Lets generate kick-ass demo content!!!

We're both here, but devel_generate is moshe's baby, so you should talk to him.

There have been ideas along these lines in the past, some even with code. Please look up these issues:
#1589056: Allow other modules to generate MEDIA FILES for devel_generate
#1010394: Use devel_create_greeking for text field instead of user_password
#1277654: Generate translated content
#1400840: Friendly role based username and passwords for dummy devel generated user accounts
#1539596: only use valid users as authors
#1228470: Support for alternate dictionaries (instead of the default greeking)
#1397582: allow the generation of content types and fields
#1242352: Don't hard code the 50 node threshold for triggering batch node generation
#1326204: Allow field generate function to be alterable
#1254790: Respect fields' maximum length setting
#1297532: drush:generate-placeholder-images: Support image "generation" from the drupal files table
#616160: Extending devel generate

We want to avoid adding dependencies on external websites / data sources, for performance and maintenance reasons.

I like the direction of your push, but we need an answer to the question of who will maintain this new functionality. If this turns out to be unmaintainable/unmaintained, then we need to be able to fall back to what we have now. With that in mind, I favor adding the required hooks to Devel (or implementing dependency injection) and implementing a separate data provider module.

There's already a module for generate images from external sources to devel: Devel images provider

For documentation purposes, here is how a module can modify generated content.

function MYMODULE_node_presave($node) {
  if (isset($node->devel_generate)) {
    switch ($node->type) {
      case 'article':
        // on my site articles are in markdown, and
        // have a few hyperlinks
        $all_matches = array();
        preg_match_all('/[^ \.][^ \.]* [^ \.][^ \.]*/', $node->body[LANGUAGE_NONE][0]['value'], $all_matches);
        $matches = $all_matches[0];
        for ($i = 0; $i < 3; $i++) {
          if (count($matches)) {
            $match = $matches[rand(0, count($matches) - 1)];
            $node->body[LANGUAGE_NONE][0]['value'] = str_replace($match, '[' . $match . '](http://example.com/)', $node->body[LANGUAGE_NONE][0]['value']);
          }
        }
        $node->body[LANGUAGE_NONE][0]['format'] = 'markdown';
        break;

      default:
        // non-articles work fine with devel_geneate defaults.
        break;
    }
  }
  return $node;
}

EDIT: I forgot to add the line "return $node". Now it works.

In my experience, each site has different "typical" content: some of my clients use wysiwyg a lot so typical pages always contain Word-type gibberish. Others like using filtered_html, and generate clean paragraphs; others have custom input formats, using markdown, as in the above example. Also, different content types often have widely different typical content, but again, this differs per project.

I normally create a master module for each of the sites I develop, and put a hook like the above in that module, so that when I generate content for any given site, the generated content corresponds to the site's typical usage.

Because each site is so different, I'm not sure what more devel_generate could do to generate kick-ass content: maybe that's the role of custom modules per project.

Cheers,

Albert.

As suggested by jmolivas, we might want to rely on Faker for generating content: https://github.com/fzaninotto/Faker

@pcambra. thanksI can also imagine some uses for that with another module I'm working on.

An update to #3: if you apply the patch at #2252965: Let third parties know it's generated content and modify it. (terms), you can create a directory with a bunch of faces in sites/all/modules/my_custom_module/user-pics and use them in generated users. The following code does that, and also trims names if they are too long:

/**
 * Generate better dummy content.
 *
 * See https://drupal.org/node/1748302.
 */
function my_custom_module_user_insert(&$edit, $account, $category) {
  // hook_user_insert can be used for fields attached to the user object,
  // but not the picture.
  if (module_exists('devel_generate') && isset($edit['devel_generate']) && is_numeric($edit['uid'])) {
    $account->field_first_name[LANGUAGE_NONE][0]['value'] = drupal_substr($account->field_first_name[LANGUAGE_NONE][0]['value'], 0, rand(10, 20)); 
    $account->field_last_name[LANGUAGE_NONE][0]['value'] = drupal_substr($account->field_last_name[LANGUAGE_NONE][0]['value'], 0, rand(10, 20));
  }
}

/**
 * Generate better dummy content.
 *
 * See https://drupal.org/node/1748302.
 */
function my_custom_module_user_presave(&$edit, $account, $category) {
  // hook_user_presave can be used for the user picture, but not
  // additional fields attached to the user object.
  if (module_exists('devel_generate') && isset($edit['devel_generate']) && is_numeric($edit['uid'])) {
    $filepath = DRUPAL_ROOT . '/' . drupal_get_path('module', 'my_custom_module') . '/devel-generate-images/user/picture';
    $files = array_values(file_scan_directory($filepath, '/.*\.png$/'));
    $file = $files[rand(0, count($files) -1)];
    $contents = file_get_contents($file->uri);
    $random = md5($filepath) . rand(1000000000, 9999999999);
    $file = file_save_data($contents, 'public://user' . $random . '.png');
    $file->uid = $account->uid;
    $file = file_save($file);
    $edit['picture'] = $file;
  }
}

I wrote this module, which aims to provide an extensible system for generating kick-ass dummy content, based on the code above: Realistic Dummy Content. I have been using it for a week now on a real project and it makes a great difference in client demos! Comments welcome.

As for devel, I think we can focus efforts at #2238645: Use Ipsum module for devel generate