'subform',
'title' => t('weblinker'),
'callback' => 'weblinker_form_page',
'type' => MENU_CALLBACK,
'callback arguments' => array(
'edit'=>$_REQUEST,
),
'access' => $access,
);
# }
if (arg(0) == 'node' && is_numeric(arg(1))) {
$node = node_load(arg(1));
if ($node->nid) {
$items[] = array(
'path' => 'node/'. arg(1) . '/weblinker',
'title' => t('weblinker'),
'callback' => 'weblinker_page',
'access' => node_access('update', $node),
'type' => MENU_LOCAL_TASK
);
}
}
return $items;
}
/**
* Insert a table listing all embedded links found in this page
*/
function weblinker_form_alter($form_id, &$form){
if($form_id == 'page_node_form'){
drupal_add_js('misc/progress.js');
drupal_add_js(drupal_get_path('module','weblinker').'/weblinker.js');
// this will cause problems between 4.7 and 5.x
/*
if(file_exists('misc/jquery.js')){
drupal_add_js('misc/jquery.js');
}
else {
drupal_add_js(drupal_get_path('module','weblinker').'/jquery.js');
}
*/
$weblinker_results = weblinker_scan_links($form['#node']);
$form['weblinker'] = array(
'#value' => theme('weblinker_found_links',$weblinker_results),
'#weight' => 5,
);
}
}
/**
* Process a page, scanning it for embedded links
*
* Default is to re-display the page with a button to assist creation of a
* corresponding weblink node.
* If op is 'auto', the weblinks will be created without any more user input.
*/
function weblinker_page($op='show_link'){
$node = node_load(arg(1));
drupal_add_js('misc/progress.js');
drupal_add_js(drupal_get_path('module','weblinker').'/weblinker.js');
drupal_set_html_head('');
$weblinker_results = weblinker_scan_links($node);
# dsm($weblinker_results);
if ($op == 'auto'){
// process the results into new nodes
weblinker_create_weblinks($weblinker_results,$node);
}
// Display the scan results above the page
$summary = theme('weblinker_found_links',$weblinker_results);
$node->body = $summary . $node->body;
return node_show($node, arg(2));
}
/**
* Find all links in the body of the node.
*
* Return a prepared array of those links, with weblinks attached if they exist
*/
function weblinker_scan_links($node){
// find all href links
$text = $node->body;
$weblinker_results = array();
$found_weblinks = array();
$local_weblinks = array();
// Thanks to Chirp Internet: www.chirp.com.au http://www.the-art-of-web.com/php/parse-links/
// there's a lookbehind assertion (? (from some truly bad input pages)
$pattern = "|(]*href=)(['\"]??)([^'\"]+?)\\2([^>]*(?)(.*)|";
if(preg_match_all($pattern."siU", $text, $matches)) {}
// $matches[3] is a list of links,
// $matches[5] is a list of titles;
# dsm($matches);
// massage the results into annotatable form
foreach($matches[3] as $ix=>$link){
$link = weblinker_scrub_url($link);
$url_parts = parse_url($link);
$local = (! isset($url_parts['host']));
$weblinks = weblinks_get($link, TRUE);
$weblink = array_pop($weblinks); // only use the first found
$details = array(
'href' => $link,
'title' => $matches[5][$ix],
'weblink' => $weblink,
'local' => $local,
);
$weblinker_results[$link] = $details;
if($details['weblink']) { // used for counting
$found_weblinks[$link] = $details['weblink'];
}
if($local){
$local_weblinks[$link] = $details['weblink'];
}
}
if($weblinker_results){
$auto_link = url('node/'. arg(1) . '/weblinker/auto');
drupal_set_message(t('Found %weblinker_count links in the page. %local_count are internal, local references. %found_count are already registered as weblinks. Import all links now - using the current pages properties.',array('%weblinker_count'=>count($weblinker_results), '%found_count' => count($found_weblinks), '%local_count' => count($local_weblinks), '!auto_link'=>$auto_link )));
}
return $weblinker_results;
}
/**
* Given an array of links, render them, with some supplimentary info ina
* table.
*
* The array may or may not have been checked at this stage. If checked, they
* will have a 'weblink' property attached to them.
*
*/
function theme_weblinker_found_links($weblinker_results){
$rows = array();
foreach($weblinker_results as $details){
$row = array(
'title' => l($details['title'],$details['href']),
'status' => ($details['remote_info']) ? $details['remote_info']['http_code'] : 'unchecked'
);
if($details['local']){
$row['status'] = 'local';
}
$class = 'status-' . $row['status'];
if($details['weblink']){ // It seems valid
if($details['weblink']->nid ){ // it exists already
$row['status'] = l('linked','node/'.$details['weblink']->nid) ;
if($details['remote_info'] && ($details['remote_info']['redirect_count'])){
$row['status'] .= '('. $details['remote_info']['redirect_count'] .' redirects)' ;
}
$class = 'weblinker-status-linked';
// Add an inline edit action.
//
// If it already exists, we only have to send the nid, and maybe the parent
if($details['weblink']->nid){
$parameter_array = array(
'nid' => $details['weblink']->nid,
'parent_nid' => $details['weblink']->parent_nid,
);
}
} else {
// No weblink exists for this yet
// Inline all the data we can
$row['status'] = l('create?','node/add/weblinks') ;
$parameter_array = $details['weblink']; // send everything we know to the subform
}
$parameters = http_build_query($parameter_array, NULL, '&'); // drupal l() escapes the ampersand for me, don't do it twice
$row['action'] = l('edit', 'subform/weblinks', array('class'=>'sub-form-link'), $parameters );
} // invalid/unchecked link
$rows[] = array('data'=>$row, 'class'=>$class);
}
$attributes = array(
'class' => 'weblinker-summary',
);
return theme('table', array_keys($row), $rows, $attributes, 'Links found in this page');
}
/**
* Given a list of links, instantiate a weblink object for each one.
*
* Does not double-up if it's already registered
*
* Pass by reference enables the verification (remote lookup) to set a flag in
* the original array.
*/
function weblinker_create_weblinks(&$link_list, $context_node){
$weblinks_vid = _weblinks_get_vid();
$weblinks_default_tid = weblinks_default_tid();
global $weblinker_count;
foreach($link_list as $ix => &$details){
if($details['local']){
continue;
}
// Check if this is already a known link
if(! empty($details['weblink'])){
# foreach($details['weblinks'] as $node){
# drupal_set_message(t("Already have a weblink entry for this, link %link", array('%nid'=>$node->nid, '%link' => $details['href'])) );
# }
}
else { // Make it on-the-fly now
# drupal_set_message(t("Going to create weblink to %link", array('%link' => $details['href'])) );
// use the FAPI rather than do it by hand
$form_id = 'weblinks';
$weblink_def = array(
'type' => 'weblinks',
'title' => strip_tags($details['title']),
'body' => $details['title'],
'status' => 1,
'url' => $details['href'],
// inheirit parent nodes owner
'name' => $context_node->name,
'uid' => $context_node->uid,
// inheirit any classifications the parent node may have had
'taxonomy' => $context_node->taxonomy ? $context_node->taxonomy : array(),
);
if(! $weblink_def['taxonomy'][$weblinks_vid]){
// stop weblinks complaining about unclassified links.
if($weblinks_default_tid){
$weblink_def['taxonomy'][$weblinks_vid] = array($weblinks_default_tid => $weblinks_default_tid);
}
else {
drupal_set_message(t('Problem initializing new weblink - there is no default term in the weblinks vocab. This can probably be fixed in the weblinks settings'),'error');
}
}
// Do remote lookups to try and retrieve the links status from the live location.
$remote_info = weblinker_fetch_remote_info($details['href']);
if($remote_info['http_code'] >= 400){
drupal_set_message(t("Remote lookup for !link failed. Request returned %http_code. This weblink is not going to be created.",array('!link' => l($details['href'],$details['href']), '%http_code' => $remote_info['http_code']) ));
}
else{
// so far so good. Fill in the retrieved details.
// Use the remote pages own description as the local body
if($remote_info['description']) {
$weblink_def['body'] = '' . trim(filter_xss($remote_info['description'])) . '
' . $weblink_def['body'];
}
if($remote_info['title']) {
$weblink_def['body'] = '' . trim(filter_xss($remote_info['title'])) . '
';
drupal_set_message(t("Actual remote title for %local_title was !remote_title.",array('%local_title' => $weblink_def['title'], '!remote_title' => l($remote_info['title'],$details['href'])) ));
if(! trim($weblink_def['title']) ) $weblink_def['title'] = trim(strip_tags($remote_info['title']));
}
// Submit the form using these values.
// Drupal 5
// drupal_execute($form_id, $weblink_def);
// Drupal 4.7
$weblink_def = (object)$weblink_def;
#dsm($weblink_def);
node_save($weblink_def);
drupal_set_message(t('Created weblink !weblink',array('!weblink' => l($weblink_def->title, 'node/'.$weblink_def->nid))));
$weblinker_count ++ ;
}
$details['remote_info'] = $remote_info; // pass this back if the renderer wants it
$details['weblink'] = $weblink_def; // pass this back if the renderer wants it
}
}
}
/**
* Retrieve the header from the named page from the remote site
*/
function weblinker_fetch_remote_info($link){
drupal_set_message(t("Remotely requesting the page at !link",array('!link' => l($link,$link)) ));
// use CURL to fetch the header response for this link
static $ch;
if(!$ch){
$ch = curl_init();
curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, TRUE); // follow remote redirects
curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 5);
}
curl_setopt($ch, CURLOPT_URL, $link);
$result = curl_exec($ch);
$info = curl_getinfo($ch);
if (curl_errno($ch)) {
drupal_set_message(curl_error($ch),'error');
} else {
//
// Scan result for title and description
// XML would be easier, but I don't want to require it'
$title_pattern = "|]*>([^<]+)|i";
$h1_pattern = "|]*>([^<]+)|i";
if(preg_match ($title_pattern, $result, $matches)){
$info['title'] = check_plain($matches[1]);
}
else if(preg_match ($h1_pattern, $result, $matches)){
$info['title'] = check_plain($matches[1]);
}
$description_pattern = '|]+name=[\'"]description[\'"][^>]+content=[\'"](.*)[\'"][^>]*>|i';
$description_pattern2 = '|]+content=[\'"](.*)[\'"][^>]+name=[\'"]description[\'"][^>]*>|i';
if(preg_match ($description_pattern, $result, $matches)){
$info['description'] = check_plain($matches[1]);
}
else if(preg_match ($description_pattern2, $result, $matches)){
$info['description'] = check_plain($matches[1]);
}
}
// do not close curl, re-use it
// curl_close($ch);
return($info);
}
function weblinks_default_tid(){
// weblinks requires a taxonomy entry for each item.
// Thus we need a default
$weblinks_vid = _weblinks_get_vid();
$weblinks_default_tid = variable_get('weblinks_default_tid',0);
if(! $weblinks_default_tid){
$weblinks_vocab = taxonomy_get_tree($weblinks_vid, 0, -1, 1);
$weblinks_default_term = array_pop($weblinks_vocab);
if($weblinks_default_term) $weblinks_default_tid = $weblinks_default_term->tid;
}
return $weblinks_default_tid;
}
/**
* A cut-down 'create weblink' edit page, suitable to embed or pop-up from other
* contexts, such as bulk action screens.
*
* Renders a form much like the node/add/weblinks one, but without the extras.
*
* Initializes from the nid given, or one found in the edit parameters array
*/
function weblinker_form_page($edit = array(), $request_nid='add'){
$nid = is_numeric($request_nid) ? $request_nid : $edit['nid'];
if($nid){
$node = node_load($nid);
}
$form = weblinker_form($node, $edit);
$output = drupal_get_form('weblinks_node_form', $form, 'node_form');
// No chrome.
print $output;
return;
}
/**
* Returns a form much like the node/add/weblinks one, but without the extras.
*
* sorta Implementation of hook_form
*/
function weblinker_form(&$node, &$edit = array()){
print_r($node);
print_r($edit);
if(! $node){
$node = (object)array('type'=>'weblinks');
}
$edit = array_merge((array)$node,$edit);
// Start with a normal weblink node edit form
// then simplify it a bit.
$form = weblinks_form(&$edit);
unset($form['body_filter']);
# $form['#action']="/node/add/weblinks";
$form['title']['#default_value'] = $edit['title'];
$form['url']['#default_value'] = $edit['url'];
$form['weight']['#type']='hidden';
$form['weight']['#default_value']=0;
//Get taxonomy to garnish this mini-form, but not all the others
// Need a fake node so taxonomy knows what to do
$form['#node'] = $node;
$form['type']['#value'] = 'weblinks';
if($edit['parent_nid']){
// preset/merge with its parents tags
$form['#node']->taxonomy = array_merge((array)$form['#node']->taxonomy, taxonomy_node_get_terms($edit['parent_nid']));
}
// just invoke the taxonomy additions
taxonomy_form_alter('weblinks_node_form', $form);
unset($form['type']); // don't let the other ones see it or they'll want to join in when drupal_get_form does its form_alter.
$form['submit'] = array('#type' => 'submit', '#value' => t('Submit'));
return $form;
}
/**
* repair any badness we find in the URLs - specifically any old redirect bounce
* pages or redundant referrers.
*/
function weblinker_scrub_url($link){
// @TODO make this configurable
$patterns = array(
'|^/click2.php\?|i' => '',
'|^http\://service\.bfast\.com/.*\?key=U|' => '',
);
foreach($patterns as $pattern => $replacement){
$link = preg_replace($pattern, $replacement, $link);
}
return $link;
}
/**
* Find and load the nid(s) that matches the given URL
*
*
* @param string $url
* @param boolean $load whether to node_load the entire weblink object while we
* are there
* @return array of partial url definitions, indexed by nid
* You probably want weblinks_get($url)[0][nid]
*/
function weblinks_get($url, $load = FALSE){
$query = "SELECT * FROM {weblinks} where url = '%s'";
$result = db_query(db_rewrite_sql($query),$url);
$nodes = array();
while ($data = db_fetch_object($result)){
$nodes[$data->nid] = $data;
if($load) $nodes[$data->nid] = node_load($data->nid);
}
return $nodes;
}