/[drupal]/contributions/modules/import_html/import_html.module

Diff of /contributions/modules/import_html/import_html.module

Parent Directory | Revision Log | View Revision Graph Revision Graph | View Patch Patch

-revision 1.71.2.15,
Wed Jul  1 02:40:56 2009 UTC
+revision 1.71.2.16,
Tue Oct  6 14:09:02 2009 UTC
 Line 1
  <?php
- // $Id: import_html.module,v 1.71.2.14 2009/05/17 15:12:46 dman Exp $
+ // $Id: import_html.module,v 1.71.2.15 2009/07/01 02:40:56 dman Exp $
  /**
   * @file Main drupal interface to import_html.
   *
   * This file contains Drupal hooks
   * and some config/preferences. Actual processing functions are in
   * import_html_process.inc
   *
   * Synopsis:
   *
   * Facility to import an existing, static HTML site structure into the Drupal
   * CMS as structured nodes.
   *
 Line 30
   *
   * import_html libraries can be used by external modules, specifically wrapper.
   * module (private development) and static.module (experimental)
   *
   * CODE:
   * Internally, there are a lot of filepath fragments being passed around.
   * Any filepath known to represent a folder (eg $source_siteroot) will end with
 Line 39
   * dividers. However, many times, small functions like 'ensure_trailing_slash()'
   * may be called. Not all of them are needed, but most of them were put in due
   * to some wierd filesystem vagery or filename encountered.
   *
   *
   * @ingroup import_html Import HTML
   *
   * @author Dan Morrison http://coders.co.nz/
   *
   */
 Line 108 
 define("IMPORT_HTML_MAX_LABEL_LENGTH", 2
  define('IMPORT_HTML_ADMIN_PATH', 'admin/build/import_html');
  define('IMPORT_HTML_GLOB_BEFORE', 'glob before');
  define('IMPORT_HTML_RECURSE_AFTER', 'recurse after ');
  /**
   * To avoid overloading the treeview display, directory listing depth recursion
   * will be truncated if the number of found files is greater than this.
   */
  define('IMPORT_HTML_MAX_FILE_LISTING_COUNT', 500);
  /**
   * If the database max_allowed_packet size is limited, then huge batch jobs
 Line 134 
 define('IMPORT_HTML_REQUIRED_MAX_ALLOWED
   */
  function import_html_menu() {
    $items[IMPORT_HTML_ADMIN_PATH] = array(
      'title' => 'Import HTML',
      'description' => "Import/Export an entire site or directory to/from static HTML",
      // Use the built-in Overview menu thing
 Line 142 
 function import_html_menu() {
      'file' => 'system.admin.inc',
      'file path' => drupal_get_path('module', 'system'),
      'access arguments' => array('access import_html'),
      'type' => MENU_NORMAL_ITEM,
    );
    $items[IMPORT_HTML_ADMIN_PATH .'/settings'] = array(
 Line 165 
 function import_html_menu() {
      'type' => MENU_NORMAL_ITEM,
    );
    $items[IMPORT_HTML_ADMIN_PATH .'/import_site'] = array(
      'title' => 'Import HTML Site',
      'description' => "Import/Export an entire site to/from static HTML",
      'weight' => -1,
      'page callback' => 'drupal_get_form',
      'page arguments' => array('import_html_process_form'),
      'file' => 'import_html_ui.inc',
      'access arguments' => array('access import_html'),
      'type' => MENU_NORMAL_ITEM,
    );
    $items[IMPORT_HTML_ADMIN_PATH .'/demo'] = array(
      'title' => 'Demo',
      'weight' => 3,
      'description' => 'Demonstrate or test HTML Import on one file.',
      'page callback' => 'drupal_get_form',
 Line 185 
 function import_html_menu() {
      'type' => MENU_NORMAL_ITEM,
    );
    $items[IMPORT_HTML_ADMIN_PATH .'/results'] = array(
      'title' => 'Import Results Summary',
      'weight' => 5,
      'description' => "View log of recent import messages",
      'page callback' => 'import_html_results',
      'file' => 'import_html_ui.inc',
      'access arguments' => array('access import_html'),
      'type' => MENU_NORMAL_ITEM,
    );
 Line 228 
 function import_html_help($path, $arg) {
        $output = t("<p>
            For background, remember to read <a href='!help_link'>the Import Html help page</a>.
            The <a href='!settings_link'>settings page</a> contains the config options.
          </p>",
          array(
            '!help_link' => url('admin/help/import_html'),
            '!settings_link' => url(IMPORT_HTML_ADMIN_PATH .'/settings'),
          )
        );
 Line 248 
 function import_html_help($path, $arg) {
        </p><p>
          If you see <em>too much</em> of the page in the 'body' area (nested navbars and layout)
          then the XSL import template or selector needs to be made more specific.
          If you see none, or not enough content in the body area, the template or
          selector needs to be adjusted to encompass the text correctly.
        </p><p>
          If you get an error or no result, the input HTML is probably too invalid to work with.
        </p><p>
          Single demo imports do not have the full context information to work with,
          so the menu or URL alias (and internal relinking) shown may not be representative of the real result.
        </p>");
 Line 282 
 function import_html_form_alter(&$form,
  /**
   * A wrapper to variable_set, variable_get to encapsulate multiple import 'profiles'
   *
   * This natively just returns the settings from the current 'default' profile,
   * but also allows the settings forms to be extended to other sets.
   *
   * Use INSTEAD OF variable_get() and it will return the 'default' or 'active' profile vars.
   *
   * @param $var Name of the variable within the currently active profile to retrieve.
   * @param $val If set, sets this variable within the profile and saves it.
   * @see import_html_current_profile();
   */
  function import_html_variable($var, $val = NULL) {
 Line 304 
 function import_html_variable($var, $val
    if (empty($import_html_profiles[$import_html_current_profile_id])) {
      // Fill in defaults (should only be needed first time, if that
      drupal_set_message('import_html_variable initing profile from nowhere - should this ever happen?');
      $import_html_profiles[$import_html_current_profile_id] = import_html_profile_defaults();
    }
    $import_html_profile = &$import_html_profiles[$import_html_current_profile_id];
 Line 314 
 function import_html_variable($var, $val
      #$import_html_profiles[$import_html_current_profile_id] = $import_html_profile;
      variable_set('import_html_profiles', $import_html_profiles);
    }
    return $import_html_profile[$var];
  }
  /**
   * Accessor for the current profile data
   *
   * Returns the data in a named import_html_profile.
   * The currently active one if not explicitly defined.
   * Profile will be padded with expected default fields if not explicitly set (to
   * assist upgrades)
   *
   * @param $profile If set, saves this data back to the saved settings. Pass FALSE to delete it.
   */
  function import_html_profile($profile_id = NULL, $profile = NULL) {
 Line 337 
 function import_html_profile($profile_id
    if (isset($profile)) {
      $import_html_profiles[$import_html_profile_id] = $profile;
      // Allow a quiet delete
      if (!$profile) {
        unset($import_html_profiles[$import_html_profile_id]);
      }
      variable_set('import_html_profiles', $import_html_profiles);
    }
 Line 348 
 function import_html_profile($profile_id
  /**
   * Accessor for a persistant profile switcher.
   *
   * @param $profile_id If set, this becomes the active profile for the duration of the request.
   * @returns the current set profile id. 'default' by default.
   */
 Line 406 
 function import_html_profile_defaults()
      'recursion_behaviour' => IMPORT_HTML_GLOB_BEFORE,
      'debug_level' => 0,
      'keep_temp_files' => FALSE,
    );
  }
 Line 425 
 function import_html_profile_defaults()
  function import_html_get_preferred_filter($as_list = FALSE) {
    static $fid;
    if ($fid) return $fid;
    if (($preferred_filter = variable_get('import_html_preferred_filter', 0)) && (!$as_list)) {
      return $preferred_filter;
    }
 Line 435 
 function import_html_get_preferred_filte
    // Detect the ID of 'Unfiltered HTML' format, if available
    // it's almost always 3, but I guess we must look it up.
    $ff = filter_formats();
    $formats = array();
    foreach ($ff as $f) {
      $formats[$f->format] = $f->name;
 Line 465 
 function import_html_get_preferred_filte
  /**
   * Returns what general 'type' a file probably is, based on suffix or mime if
   * available.
   *
   * @returns one of the defined 'file_classes' : page|image|resource|document ... or null if unknown
   *
   * This is mainly used for UI coloring, so is not totally cannonic. HTML-or-not
   * is all that really counts.
   *
   * @see $_import_html_file_classes
   */
  function import_html_guess_file_class($filename) {
 Line 488 
 function import_html_guess_file_class($f
      if ($mime_type == 'application') return 'document'; // gross generalization
      # return 'resource';
    }
    // Some file mirrors (wget or myself) may have produced odd filenames
    // strip URL args like # and ? off it
    $filename = preg_replace('|[\?\#].*$|', '', $filename);
 Line 497 
 function import_html_guess_file_class($f
      // assume no suffix at all is a html page
      return 'html';
    }
    $extension = pathinfo($filename, PATHINFO_EXTENSION);
    return @$_import_html_file_classes[strtolower($extension)];
  }
 Line 521 
 function import_html_theme() {
  /**
   * Implementation of hook_elements().
   *
   * Declare our custom pseudo for item
   */
  function import_html_elements() {
 Line 534 
 function import_html_elements() {
  ///////////////////////////////////////////////////////
  // Batch operations are really part of import_html_process, but
  // need to be in this module file so the batch runner can FIND the jobs
  // Batch ops do not allow inc file inclusions AFAIK.
  /**
   * Return a batch set containing the instructions to run over many files.
   *
   * A batch set is a queue consisting of multiple batch jobs.
   *
   * @see import_html_import_file_batch_job()
   *
   * @param $file_list
   * @param $profile context settings.
   */
 Line 575 
 function import_html_import_file_batch_j
    $results = import_html_import_file($rel_path, $profile);
    if (! $results) {
      import_html_debug(
        'Failed to get any results from the attempted analysis of %rel_path.
          The source file path was probably unavailable, invalid or incorrect.',
        array('%rel_path' => $rel_path),
        WATCHDOG_ERROR
      );
      return FALSE;
    }
    foreach ($results as $summary) {
      // Almost useless loop, usually just one node per file
      if (! empty($summary['node'])) {
        $batch_context['message'] = "Processed <span class='rel-path'>$rel_path</span> : <span class='node-title'>" . $summary['node']->title ."</span>";
 Line 610 
 function import_html_import_file_batch_j
  /**
   * Called by the batch queue, invoke import_html_import_directory() and update
   * the batch context with the status
   *
   * @param $rel_path
   * @param $profile
   */
 Line 618 
 function import_html_import_directory_ba
    // Processing a directory when in batch context means adding batch jobs to the end of the current process
    // - that in turn may add more jobs.
    // List the files we can, and add them as jobs.
    import_html_debug("Starting batch import dir job '$rel_path' ", array('%rel_path' => $rel_path), WATCHDOG_DEBUG );
    $working_path = $profile['source_siteroot'] . $rel_path;
    // Scan the given directory (NOT deep), add the files, and the subdirectories.
    // Processing the subdir will do the recursion itself
    $dir_structure = import_html_file_scan_directory($working_path, ".*", array('.', '..', 'CVS'), 0, FALSE, 'filename', 0, NULL, 1);
 Line 662 
 function import_html_batch_import_finish
    $duration = time() - variable_get('import_html_last_import_timestamp', time());
    // results returns a list of node path -titles . Convert to something we can theme
    array_walk(
      $results,
      create_function('&$title, $link', '$title = array("href" => $link, "title" => url($link) ." : ". $title);')
    );
    import_html_debug(
      '<strong>Batch import completed</strong> in %duration. %count items processed (%average) : !links',
      array(
 Line 675 
 function import_html_batch_import_finish
        '%count' => count($results),
        '%average' => count($results) ? format_interval($duration / count($results)) .' per item' : "no time details",
        '!links' => theme('links', array_values($results), 'tree'),
      ),
      WATCHDOG_NOTICE
    );
    #dpm($results);
  }

 Legend:



Removed from v.1.71.2.15
 


changed lines


 
Added in v.1.71.2.16
 Legend:



Removed from v.1.71.2.15
 


changed lines


 
Added in v.1.71.2.16
-Removed from v.1.71.2.15
+Added in v.1.71.2.16

	ViewVC Help
Powered by ViewVC 1.1.2