Single and double quote too greedy

tempestphp · Mar 20, 2024 · 3989089 · 3989089
1 parent 0a869fc
commit 3989089
Show file tree

Hide file tree

Showing 17 changed files with 1,541 additions and 87 deletions.
diff --git a/src/Languages/JavaScript/Patterns/JsDoubleQuoteValuePattern.php b/src/Languages/JavaScript/Patterns/JsDoubleQuoteValuePattern.php
@@ -13,22 +13,13 @@
     input: 'return "hello"',
     output: 'hello',
 )]
-
-#[PatternTest(
-    input: 'return "hello
-    
-    world"',
-    output: 'hello
-    
-    world',
-)]
 final readonly class JsDoubleQuoteValuePattern implements Pattern
 {
     use IsPattern;
 
     public function getPattern(): string
     {
-        return '"(?<match>(\n|.)*?)"';
+        return '"(?<match>.*?)"';
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/JavaScript/Patterns/JsSingleQuoteValuePattern.php b/src/Languages/JavaScript/Patterns/JsSingleQuoteValuePattern.php
@@ -13,21 +13,13 @@
     input: "return 'hello';",
     output: 'hello',
 )]
-#[PatternTest(
-    input: "return 'hello
-    
-    world';",
-    output: 'hello
-    
-    world',
-)]
 final readonly class JsSingleQuoteValuePattern implements Pattern
 {
     use IsPattern;
 
     public function getPattern(): string
     {
-        return "'(?<match>(\n|.)*?)'";
+        return "'(?<match>.*?)'";
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/Json/Patterns/JsonDoubleQuoteValuePattern.php b/src/Languages/Json/Patterns/JsonDoubleQuoteValuePattern.php
@@ -16,7 +16,7 @@
 
     public function getPattern(): string
     {
-        return '(?<match>\"(.|\n)*?\")';
+        return '(?<match>\".*?\")';
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/Php/Patterns/DoubleQuoteValuePattern.php b/src/Languages/Php/Patterns/DoubleQuoteValuePattern.php
@@ -13,22 +13,13 @@
     input: 'return "hello"',
     output: 'hello',
 )]
-
-#[PatternTest(
-    input: 'return "hello
-    
-    world"',
-    output: 'hello
-    
-    world',
-)]
 final readonly class DoubleQuoteValuePattern implements Pattern
 {
     use IsPattern;
 
     public function getPattern(): string
     {
-        return '"(?<match>(\n|.)*?)"';
+        return '"(?<match>.*?)"';
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/Php/Patterns/SingleQuoteValuePattern.php b/src/Languages/Php/Patterns/SingleQuoteValuePattern.php
@@ -13,21 +13,13 @@
     input: "return 'hello';",
     output: 'hello',
 )]
-#[PatternTest(
-    input: "return 'hello
-    
-    world';",
-    output: 'hello
-    
-    world',
-)]
 final readonly class SingleQuoteValuePattern implements Pattern
 {
     use IsPattern;
 
     public function getPattern(): string
     {
-        return "'(?<match>(\n|.)*?)'";
+        return "'(?<match>.*?)'";
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/Php/PhpLanguage.php b/src/Languages/Php/PhpLanguage.php
@@ -131,10 +131,6 @@ public function getPatterns(): array
             new MultilineSingleDocCommentPattern(),
             new SinglelineCommentPattern(),
 
-            // VALUES
-            new SingleQuoteValuePattern(),
-            new DoubleQuoteValuePattern(),
-
             // TYPES
             new AttributeTypePattern(),
             new ImplementsPattern(),
@@ -158,6 +154,10 @@ public function getPatterns(): array
             new FunctionCallPattern(),
             new ConstantPropertyPattern(),
             new ConstantNamePattern(),
+
+            // VALUES
+            new SingleQuoteValuePattern(),
+            new DoubleQuoteValuePattern(),
         ];
     }
 }
diff --git a/src/Languages/Sql/Patterns/SqlDoubleQuoteValuePattern.php b/src/Languages/Sql/Patterns/SqlDoubleQuoteValuePattern.php
@@ -10,18 +10,13 @@
 use Tempest\Highlight\Tokens\TokenType;
 
 #[PatternTest(input: 'bar = "baz"', output: 'baz')]
-#[PatternTest(input: 'bar = "ba
-
-z"', output: 'ba
-
-z')]
 final readonly class SqlDoubleQuoteValuePattern implements Pattern
 {
     use IsPattern;
 
     public function getPattern(): string
     {
-        return '\"(?<match>(.|\n)*?)\"';
+        return '\"(?<match>.*?)\"';
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/Sql/Patterns/SqlSingleQuoteValuePattern.php b/src/Languages/Sql/Patterns/SqlSingleQuoteValuePattern.php
@@ -10,18 +10,13 @@
 use Tempest\Highlight\Tokens\TokenType;
 
 #[PatternTest(input: "bar = 'baz'", output: 'baz')]
-#[PatternTest(input: "bar = 'ba
-
-z'", output: 'ba
-
-z')]
 final readonly class SqlSingleQuoteValuePattern implements Pattern
 {
     use IsPattern;
 
     public function getPattern(): string
     {
-        return '\'(?<match>(.|\n)*?)\'';
+        return '\'(?<match>.*?)\'';
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/Yaml/Patterns/YamlDoubleQuoteValuePattern.php b/src/Languages/Yaml/Patterns/YamlDoubleQuoteValuePattern.php
@@ -16,7 +16,7 @@
 
     public function getPattern(): string
     {
-        return '\"(?<match>(.|\n)*?)\"';
+        return '\"(?<match>.*?)\"';
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Languages/Yaml/Patterns/YamlSingleQuoteValuePattern.php b/src/Languages/Yaml/Patterns/YamlSingleQuoteValuePattern.php
@@ -16,7 +16,7 @@
 
     public function getPattern(): string
     {
-        return '\'(?<match>(.|\n)*?)\'';
+        return '\'(?<match>.*?)\'';
     }
 
     public function getTokenType(): TokenType

diff --git a/src/Tokens/GroupTokens.php b/src/Tokens/GroupTokens.php
@@ -6,6 +6,10 @@
 
 final readonly class GroupTokens
 {
+    /**
+     * @param \Tempest\Highlight\Tokens\Token[] $tokens
+     * @return \Tempest\Highlight\Tokens\Token[]
+     */
     public function __invoke(array $tokens): array
     {
         // Sort tokens in the right order
@@ -25,7 +29,7 @@ public function __invoke(array $tokens): array
             $token = $token->cloneWithoutParent();
 
             foreach ($tokens as $compareKey => $compareToken) {
-                if ($token->contains($compareToken)) {
+                if ($token->containsOrOverlaps($compareToken)) {
                     if ($token->canContain($compareToken)) {
                         $token->addChild($compareToken);
                     }

diff --git a/src/Tokens/Token.php b/src/Tokens/Token.php
@@ -33,12 +33,12 @@ public function equals(Token $other): bool
         //            && $this->type === $other->type;
     }
 
-    public function contains(Token $other): bool
+    public function containsOrOverlaps(Token $other): bool
     {
         return
             ! $this->equals($other)
             && $this->start <= $other->start
-            && $this->end >= $other->end;
+            && $other->start < $this->end;
     }
 
     public function addChild(Token $child): void

diff --git a/tests/Languages/Php/PhpLanguageTest.php b/tests/Languages/Php/PhpLanguageTest.php
@@ -29,6 +29,18 @@ public static function data(): array
             ["public string|\Stringable \$message;", '<span class="hl-keyword">public</span> <span class="hl-type">string|\<span class="hl-type">Stringable</span></span> <span class="hl-property">$message</span>;'],
             ['for($x = 0; $x < 150; $x++) {', '<span class="hl-keyword">for</span>($x = 0; $x &lt; 150; $x++) {'],
             ["'namespace ';", "'<span class=\"hl-value\">namespace </span>';"],
+            [
+                "// We'll
+
+foo()
+
+// We'll",
+                '<span class="hl-comment">// We\'ll</span>
+
+<span class="hl-property">foo</span>()
+
+<span class="hl-comment">// We\'ll</span>',
+            ],
         ];
     }
 }
diff --git a/tests/TokenTest.php b/tests/TokenTest.php
@@ -26,7 +26,7 @@ public function test_contains(): void
             type: TokenType::VALUE,
         );
 
-        $this->assertTrue($a->contains($b));
-        $this->assertFalse($b->contains($a));
+        $this->assertTrue($a->containsOrOverlaps($b));
+        $this->assertFalse($b->containsOrOverlaps($a));
     }
 }
diff --git a/tests/index.php b/tests/index.php
@@ -20,7 +20,7 @@
 
 $markdown = new MarkdownConverter($environment);
 
-$contents = $markdown->convert(file_get_contents(__DIR__ . '/test.md'))->getContent();
+$contents = $markdown->convert(file_get_contents(__DIR__ . '/test-big.md'))->getContent();
 
 ?>