\n\nHi");
+
+        $this->assertSame("<p>Hello_world</p>\n\n<p>Hi</p>", $html);
+    }
+
     #[Test]
     public function test_lex_with_asterisk(): void
     {
         $html = (string) new Parser(highlighter: null, rules: [new ItalicRule()])->parse('*italic*');
 
         $this->assertSame('<em>italic</em>', $html);
     }
+
+    #[Test]
+    public function test_asterisk_must_be_terminated(): void
+    {
+        $html = (string) new Parser(highlighter: null, rules: [new NewLineRule(), new ItalicRule(), new ParagraphRule()])->parse("Hello*world\n\nHi");
+
+        $this->assertSame("<p>Hello*world</p>\n\n<p>Hi</p>", $html);
+    }
 }