From b2411deec4b1732e73754f69eeaf19c0bd7c78f7 Mon Sep 17 00:00:00 2001
From: Brion Vibber <brion@users.mediawiki.org>
Date: Mon, 27 Sep 2004 06:23:33 +0000
Subject: [PATCH] Far from finished in-place UTF-8 wiki converter. I had this
 code lying around for a while and wanted to check it in before I forget about
 it.

---
 maintenance/convertUtf8.php | 199 ++++++++++++++++++++++++++++++++++++
 1 file changed, 199 insertions(+)
 create mode 100644 maintenance/convertUtf8.php

diff --git a/maintenance/convertUtf8.php b/maintenance/convertUtf8.php
new file mode 100644
index 0000000000..3281d6b6ff
--- /dev/null
+++ b/maintenance/convertUtf8.php
@@ -0,0 +1,199 @@
+<?php
+
+die("This file is not complete; it's checked in so I don't forget it.");
+
+/*
+UTF-8 conversion of DOOOOOOOM
+
+1. Lock the wiki
+2. Make a convertlist of all pages
+3. Enable CONVERTLOCK mode and switch to UTF-8
+4. As quick as possible, convert the cur, images, *links, user, etc tables. Clear cache tables.
+5. Unlock the wiki. Attempts to access pages on the convertlist will be trapped to read-only.
+6. Go through the list, fixing up old revisions. Remove pages from the convertlist.
+*/
+
+
+if(function_exists("iconv")) {
+	# There are likely to be Windows code page 1252 chars in there.
+	# Convert them to the proper UTF-8 chars if possible.
+	function toUtf8($string) {
+		return wfStrencode(iconv("CP1252", "UTF-8", $string));
+	}
+} else {
+	# Will work from plain iso 8859-1 and may corrupt these chars
+	function toUtf8($string) {
+		return wfStrencode(utf8_encode($string));
+	}
+}
+
+
+
+# user table
+$sql = "SELECT user_id,user_name,user_real_name,user_options FROM user";
+$res = wfQuery( $sql, DB_WRITE );
+print "Converting " . wfNumResults( $res ) . " user accounts:\n";
+$n = 0;
+while( $s = wfFetchObject( $res ) ) {
+	$uname = toUtf8( $s->user_name );
+	$ureal = toUtf8( $s->user_real_name );
+	$uoptions = toUtf8( $s->user_options );
+	if( $uname != wfStrencode( $s->user_name ) ||
+		$ureal != wfStrencode( $s->user_real_name ) ||
+		$uoptions != wfStrencode( $s->user_options ) ) {
+		$now = wfTimestampNow();
+		$sql = "UPDATE user
+			SET user_name='$uname',user_real_name='$ureal',
+				user_options='$uoptions',user_touched='$now'
+			WHERE user_id={$s->user_id}";
+		wfQuery( $sql, DB_WRITE );
+		$wgMemc->delete( "$wgDBname:user:id:{$s->user_id}" );
+		$u++;
+	}
+	if( ++$n % 100 == 0 ) print "$n\n";
+}
+wfFreeResult( $res );
+if( $n ) {
+	printf("%2.02%% required conversion.\n\n", $u / $n);
+} else {
+	print "None?\n\n";
+}
+
+# ipblocks
+$sql = "SELECT DISTINCT ipb_reason FROM ipblocks";
+$res = wfQuery( $sql, DB_WRITE );
+print "Converting " . wfNumResults( $res ) . " IP block comments:\n";
+$n = 0;
+while( $s = wfFetchObject( $res ) ) {
+	$ucomment = toUtf8($s->ipb_reason);
+	$ocomment = wfStrencode( $s->ipb_reason );
+	if( $u != $o ) {
+		$sql = "UPDATE ipblocks SET ipb_reason='$ucomment' WHERE ipb_reason='$ocomment'";
+		wfQuery( $sql, DB_WRITE );
+		$u++;
+	}
+	if( ++$n % 100 == 0 ) print "$n\n";
+}
+wfFreeResult( $res );
+if( $n ) {
+	printf("%2.02%% required conversion.\n\n", $u / $n);
+} else {
+	print "None?\n\n";
+}
+
+# image
+$sql = "SELECT img_name,img_description,img_user_text FROM image";
+	img_name --> also need to rename files
+	img_description
+	img_user_text
+
+oldimage
+	oi_name
+	oi_archive_name --> also need to rename files
+	oi_user_text
+
+recentchanges
+	rc_user_text
+	rc_title
+	rc_comment
+
+# searchindex
+print "Clearing searchindex... don't forget to rebuild it.\n";
+$sql = "DELETE FROM searchindex";
+wfQuery( $sql, DB_WRITE );
+
+# linkscc
+print "Clearing linkscc...\n";
+$sql = "DELETE FROM linkscc";
+wfQuery( $sql, DB_WRITE );
+
+# querycache: just rebuild these
+print "Clearing querycache...\n";
+$sql = "DELETE FROM querycache";
+wfQuery( $sql, DB_WRITE );
+
+# objectcache
+print "Clearing objectcache...\n";
+$sql = "DELETE FROM objectcache";
+wfQuery( $sql, DB_WRITE );
+
+
+function unicodeLinks( $table, $field ) {
+	$sql = "SELECT DISTINCT $field FROM $table WHERE $field RLIKE '[\x80-\xff]'";
+	$res = wfQuery( $sql, DB_WRITE );
+	print "Converting " . wfNumResults( $res ) . " from $table:\n";
+	$n = 0;
+	while( $s = wfFetchObject( $res ) ) {
+		$ulink = toUtf8( $s->$field );
+		$olink = wfStrencode( $s->$field );
+		$sql = "UPDATE $table SET $field='$ulink' WHERE $field='$olink'";
+		wfQuery( $sql, DB_WRITE );
+		if( ++$n % 100 == 0 ) print "$n\n";
+	}
+	wfFreeResult( $res );
+	print "Done.\n\n";
+}
+unicodeLinks( "brokenlinks", "bl_to" );
+unicodeLinks( "imagelinks", "il_to" );
+unicodeLinks( "categorylinks", "cl_to" );
+
+
+# The big guys...
+$sql = "SELECT cur_id,cur_namespace,cur_title,cur_text,cur_user_text FROM cur
+WHERE cur_title rlike '[\x80-\xff]' OR cur_comment rlike '[\x80-\xff]'
+OR cur_user_text rlike '[\x80-\xff]' OR cur_text rlike '[\x80-\xff]'";
+$res = wfQuery( $sql, DB_WRITE );
+print "Converting " . wfNumResults( $res ) . " cur pages:\n";
+$n = 0;
+while( $s = wfFetchObject( $res ) ) {
+	$utitle = toUtf8( $s->cur_title );
+	$uuser = toUtf8( $s->cur_user_text );
+	$ucomment = toUtf8( $s->cur_comment );
+	$utext = toUtf8( $s->cur_text );
+	$now = wfTimestampNow();
+	
+	$sql = "UPDATE cur
+		SET cur_title='$utitle',cur_user_text='$uuser',
+			cur_comment='$ucomment',cur_text='$utext'
+		WHERE cur_id={$s->cur_id}";
+	wfQuery( $sql, DB_WRITE );
+	#$wgMemc->delete( "$wgDBname:user:id:{$s->user_id}" );
+	
+	$otitle = wfStrencode( $s->cur_title );
+	if( $otitle != $utitle ) {
+		# Also update titles in watchlist and old
+		$sql = "UPDATE old SET old_title='$utitle'
+			WHERE old_namespace={$s->cur_namespace} AND old_title='$otitle'";
+		wfQuery( $sql, DB_WRITE );
+		
+		$ns = IntVal( $s->cur_namespace) & ~1;
+		$sql = "UPDATE watchlist SET wl_title='$utitle'
+			WHERE wl_namespace=$ns AND wl_title='$otitle'";
+		wfQuery( $sql, DB_WRITE );
+		$u++;
+	}
+	
+	if( ++$n % 100 == 0 ) print "$n\n";
+}
+wfFreeResult( $res );
+if( $n ) {
+	printf("Updated old/watchlist titles on %2.02%%.\n\n", $u / $n);
+} else {
+	print "Didn't update any old/watchlist titles.\n\n";
+}
+
+/*
+old
+	old_title
+	old_text -> may be gzipped
+	old_comment
+	old_user_text
+
+archive
+	ar_title
+	ar_text -> may be gzipped
+	ar_comment
+	ar_user_text
+*/
+
+?>
\ No newline at end of file
-- 
2.20.1