Generate hierarchical books

oops. thats a self reference. i think you copied wrong link or posted into wrong window.

Yes, that would be great. Note that book.module changed substantially in D6.

Subscribing. This will probably be necessary for #143805: Auto-populate handbooks to be completed.

patches must be for d7 and then backported.

Pushing to 7.x-1.x. Also marked #307562: Patch: Generate bookpages with parents as a duplicate of this issue.

A D7 patch for this would be great.

I'd like to see this:

1. A checkbox on admin/config/development/generate/content - "Add nodes to existing books"
2. A checkbox on admin/config/development/generate/content - "Create new books"

And I'll supply the patch if I get a chance, but would be even happier to test out someone else's patch.

Here's a patch that:

1. Adds a new file to devel_generate called book.devel_generate.inc that implements the book functionality
2. Adds radio buttons to devel_generate_content_form() to let you choose how you want to associate generated "book allowed" nodes with books

If you choose to add nodes to existing books, nodes will go in there 1 level deep. I'd like to get this patch in and then enhance that if anyone cares about getting book data further into the hierarchy.

To apply:

1. Place patch file in modules/devel/devel_generate
2. patch < devel-generate-book-303143-13.patch

The patch will create the new file for you.

Here's another attempt with a patch that can be run with patch -p 0 < devel-generate-book-303143-15.patch

Setting to "needs review".

Here's a patch that also lets you choose "max depth" for book pages, so you can generate books of any depth (up to MENU_MAX_DEPTH, of course).

This patch supersedes the previous ones.

The only thing missing here would be some javascript to disable the max depth field when either not creating books or creating only new books.

This would be nice to get into devel so other folks could test out the core patch in #589440: Reordering fails with more than 31 book pages in a book.

That ORDER BY RAND() looks like a performance killer. We have solved similar problem in devel_generate elsewhere (comment parents?).

Seems like a good patch otherwise. Needs a code and functionality review from someone. That javascript is nice to have but not a prerequisite.

Yeah, I'll look into what other ways there are of doing that. I did it differently at first but found an ORDER BY RAND() somewhere in devel and thought that it made the code simpler.

Looking back at the rest of the devel_generate code, ORDER BY RAND() is still around for comment parents:

    switch ($i % 3) {
      case 1:
        $comment->pid = db_query_range("SELECT cid FROM {comment} WHERE pid = 0 AND nid = :nid ORDER BY RAND()", 0, 1, array(':nid' => $comment->nid))->fetchField();
        break;
      case 2:
        $comment->pid = db_query_range("SELECT cid FROM {comment} WHERE pid > 0 AND nid = :nid ORDER BY RAND()", 0, 1, array(':nid' => $comment->nid))->fetchField();
        break;
      default:
        $comment->pid = 0;
    }

and taxonomy:

    $sql = "SELECT tid FROM {taxonomy_term_data} WHERE vid = :vid ORDER BY RAND()";

I think it makes sense to commit this and then possibly open a separate issue to remove the ORDER BY RAND() calls, but since we're just generating random content, I don't think it's a big deal for this to be slow. It's really only slow, as far as I can tell from doing some reading, when the tables get really large.

Tables getting really big is a use case we support heartily.

See devel_generate_terms() and _taxonomy_devel_generate()

#836330: Generate menu and menu links has some info on this -- I'll use that as a reference.

ORDER BY RAND() is out; looping around mt_rand() looking for a candidate plid is in.

This strategy runs the risk of causing lots of attempts to find a good candidate plid when the menu_links table has interspersed items that don't belong to books. We first look for the min and max mlid for a book-related menu_link and then take stabs in that range in order to find a good book parent. Also, if there are lots of deep book links and you want to generate only shallow links, you'll see the same thing where it takes a lot of attempts to find a valid plid.

Still, though, it seems better than ORDER BY RAND(), and it's more D7-like.

To apply, put it in sites/all/modules/devel and run patch -p 0 < devel-generate-book-303143-23.patch

The previous patch caused an error during validation of the devel_generate_content form if the book module wasn't enabled.

To apply, put it in sites/all/modules/devel and run patch -p 0 < devel-generate-book-303143-24.patch

Code looks good. Would be good to get someone to try it and assure it works. Also, would be good to get integration into drush genc commmand.

@clemens.tolboom - That last patch is from the good old CVS days. Here's a re-rolled one that applies with one whitespace error line. I manually re-created it and haven't tested it out yet, but you should be able to apply with:

   git apply devel-generate-book-303143-27-D7.patch

Make sure to put the patch in sites/all/modules/devel (or wherever devel is on your site).

And here's a patch that gets rid of the whitespace issue so it should apply cleanly with:

   git apply devel-generate-book-303143-28-D7.patch

CNW for drush integration

@clemens.tolboom - I'm not seeing the drush genc options in your patch. I started working on this also but stopped when the book module didn't magically add its stuff to the newly created node in "drush genc" mode.

When you re-roll your patch with the genc options, please stick the comment number in the part of the patch name before the "-D7" - I'll test it out and see if we can work together to finish this up.

@clemens.tolboom - Generating top level books appears to work via drush genc with:

drush --types="book" --book-options=create-books genc 3

but, adding to books isn't working:

drush --types="book" --book-options=add-to-books genc 60

should add 60 child book pages. Instead, it adds 60 book pages that are not associated with any book.

For books that are created via admin/config/development/generate/content here's what we see in the menu_links table:

mysql> select menu_name, mlid, plid, link_path, module, link_title from menu_links where menu_name like 'book%';
+-------------+------+------+-----------+--------+---------------+
| menu_name   | mlid | plid | link_path | module | link_title    |
+-------------+------+------+-----------+--------+---------------+
| book-toc-15 |  363 |    0 | node/15   | book   | Molior Scisco |
| book-toc-16 |  364 |    0 | node/16   | book   | Abbas Ut      |
| book-toc-17 |  365 |    0 | node/17   | book   | Exerci Natu   |
+-------------+------+------+-----------+--------+---------------+
3 rows in set (0.00 sec)

If you generate 3 top level books with drush genc, here's what you'll see:

mysql> select menu_name, mlid, plid, link_path, module, link_title from menu_links where menu_name like 'book%';
+-------------+------+------+-----------+--------+-------------------------------------------+
| menu_name   | mlid | plid | link_path | module | link_title                                |
+-------------+------+------+-----------+--------+-------------------------------------------+
| book-toc-18 |  366 |    0 | node/18   | menu   | Erat Meus Quis                            |
| book-toc-19 |  367 |    0 | node/19   | menu   | Comis Consequat Ideo Pertineo Venio Vicis |
| book-toc-20 |  368 |    0 | node/20   | menu   | Iusto Probo Sudo                          |
+-------------+------+------+-----------+--------+-------------------------------------------+
3 rows in set (0.00 sec)

Note that the 'module' that owns the menu links is menu in the drush genc case, but it's book when done via the devel_generate gui. So, it looks like the book module is not inserting itself into the node creation process during drush genc.

Closing for lack of activity. Feel free to reopen if the issue still exists