Version Notes
Added missing shell script in installation package. As usual, you can find the shell script in the folder "shell".
Download this release
Release Info
Developer | Matthias Kerstner |
Extension | BothInteract_CleanProductTexts |
Version | 1.1.0 |
Comparing to | |
See all releases |
Code changes from version 1.0.0 to 1.1.0
app/code/community/BothInteract/CleanProductTexts/Model/Observer.php
CHANGED
@@ -38,29 +38,40 @@ class BothInteract_CleanProductTexts_Model_Observer {
|
|
38 |
. mb_strtoupper($product->getTypeId())
|
39 |
. ' product ' . $product->getId());
|
40 |
|
|
|
|
|
41 |
foreach ($cleaningOptions as $cleaningOption) {
|
42 |
$this->logToFile('Checking option ' . $cleaningOption);
|
43 |
|
44 |
// remove control characters for description and short description
|
45 |
if ($cleaningOption == BothInteract_CleanProductTexts_Model_System_Config_Source_View::$VALUE_CLEANING_TYPE_CONTROL_CHARACTERS) {
|
46 |
-
// @see http://unicode-table.com/en/
|
47 |
-
$this->logToFile('Cleaning product ' . $product->getId());
|
48 |
|
|
|
49 |
$regexp = '/[^\PC\s]/u'; //i.e. /[\x00—\x1F\x80-\x9f]
|
50 |
-
|
51 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
}
|
53 |
}
|
54 |
|
55 |
-
if (
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
|
|
|
|
64 |
}
|
65 |
|
66 |
$this->logToFile('Successfully handled product ' . $product->getId());
|
38 |
. mb_strtoupper($product->getTypeId())
|
39 |
. ' product ' . $product->getId());
|
40 |
|
41 |
+
$isChanged = false;
|
42 |
+
|
43 |
foreach ($cleaningOptions as $cleaningOption) {
|
44 |
$this->logToFile('Checking option ' . $cleaningOption);
|
45 |
|
46 |
// remove control characters for description and short description
|
47 |
if ($cleaningOption == BothInteract_CleanProductTexts_Model_System_Config_Source_View::$VALUE_CLEANING_TYPE_CONTROL_CHARACTERS) {
|
|
|
|
|
48 |
|
49 |
+
// @see http://unicode-table.com/en/
|
50 |
$regexp = '/[^\PC\s]/u'; //i.e. /[\x00—\x1F\x80-\x9f]
|
51 |
+
|
52 |
+
if (preg_match($regexp, $product->getDescription()) ||
|
53 |
+
preg_match($regexp, $product->getShortDescription())) { // check to avoid unnecessary save()
|
54 |
+
$this->logToFile('Cleaning product ' . $product->getId());
|
55 |
+
$product->setDescription(preg_replace('/\x0b|\x0c/', '', preg_replace($regexp, '', $product->getDescription())));
|
56 |
+
$product->setShortDescription(preg_replace('/\x0b|\x0c/', '', preg_replace($regexp, '', $product->getShortDescription())));
|
57 |
+
$isChanged = true;
|
58 |
+
} else {
|
59 |
+
$this->logToFile('Ignoring product ' . $product->getId());
|
60 |
+
}
|
61 |
}
|
62 |
}
|
63 |
|
64 |
+
if ($isChanged) {
|
65 |
+
if (Mage::getStoreConfig(self::$_MODULE_NAMESPACE
|
66 |
+
. '/general/is_simulation')) {
|
67 |
+
$this->logToFile('************************************');
|
68 |
+
$this->logToFile('SIMULATION: Not saving product '
|
69 |
+
. $product->getId());
|
70 |
+
$this->logToFile('************************************');
|
71 |
+
} else {
|
72 |
+
$this->logToFile('Saving product ' . $product->getId());
|
73 |
+
$product->save();
|
74 |
+
}
|
75 |
}
|
76 |
|
77 |
$this->logToFile('Successfully handled product ' . $product->getId());
|
package.xml
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
<?xml version="1.0"?>
|
2 |
<package>
|
3 |
<name>BothInteract_CleanProductTexts</name>
|
4 |
-
<version>1.
|
5 |
<stability>stable</stability>
|
6 |
<license uri="http://www.opensource.org/licenses/mit-license.php">MIT License (MITL)</license>
|
7 |
<channel>community</channel>
|
@@ -22,11 +22,11 @@
|
|
22 |
<p>In addition this extension enabled you to logs messages to a custom log file that can be specified in the configuration options to your log directory under <i>General</i> / <i>Log File</i>.</p> 
|
23 |

|
24 |
<p>So you have the option to easily track events handled by this extension and check for possible problems.</p></description>
|
25 |
-
<notes>
|
26 |
<authors><author><name>Matthias Kerstner</name><user>mkbothinteract</user><email>matthias@both-interact.com</email></author></authors>
|
27 |
-
<date>2015-
|
28 |
-
<time>
|
29 |
-
<contents><target name="magecommunity"><dir name="BothInteract"><dir name="CleanProductTexts"><dir name="Model"><file name="Observer.php" hash="
|
30 |
<compatible/>
|
31 |
<dependencies><required><php><min>5.3.0</min><max>5.6.6</max></php></required></dependencies>
|
32 |
</package>
|
1 |
<?xml version="1.0"?>
|
2 |
<package>
|
3 |
<name>BothInteract_CleanProductTexts</name>
|
4 |
+
<version>1.1.0</version>
|
5 |
<stability>stable</stability>
|
6 |
<license uri="http://www.opensource.org/licenses/mit-license.php">MIT License (MITL)</license>
|
7 |
<channel>community</channel>
|
22 |
<p>In addition this extension enabled you to logs messages to a custom log file that can be specified in the configuration options to your log directory under <i>General</i> / <i>Log File</i>.</p> 
|
23 |

|
24 |
<p>So you have the option to easily track events handled by this extension and check for possible problems.</p></description>
|
25 |
+
<notes>Added missing shell script in installation package. As usual, you can find the shell script in the folder "shell".</notes>
|
26 |
<authors><author><name>Matthias Kerstner</name><user>mkbothinteract</user><email>matthias@both-interact.com</email></author></authors>
|
27 |
+
<date>2015-04-07</date>
|
28 |
+
<time>13:36:12</time>
|
29 |
+
<contents><target name="magecommunity"><dir name="BothInteract"><dir name="CleanProductTexts"><dir name="Model"><file name="Observer.php" hash="fd3b0a87136f323fa55746487714d4c9"/><dir name="System"><dir name="Config"><dir name="Source"><file name="View.php" hash="06c8cab69af1427591ccd80a5ff84b85"/></dir></dir></dir></dir><dir name="etc"><file name="adminhtml.xml" hash="09afe6511e6b0c1ee544c2ca87d59474"/><file name="config.xml" hash="8b3dceb0487024d8822a504af9c0241e"/><file name="system.xml" hash="9250f6bd8e3089b823c46138e1d6622f"/></dir></dir></dir></target><target name="mageetc"><dir name="modules"><file name="BothInteract_CleanProductTexts.xml" hash="c685308779c1427921191dd9860627a1"/></dir></target><target name="mage"><dir><dir name="shell"><file name="clean_product_texts.php" hash="be3c1e00b6705bb6d2bc9ce4f1546c1d"/></dir></dir></target></contents>
|
30 |
<compatible/>
|
31 |
<dependencies><required><php><min>5.3.0</min><max>5.6.6</max></php></required></dependencies>
|
32 |
</package>
|
shell/clean_product_texts.php
ADDED
@@ -0,0 +1,145 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?php
|
2 |
+
|
3 |
+
require_once 'abstract.php';
|
4 |
+
|
5 |
+
/**
|
6 |
+
* @author Matthias Kerstner <matthias@both-interact.com>
|
7 |
+
* @version 1.0.0
|
8 |
+
* @copyright (c) 2015, Both Interact GmbH
|
9 |
+
*/
|
10 |
+
class Mage_Shell_Clean_Product_Texts extends Mage_Shell_Abstract {
|
11 |
+
|
12 |
+
const PAGE_SIZE = 100;
|
13 |
+
|
14 |
+
/**
|
15 |
+
* Parse $string with comma separated values and return array.
|
16 |
+
*
|
17 |
+
* @param string $string
|
18 |
+
* @return array
|
19 |
+
*/
|
20 |
+
protected function _parseString($string) {
|
21 |
+
$values = array();
|
22 |
+
if (!empty($string)) {
|
23 |
+
$values = explode(',', $string);
|
24 |
+
$values = array_map('trim', $values);
|
25 |
+
}
|
26 |
+
return $values;
|
27 |
+
}
|
28 |
+
|
29 |
+
/**
|
30 |
+
* Run script based on CL options specified.
|
31 |
+
*/
|
32 |
+
public function run() {
|
33 |
+
if ($this->getArg('products')) {
|
34 |
+
|
35 |
+
// switch to admin event area
|
36 |
+
Mage::app()->addEventArea('admin');
|
37 |
+
|
38 |
+
// product model observer to be called on products
|
39 |
+
$productModelObserver = new BothInteract_CleanProductTexts_Model_Observer();
|
40 |
+
|
41 |
+
// allowed attribute types
|
42 |
+
$types = array('varchar', 'text', 'decimal', 'datetime', 'int');
|
43 |
+
|
44 |
+
// attribute sets array
|
45 |
+
$attributeSets = array();
|
46 |
+
|
47 |
+
// user defined attribute ids
|
48 |
+
$entityType = Mage::getModel('eav/entity_type')
|
49 |
+
->loadByCode('catalog_product');
|
50 |
+
|
51 |
+
$attributeCollection = $entityType
|
52 |
+
->getAttributeCollection()
|
53 |
+
->addFilter('is_user_defined', '1')
|
54 |
+
->getItems();
|
55 |
+
$attrIds = array();
|
56 |
+
foreach ($attributeCollection as $attribute) {
|
57 |
+
$attrIds[] = $attribute->getId();
|
58 |
+
}
|
59 |
+
$userDefined = implode(',', $attrIds);
|
60 |
+
|
61 |
+
// product collection based on attribute filters
|
62 |
+
$collection = Mage::getModel('catalog/product')->getCollection();
|
63 |
+
$entityTable = $collection
|
64 |
+
->getTable(Mage::getModel('eav/entity_type')
|
65 |
+
->loadByCode('catalog_product')
|
66 |
+
->getEntityTable());
|
67 |
+
|
68 |
+
// load specific product_types (currently supported: simple, configurable)
|
69 |
+
$cliTypes = $this->getArg('types');
|
70 |
+
if (!empty($cliTypes) && $cliTypes != 'all') {
|
71 |
+
$productTypes = $this->_parseString($this->getArg('types'));
|
72 |
+
|
73 |
+
foreach ($productTypes as $k => $productType) {
|
74 |
+
if (!in_array($productType, array('simple', 'configurable'))) {
|
75 |
+
unset($productTypes[$k]);
|
76 |
+
}
|
77 |
+
}
|
78 |
+
$collection->addAttributeToFilter('type_id', array('in' => $productTypes));
|
79 |
+
}
|
80 |
+
|
81 |
+
// load product IDs specified only
|
82 |
+
if ($this->getArg('products') != 'all') {
|
83 |
+
if ($ids = $this->_parseString($this->getArg('products'))) {
|
84 |
+
$collection->addAttributeToFilter('entity_id', array('in' => $ids));
|
85 |
+
}
|
86 |
+
}
|
87 |
+
|
88 |
+
$collection->setPageSize(self::PAGE_SIZE);
|
89 |
+
|
90 |
+
$pages = $collection->getLastPageNumber();
|
91 |
+
$currentPage = 1;
|
92 |
+
|
93 |
+
//light product collection iterating
|
94 |
+
while ($currentPage <= $pages) {
|
95 |
+
|
96 |
+
echo 'Processing products page ' . $currentPage . ' of ' . $pages
|
97 |
+
. '...' . PHP_EOL;
|
98 |
+
|
99 |
+
$collection->setCurPage($currentPage);
|
100 |
+
$collection->load();
|
101 |
+
|
102 |
+
foreach ($collection->getItems() as $item) {
|
103 |
+
|
104 |
+
// load product to manipulate
|
105 |
+
$product = Mage::getModel('catalog/product')
|
106 |
+
->load($item->getId());
|
107 |
+
|
108 |
+
// manually call our observer to update child products
|
109 |
+
$productModelObserver->processProduct($product);
|
110 |
+
}
|
111 |
+
|
112 |
+
$currentPage++;
|
113 |
+
$collection->clear();
|
114 |
+
}
|
115 |
+
|
116 |
+
echo 'Done!' . PHP_EOL;
|
117 |
+
} else {
|
118 |
+
echo $this->usageHelp();
|
119 |
+
}
|
120 |
+
}
|
121 |
+
|
122 |
+
/**
|
123 |
+
* Retrieve Usage Help Message.
|
124 |
+
*/
|
125 |
+
public function usageHelp() {
|
126 |
+
return <<<USAGE
|
127 |
+
|
128 |
+
Usage: php -f clean_product_texts -- [options]
|
129 |
+
|
130 |
+
--products all Clean all products
|
131 |
+
--products <product_ids> Clean products by IDs
|
132 |
+
--types all Clean all product types
|
133 |
+
--types <product_types> Clean only product types specified
|
134 |
+
help Show this help
|
135 |
+
|
136 |
+
<product_ids> Comma separated IDs of products
|
137 |
+
<product_types> Comma separated list of product types: simple and/or configurable
|
138 |
+
|
139 |
+
USAGE;
|
140 |
+
}
|
141 |
+
|
142 |
+
}
|
143 |
+
|
144 |
+
$shell = new Mage_Shell_Clean_Product_Texts();
|
145 |
+
$shell->run();
|