| 1 |
<?php
|
| 2 |
/**
|
| 3 |
* @abstract Update module for feedapi_node.module.
|
| 4 |
*
|
| 5 |
* Configures and converts an existing feedapi installation to use
|
| 6 |
* cross feed duplicate checking. When cross feed duplicate checking is enabled,
|
| 7 |
* feedapi does not create a feed item, if it exists on ANY feed, whereas normally,
|
| 8 |
* duplicates are only searched for _within_ the feed item's feed.
|
| 9 |
*
|
| 10 |
* This module converts all existing feeds to use cross feed duplicate checking on cron.
|
| 11 |
*
|
| 12 |
*/
|
| 13 |
|
| 14 |
define("FEEDAPI_NODE_UPDATE_CRON_TIMEOUT", 60);
|
| 15 |
|
| 16 |
function feedapi_node_update_cron() {
|
| 17 |
$worked = FALSE;
|
| 18 |
$start_time = time();
|
| 19 |
while (time() < (FEEDAPI_NODE_UPDATE_CRON_TIMEOUT + $start_time) ) {
|
| 20 |
// Set all feedapi enabled content types to x feed de-duping.
|
| 21 |
if (!variable_get('feedapi_node_update_node_types', 0)) {
|
| 22 |
variable_set('feedapi_node_update_node_types', 1);
|
| 23 |
$worked = TRUE;
|
| 24 |
$types = array_keys(node_get_types('types'));
|
| 25 |
foreach ($types as $type) {
|
| 26 |
if ($settings = _feedapi_get_settings(array('node_type' => $type))) {
|
| 27 |
$settings['processors']['feedapi_node']['x_dedupe'] = 1;
|
| 28 |
_feedapi_store_settings(array('node_type' => $type), $settings);
|
| 29 |
}
|
| 30 |
}
|
| 31 |
}
|
| 32 |
|
| 33 |
// Set all feeds to x feed de-duping.
|
| 34 |
$result = db_query_range(
|
| 35 |
'SELECT DISTINCT n.title, n.type, n.nid
|
| 36 |
FROM {feedapi_node_item_feed} ff
|
| 37 |
JOIN {node} n ON n.nid = ff.feed_nid
|
| 38 |
WHERE n.nid > %d
|
| 39 |
ORDER BY n.nid ASC', variable_get('feedapi_node_update_last_nid', 0), 0, 100);
|
| 40 |
$last_nid = 0;
|
| 41 |
while ($feed = db_fetch_object($result)) {
|
| 42 |
$worked = TRUE;
|
| 43 |
if ($settings = _feedapi_get_settings(array('node_type' => $feed->type, 'nid' => $feed->nid))) {
|
| 44 |
$settings['processors']['feedapi_node']['x_dedupe'] = 1;
|
| 45 |
_feedapi_store_settings(array('node_type' => $feed->type, 'nid' => $feed->nid), $settings);
|
| 46 |
}
|
| 47 |
else {
|
| 48 |
watchdog('feedapi_node_update', t("No settings for feed found."), WATCHDOG_ERROR, l($feed->title, "node/". $feed->nid));
|
| 49 |
}
|
| 50 |
$last_nid = $feed->nid;
|
| 51 |
$results = TRUE;
|
| 52 |
}
|
| 53 |
if ($last_nid) {
|
| 54 |
variable_set('feedapi_node_update_last_nid', $last_nid);
|
| 55 |
}
|
| 56 |
else {
|
| 57 |
break;
|
| 58 |
}
|
| 59 |
}
|
| 60 |
|
| 61 |
// Remove x feed duplicate feed items.
|
| 62 |
foreach (array('url', 'guid') as $field) {
|
| 63 |
$result = db_query(
|
| 64 |
"SELECT *,
|
| 65 |
GROUP_CONCAT(fni.nid ORDER BY fni.nid ASC SEPARATOR ',') AS feed_item_nids,
|
| 66 |
COUNT(*) AS cnt
|
| 67 |
FROM {feedapi_node_item} fni
|
| 68 |
GROUP BY fni.%s
|
| 69 |
ORDER BY cnt DESC", $field);
|
| 70 |
while ($dupe_set = db_fetch_object($result)) {
|
| 71 |
// drupal_set_message(dprint_r($dupe_set, true));
|
| 72 |
if ($dupe_set->cnt < 2) {
|
| 73 |
break;
|
| 74 |
}
|
| 75 |
if (time() > (FEEDAPI_NODE_UPDATE_CRON_TIMEOUT + $start_time)) {
|
| 76 |
break;
|
| 77 |
}
|
| 78 |
$worked = TRUE;
|
| 79 |
$feed_item_nids = explode(',', $dupe_set->feed_item_nids);
|
| 80 |
$first_nid = array_shift($feed_item_nids);
|
| 81 |
$first_node = node_load($first_nid);
|
| 82 |
|
| 83 |
// Get feeds associated to all feed items except the remaining feed item.
|
| 84 |
// We need to associate those feeds to the remaining feed item.
|
| 85 |
$r = db_query('SELECT feed_nid FROM feedapi_node_item_feed WHERE feed_item_nid IN (%s)', implode(", ", $feed_item_nids));
|
| 86 |
|
| 87 |
while ($feed = db_fetch_object($r)) {
|
| 88 |
$first_node->feedapi_node->feed_nids[$feed->feed_nid] = $feed->feed_nid;
|
| 89 |
}
|
| 90 |
node_save($first_node);
|
| 91 |
|
| 92 |
// Delete all feed item nodes except the first one.
|
| 93 |
foreach ($feed_item_nids as $nid) {
|
| 94 |
// drupal_set_message("DELETE $nid");
|
| 95 |
node_delete($nid);
|
| 96 |
}
|
| 97 |
}
|
| 98 |
}
|
| 99 |
|
| 100 |
if (!$worked) {
|
| 101 |
watchdog('feedapi_node_update', t("Ran feedapi_node_update_cron() but update already done. You can turn off feedapi_node_update module now."));
|
| 102 |
}
|
| 103 |
}
|