<?php

declare(strict_types=1);

namespace League\HTMLToMarkdown\Converter;

use League\HTMLToMarkdown\Backticks;
use League\HTMLToMarkdown\ElementInterface;
use League\HTMLToMarkdown\NoBreakSpaces;

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, other than those which keep text on the same line
        $markdown = \preg_replace('~[^\S' . NoBreakSpaces::CHARACTERS . ']+~u', ' ', $markdown);
        \assert(\is_string($markdown));

        // Escape the following characters: '*', '_', '[', ']', '|' and '\'
        $isEscaped = ($parent = $element->getParent()) && $parent->getTagName() !== 'div';
        if ($isEscaped) {
            $markdown = \preg_replace('~([*_\\[\\]|\\\\])~u', '\\\\$1', $markdown);
            \assert(\is_string($markdown));
        }

        $markdown = \preg_replace('~^#~u', '\\\\#', $markdown);
        \assert(\is_string($markdown));

        if (\preg_match('~^\s++\z~u', $markdown) === 1) {
            $next = $element->getNext();
            if (! $next || $next->isBlock() || \in_array($next->getTagName(), ['pre', 'table'], true)) {
                $markdown = '';
            }
        }

        $markdown = \htmlspecialchars($markdown, ENT_NOQUOTES, 'UTF-8');

        // A pipe could otherwise make a table out of its line, cutting through any code span on the line above
        if (! $isEscaped) {
            $markdown = \str_replace('|', '&#124;', $markdown);
        }

        return Backticks::escapeText($markdown);
    }

    /**
     * @return string[]
     */
    public function getSupportedTags(): array
    {
        return ['#text'];
    }
}