Libretexts: get real cover page

[benoit74]

Currently, to build index (#55) and soon detailed licensing (#54) and probably for the table of content (#56) we need to get the URL and ID of the "cover page" of current book.

In #55, we've built this logic by going up into the tree and search for a page with "article": "topic-category" property (see get_page_parent_book_id logic).

Real logic used online on libretexts.org is based on tags, in getCoverpage function of https://cdn.libretexts.net/github/LibreTextsMain/Miscellaneous/reuse.js:

    /**
     * Locates the parent page that is the coverpage, if it exists
     * @param url - page to look up the coverpage for
     * @returns {Promise<string>} - path to the coverpage
     */
    async function getCoverpage(url = window.location.href) {
        if (typeof getCoverpage.coverpage === 'undefined') {
            const urlArray = url.replace("?action=edit", "").split("/");
            for (let i = urlArray.length; i > 3; i--) {
                let path = urlArray.slice(3, i).join("/");
                if (!path)
                    break;
                let response = await LibreTexts.authenticatedFetch(path, 'tags?dream.out.format=json');
                let tags = await response.json();
                if (tags.tag) {
                    if (tags.tag.length) {
                        tags = tags.tag.map((tag) => tag["@value"]);
                    }
                    else {
                        tags = tags.tag["@value"];
                    }
                    if (tags.includes("coverpage:yes") || tags.includes("coverpage:toc") || tags.includes("coverpage:nocommons")) {
                        getCoverpage.coverpage = path;
                        break;
                    }
                }
            }
        }
        return getCoverpage.coverpage;
    }

As one can see, this code walks up the tree of pages and look for first page with coverpage:yes or coverpage:toc or coverpage:nocommons tags.

Obviously, this is something we might have to consider at some point (probably easier to compute on the tree of pages at crawl time).

To be analyzed at least, if not integrated into the crawler (finding all page tags means doing a query to retrieve them, we do not use them ATM, and this will have an impact on crawl time).

[benoit74]

In fact, we need to retrieve tags only for the backmatter pages and their parents. Quite a limited impact, so let's do it now.