Adds trees of pages to the Apostrophe content management system


apostrophe-pages adds page rendering to the Apostrophe content management system. The apostrophe-pages module makes it easy to serve pages, fetching the requested page and making its content areas available to your Nunjucks template along with any other attributes of interest.

apostrophe-pages also provides a UI for adding pages, deleting pages and changing page settings such as the title. As part of that, apostrophe-pages provides functions to fetch ancestor and descendant pages to any desired depth.

By default apostrophe-pages renders page templates found in its views folder depending on the template property of the page. You can override the location where one or all of these are found in order to provide custom templates for your project's needs.

Setting up apostrophe-pages is easy once you have the main apos object. See the apostrophe-wiki project for a more complete sample app. In this example we want to render any URL that matches a page slug as a page:

var apos = require('apostrophe')({ ..., partialPaths: [ '/views/global' ]});
var pages = require('apostrophe-pages')({ apos: apos, templatePath: __dirname + '/views/pages' });

// Your app specific routes should come first.
app.get('/mything', function(req, res) { ... my non-Apostrophe-related stuff ... });

// Now try to serve anything else as a page.
app.get('*', pages.serve());

Notice that we set partialPaths to provide global layout templates that can be extended by our page templates, and set templatePath to specify where our page templates area.

The constructor for the pages module accepts a types option. In its simplest form, a "page type" is just a template name and a label. If you do not specify a types parameter, you get a single type with the name default and the label Default. this causes views/default.html to be loaded to render pages with that type.

However page types can also be extended with custom behavior. See the apostrophe-blog module for an example with all the trimmings.

To facilitate code reuse, page types can have a single "superclass" from which they inherit behavior in both the server- and browser-side JavaScript code.

Here is a simple example of specifying the types option:

types: [ { name: 'default', label: 'Default (Two Column)' }, { name: 'onecolumn', label: 'One Column' }]

You can also add types later, for instance when initializing other modules such as our blog module. See pages.addType.

pages.serve has options that can be used to override its behavior in many ways. Complete documentation for the pages.serve function is provided at the top of the function in index.js (TODO: work on publishing this as jsdoc).

Your page templates will want to render areas. Just use the page objects passed to you to access them and call the aposArea helper available in anything rendered via apos.partial, which includes page templates:

{{ aposArea(page, 'main') }}
{{ aposArea(global, 'footer') }}

Pages may have a path, separate from their slug, which expresses their relationship to other pages. Paths look like this: this:


The path always reflects the relationship between the pages no matter what the slug may be changed to by the user. (Clients routinely shorten slugs to make URLs easier to publish in print, but don't want to lose the relationships between pages.)

Note that paths do not have a leading /. Multiple roots are permitted, but typically there is one root page with the path home.

Pages also have a rank, which determines their ordering among the children of a particular page.

pages.getAncestors and pages.getDescendants can be used to fetch the ancestors and descendants of a page. pages.getAncestors(req, page, callback) delivers ancestor pages to its callback, in order beginning with the root page. pages.getDescendants(req, page, callback) delivers the children of the page, in order by rank. req must be passed to provide a context for permissions (req.user) and, potentially, for caching during the lifetime of a single request. It must be either a real Express request object or some other object acceptable to your apos.permissions function.

You can optionally specify a depth:

pages.getDescendants(req, page, { depth: 2 }, callback)

In this case your callback still receives an array of the immediate children of page. However, each of those pages has a children property containing an array of its children.

For performance reasons, pages.getAncestors and pages.getDescendants do not return the items property. Typically only the slug and title properties are necessary to build navigation. If necessary you may use the slug property of a page to fetch the entire page with its items, via apos.getPage.

pages.serve automatically fetches the ancestors of the page into the ancestors property of the page object given to the page template. In addition, the children of the page are available in the children property. And the children of the home page (whether the current page is the home page or not) are available in the tabs property. Also, the peers of the current page (children of the same parent) are available in the peers property.

If you need to see the descendants of the current page to a greater depth, set the descendantOptions option when calling pages.serve:

  descendantOptions: {
    depth: 2

You can then look at the children property of each entry in page.children, and so on.

To do the same thing for descendants of the home page, set:

  tabOptions: {
    depth: 2

You can add a .children array to each ancestor in order to implement accordion navigation, in which each ancestor's children are also displayed:

  ancestorOptions: {
    children: true

You can also shut off ancestors, descendants or tabs entirely if you're not interested:

  ancestors: false
  tabs: false,
  descendants: false,

Typically "tree pages" (pages that are part of the tree, i.e. the home page and its descendants) are displayed and browsed like nested folders. But from time to time it is useful to fetch pages based on taxonomy instead.

The pages.getByTag method returns all tree pages on the site that have a specified tag:

pages.getByTag(req, 'green', function(err, results) { ... });

And pages.getByTags returns all pages with at least one of the tags in an array:

pages.getByTags(req, ['green', 'blue'], function(err, results) { ... });

When pages are fetched by tag, they are sorted alphabetically by title.

If you already have an array of pages, for instance the children or peers of the current page, it can be useful to filter them by tag. You can do that with pages.filterByTag:

pages.filterByTag(children, 'tag')
pages.filterByTags(children, ['green', 'blue'])

Again, filterByTags returns pages with at least one of the specified tags.

Note that these functions do not take a callback. They return the pages directly, since they are just filtering an existing array of pages based on their metadata.

Most sites require that some extra data be loaded along with pages. The data to be loaded is often dependent on the site structure and the unique needs of the project. The pages.serve function supports this via the load option.

The load option accepts an array made up of page slugs and functions. Any strings found in this array are assumed to be the slugs of pages, usually "virtual pages" whose slugs do not start with a leading / and are not directly reachable by navigating the site. Such pages are loaded and added to the req.extras property. All properties of req.extras are then made available to your page templates. So if you use the virtual page global to hold a shared global footer area, you can access it as global.footer in your page templates.

In addition, loaders can be asynchronous functions that modify the req object in their own ways. Loaders receive the req object as their first parameter and a callback to be invoked on completion as their second parameter. The req object will have a page property containing the page that matched the slug, if any, and a remainder property matching additional content in the URL after the slug if the page is greedy, as explained below.

Many sites need to go beyond a simple tree of pages, implementing experiences like blogs and catalogs that require "subpages" to exist for every product or blog post, and URLs that contain elements other than page slugs, such as the date and slug of a blog post. This is easily implemented using greedy pages.

While the req.page page property is set only if the slug exactly matches a page, the req.bestPage property is set to the page whose slug comes closest to matching the page. To be the "best page," a page must meet the following conditions:

Either (1) it matches the requested URL exactly (in which case req.page and req.bestPage will be the same page), or (2) it matches the longest portion of the URL ending in a /.


Assume there are pages with the slugs /blog and /blog/credits in the databsae.

  1. If the user requests /blog and there is a page with the slug /blog, both req.page and req.bestPage will be set to the /blog page object. req.remainder will be the empty string.
  2. If the user requests /blog/2013/01/01/i-like-kittens, req.page will be undefined, and req.bestPage will be set to the /blog page object. req.remainder will be set to /2013/01/01/i-like-kittens.
  3. If the user requests /blog/credits, req.page and req.bestPage will both be set to the /blog/credits page object. req.remainder will be the empty string.
  4. If the user requests /blog/credits/paul, req.page will be undefined, and req.bestPage will be set to the /blog/credits page object. req.remainder will be set to /paul.
  5. For consistency, if the URL ends with a trailing /, this is not included in req.remainder.

This approach allows a mix of "ordinary" subpages and subpages implemented by custom logic in a load function that examines req.remainder and req.bestPage to decide what to do.

A loader that decides a page should be rendered after all based on a partial match should set req.page to req.bestPage. Otherwise the page is considered to be a 404.

You could implement a blog with custom behavior for different values of remainder entirely by setting properties of req.extras and examining them in your template.

But it is often easier to use an entirely different template, for instance to render a blog post's permalink page differently from the main index of a blog.

To achieve that in your load function, just set req.type to the template you want to render:

req.type = 'blogPost';

Note that you can set req.type to notfound to display the standard "404 not found" template for the project.

apostrophe-pages provides a full user interface for creating, modifying and removing pages. To enable it, just insert the appropriate markup into your page layout:

{{ aposEditPage({ page: page, edit: edit, root: '/' }) }}

This helper function inserts the page-related buttons at that point and also the necessary browser-side JavaScript to power them.

If you change the slug (URL) of a page via the Page Settings button, that doesn't tell Google and other search engines that the page has moved. So as a convenience, apostrophe-pages automatically tracks the old URLs and provides redirects to the new URLs. Of course, if a new page is created at the old URL, that page wins and the old redirect is not used.

The apostrophe-search module also provides a sitewide search facility. This is implemented by a page loader function that kicks in for pages (usually just one) with the search type.

Simple filters are provided to include or exclude results. There is a checkbox for each searchable instance type (such as blogPost or event) and for regular pages. You can override these with the searchLoader option to the pages module. For instance, in app.js in a project using apostrophe-site for configuration:

pages: {
  searchLoader: [
      name: 'page',
      label: 'Pages'
      name: 'blogPost',
      label: 'Blog Posts'

Any searchable document whose page type does not have a specific filter is toggled by the page filter. If you do not include a filter with the name page then such documents will not be visible in search results. However this is frontend filtering and should not be relied upon to secure information by keeping it out of search. For that, if you are subclassing snippets, you may use the searchable option when configuring the module.

Apostrophe's "page menu" in the lower left corner is designed to be context-sensitive. Its contents can be overridden for the current page by setting req.contextMenu in a page loader function.

Here's an example:

req.contextMenu = [
    name: 'new-monkey',
    label: 'New Monkey'
    name: 'edit-page',
    label: 'Page Settings'
    name: 'versions-page',
    label: 'Page Versions'
    name: 'delete-page',
    label: 'Move to Trash'
    name: 'reorganize-page',
    label: 'Reorganize'

Note that if we set this property, any options we don't spell out will NOT appear. So you can use this option to hide standard choices like "Reorganize."

The standard choices are as shown above beginning with edit-page.

"How do I implement my new-monkey option?"

The context menu outputs links with data attributes based on the name property. The links also get classes based on this property as you can see via "inspect element" in your browser.

Consider this browser-side JavaScript, which invites the user to create a new page of type monkey when the above link is clicked:

  $('body').on('click', '[data-new-monkey]', function() {
    // Get metadata about the current page we're looking at 
    var page = apos.data.aposPages.page;
    // Grab the title of the link for use at the top of the modal dialog 
    var title = $(this).text();
    // Pop open the "new page" dialog with the page type set to 
    // "monkey" 
    var $el = aposPages.newPage(page.slug, { type: 'monkey', title: title });
    return false;