Fix #30, bug where the package cannot properly parse a file that contains Markdown code blocks with front matter in them

src/ComplexMarkdownParser.php

@@ -0,0 +1,99 @@
+<?php
+
+namespace Spatie\YamlFrontMatter;
+
+use Symfony\Component\Yaml\Yaml;
+
+class ComplexMarkdownParser
+{
+    protected $content;
+    protected $lines;
+
+    public $frontMatterStartLine;
+    public $frontMatterEndLine;
+
+    public function __construct(string $content)
+    {
+        $this->content = $content;
+        $this->lines = explode("\n", $this->content);
+    }
+
+    public function parse(): Document
+    {
+        $this->findFrontMatterStartAndEndLineNumbers();
+
+        if (!$this->hasFrontMatter()) {
+            return new Document([], $this->content);
+        }
+
+        $matter = $this->getFrontMatter();
+        $body = $this->getBody();
+
+        $matter = Yaml::parse($matter);
+
+        return new Document($matter, $body);
+    }
+
+    protected function findFrontMatterStartAndEndLineNumbers()
+    {
+        foreach ($this->lines as $lineNumber => $lineContents) {
+            if ($this->isFrontMatterControlBlock($lineContents)) {
+                $this->setFrontMatterLineNumber($lineNumber);
+            }
+        }
+    }
+
+    protected function setFrontMatterLineNumber(int $lineNumber)
+    {
+        if (!isset($this->frontMatterStartLine)) {
+            $this->frontMatterStartLine = $lineNumber;
+            return;
+        }
+
+        if (!isset($this->frontMatterEndLine)) {
+            $this->frontMatterEndLine = $lineNumber;
+        }
+    }
+
+    protected function getFrontMatter(): string
+    {
+        $matter = [];
+        foreach ($this->lines as $lineNumber => $lineContents) {
+            if ($lineNumber <= $this->frontMatterEndLine) {
+                if (!$this->isFrontMatterControlBlock($lineContents)) {
+                    $matter[] = $lineContents;
+                }
+            }
+        }
+        return implode("\n", $matter);
+    }
+
+    protected function getBody(): string
+    {
+        $body = [];
+        foreach ($this->lines as $lineNumber => $lineContents) {
+            if ($lineNumber > $this->frontMatterEndLine) {
+                $body[] = $lineContents;
+            }
+        }
+        return implode("\n", $this->trimBody($body));
+    }
+
+    protected function trimBody(array $body): array
+    {
+        if (trim($body[0]) === '') {
+            unset($body[0]);
+        }
+        return $body;
+    }
+
+    protected function hasFrontMatter(): bool
+    {
+        return ($this->frontMatterStartLine !== null) && ($this->frontMatterEndLine !== null);
+    }
+
+    protected function isFrontMatterControlBlock(string $line): bool
+    {
+        return substr($line, 0, 3) === '---';
+    }
+}

src/YamlFrontMatter.php

@@ -23,6 +23,14 @@ public static function parse(string $content): Document
         return new Document($matter, $body);
     }
 
+    /**
+     * A parser that can handle Markdown that contains Markdown.
+     */
+    public static function markdownCompatibleParse(string $content): Document
+    {
+        return (new ComplexMarkdownParser($content))->parse();
+    }
+
     public static function parseFile(string $path): Document
     {
         return static::parse(

tests/YamlFrontMatterMarkdownCompatibleParseTest.php

@@ -0,0 +1,100 @@
+<?php
+
+namespace Spatie\YamlFrontMatter\Tests;
+
+use Spatie\YamlFrontMatter\Document;
+use Spatie\YamlFrontMatter\YamlFrontMatter;
+use PHPUnit\Framework\TestCase;
+
+class YamlFrontMatterMarkdownCompatibleParseTest extends TestCase
+{
+    /** @test */
+    public function it_can_parse_simple_front_matter_from_a_file()
+    {
+        $document = YamlFrontMatter::markdownCompatibleParse(
+            file_get_contents(__DIR__.'/document.md')
+        );
+
+        $this->assertInstanceOf(Document::class, $document);
+        $this->assertEquals(['foo' => 'bar'], $document->matter());
+        $this->assertStringContainsString('Lorem ipsum.', $document->body());
+    }
+
+    /** @test */
+    public function it_can_parse_complex_front_matter_from_a_file()
+    {
+        $document = YamlFrontMatter::markdownCompatibleParse(
+            file_get_contents(__DIR__.'/meta-document.md')
+        );
+
+        $this->assertInstanceOf(Document::class, $document);
+        $this->assertEquals(['foo' => 'bar'], $document->matter());
+        $this->assertStringContainsString('Lorem ipsum.', $document->body());
+    }
+
+    /** @test */
+    public function it_separates_the_front_matter_from_the_body()
+    {
+        $document = YamlFrontMatter::markdownCompatibleParse(
+           "---\ntitle: Front Matter\n---\n\nLorem ipsum."
+        );
+
+        $this->assertInstanceOf(Document::class, $document);
+
+        // This implicitly asserts that the front matter does not contain any markdown
+        $this->assertEquals(['title' => 'Front Matter'], $document->matter());
+        // This implicitly asserts that the body does not contain any front matter remnants
+        $this->assertEquals('Lorem ipsum.', $document->body());
+    }
+
+    /** @test */
+    public function it_leaves_string_without_front_matter_intact()
+    {
+        $document = YamlFrontMatter::markdownCompatibleParse(
+            "Lorem ipsum."
+        );
+
+        $this->assertInstanceOf(Document::class, $document);
+        $this->assertEmpty($document->matter());
+        $this->assertEquals('Lorem ipsum.', $document->body());
+    }
+
+    /** @test */
+    public function it_can_parse_a_file_partial_front_matter()
+    {
+        // If there is only one YAML control block, (---) the front matter is invalid
+        // and the document should be interpreted as having no front matter.
+
+        $document = YamlFrontMatter::markdownCompatibleParse(
+           "---\ntitle: Front Matter\n\nLorem ipsum."
+        );
+
+        $this->assertInstanceOf(Document::class, $document);
+        $this->assertEmpty($document->matter());
+        $this->assertEquals("---\ntitle: Front Matter\n\nLorem ipsum.", $document->body());
+    }
+
+    /** @test */
+    public function it_can_parse_a_string_with_unix_line_endings()
+    {
+        $document = YamlFrontMatter::markdownCompatibleParse(
+           "---\nfoo: bar\n---\n\nLorem ipsum."
+        );
+
+        $this->assertInstanceOf(Document::class, $document);
+        $this->assertEquals(['foo' => 'bar'], $document->matter());
+        $this->assertStringContainsString('Lorem ipsum.', $document->body());
+    }
+
+    /** @test */
+    public function it_can_parse_a_string_with_windows_line_endings()
+    {
+        $document = YamlFrontMatter::markdownCompatibleParse(
+           "---\r\nfoo: bar\r\n---\r\n\r\nLorem ipsum."
+        );
+
+        $this->assertInstanceOf(Document::class, $document);
+        $this->assertEquals(['foo' => 'bar'], $document->matter());
+        $this->assertStringContainsString('Lorem ipsum.', $document->body());
+    }
+}

tests/meta-document.md

@@ -0,0 +1,17 @@
+---
+foo: bar
+---
+
+Lorem ipsum.
+
+A paragraph in a Markdown Post.
+This file contains a Markdown code block that the original parser does not handle.
+See https://github.com/spatie/yaml-front-matter/discussions/30.
+
+```
+---
+title: The Title of Markdown Code
+---
+
+A paragraph in Markdown Code
+```