From 1cf86c3050d14e6584ca669bb2f603a529a55889 Mon Sep 17 00:00:00 2001
From: Joao Inacio <joao.inacio@ez.no>
Date: Tue, 26 Jul 2016 15:52:11 +0100
Subject: [PATCH 1/3] EZP-26058: update script for non-breaking spaces in
 ezxmltext

---
 update/common/scripts/updatenbxmlcontents.php | 117 ++++++++++++++++++
 1 file changed, 117 insertions(+)
 create mode 100644 update/common/scripts/updatenbxmlcontents.php

diff --git a/update/common/scripts/updatenbxmlcontents.php b/update/common/scripts/updatenbxmlcontents.php
new file mode 100644
index 00000000000..e7b8db92ffc
--- /dev/null
+++ b/update/common/scripts/updatenbxmlcontents.php
@@ -0,0 +1,117 @@
+<?php
+/**
+ * File containing the updatenbxmlcontents script.
+ *
+ * @copyright Copyright (C) eZ Systems AS. All rights reserved.
+ * @license For full copyright and license information view LICENSE file distributed with this source code.
+ * @version //autogentag//
+ */
+
+require_once 'autoload.php';
+$cli = eZCLI::instance();
+
+$script = eZScript::instance(
+    array(
+        'description' => "Updates non-break space encoding in ezxml contents. See issue EZP-18220\n",
+        'use-session' => true,
+        'use-modules' => false,
+        'use-extensions' => true
+    )
+);
+$script->startup();
+
+$options = $script->getOptions(
+    "[dry-run][n][v][iteration-sleep:][iteration-limit:]",
+    "",
+    array(
+        'dry-run' => 'Dry run',
+        'iteration-sleep' => 'Sleep duration between batches, in seconds (default: 1)',
+        'iteration-limit' => 'Batch size (default: 100)',
+        'n' => 'Do not wait 30 seconds before starting',
+    )
+);
+$optDryRun = (bool)$options['dry-run'];
+$optIterationSleep = (int)$options['iteration-sleep'] ?: 1;
+$optIterationLimit = (int)$options['iteration-limit'] ?: 100;
+$verboseLevel = $script->verboseOutputLevel();
+
+$limit = array(
+    "offset" => 0,
+    "limit" => $optIterationLimit,
+);
+
+$script->initialize();
+$db = eZDB::instance();
+
+if ( $optDryRun ) {
+    $cli->warning( "dry-run mode" );
+}
+
+/**
+ * Updates non-breaking spaces from existing "&amp;nbsp;" to proper "\xC2\xA0"
+ *
+ * @param array $attribute
+ */
+function updateEzxmlNonbreakSpaces( $attribute, $optDryRun, $verbose )
+{
+    $id = $attribute['id'];
+    $contentId = $attribute['contentobject_id'];
+    $version = $attribute['version'];
+    $xmlData = $attribute['data_text'];
+
+    $pattern = '/(<paragraph[^>]*\>)(.*)&amp;nbsp;(.*)(<\/paragraph>)/';
+    $replace = "\\1\\2\xC2\xA0\\3\\4";
+    do {
+        $xmlData = preg_replace( $pattern, $replace, $xmlData, -1, $countReplaced );
+    } while ($countReplaced > 0);
+
+    if ( $verbose ) {
+        eZCLI::instance()->output( "Updating data for content #$contentId (ver. $version) ..." );
+    }
+    if ( !$optDryRun ) {
+        eZDB::instance()->query( "UPDATE ezcontentobject_attribute SET data_text='$xmlData' WHERE id='$id'" );
+    }
+}
+
+if ( !$options['n'] )
+{
+    $cli->output();
+    $cli->warning( "You have 30 seconds to break the script before actual processing starts (press Ctrl-C)." );
+    $cli->warning( "Execute the script with '-n' switch to skip this delay." );
+    sleep( 30 );
+}
+
+$attributeCount = $db->arrayQuery(
+    "SELECT count(id) as count " .
+    "FROM ezcontentobject_attribute attr " .
+    "WHERE data_type_string='ezxmltext' AND data_text LIKE '%&amp;nbsp;%' "
+);
+$attributeCount = $attributeCount[0]['count'];
+
+$cli->output( "Number of xml attributes to update: " . $attributeCount );
+
+// main loop
+do {
+    $rows = $db->arrayQuery(
+        "SELECT id, contentobject_id, version, data_text " .
+        "FROM ezcontentobject_attribute attr " .
+        "WHERE data_type_string='ezxmltext' AND data_text LIKE '%&amp;nbsp;%' ",
+        $limit
+    );
+
+    $db->begin();
+    foreach ( $rows as $attribute )
+    {
+        updateEzxmlNonbreakSpaces( $attribute, $optDryRun, $verboseLevel );
+    }
+    $db->commit();
+
+    $cli->output(".");
+
+    $limit["offset"] += $optIterationLimit;
+    sleep( $optIterationSleep );
+} while ( count($rows) == $optIterationLimit );
+
+$cli->output( "Update has been completed." );
+
+$script->shutdown();

From d41a123176b6ffefd2bdfa632bed3dc4eeb7049e Mon Sep 17 00:00:00 2001
From: Joao Inacio <joao.inacio@ez.no>
Date: Thu, 28 Jul 2016 15:15:45 +0100
Subject: [PATCH 2/3] fixup! EZP-26058: update script for non-breaking spaces
 in ezxmltext

---
 update/common/scripts/updatenbxmlcontents.php | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/update/common/scripts/updatenbxmlcontents.php b/update/common/scripts/updatenbxmlcontents.php
index e7b8db92ffc..3bb76b380d8 100644
--- a/update/common/scripts/updatenbxmlcontents.php
+++ b/update/common/scripts/updatenbxmlcontents.php
@@ -31,8 +31,8 @@
     )
 );
 $optDryRun = (bool)$options['dry-run'];
-$optIterationSleep = (int)$options['iteration-sleep'] ?: 1;
-$optIterationLimit = (int)$options['iteration-limit'] ?: 100;
+$optIterationSleep = $options['iteration-sleep'] ? (int)$options['iteration-sleep'] : 1;
+$optIterationLimit = $options['iteration-limit'] ? (int)$options['iteration-limit'] : 100;
 $verboseLevel = $script->verboseOutputLevel();
 
 $limit = array(
@@ -43,7 +43,8 @@
 $script->initialize();
 $db = eZDB::instance();
 
-if ( $optDryRun ) {
+if ( $optDryRun )
+
     $cli->warning( "dry-run mode" );
 }
 

From 25bac64394e96f22def07fd6d634fb922e6edc3a Mon Sep 17 00:00:00 2001
From: Joao Inacio <joao.inacio@ez.no>
Date: Thu, 28 Jul 2016 15:33:36 +0100
Subject: [PATCH 3/3] fixup! EZP-26058: update script for non-breaking spaces
 in ezxmltext

---
 update/common/scripts/updatenbxmlcontents.php | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/update/common/scripts/updatenbxmlcontents.php b/update/common/scripts/updatenbxmlcontents.php
index 3bb76b380d8..603a3eefee1 100644
--- a/update/common/scripts/updatenbxmlcontents.php
+++ b/update/common/scripts/updatenbxmlcontents.php
@@ -44,7 +44,7 @@
 $db = eZDB::instance();
 
 if ( $optDryRun )
-
+{
     $cli->warning( "dry-run mode" );
 }
 
@@ -60,8 +60,10 @@ function updateEzxmlNonbreakSpaces( $attribute, $optDryRun, $verbose )
     $version = $attribute['version'];
     $xmlData = $attribute['data_text'];
 
-    $pattern = '/(<paragraph[^>]*\>)(.*)&amp;nbsp;(.*)(<\/paragraph>)/';
-    $replace = "\\1\\2\xC2\xA0\\3\\4";
+    $matchTags = implode('|', array( 'paragraph', 'header') );
+    $pattern = '/(<(?<tag>' . $matchTags . ')[^>]*\>)(.*)&amp;nbsp;(.*)(<\/(?P=tag)>)/';
+    $replace = "\\1\\3\xC2\xA0\\4\\5";
+
     do {
         $xmlData = preg_replace( $pattern, $replace, $xmlData, -1, $countReplaced );
     } while ($countReplaced > 0);