@publiwrite/html-to-epub/lib/html.utils.js

A library to generate EPUB from HTML. Inspired by epub-gen.

import { visit } from 'unist-util-visit';
import { v4 as uuid } from 'uuid';
import mime from 'mime';
import { unified } from "unified";
import rehypeParse from "rehype-parse";
import rehypeStringify from "rehype-stringify";
const loadHtml = (content, plugins) => unified()
    .use(rehypeParse, { fragment: true })
    .use(plugins)
    // Voids: [] is required for epub generation, and causes little/no harm for non-epub usage
    .use(rehypeStringify, { allowDangerousHtml: true, voids: [] })
    .processSync(content)
    .toString();
const validateElements = (node, options, index) => {
    const attrs = node.properties;
    if (["img", "br", "hr"].includes(node.tagName)) {
        if (node.tagName === "img") {
            node.properties.alt = node.properties.alt || "image-placeholder";
        }
    }
    for (const k of Object.keys(attrs)) {
        if (options.allowedAttributes.includes(k)) {
            if (k === "type") {
                if (attrs[k] !== "script") {
                    delete node.properties[k];
                }
            }
        }
        else {
            delete node.properties[k];
        }
    }
    if (options.version === 2) {
        if (!options.allowedXhtml11Tags.includes(node.tagName)) {
            if (options.verbose) {
                console.log("Warning (content[" + index + "]):", node.tagName, "tag isn't allowed on EPUB 2/XHTML 1.1 DTD.");
            }
            node.tagName = "div";
        }
    }
};
const processImgTags = (node, options, dir) => {
    if (!["img", "input"].includes(node.tagName)) {
        return;
    }
    const url = node.properties.src;
    if (url === undefined || url === null) {
        return;
    }
    let extension, id;
    const image = options.images.find((element) => element.url === url);
    if (image) {
        id = image.id;
        extension = image.extension;
    }
    else {
        id = uuid();
        const mediaType = mime.getType(url.replace(/\?.*/, ""));
        if (mediaType === null) {
            if (options.verbose) {
                console.error("[Image Error]", `The image can't be processed : ${url}`);
            }
            return;
        }
        extension = mime.getExtension(mediaType);
        if (extension === null) {
            if (options.verbose) {
                console.error("[Image Error]", `The image can't be processed : ${url}`);
            }
            return;
        }
        options.images.push({ id, url, dir, mediaType, extension });
    }
    node.properties.src = `images/${id}.${extension}`;
};
export const processHtmlContent = (content, options, dir) => {
    const html = loadHtml(content, [
        () => (tree) => {
            visit(tree, "element", (node, index) => validateElements(node, options, index));
        },
        () => (tree) => {
            visit(tree, "element", (node) => processImgTags(node, options, dir));
        },
    ]);
    return html;
};