Upstream thephpleague/html-to-markdown state of the touched files before aa29149d30

src/Converter/EmphasisConverter.php

@@ -0,0 +1,73 @@
+<?php
+
+declare(strict_types=1);
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\Coerce;
+use League\HTMLToMarkdown\Configuration;
+use League\HTMLToMarkdown\ConfigurationAwareInterface;
+use League\HTMLToMarkdown\ElementInterface;
+
+class EmphasisConverter implements ConverterInterface, ConfigurationAwareInterface
+{
+    /** @var Configuration */
+    protected $config;
+
+    protected function getNormTag(?ElementInterface $element): string
+    {
+        if ($element !== null && ! $element->isText()) {
+            $tag = $element->getTagName();
+            if ($tag === 'i' || $tag === 'em') {
+                return 'em';
+            }
+
+            if ($tag === 'b' || $tag === 'strong') {
+                return 'strong';
+            }
+        }
+
+        return '';
+    }
+
+    public function setConfig(Configuration $config): void
+    {
+        $this->config = $config;
+    }
+
+    public function convert(ElementInterface $element): string
+    {
+        $tag   = $this->getNormTag($element);
+        $value = $element->getValue();
+
+        if (! \trim($value)) {
+            return $value;
+        }
+
+        if ($tag === 'em') {
+            $style = Coerce::toString($this->config->getOption('italic_style'));
+        } else {
+            $style = Coerce::toString($this->config->getOption('bold_style'));
+        }
+
+        $prefix = \ltrim($value) !== $value ? ' ' : '';
+        $suffix = \rtrim($value) !== $value ? ' ' : '';
+
+        /* If this node is immediately preceded or followed by one of the same type don't emit
+         * the start or end $style, respectively. This prevents <em>foo</em><em>bar</em> from
+         * being converted to *foo**bar* which is incorrect. We want *foobar* instead.
+         */
+        $preStyle  = $this->getNormTag($element->getPreviousSibling()) === $tag ? '' : $style;
+        $postStyle = $this->getNormTag($element->getNextSibling()) === $tag ? '' : $style;
+
+        return $prefix . $preStyle . \trim($value) . $postStyle . $suffix;
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags(): array
+    {
+        return ['em', 'i', 'strong', 'b'];
+    }
+}

src/Converter/TextConverter.php

@@ -0,0 +1,48 @@
+<?php
+
+declare(strict_types=1);
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class TextConverter implements ConverterInterface
+{
+    public function convert(ElementInterface $element): string
+    {
+        $markdown = $element->getValue();
+
+        // Remove leftover \n at the beginning of the line
+        $markdown = \ltrim($markdown, "\n");
+
+        // Replace sequences of invisible characters with spaces
+        $markdown = \preg_replace('~\s+~u', ' ', $markdown);
+        \assert(\is_string($markdown));
+
+        // Escape the following characters: '*', '_', '[', ']' and '\'
+        if (($parent = $element->getParent()) && $parent->getTagName() !== 'div') {
+            $markdown = \preg_replace('~([*_\\[\\]\\\\])~u', '\\\\$1', $markdown);
+            \assert(\is_string($markdown));
+        }
+
+        $markdown = \preg_replace('~^#~u', '\\\\#', $markdown);
+        \assert(\is_string($markdown));
+
+        if ($markdown === ' ') {
+            $next = $element->getNext();
+            if (! $next || $next->isBlock()) {
+                $markdown = '';
+            }
+        }
+
+        return \htmlspecialchars($markdown, ENT_NOQUOTES, 'UTF-8');
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags(): array
+    {
+        return ['#text'];
+    }
+}

tests/HtmlConverterTest.php

@@ -163,8 +163,6 @@ class HtmlConverterTest extends TestCase
         $this->assertHtmlGivesMarkdown('<a href="google.com">google.com</a>', '[google.com](google.com)');
         $this->assertHtmlGivesMarkdown('<a href="https://www.google.com">https://www.google.com</a>', '<https://www.google.com>');
         $this->assertHtmlGivesMarkdown('<a href="ftp://files.example.com">ftp://files.example.com</a>', '<ftp://files.example.com>');
-        $this->assertHtmlGivesMarkdown('<a href="https://example.com/?a=1&amp;b=2">https://example.com/?a=1&amp;b=2</a>', '<https://example.com/?a=1&b=2>');
-        $this->assertHtmlGivesMarkdown('<a href="https://example.com/&lt;b&gt;">https://example.com/&lt;b&gt;</a>', '[https://example.com/&lt;b&gt;](https://example.com/<b>)');
         $this->assertHtmlGivesMarkdown('<a href="mailto:test@example.com">test@example.com</a>', '<test@example.com>');
         $this->assertHtmlGivesMarkdown('<a href="mailto:test+foo@example.bar-baz.com">test+foo@example.bar-baz.com</a>', '<test+foo@example.bar-baz.com>');