const TurndownService = require('@joplin/turndown');
const turndownPluginGfm = require('@joplin/turndown-plugin-gfm').gfm;
import markdownUtils from './markdownUtils';
const pdfUrlRegex = /[\s\S]*?\.pdf$/i;
export interface ParseOptions {
anchorNames?: string[];
preserveImageTagsWithSize?: boolean;
preserveNestedTables?: boolean;
baseUrl?: string;
disableEscapeContent?: boolean;
convertEmbeddedPdfsToLinks?: boolean;
}
export default class HtmlToMd {
public parse(html: string, options: ParseOptions = {}) {
const turndownOpts: any = {
headingStyle: 'atx',
anchorNames: options.anchorNames ? options.anchorNames.map(n => n.trim().toLowerCase()) : [],
codeBlockStyle: 'fenced',
preserveImageTagsWithSize: !!options.preserveImageTagsWithSize,
preserveNestedTables: !!options.preserveNestedTables,
bulletListMarker: '-',
emDelimiter: '*',
strongDelimiter: '**',
// If soft-breaks are enabled, lines need to end with two or more spaces for
// trailing
s to render. See
// https://github.com/laurent22/joplin/issues/8430
br: ' ',
disableEscapeContent: 'disableEscapeContent' in options ? options.disableEscapeContent : false,
};
if (options.convertEmbeddedPdfsToLinks) {
// Turndown ignores empty