Move HTML to Markdown library to Composer

vendor/league/html-to-markdown/src/Converter/BlockquoteConverter.php

@@ -0,0 +1,44 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class BlockquoteConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        // Contents should have already been converted to Markdown by this point,
+        // so we just need to add '>' symbols to each line.
+
+        $markdown = '';
+
+        $quote_content = trim($element->getValue());
+
+        $lines = preg_split('/\r\n|\r|\n/', $quote_content);
+
+        $total_lines = count($lines);
+
+        foreach ($lines as $i => $line) {
+            $markdown .= '> ' . $line . "\n";
+            if ($i + 1 === $total_lines) {
+                $markdown .= "\n";
+            }
+        }
+
+        return $markdown;
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('blockquote');
+    }
+}

vendor/league/html-to-markdown/src/Converter/CodeConverter.php

@@ -0,0 +1,62 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class CodeConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $language = null;
+
+        // Checking for language class on the code block
+        $classes = $element->getAttribute('class');
+
+        if ($classes) {
+            // Since tags can have more than one class, we need to find the one that starts with 'language-'
+            $classes = explode(' ', $classes);
+            foreach ($classes as $class) {
+                if (strpos($class, 'language-') !== false) {
+                    // Found one, save it as the selected language and stop looping over the classes.
+                    // The space after the language avoids gluing the actual code with the language tag
+                    $language = str_replace('language-', '', $class) . ' ';
+                    break;
+                }
+            }
+        }
+
+        $markdown = '';
+        $code = html_entity_decode($element->getChildrenAsString());
+
+        // In order to remove the code tags we need to search for them and, in the case of the opening tag
+        // use a regular expression to find the tag and the other attributes it might have
+        $code = preg_replace('/<code\b[^>]*>/', '', $code);
+        $code = str_replace('</code>', '', $code);
+
+        // Checking if the code has multiple lines
+        $lines = preg_split('/\r\n|\r|\n/', $code);
+        if (count($lines) > 1) {
+            // Multiple lines detected, adding three backticks and newlines
+            $markdown .= '```' . $language . "\n" . $code . "\n" . '```';
+        } else {
+            // One line of code, wrapping it on one backtick.
+            $markdown .= '`' . $language . $code . '`';
+        }
+
+        return $markdown;
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('code');
+    }
+}

vendor/league/html-to-markdown/src/Converter/CommentConverter.php

@@ -0,0 +1,26 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class CommentConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        return '';
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('#comment');
+    }
+}

vendor/league/html-to-markdown/src/Converter/ConverterInterface.php

@@ -0,0 +1,20 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+interface ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element);
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags();
+}

vendor/league/html-to-markdown/src/Converter/DefaultConverter.php

@@ -0,0 +1,50 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\Configuration;
+use League\HTMLToMarkdown\ConfigurationAwareInterface;
+use League\HTMLToMarkdown\ElementInterface;
+
+class DefaultConverter implements ConverterInterface, ConfigurationAwareInterface
+{
+    const DEFAULT_CONVERTER = '_default';
+
+    /**
+     * @var Configuration
+     */
+    protected $config;
+
+    /**
+     * @param Configuration $config
+     */
+    public function setConfig(Configuration $config)
+    {
+        $this->config = $config;
+    }
+
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        // If strip_tags is false (the default), preserve tags that don't have Markdown equivalents,
+        // such as <span> nodes on their own. C14N() canonicalizes the node to a string.
+        // See: http://www.php.net/manual/en/domnode.c14n.php
+        if ($this->config->getOption('strip_tags', false)) {
+            return $element->getValue();
+        }
+
+        return html_entity_decode($element->getChildrenAsString());
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array(self::DEFAULT_CONVERTER);
+    }
+}

vendor/league/html-to-markdown/src/Converter/DivConverter.php

@@ -0,0 +1,45 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\Configuration;
+use League\HTMLToMarkdown\ConfigurationAwareInterface;
+use League\HTMLToMarkdown\ElementInterface;
+
+class DivConverter implements ConverterInterface, ConfigurationAwareInterface
+{
+    /**
+     * @var Configuration
+     */
+    protected $config;
+
+    /**
+     * @param Configuration $config
+     */
+    public function setConfig(Configuration $config)
+    {
+        $this->config = $config;
+    }
+
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        if ($this->config->getOption('strip_tags', false)) {
+            return $element->getValue() . "\n\n";
+        }
+
+        return html_entity_decode($element->getChildrenAsString());
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('div');
+    }
+}

vendor/league/html-to-markdown/src/Converter/EmphasisConverter.php

@@ -0,0 +1,57 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\Configuration;
+use League\HTMLToMarkdown\ConfigurationAwareInterface;
+use League\HTMLToMarkdown\ElementInterface;
+
+class EmphasisConverter implements ConverterInterface, ConfigurationAwareInterface
+{
+    /**
+     * @var Configuration
+     */
+    protected $config;
+
+    /**
+     * @param Configuration $config
+     */
+    public function setConfig(Configuration $config)
+    {
+        $this->config = $config;
+    }
+
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $tag = $element->getTagName();
+        $value = $element->getValue();
+
+        if (!trim($value)) {
+            return '';
+        }
+
+        if ($tag === 'i' || $tag === 'em') {
+            $style = $this->config->getOption('italic_style');
+        } else {
+            $style = $this->config->getOption('bold_style');
+        }
+
+        $prefix = ltrim($value) !== $value ? ' ' : '';
+        $suffix = rtrim($value) !== $value ? ' ' : '';
+
+        return $prefix . $style . trim($value) . $style . $suffix;
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('em', 'i', 'strong', 'b');
+    }
+}

vendor/league/html-to-markdown/src/Converter/HardBreakConverter.php

@@ -0,0 +1,41 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\Configuration;
+use League\HTMLToMarkdown\ConfigurationAwareInterface;
+use League\HTMLToMarkdown\ElementInterface;
+
+class HardBreakConverter implements ConverterInterface, ConfigurationAwareInterface
+{
+    /**
+     * @var Configuration
+     */
+    protected $config;
+
+    /**
+     * @param Configuration $config
+     */
+    public function setConfig(Configuration $config)
+    {
+        $this->config = $config;
+    }
+
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        return $this->config->getOption('hard_break') ? "\n" : "  \n";
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('br');
+    }
+}

vendor/league/html-to-markdown/src/Converter/HeaderConverter.php

@@ -0,0 +1,78 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\Configuration;
+use League\HTMLToMarkdown\ConfigurationAwareInterface;
+use League\HTMLToMarkdown\ElementInterface;
+
+class HeaderConverter implements ConverterInterface, ConfigurationAwareInterface
+{
+    const STYLE_ATX = 'atx';
+    const STYLE_SETEXT = 'setext';
+
+    /**
+     * @var Configuration
+     */
+    protected $config;
+
+    /**
+     * @param Configuration $config
+     */
+    public function setConfig(Configuration $config)
+    {
+        $this->config = $config;
+    }
+
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $level = (int) substr($element->getTagName(), 1, 1);
+        $style = $this->config->getOption('header_style', self::STYLE_SETEXT);
+
+        if (($level === 1 || $level === 2) && !$element->isDescendantOf('blockquote') && $style === self::STYLE_SETEXT) {
+            return $this->createSetextHeader($level, $element->getValue());
+        }
+
+        return $this->createAtxHeader($level, $element->getValue());
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('h1', 'h2', 'h3', 'h4', 'h5', 'h6');
+    }
+
+    /**
+     * @param int    $level
+     * @param string $content
+     *
+     * @return string
+     */
+    private function createSetextHeader($level, $content)
+    {
+        $length = function_exists('mb_strlen') ? mb_strlen($content, 'utf-8') : strlen($content);
+        $underline = ($level === 1) ? '=' : '-';
+
+        return $content . "\n" . str_repeat($underline, $length) . "\n\n";
+    }
+
+    /**
+     * @param int    $level
+     * @param string $content
+     *
+     * @return string
+     */
+    private function createAtxHeader($level, $content)
+    {
+        $prefix = str_repeat('#', $level) . ' ';
+
+        return $prefix . $content . "\n\n";
+    }
+}

vendor/league/html-to-markdown/src/Converter/HorizontalRuleConverter.php

@@ -0,0 +1,26 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class HorizontalRuleConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        return "- - - - - -\n\n";
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('hr');
+    }
+}

vendor/league/html-to-markdown/src/Converter/ImageConverter.php

@@ -0,0 +1,35 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class ImageConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $src = $element->getAttribute('src');
+        $alt = $element->getAttribute('alt');
+        $title = $element->getAttribute('title');
+
+        if ($title !== '') {
+            // No newlines added. <img> should be in a block-level element.
+            return '![' . $alt . '](' . $src . ' "' . $title . '")';
+        }
+
+        return '![' . $alt . '](' . $src . ')';
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('img');
+    }
+}

vendor/league/html-to-markdown/src/Converter/LinkConverter.php

@@ -0,0 +1,52 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class LinkConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $href = $element->getAttribute('href');
+        $title = $element->getAttribute('title');
+        $text = trim($element->getValue());
+
+        if ($title !== '') {
+            $markdown = '[' . $text . '](' . $href . ' "' . $title . '")';
+        } elseif ($href === $text && $this->isValidAutolink($href)) {
+            $markdown = '<' . $href . '>';
+        } else {
+            $markdown = '[' . $text . '](' . $href . ')';
+        }
+
+        if (!$href) {
+            $markdown = html_entity_decode($element->getChildrenAsString());
+        }
+
+        return $markdown;
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('a');
+    }
+
+    /**
+     * @param string $href
+     *
+     * @return bool
+     */
+    private function isValidAutolink($href)
+    {
+        return preg_match('/^[A-Za-z][A-Za-z0-9.+-]{1,31}:[^<>\x00-\x20]*/i', $href) === 1;
+    }
+}

vendor/league/html-to-markdown/src/Converter/ListBlockConverter.php

@@ -0,0 +1,26 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class ListBlockConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        return $element->getValue() . "\n";
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('ol', 'ul');
+    }
+}

vendor/league/html-to-markdown/src/Converter/ListItemConverter.php

@@ -0,0 +1,47 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class ListItemConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        // If parent is an ol, use numbers, otherwise, use dashes
+        $list_type = $element->getParent()->getTagName();
+
+        // Add spaces to start for nested list items
+        $level = $element->getListItemLevel($element);
+
+        $prefixForParagraph = str_repeat('  ', $level + 1);
+        $value = trim(implode("\n" . $prefixForParagraph, explode("\n", trim($element->getValue()))));
+
+        // If list item is the first in a nested list, add a newline before it
+        $prefix = '';
+        if ($level > 0 && $element->getSiblingPosition() === 1) {
+            $prefix = "\n";
+        }
+
+        if ($list_type === 'ul') {
+            return $prefix . '- ' . $value . "\n";
+        }
+
+        $number = $element->getSiblingPosition();
+
+        return $prefix . $number . '. ' . $value . "\n";
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('li');
+    }
+}

vendor/league/html-to-markdown/src/Converter/ParagraphConverter.php

@@ -0,0 +1,124 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class ParagraphConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $value = $element->getValue();
+
+        $markdown = '';
+
+        $lines = preg_split('/\r\n|\r|\n/', $value);
+        foreach ($lines as $line) {
+            /*
+             * Some special characters need to be escaped based on the position that they appear
+             * The following function will deal with those special cases.
+             */
+            $markdown .= $this->escapeSpecialCharacters($line);
+            $markdown .= "\n";
+        }
+
+        return trim($markdown) !== '' ? rtrim($markdown) . "\n\n" : '';
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('p');
+    }
+
+    /**
+     * @param string $line
+     *
+     * @return string
+     */
+    private function escapeSpecialCharacters($line)
+    {
+        $line = $this->escapeFirstCharacters($line);
+        $line = $this->escapeOtherCharacters($line);
+        $line = $this->escapeOtherCharactersRegex($line);
+
+        return $line;
+    }
+
+    /**
+     * @param string $line
+     *
+     * @return string
+     */
+    private function escapeFirstCharacters($line)
+    {
+        $escapable = array(
+            '>',
+            '- ',
+            '+ ',
+            '--',
+            '~~~',
+            '---',
+            '- - -'
+        );
+
+        foreach ($escapable as $i) {
+            if (strpos(ltrim($line), $i) === 0) {
+                // Found a character that must be escaped, adding a backslash before
+                return '\\' . ltrim($line);
+            }
+        }
+
+        return $line;
+    }
+
+    /**
+     * @param string $line
+     *
+     * @return string
+     */
+    private function escapeOtherCharacters($line)
+    {
+        $escapable = array(
+            '<!--'
+        );
+
+        foreach ($escapable as $i) {
+            if (strpos($line, $i) !== false) {
+                // Found an escapable character, escaping it
+                $line = substr_replace($line, '\\', strpos($line, $i), 0);
+            }
+        }
+
+        return $line;
+    }
+
+    /**
+     * @param string $line
+     *
+     * @return string
+     */
+    private function escapeOtherCharactersRegex($line)
+    {
+        $regExs = array(
+            // Match numbers ending on ')' or '.' that are at the beginning of the line.
+            '/^[0-9]+(?=\)|\.)/'
+        );
+
+        foreach ($regExs as $i) {
+            if (preg_match($i, $line, $match)) {
+                // Matched an escapable character, adding a backslash on the string before the offending character
+                $line = substr_replace($line, '\\', strlen($match[0]), 0);
+            }
+        }
+
+        return $line;
+    }
+}

vendor/league/html-to-markdown/src/Converter/PreformattedConverter.php

@@ -0,0 +1,59 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class PreformattedConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $markdown = '';
+
+        $pre_content = html_entity_decode($element->getChildrenAsString());
+        $pre_content = str_replace(array('<pre>', '</pre>'), '', $pre_content);
+
+        /*
+         * Checking for the code tag.
+         * Usually pre tags are used along with code tags. This conditional will check for already converted code tags,
+         * which use backticks, and if those backticks are at the beginning and at the end of the string it means
+         * there's no more information to convert.
+         */
+
+        $firstBacktick = strpos(trim($pre_content), '`');
+        $lastBacktick = strrpos(trim($pre_content), '`');
+        if ($firstBacktick === 0 && $lastBacktick === strlen(trim($pre_content)) - 1) {
+            return $pre_content;
+        }
+
+        // If the execution reaches this point it means it's just a pre tag, with no code tag nested
+
+        // Normalizing new lines
+        $pre_content = preg_replace('/\r\n|\r|\n/', PHP_EOL, $pre_content);
+
+        // Checking if the string has multiple lines
+        $lines = preg_split('/\r\n|\r|\n/', $pre_content);
+        if (count($lines) > 1) {
+            // Multiple lines detected, adding three backticks and newlines
+            $markdown .= '```' . "\n" . $pre_content . "\n" . '```';
+        } else {
+            // One line of code, wrapping it on one backtick.
+            $markdown .= '`' . $pre_content . '`';
+        }
+
+        return $markdown;
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('pre');
+    }
+}

vendor/league/html-to-markdown/src/Converter/TextConverter.php

@@ -0,0 +1,46 @@
+<?php
+
+namespace League\HTMLToMarkdown\Converter;
+
+use League\HTMLToMarkdown\ElementInterface;
+
+class TextConverter implements ConverterInterface
+{
+    /**
+     * @param ElementInterface $element
+     *
+     * @return string
+     */
+    public function convert(ElementInterface $element)
+    {
+        $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);
+
+        // Escape the following characters: '*', '_', '[', ']' and '\'
+        $markdown = preg_replace('~([*_\\[\\]\\\\])~u', '\\\\$1', $markdown);
+
+        $markdown = preg_replace('~^#~u', '\\\\#', $markdown);
+
+        if ($markdown === ' ') {
+            $next = $element->getNext();
+            if (!$next || $next->isBlock()) {
+                $markdown = '';
+            }
+        }
+
+        return $markdown;
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getSupportedTags()
+    {
+        return array('#text');
+    }
+}