| 1 |
<?php // -*-php-*-
|
| 2 |
/* $Id: fooaggregator.module,v 1.3 2007/07/15 15:02:13 msameer Exp $ */
|
| 3 |
|
| 4 |
require_once('fooaggregator_feed.inc');
|
| 5 |
require_once('fooaggregator_item.inc');
|
| 6 |
require_once('fooaggregator_views.inc');
|
| 7 |
|
| 8 |
// todo:
|
| 9 |
// don't update checked if the feed fails.
|
| 10 |
// set up a custom user agent ;-)
|
| 11 |
// update items if the feed has updated ones ?
|
| 12 |
// a way to set the defaults (admin -> content type -> aggregator feed)
|
| 13 |
// a way to support libcurl.
|
| 14 |
// import/export: opml or CSV
|
| 15 |
// cron should process published nodes only. And unpublish them if there's an error.
|
| 16 |
|
| 17 |
/**
|
| 18 |
* Implementation of hook_node_info()
|
| 19 |
*/
|
| 20 |
function fooaggregator_node_info() {
|
| 21 |
return array(
|
| 22 |
'fooaggregator_item' => array(
|
| 23 |
'name' => t('RSS feed item.'),
|
| 24 |
'module' => 'fooaggregator_item',
|
| 25 |
'description' => t('RSS feed item.'),
|
| 26 |
),
|
| 27 |
'fooaggregator_feed' => array(
|
| 28 |
'name' => t('RSS feed'),
|
| 29 |
'module' => 'fooaggregator_feed',
|
| 30 |
'description' => t('RSS feed.'),
|
| 31 |
),
|
| 32 |
);
|
| 33 |
}
|
| 34 |
|
| 35 |
/**
|
| 36 |
* Implementation of hook_menu()
|
| 37 |
*/
|
| 38 |
function fooaggregator_menu($may_cache) {
|
| 39 |
$items = array();
|
| 40 |
|
| 41 |
if ($may_cache) {
|
| 42 |
$items[] = array(
|
| 43 |
'path' => 'admin/content/fooaggregator/settings',
|
| 44 |
'callback' => 'drupal_get_form',
|
| 45 |
'callback arguments' => array('fooaggregator_admin_settings'),
|
| 46 |
'access' => user_access('administer fooaggregator'),
|
| 47 |
'title' => t('Settings'),
|
| 48 |
'type' => MENU_LOCAL_TASK,
|
| 49 |
'weight' => 2,
|
| 50 |
);
|
| 51 |
}
|
| 52 |
else if ((arg(0) == 'node') && (is_numeric(arg(1))) && arg(2)) {
|
| 53 |
$node = node_load(arg(1));
|
| 54 |
if ($node->type == 'fooaggregator_feed') {
|
| 55 |
switch (arg(2)) {
|
| 56 |
case 'clear':
|
| 57 |
$items[] = array(
|
| 58 |
'path' => 'node/'.$node->nid.'/clear',
|
| 59 |
'callback' => 'fooaggregator_clear',
|
| 60 |
'callback arguments' => array(arg(1)),
|
| 61 |
'access' => user_access('administer fooaggregator'),
|
| 62 |
'title' => t('Clear feed items'),
|
| 63 |
'type' => MENU_CALLBACK,
|
| 64 |
);
|
| 65 |
break;
|
| 66 |
case 'refresh':
|
| 67 |
$items[] = array(
|
| 68 |
'path' => 'node/'.$node->nid.'/refresh',
|
| 69 |
'callback' => 'fooaggregator_refresh',
|
| 70 |
'callback arguments' => array(arg(1)),
|
| 71 |
'access' => user_access('administer fooaggregator'),
|
| 72 |
'title' => t('Refresh the feed and update it in case of new items.'),
|
| 73 |
'type' => MENU_CALLBACK,
|
| 74 |
);
|
| 75 |
break;
|
| 76 |
}
|
| 77 |
}
|
| 78 |
}
|
| 79 |
|
| 80 |
return $items;
|
| 81 |
}
|
| 82 |
|
| 83 |
/**
|
| 84 |
* Implementation of hook_nodeapi()
|
| 85 |
*
|
| 86 |
* We delete the feed items belonging to a feed if that feed is being deleted.
|
| 87 |
*/
|
| 88 |
function fooaggregator_nodeapi(&$node, $op, $a3 = NULL, $a4 = NULL) {
|
| 89 |
if ($node->type == 'fooaggregator_feed') {
|
| 90 |
switch ($op) {
|
| 91 |
case 'delete':
|
| 92 |
$result = db_query("select nid from {fooaggregator_item} where fid=%d", $node->nid);
|
| 93 |
while ($o = db_fetch_object($result)) {
|
| 94 |
node_delete($o->nid);
|
| 95 |
}
|
| 96 |
break;
|
| 97 |
}
|
| 98 |
}
|
| 99 |
}
|
| 100 |
|
| 101 |
/**
|
| 102 |
* Implementation of hook_perm()
|
| 103 |
*/
|
| 104 |
function fooaggregator_perm() {
|
| 105 |
return array('administer fooaggregator', 'access fooaggregator');
|
| 106 |
}
|
| 107 |
|
| 108 |
/**
|
| 109 |
* Implementation of hook_access();
|
| 110 |
*/
|
| 111 |
function fooaggregator_access($op, $node) {
|
| 112 |
switch ($op) {
|
| 113 |
case 'view':
|
| 114 |
return user_access('access fooaggregator');
|
| 115 |
case 'create':
|
| 116 |
case 'update':
|
| 117 |
case 'delete':
|
| 118 |
return user_access('administer fooaggregator');
|
| 119 |
}
|
| 120 |
}
|
| 121 |
|
| 122 |
|
| 123 |
/**
|
| 124 |
* Implementation of hook_cron()
|
| 125 |
*
|
| 126 |
* This will refresh our feeds.
|
| 127 |
*/
|
| 128 |
function fooaggregator_cron() {
|
| 129 |
$limit = variable_get('fooaggregator_per_cron', 5);
|
| 130 |
$parser = _fooaggregator_parser_load();
|
| 131 |
|
| 132 |
if ($limit == -1) {
|
| 133 |
$result = db_query("select n.nid as nid from {node} n inner join {fooaggregator_feed} f on f.nid=n.nid where n.status=1 and f.refresh > 0 and f.checked+f.refresh < %d order by RAND()", time());
|
| 134 |
}
|
| 135 |
else {
|
| 136 |
$result = db_query_range("select n.nid as nid from {node} n inner join {fooaggregator_feed} f on f.nid=n.nid where n.status=1 and f.refresh > 0 and f.checked+f.refresh < %d order by RAND()", time(), 0, $limit);
|
| 137 |
}
|
| 138 |
|
| 139 |
while ($res = db_fetch_array($result)) {
|
| 140 |
$feed = node_load($res['nid'], NULL, true);
|
| 141 |
$parser['callback']($feed);
|
| 142 |
}
|
| 143 |
}
|
| 144 |
|
| 145 |
/**
|
| 146 |
* Download a feed logging any problems via the drupal watchdog system.
|
| 147 |
* This is based on the core aggregator module.
|
| 148 |
*
|
| 149 |
* We will include $feed->etag and $feed->modified if they exist in the request.
|
| 150 |
* This will also update $feed->checked and will change $feed->url if we are being redirected from the old URL.
|
| 151 |
* It will also update $feed->etag and $feed->modified from the values in the corresponding HTTP headers.
|
| 152 |
*
|
| 153 |
* @param $feed a feed object.
|
| 154 |
* @param $xml where to put the contents of the feed.
|
| 155 |
* @return false if we have an error and it will be logged otherwise true even if the feed has no new content.
|
| 156 |
*/
|
| 157 |
function _fooaggregator_download(&$feed, &$xml) {
|
| 158 |
$headers = array();
|
| 159 |
|
| 160 |
if ($feed->etag) {
|
| 161 |
$headers['If-None-Match'] = $feed->etag;
|
| 162 |
}
|
| 163 |
|
| 164 |
if ($feed->modified) {
|
| 165 |
$headers['If-Modified-Since'] = gmdate('D, d M Y H:i:s', $feed->modified) .' GMT';
|
| 166 |
}
|
| 167 |
|
| 168 |
$result = drupal_http_request($feed->url, $headers);
|
| 169 |
|
| 170 |
switch ($result->code) {
|
| 171 |
case 304:
|
| 172 |
_fooaggregator_log_notice(t('No new content for %url', array('%url' => $feed->url)));
|
| 173 |
$feed->checked = time();
|
| 174 |
break;
|
| 175 |
|
| 176 |
case 301:
|
| 177 |
_fooaggregator_log_warning(t('Updated URL for feed from %old to %new', array(
|
| 178 |
'%old' => $feed->url,
|
| 179 |
'%new' => $result->redirect_url,
|
| 180 |
)));
|
| 181 |
$feed->url = $result->redirect_url;
|
| 182 |
|
| 183 |
case 200:
|
| 184 |
case 302:
|
| 185 |
case 307:
|
| 186 |
$feed->checked = time();
|
| 187 |
if (($result->headers['Content-Length']) && ($result->headers['Content-Length'] == 0)) {
|
| 188 |
break;
|
| 189 |
}
|
| 190 |
|
| 191 |
if ($result->data == '') {
|
| 192 |
break;
|
| 193 |
}
|
| 194 |
|
| 195 |
$feed->updated = $feed->checked;
|
| 196 |
$xml = $result->data;
|
| 197 |
break;
|
| 198 |
default:
|
| 199 |
_fooaggregator_log_error(t('HTTP error %number - "%error" while getting %url', array(
|
| 200 |
'%number' => $result->code,
|
| 201 |
'%error' => $result->error,
|
| 202 |
'%url' => $feed->url,
|
| 203 |
)));
|
| 204 |
$feed->checked = time();
|
| 205 |
$result = Null;
|
| 206 |
return false;
|
| 207 |
}
|
| 208 |
|
| 209 |
// update etag and modified
|
| 210 |
if ($result->headers['Last-Modified']) {
|
| 211 |
$feed->modified = strtotime($result->headers['Last-Modified']);
|
| 212 |
}
|
| 213 |
|
| 214 |
$feed->etag = $result->headers['ETag'];
|
| 215 |
|
| 216 |
$result = Null;
|
| 217 |
|
| 218 |
return true;
|
| 219 |
}
|
| 220 |
|
| 221 |
/**
|
| 222 |
* log a warning via watchdog
|
| 223 |
*
|
| 224 |
* @param $error a string to log
|
| 225 |
*/
|
| 226 |
function _fooaggregator_log_warning($error) {
|
| 227 |
watchdog('fooaggregator', $error, WATCHDOG_WARNING);
|
| 228 |
}
|
| 229 |
|
| 230 |
/**
|
| 231 |
* log an error via watchdog
|
| 232 |
*
|
| 233 |
* @param $error a string to log
|
| 234 |
*/
|
| 235 |
function _fooaggregator_log_error($error) {
|
| 236 |
watchdog('fooaggregator', $error, WATCHDOG_ERROR);
|
| 237 |
}
|
| 238 |
|
| 239 |
/**
|
| 240 |
* log a notice via watchdog
|
| 241 |
*
|
| 242 |
* @param $error a string to log
|
| 243 |
*/
|
| 244 |
function _fooaggregator_log_notice($error) {
|
| 245 |
watchdog('fooaggregator', $error, WATCHDOG_NOTICE);
|
| 246 |
}
|
| 247 |
|
| 248 |
/**
|
| 249 |
* Check the existance of a guid for a feed item.
|
| 250 |
*
|
| 251 |
* @param $guid the guid to check for
|
| 252 |
* @param $fid the feed id
|
| 253 |
* @return the node ID or 0 if nothing exists.
|
| 254 |
*/
|
| 255 |
function _fooaggregator_guid_exists($guid, $fid) {
|
| 256 |
$nid = db_result(db_query("select nid from {fooaggregator_item} where fid=%d and guid='%s' limit 1", $fid, $guid));
|
| 257 |
return $nid === FALSE ? 0 : $nid;
|
| 258 |
}
|
| 259 |
|
| 260 |
/**
|
| 261 |
* This is the settings page.
|
| 262 |
*/
|
| 263 |
function fooaggregator_admin_settings() {
|
| 264 |
$form = array();
|
| 265 |
$form['fooaggregator_per_cron'] = array(
|
| 266 |
'#title' => t('Number of feeds to update per cron run'),
|
| 267 |
'#type' => 'textfield',
|
| 268 |
'#description' => t('-1 for all. 0 to disable the cron refresh. Decrease this if you get out of memory problems.'),
|
| 269 |
'#default_value' => variable_get('fooaggregator_per_cron', 5),
|
| 270 |
);
|
| 271 |
|
| 272 |
$users = array();
|
| 273 |
|
| 274 |
$res = db_query("select uid, name from {users}");
|
| 275 |
while ($o = db_fetch_object($res)) {
|
| 276 |
$users[$o->uid] = $o->name;
|
| 277 |
}
|
| 278 |
$users[0] = variable_get('anonymous', t('Anonymous'));
|
| 279 |
|
| 280 |
$form['fooaggregator_uid'] = array(
|
| 281 |
'#title' => t('Owner of the nodes created'),
|
| 282 |
'#type' => 'select',
|
| 283 |
'#options' => $users,
|
| 284 |
'#default_value' => variable_get('fooaggregator_uid', 1),
|
| 285 |
);
|
| 286 |
|
| 287 |
$formats = filter_formats();
|
| 288 |
$filters = array();
|
| 289 |
foreach ($formats as $format) {
|
| 290 |
$filters[$format->format] = check_plain($format->name);
|
| 291 |
}
|
| 292 |
$form['fooaggregator_filter'] = array(
|
| 293 |
'#title' => t('Default input format'),
|
| 294 |
'#type' => 'select',
|
| 295 |
'#options' => $filters,
|
| 296 |
'#default_value' => variable_get('fooaggregator_filter', 1),
|
| 297 |
);
|
| 298 |
|
| 299 |
$form['fooaggregator_parser'] = array(
|
| 300 |
'#title' => t('The parser to use'),
|
| 301 |
'#type' => 'select',
|
| 302 |
'#options' => _fooaggregator_parsers_list(),
|
| 303 |
'#default_value' => variable_get('fooaggregator_parser', 'simplepie'),
|
| 304 |
);
|
| 305 |
|
| 306 |
$form['fooaggregator_update_items']= array(
|
| 307 |
'#title' => t('Update feed items'),
|
| 308 |
'#type' => 'checkbox',
|
| 309 |
'#default_value' => variable_get('fooaggregator_update_items', 1),
|
| 310 |
'#description' => t('Uncheck this to disable updating existing nodes even if they were changed in the feed.'),
|
| 311 |
);
|
| 312 |
return system_settings_form($form);
|
| 313 |
}
|
| 314 |
|
| 315 |
/**
|
| 316 |
* list the parsers in the parsers subdirectory
|
| 317 |
*
|
| 318 |
* @return a hash table of the parsers or an empty array in case of error.
|
| 319 |
*/
|
| 320 |
function _fooaggregator_parsers_list() {
|
| 321 |
$res = array();
|
| 322 |
$path = dirname(__FILE__)."/parsers/";
|
| 323 |
$dir = opendir($path);
|
| 324 |
if (!$dir) {
|
| 325 |
drupal_set_message(t('Cannot scan the parsers subdirectory'), 'error');
|
| 326 |
return $res;
|
| 327 |
}
|
| 328 |
|
| 329 |
while (($file = readdir($dir)) !== false) {
|
| 330 |
if (is_file($path.'/'.$file.'/parser.inc')) {
|
| 331 |
$res[$file] = $file;
|
| 332 |
}
|
| 333 |
|
| 334 |
}
|
| 335 |
closedir($dir);
|
| 336 |
|
| 337 |
return $res;
|
| 338 |
}
|
| 339 |
|
| 340 |
/**
|
| 341 |
* Load the configured parser.
|
| 342 |
*
|
| 343 |
* @return A hash table with 2 keys: name and callback.
|
| 344 |
*/
|
| 345 |
function _fooaggregator_parser_load() {
|
| 346 |
$parser = variable_get('fooaggregator_parser', 'simplepie');
|
| 347 |
require_once drupal_get_path('module', 'fooaggregator')."/parsers/${parser}/parser.inc";
|
| 348 |
|
| 349 |
return array('name' => $parser, 'callback' => "_fooaggregator_${parser}_refresh");
|
| 350 |
}
|
| 351 |
|
| 352 |
/**
|
| 353 |
* implementation of hook_node_operations()
|
| 354 |
*/
|
| 355 |
function fooaggregator_node_operations() {
|
| 356 |
$operations = array(
|
| 357 |
'clear feeds' => array(
|
| 358 |
'label' => t('Clear Feeds'),
|
| 359 |
'callback' => 'fooaggregator_operations_clear',
|
| 360 |
),
|
| 361 |
);
|
| 362 |
return $operations;
|
| 363 |
}
|
| 364 |
|
| 365 |
/**
|
| 366 |
* a node operations callback
|
| 367 |
* clear multiple feeds
|
| 368 |
*
|
| 369 |
* @param $feeds array of feed IDs
|
| 370 |
*/
|
| 371 |
function fooaggregator_operations_clear($feeds) {
|
| 372 |
// MSameer: todo: performance...
|
| 373 |
//todo: a confirmation dialog
|
| 374 |
foreach ($feeds as $fid) {
|
| 375 |
$result = db_query("select nid from {node} where nid in (select nid from {fooaggregator_item} where fid=%d)", $fid);
|
| 376 |
while ($o = db_fetch_object($result)) {
|
| 377 |
node_delete($o->nid);
|
| 378 |
}
|
| 379 |
|
| 380 |
fooaggregator_feed_clear($fid);
|
| 381 |
drupal_set_message(t("feed $fid cleared."));
|
| 382 |
}
|
| 383 |
}
|
| 384 |
|
| 385 |
/**
|
| 386 |
* A menu callback.
|
| 387 |
* Clear a feed items given a feed ID
|
| 388 |
* This function will also redirect to the feed node.
|
| 389 |
*
|
| 390 |
* @param $fid a feed ID
|
| 391 |
*/
|
| 392 |
function fooaggregator_clear($fid) {
|
| 393 |
// todo: a confirmation dialog.
|
| 394 |
$result = db_query("select nid from {node} where nid in (select nid from {fooaggregator_item} where fid=%d)", $fid);
|
| 395 |
while ($o = db_fetch_object($result)) {
|
| 396 |
node_delete($o->nid);
|
| 397 |
}
|
| 398 |
|
| 399 |
fooaggregator_feed_clear($fid);
|
| 400 |
drupal_set_message(t('The feed was cleared.'));
|
| 401 |
drupal_goto('node/'.$fid);
|
| 402 |
}
|
| 403 |
|
| 404 |
|
| 405 |
/**
|
| 406 |
* Refresh a feed.
|
| 407 |
*
|
| 408 |
* @param $fid a feed ID
|
| 409 |
*/
|
| 410 |
function fooaggregator_refresh($fid) {
|
| 411 |
$parser = _fooaggregator_parser_load();
|
| 412 |
|
| 413 |
$message = "";
|
| 414 |
$feed = node_load($fid);
|
| 415 |
if ($parser['callback']($feed)) {
|
| 416 |
$message = t('Updated items for feed %title', array('%title' => $feed->title));
|
| 417 |
}
|
| 418 |
else {
|
| 419 |
$message = t('Failed to update items for feed %title', array('%title' => $feed->title));
|
| 420 |
}
|
| 421 |
drupal_set_message($message);
|
| 422 |
_fooaggregator_log_notice($message);
|
| 423 |
drupal_goto('node/'.$fid);
|
| 424 |
}
|
| 425 |
|
| 426 |
/**
|
| 427 |
* Implementation of hook_link()
|
| 428 |
*/
|
| 429 |
function fooaggregator_link($type, $node = NULL, $teaser = FALSE) {
|
| 430 |
$links = array();
|
| 431 |
|
| 432 |
if (($type == 'node') && ($node->type == 'fooaggregator_item')){
|
| 433 |
$links['fooaggregator_view_original'] = array(
|
| 434 |
'title' => t('view original'),
|
| 435 |
'href' => $node->link,
|
| 436 |
);
|
| 437 |
}
|
| 438 |
else if (($type == 'node') && ($node->type == 'fooaggregator_feed')) {
|
| 439 |
// Too bad they are showing at the end of the page if we use the fooaggregator_page module.
|
| 440 |
// I guess it has to be fixed using a theme hack or something.
|
| 441 |
if (user_access('administer aggregator')) {
|
| 442 |
$links['fooaggregator_clear'] = array(
|
| 443 |
'title' => t('clear items'),
|
| 444 |
'href' => 'node/'.$node->nid.'/clear',
|
| 445 |
);
|
| 446 |
$links['fooaggregator_refresh'] = array(
|
| 447 |
'title' => t('refresh feed'),
|
| 448 |
'href' => 'node/'.$node->nid.'/refresh',
|
| 449 |
);
|
| 450 |
}
|
| 451 |
}
|
| 452 |
|
| 453 |
return $links;
|
| 454 |
}
|
Parent Directory
| 
Revision Log
| 
Revision Graph