Isarra has uploaded a new change for review. https://gerrit.wikimedia.org/r/87691
Change subject: CHANGES - ashley this needs a sanity check please. ...................................................................... CHANGES - ashley this needs a sanity check please. More descriptive comments, better redel support (I hope), include user ids with everything... Change-Id: I7705dae5418f35b5459a4d5b2285becab0981b8f --- M grabDeletedFiles.php M grabDeletedText.php M grabFiles.php M grabImages.php M grabInterwikiMap.php M grabLogs.php M grabNamespaceInfo.php M grabNewText.php M grabText.php M grabUserGroups.php M mediawikibot.class.php 11 files changed, 34 insertions(+), 22 deletions(-) git pull ssh://gerrit.wikimedia.org:29418/mediawiki/tools/grabbers refs/changes/91/87691/1 diff --git a/grabDeletedFiles.php b/grabDeletedFiles.php old mode 100644 new mode 100755 index 666cd49..decc006 --- a/grabDeletedFiles.php +++ b/grabDeletedFiles.php @@ -10,7 +10,7 @@ * @note Based on code by Jack Phoenix and Edward Chernenko. */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); diff --git a/grabDeletedText.php b/grabDeletedText.php old mode 100644 new mode 100755 index c014c56..135069f --- a/grabDeletedText.php +++ b/grabDeletedText.php @@ -11,7 +11,7 @@ * @date 1 January 2013 */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); diff --git a/grabFiles.php b/grabFiles.php index 56d5921..8e98a5c 100755 --- a/grabFiles.php +++ b/grabFiles.php @@ -10,7 +10,7 @@ * @note Based on code by Misza, Jack Phoenix and Edward Chernenko. */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); diff --git a/grabImages.php b/grabImages.php old mode 100644 new mode 100755 index 8fffc21..f4c2411 --- a/grabImages.php +++ b/grabImages.php @@ -11,12 +11,8 @@ * @date 20 June 2012 */ -/** - * Set the correct include path for PHP so that we can run this script from - * $IP/extensions/ShoutWikiMaintenance and we don't need to move this file to - * $IP/maintenance/. - */ -# ini_set( 'include_path', dirname( __FILE__ ) . '/../../maintenance' ); +# Because we're in core/grabbers instead of core/maintenance +ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); diff --git a/grabInterwikiMap.php b/grabInterwikiMap.php index ab7aedf..434c6b2 100755 --- a/grabInterwikiMap.php +++ b/grabInterwikiMap.php @@ -20,7 +20,7 @@ * - Edward Chernenko <[email protected]> (MediaWikiDumper 1.1.5, interwiki.pl) */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); diff --git a/grabLogs.php b/grabLogs.php index 0c8e46c..c386a8d 100755 --- a/grabLogs.php +++ b/grabLogs.php @@ -13,7 +13,7 @@ * - Edward Chernenko <[email protected]> (MediaWikiDumper 1.1.5, logs.pl) */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); @@ -71,6 +71,7 @@ 'list' => 'logevents', 'lelimit' => 'max', 'ledir' => 'newer', + 'leprop' => 'ids|title|type|user|userid|timestamp|comment|details', ); if ( $carlb ) { # Tone this down a bit @@ -157,6 +158,7 @@ } if ( isset( $entry['userhidden'] ) ) { $entry['user'] = 'User hidden'; + $entry['userid'] = 0; $revdeleted = $revdeleted | LogPage::DELETED_USER; } @@ -182,7 +184,7 @@ 'type' => $entry['type'], 'action' => $entry['action'], 'timestamp' => $ts, - 'user' => 0, # Unsassigned + 'user' => $entry['userid'], 'user_text' => $entry['user'], 'namespace' => $ns, 'title' => $title, diff --git a/grabNamespaceInfo.php b/grabNamespaceInfo.php index 4d33e91..94695b5 100755 --- a/grabNamespaceInfo.php +++ b/grabNamespaceInfo.php @@ -10,7 +10,7 @@ * @date 1 January 2013 */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( "Maintenance.php" ); diff --git a/grabNewText.php b/grabNewText.php index 281c102..a6982e4 100755 --- a/grabNewText.php +++ b/grabNewText.php @@ -11,7 +11,7 @@ * @date 1 January 2013 */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); diff --git a/grabText.php b/grabText.php index 3228fcd..f92af6d 100755 --- a/grabText.php +++ b/grabText.php @@ -11,7 +11,7 @@ * @date 1 January 2013 */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); @@ -193,7 +193,7 @@ $localID = $pageID; $titleIsPresent = false; - $this->output( "Processing page $pageID: $title...\n" ); + $this->output( "Processing page $pageID: $title\n" ); # Trim and convert displayed title to database page title if( $ns != 0 ) { diff --git a/grabUserGroups.php b/grabUserGroups.php index 8879094..e9ac3b5 100755 --- a/grabUserGroups.php +++ b/grabUserGroups.php @@ -7,7 +7,7 @@ * @author Kunal Mehta <[email protected]> */ -# Because we're not in maintenance +# Because we're in core/grabbers instead of core/maintenance ini_set( 'include_path', dirname( __FILE__ ) . '/../maintenance' ); require_once( 'Maintenance.php' ); diff --git a/mediawikibot.class.php b/mediawikibot.class.php old mode 100644 new mode 100755 index 82b8aed..028e973 --- a/mediawikibot.class.php +++ b/mediawikibot.class.php @@ -106,11 +106,11 @@ /** Constructor */ - public function __construct( + public function __construct( $url = 'http://example.com/w/api.php', $format = 'php', - $username = 'bot', - $password = 'passwd', + $username = 'bot', + $password = 'passwd', $useragent = 'WikimediaBot Framework by JKH', $cookies = 'cookies.tmp' ) { @@ -122,7 +122,7 @@ define( 'PASSWORD', $password ); define( 'USERAGENT', $useragent ); define( 'COOKIES', $cookies ); - + } /** Dynamic method server @@ -204,6 +204,20 @@ $url = $this->api_url( $method ); # get the data $data = $this->curl_post( $url, $params, $multipart ); + # check data for grabbers; shut up loops are confusing it's too early. + if ( !isset( $data[$method] ) ) { + echo "API error: no results; retrying in 5s\n"; + sleep( 5 ); + $data = $this->curl_post( $url, $params, $multipart ); + if ( !isset( $data[$method] ) ) { + echo "API error: no results; retrying in 30s\n"; + sleep( 30 ); + $data = $this->curl_post( $url, $params, $multipart ); + if ( !isset( $data[$method] ) ) { + echo "API error: no results found.\n"; + } + } + } # set smwinfo $this->$method = $data; # return the data @@ -290,7 +304,7 @@ */ private function urlize_params( $params ) { # url-ify the data for POST - $urlstring = ''; + $urlstring = ''; foreach ( $params as $key => $value ) { $urlstring .= $key . '=' . $value . '&'; } -- To view, visit https://gerrit.wikimedia.org/r/87691 To unsubscribe, visit https://gerrit.wikimedia.org/r/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I7705dae5418f35b5459a4d5b2285becab0981b8f Gerrit-PatchSet: 1 Gerrit-Project: mediawiki/tools/grabbers Gerrit-Branch: master Gerrit-Owner: Isarra <[email protected]> _______________________________________________ MediaWiki-commits mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
