@rr0/cms/dist/source/SourceFactory.js

RR0 Content Management System (CMS)

import path from "path";
import { Level2Date as EdtfDate } from "@rr0/time";
import { JSDOM } from "jsdom";
import { FileContents } from "@javarome/fileutil";
/**
 * Create Source objects.
 */
export class SourceFactory {
    constructor(dataService, http, baseUrl, options, time) {
        this.dataService = dataService;
        this.http = http;
        this.baseUrl = baseUrl;
        this.options = options;
        this.time = time;
    }
    /**
     * Create a Source object from an anchor's URL.
     *
     * @param context
     * @param href The anchor's URL string.
     */
    async create(context, href) {
        return href.startsWith("http") ? await this.createExternal(context, href) : await this.createInternal(context, href);
    }
    /**
     * Create a Source object from a link referencing an internal page of the site.
     *
     * @param context
     * @param href
     */
    async createInternal(context, href) {
        var _a;
        if (path.dirname(href).startsWith("/")) {
            href = href.substring(1);
        }
        const hashPos = href.lastIndexOf("#");
        let hash;
        if (hashPos > 0) {
            hash = href.substring(hashPos + 1);
            href = href.substring(0, hashPos);
        }
        else {
            hash = undefined;
        }
        const ext = path.extname(href);
        let source;
        const sourceTypes = ["article", "book", "website",
            undefined // TODO: Remove undefined when type is set in all .json files
        ];
        switch (ext) {
            case ".htm":
            case ".html":
                source = await this.fromPage(href, hash);
                break;
            case ".json":
                const sources = await this.dataService.getFromDir(path.dirname(href), sourceTypes, [path.basename(href)]);
                source = sources === null || sources === void 0 ? void 0 : sources[0];
                break;
            default: {
                const sources = await this.dataService.getFromDir(ext ? path.dirname(href) : href, sourceTypes, ["index.json", "people.json"]);
                source = sources === null || sources === void 0 ? void 0 : sources[0];
                if (!source) {
                    source = await this.fromPage(path.join(href, "index.html"), hash);
                }
            }
        }
        const publication = source.publication;
        if (publication && !publication.time) {
            publication.time = (_a = this.time.contextFromFileName(context, href)) === null || _a === void 0 ? void 0 : _a.date;
        }
        if (hash) {
            source.index = hash;
        }
        return source;
    }
    /**
     * Create a Source object from a link referencing a page outside the site.
     *
     * @param context
     * @param href
     */
    async createExternal(context, href) {
        const resOut = {};
        let title;
        let lastModif;
        let publisher;
        try {
            const doc = await this.http.get(new URL(href), {}, resOut);
            href = resOut.url || href;
            title = doc.querySelector("title").textContent;
            lastModif = resOut.headers.get("last-modified");
        }
        catch (e) {
            context.error("Could not fetch source from " + href, e.message);
            title = href;
        }
        publisher = resOut.headers.get("host");
        const time = lastModif ? EdtfDate.fromDate(new Date(lastModif)) : context.time.date;
        const publication = { publisher, time };
        return {
            title,
            url: href,
            events: [],
            previousSourceRefs: [],
            publication
        };
    }
    async fromPage(href, hash = "") {
        var _a;
        const filePath = path.extname(href) ? href : path.join(href, "index.html");
        const fileContents = FileContents.read(filePath);
        const doc = new JSDOM(fileContents.contents).window.document.documentElement;
        const url = new URL(href, this.baseUrl);
        if (hash) {
            url.hash = hash;
        }
        const values = Array.from(this.dataService.pathToData.values());
        const data = values.find(value => { var _a; return filePath.startsWith((_a = value[0]) === null || _a === void 0 ? void 0 : _a.dirName); });
        let title;
        if (data) {
            title = data[0].title;
        }
        else {
            title = doc.querySelector("title").textContent;
        }
        const publisher = (_a = doc.querySelector("meta[name='copyright']")) === null || _a === void 0 ? void 0 : _a.getAttribute("content");
        const authors = Array.from(doc.querySelectorAll("meta[name='author']")).map(meta => meta.getAttribute("content"));
        return {
            title,
            url: url.href,
            events: [],
            previousSourceRefs: [],
            authors,
            publication: { publisher, time: undefined }
        };
    }
}