Add parameters to importTextFiles.php

- Add --bot to mark edits as bot edits when --rc is specified
- Add --overwrite to overwrite existing pages with changes.
This respects the --use-timestamp option by only overwriting if
the file is newer than the latest revision on the destination page.
- Add --prefix for specifying a prefix
- Add --rc to add an edit entry to Special:RecentChanges

This is a GCI task.

Change-Id: I5acf829409853e2b311ae6c1c75a009fef91ceeb

diff --git a/maintenance/importTextFiles.php b/maintenance/importTextFiles.php
index b26eaa5..14d8420 100644 (file)
--- a/maintenance/importTextFiles.php
+++ b/maintenance/importTextFiles.php
@@ -38,13 +38,22 @@ class ImportTextFiles extends Maintenance {
                $this->addOption( 'summary', 'Specify edit summary for the edits', false, true, 's' );
                $this->addOption( 'use-timestamp', 'Use the modification date of the text file ' .
                        'as the timestamp for the edit' );
-               $this->addArg( 'titles', 'Titles of article to edit' );
+               $this->addOption( 'overwrite', 'Overwrite existing pages. If --use-timestamp is passed, this ' .
+                       'will only overwrite pages if the file has been modified since the page was last modified.' );
+               $this->addOption( 'prefix', 'A string to place in front of the file name', false, true, 'p' );
+               $this->addOption( 'bot', 'Mark edits as bot edits in the recent changes list.' );
+               $this->addOption( 'rc', 'Place revisions in RecentChanges.' );
+               $this->addArg( 'files', 'Files to import' );
        }
 
        public function execute() {
                $userName = $this->getOption( 'user', false );
                $summary = $this->getOption( 'summary', 'Imported from text file' );
                $useTimestamp = $this->hasOption( 'use-timestamp' );
+               $rc = $this->hasOption( 'rc' );
+               $bot = $this->hasOption( 'bot' );
+               $overwrite = $this->hasOption( 'overwrite' );
+               $prefix = $this->getOption( 'prefix', '' );
 
                // Get all the arguments. A loop is required since Maintenance doesn't
                // suppport an arbitrary number of arguments.
@@ -59,7 +68,7 @@ class ImportTextFiles extends Maintenance {
                };
 
                $count = count( $files );
-               $this->output( "Creating $count pages...\n" );
+               $this->output( "Importing $count pages...\n" );
 
                if ( $userName === false ) {
                        $user = User::newSystemUser( 'Maintenance script', array( 'steal' => true ) );
@@ -81,46 +90,103 @@ class ImportTextFiles extends Maintenance {
                $skipCount = 0;
 
                foreach ( $files as $file => $text ) {
-                       $pageName = pathinfo( $file, PATHINFO_FILENAME );
+                       $pageName = $prefix . pathinfo( $file, PATHINFO_FILENAME );
+                       $timestamp = $useTimestamp ? wfTimestamp( TS_UNIX, filemtime( $file ) ) : wfTimestampNow();
+
                        $title = Title::newFromText( $pageName );
+                       $exists = $title->exists();
+                       $oldRevID = $title->getLatestRevID();
+                       $oldRev = $oldRevID ? Revision::newFromId( $oldRevID ) : null;
+
                        if ( !$title ) {
                                $this->error( "Invalid title $pageName. Skipping.\n" );
                                $skipCount++;
                                continue;
                        }
 
-                       if ( $title->exists() ) {
-                               $actualTitle = $title->getPrefixedText();
-                               $this->output( "Title $pageName already exists. Skipping.\n" );
-                               $skipCount++;
-                               continue;
-                       }
-
                        $actualTitle = $title->getPrefixedText();
 
+                       if ( $exists ) {
+                               $touched = wfTimestamp( TS_UNIX, $title->getTouched() );
+                               if ( !$overwrite ) {
+                                       $this->output( "Title $actualTitle already exists. Skipping.\n" );
+                                       $skipCount++;
+                                       continue;
+                               } elseif ( $useTimestamp && intval( $touched ) >= intval( $timestamp ) ) {
+                                       $this->output( "File for title $actualTitle has not been modified since the " .
+                                               "destination page was touched. Skipping.\n" );
+                                       $skipCount++;
+                                       continue;
+                               }
+                       }
+
                        $rev = new WikiRevision( ConfigFactory::getDefaultInstance()->makeConfig( 'main' ) );
-                       $rev->setText( $text );
+                       $rev->setText( rtrim( $text ) );
                        $rev->setTitle( $title );
                        $rev->setUserObj( $user );
                        $rev->setComment( $summary );
-                       if ( $useTimestamp ) {
-                               $rev->setTimestamp( wfTimestamp( TS_UNIX, filemtime( $file ) ) );
-                       } else {
-                               $rev->setTimestamp( wfTimestampNow() );
+                       $rev->setTimestamp( $timestamp );
+
+                       if ( $exists && $overwrite && $rev->getContent()->equals( $oldRev->getContent() ) ) {
+                               $this->output( "File for title $actualTitle contains no changes from the current " .
+                                       "revision. Skipping.\n" );
+                               $skipCount++;
+                               continue;
                        }
 
                        $status = $rev->importOldRevision();
+                       $newId = $title->getLatestRevID();
+
                        if ( $status ) {
-                               $this->output( "Successfully created $actualTitle\n" );
+                               $action = $exists ? 'updated' : 'created';
+                               $this->output( "Successfully $action $actualTitle\n" );
                                $successCount++;
                        } else {
-                               $actualTitle = $title->getPrefixedText();
-                               $this->output( "Failed to create $actualTitle\n" );
+                               $action = $exists ? 'update' : 'create';
+                               $this->output( "Failed to $action $actualTitle\n" );
                                $failCount++;
                                $exit = 1;
                        }
+
+                       // Create the RecentChanges entry if necessary
+                       if ( $rc && $status ) {
+                               if ( $exists ) {
+                                       if ( is_object( $oldRev ) ) {
+                                               $oldContent = $oldRev->getContent();
+                                               RecentChange::notifyEdit(
+                                                       $timestamp,
+                                                       $title,
+                                                       $rev->getMinor(),
+                                                       $user,
+                                                       $summary,
+                                                       $oldRevID,
+                                                       $oldRev->getTimestamp(),
+                                                       $bot,
+                                                       '',
+                                                       $oldContent ? $oldContent->getSize() : 0,
+                                                       $rev->getContent()->getSize(),
+                                                       $newId,
+                                                       1 /* the pages don't need to be patrolled */
+                                               );
+                                       }
+                               } else {
+                                       RecentChange::notifyNew(
+                                               $timestamp,
+                                               $title,
+                                               $rev->getMinor(),
+                                               $user,
+                                               $summary,
+                                               $bot,
+                                               '',
+                                               $rev->getContent()->getSize(),
+                                               $newId,
+                                               1
+                                       );
+                               }
+                       }
                }
-               $this->output( "Done! $successCount successfully created, $skipCount skipped.\n" );
+
+               $this->output( "Done! $successCount succeeded, $skipCount skipped.\n" );
                if ( $exit ) {
                        $this->error( "Import failed with $failCount failed pages.\n", $exit );
                }