Complete run.php using new classes

run.php

@@ -4,12 +4,56 @@ declare(strict_types=1);
 
 require __DIR__ . '/vendor/autoload.php';
 
-$parser = new LogseqGem\Parser;
+$opts = getopt('o:', rest_index: $rest_index);
+$rest = array_slice($argv, $rest_index);
 
-$page = file_get_contents("/Users/annika/sync/BTSync/logseq/pages/Gemlog.md");
-$page = file_get_contents("/Users/annika/sync/BTSync/logseq/pages/Publishing to the Gemlog with logseq.md");
+$gemlog_input_path = $rest[0] ?? null;
+$gemlog_output_path = $opts['o'] ?? null;
 
-$document = $parser->parse($page);
-foreach ($document->iterator() as $node) {
-    echo 'Current node: ' . get_class($node) . "\n";
+if (is_null($gemlog_input_path) || !file_exists($gemlog_input_path)) {
+    throw new \Exception('Please provide an input gemlog file listing some articles. Got: ' . $gemlog_input_path);
+}
+
+if (is_null($gemlog_output_path) || !is_dir($gemlog_output_path)) {
+    throw new \Exception('Please provide a destination directory using -o');
+}
+
+$dir = dirname($gemlog_input_path);
+
+$gemlog = new LogseqGem\GemlogParser;
+$article = new LogseqGem\ArticleParser;
+
+$gemlog->load($gemlog_input_path);
+$titles = $gemlog->getTitles();
+
+foreach ($titles as $title) {
+    $article_path = $dir . '/' . $title . '.md';
+    $article->load($article_path);
+    $gemtext = $article->convert();
+
+    if (!$article->isPublished() || empty($gemtext)) {
+        continue;
+    }
+
+    $sanitize_title = strtolower($title);
+    $sanitize_title = str_replace(' ', '-', $sanitize_title);
+    $sanitize_title .= '.gmi';
+
+    $output_path = $gemlog_output_path . '/' . $sanitize_title;
+
+    echo "Writing to ", $sanitize_title, "...\n";
+
+    $date = strtotime($article->getProperties()['date'] ?? null);
+    if (empty($date)) {
+        throw new Exception('Article ' . $title . ' has no date property');
+    }
+
+    $output = sprintf(
+        "# %s\n\n%s\n\n%s\n\n=> /gemlog.gmi gemlog\n=> / home\n",
+        $title,
+        date('j F Y', $date),
+        $gemtext
+    );
+
+    file_put_contents($output_path, $output);
 }

src/ArticleParser.php

@@ -7,20 +7,42 @@ namespace LogseqGem;
 use League\CommonMark\Node\NodeWalker;
 use League\CommonMark\Node\Inline\Text;
 use League\CommonMark\Extension\CommonMark\Node\Inline\Link;
+use League\CommonMark\Extension\CommonMark\Node\Block\ListBlock;
 
 class ArticleParser extends Parser {
-    public function convert() {
-        $gemtext = [];
+    private ?array $properties;
 
-        $link = 0;
+    public function convert(): string {
+        $gemtext = [];
+        $this->properties = [];
+
+        $in_frontmatter = true;
 
         $walker = $this->document->walker();
         while ($event = $walker->next()) {
             $entering = $event->isEntering();
             $node = $event->getNode();
 
+            if ($entering && $node instanceof ListBlock) {
+                $in_frontmatter = false;
+                continue;
+            }
+
             if ($entering && $node instanceof Text) {
-                $gemtext[] = $node->getLiteral();
+                $text = $node->getLiteral();
+
+                if ($in_frontmatter) {
+                    if (preg_match('/^([A-Za-z-]+):: (.*)/', $text, $matches)) {
+                        $this->properties[$matches[1]] = $matches[2];
+                        continue;
+                    } else {
+                        $in_frontmatter = false;
+                    }
+                }
+
+                $gemtext[] = $text;
+
+                continue;
             }
 
             if ($entering && $node instanceof Link) {
@@ -30,9 +52,22 @@ class ArticleParser extends Parser {
                 }
 
                 $gemtext[] = sprintf('=> %s %s', $node->getUrl(), $label->getLiteral());
+
+                continue;
             }
         }
 
         return implode("\n\n", $gemtext);
     }
+
+    public function getProperties(): array {
+        if (is_null($this->properties)) {
+            throw new \Exception("Cannot get properties before converting a document");
+        }
+        return $this->properties;
+    }
+
+    public function isPublished(): bool {
+        return (($this->getProperties()['status'] ?? '') === 'published');
+    }
 }

tests/ArticleParserTest.php

@@ -0,0 +1,98 @@
+<?php
+
+declare(strict_types=1);
+
+namespace LogseqGem\Tests;
+
+use InvalidArgumentException;
+use Exception;
+use PHPUnit\Framework\TestCase;
+use org\bovigo\vfs\vfsStream;
+use org\bovigo\vfs\vfsStreamDirectory;
+
+use LogseqGem\ArticleParser;
+use RuntimeException;
+use SebastianBergmann\RecursionContext\InvalidArgumentException as RecursionContextInvalidArgumentException;
+use PHPUnit\Framework\ExpectationFailedException;
+
+class ArticleParserTest extends TestCase {
+    private vfsStreamDirectory $root;
+
+    private ArticleParser $parser;
+
+    public function setUp(): void {
+        $this->parser = new ArticleParser;
+        $this->root = vfsStream::setup();
+    }
+
+    private function setArticleText(string $text) {
+        vfsStream::newFile('article.md')
+            ->withContent($text)
+            ->at($this->root);
+    }
+
+    public function testConvertSimple() {
+        $this->setArticleText("- This is a test line.");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->assertSame('This is a test line.', $this->parser->convert());
+    }
+
+    public function testConvertLink() {
+        $this->setArticleText("- [foo](bar)");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->assertSame('=> bar foo', $this->parser->convert());
+    }
+
+    public function testMultiParagraph() {
+        $this->setArticleText("- Block 1\n- Block 2");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->assertSame("Block 1\n\nBlock 2", $this->parser->convert());
+    }
+
+    public function testEmpty() {
+        $this->setArticleText("-");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->assertSame("", $this->parser->convert());
+    }
+
+    public function testInterleavedParagraphsAndLinks() {
+        $this->setArticleText("- Block 1\n- [foo](bar)\n- Block 2");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->assertSame("Block 1\n\n=> bar foo\n\nBlock 2", $this->parser->convert());
+    }
+
+    public function testSequentialLinks() {
+        $this->markTestSkipped('not yet implemented');
+        $this->setArticleText("\n- [foo](bar)\n- [bar](foo)");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->assertSame("=> bar foo\n=> foo bar", $this->parser->convert());
+    }
+
+    public function testFrontmatter() {
+        $this->setArticleText("foo:: bar\nanother:: prop\n\n- first line");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->assertSame("first line", $this->parser->convert());
+        $this->assertSame(["foo" => "bar", "another" => "prop"], $this->parser->getProperties());
+    }
+
+    public function testDraftIsNotPublished() {
+        $this->setArticleText("status:: draft\n\n- my article");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->parser->convert();
+        $this->assertFalse($this->parser->isPublished());
+    }
+
+    public function testMissingStatusIsNotPublished() {
+        $this->setArticleText("status:: draft\n\n- my article");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->parser->convert();
+        $this->assertFalse($this->parser->isPublished());
+    }
+
+    public function testPublished() {
+        $this->setArticleText("status:: published\n\n- my article");
+        $this->parser->load(vfsStream::url('root/article.md'));
+        $this->parser->convert();
+        $this->assertTrue($this->parser->isPublished());
+    }
+}