midcom_services_i18n::html_entity_decode() - Code Metrics - Inspection of "Fix erro checking" - flack/openpsa - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 07213d...3b1a54 )

by Andreas

created 2022-07-01 12:03 UTC

midcom_services_i18n::html_entity_decode() A

↳ Parent: midcom_services_i18n

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	1

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	2
CRAP Score	1

Importance

Changes

Metric	Value
cc	1
eloc	1
nc	1
nop	1
dl	0
loc	3
ccs	2
cts	2
cp	1
crap	1
rs	10
c	0
b	0
f	0

<?php
/**
 * @package midcom.services
 * @author The Midgard Project, http://www.midgard-project.org
 * @copyright The Midgard Project, http://www.midgard-project.org
 * @license http://www.gnu.org/licenses/lgpl.html GNU Lesser General Public License
 */

use Symfony\Component\Intl\Intl;
use Symfony\Component\Intl\Locales;
use Symfony\Component\HttpFoundation\RequestStack;
use Symfony\Component\HttpFoundation\Request;
use Symfony\Component\Translation\Translator;

/**
 * This is a basic MidCOM Service which provides an interfaces to the
 * various I18n facilities of MidCOM.
 *
 * The I18n service serves as a central access point for all aspects
 * around internationalization and localization. It provides auto-detection
 * of language data using HTTP Content-Negotiation along with a cookie-based
 * fallback.
 *
 * This class is able to run independently from midcom_application
 * due to the fact that it is used in the cache_hit code.
 *
 * Use this class to set the language preferences (charset and locale) and to gain
 * access to the l10n string databases. A few helpers which can be used to ease
 * translation work (like charset conversion) are in here as well.
 *
 * All language codes used here are ISO 639-1 two-letter codes.
 *
 * @package midcom.services
 */
class midcom_services_i18n
{
    /**
     * @var string
     */
    private $_fallback_language;

    /**
     * @var midcom_services_i18n_l10n[]
     */
    private $_obj_l10n = [];

    /**
     * @var string
     */
    private $_current_language;

    /**
     * @var string
     */
    private $_current_charset = 'utf-8';

    /**
     * Initialize the available i18n framework by determining the desired language
     * from these different sources: HTTP Content Negotiation, Client side language cookie.
     *
     * Its two parameters set the default language in case that none is supplied
     * via HTTP Content Negotiation or through Cookies.
     *
     * The default language set on startup is currently hardcoded to 'en',
     * you should override it after initialization, if you want something
     * else using the setter methods below.
     *
     * The fallback language is read from the MidCOM configuration directive
     * <i>i18n_fallback_language</i>.
     */
    public function __construct(RequestStack $request_stack, string $fallback_language)
    {
        $this->_fallback_language = $fallback_language;

        $found = false;
        if ($request = $request_stack->getCurrentRequest()) {
            $found = $this->_read_cookie($request) || $this->_read_http_negotiation($request);
        }
        if (!$found) {
            $this->set_language($fallback_language);
        }
    }

    /**
     * Try to pull the user's preferred language and
     * character set out of a cookie named "midcom_services_i18n".
     */
    private function _read_cookie(Request $request) : bool
    {
        if (!$request->cookies->has('midcom_services_i18n')) {
            return false;
        }

        $rawdata = base64_decode($request->cookies->get('midcom_services_i18n'));
        $array = unserialize($rawdata);

        if (   !array_key_exists('language', $array)
            || !array_key_exists('charset', $array)) {
            debug_add("Rejecting cookie, it seems invalid.");
            return false;
        }
        $this->set_charset($array['charset']);
        return $this->set_language($array['language']);
    }

    /**
     * Pull available language out of the HTTP Headers
     *
     * q-parameters for prioritization are supported.
     */
    private function _read_http_negotiation(Request $request) : bool
    {
        foreach ($request->getLanguages() as $lang) {
            // we can't use strings like en_US, so we only use the first two characters
            $lang = substr($lang, 0, 2);
            if ($this->set_language($lang)) {
                return true;
            }
        }
        return false;
    }

    public function get_translator(array $prefixes) : Translator
    {
        $locale = $this->get_current_language();
        $translator = new Translator($locale);
        foreach ($prefixes as $prefix) {
            $translator->addResource('xlf', $prefix . $locale . '.xlf', $locale);
        }
        return $translator;
    }

    /**
     * Set output character set.
     */
    public function set_charset(string $charset)
    {
        $this->_current_charset = strtolower($charset);
    }

    /**
     * Set output language.
     *
     * This will set the character encoding to the language's default
     * encoding and will also set the system locale to the one
     * specified in the language database.
     *
     * If you want another character encoding as the default one, you
     * have to override it manually using midcom_services_i18n::set_charset()
     * after calling this method.
     *
     * @param string $lang    Language ISO 639-1 code
     */
    public function set_language(string $lang) : bool
    {
        try {
            Locales::getName($lang);
        } catch (Exception $e) {
            debug_add("Language {$lang} not found:" . $e->getMessage(), MIDCOM_LOG_ERROR);
            return false;
        }

        $this->_current_language = $lang;

        setlocale(LC_ALL, $lang);
        if (Intl::isExtensionLoaded()) {
            Locale::setDefault($lang);
        }

        foreach ($this->_obj_l10n as $object) {
            $object->set_language($lang);
        }
        return true;
    }

    /**
     * Set the fallback language.
     */
    public function set_fallback_language(string $lang)
    {
        $this->_fallback_language = $lang;
        foreach ($this->_obj_l10n as $object) {
            $object->set_fallback_language($lang);
        }
    }

    /**
     * Returns the current language code
     */
    public function get_current_language() : string
    {
        return $this->_current_language;
    }

    /**
     * Returns the current fallback language code
     */
    public function get_fallback_language() : string
    {
        return $this->_fallback_language;
    }

    /**
     * Returns the current character set
     */
    public function get_current_charset() : string
    {
        return $this->_current_charset;
    }

    /**
     * Returns a l10n class instance which can be used to
     * access the localization data of the current component.
     *
     * If loading failed, midcom_error is thrown, otherwise the l10n
     * db cache is populated accordingly.
     *
     * @see midcom_services_i18n_l10n
     */
    public function get_l10n(string $component = 'midcom') : midcom_services_i18n_l10n
    {
        if (!array_key_exists($component, $this->_obj_l10n)) {
            $this->_obj_l10n[$component] = new midcom_services_i18n_l10n($component, $this->_current_language, $this->_fallback_language);
        }

        return $this->_obj_l10n[$component];
    }

    /**
     * Returns a translated string using the l10n database specified in the function
     * arguments.
     *
     * @see midcom_services_i18n_l10n::get()
     */
    public function get_string(string $stringid, string $component) : string
    {
        try {
            return $this->get_l10n($component)->get($stringid);
        } catch (midcom_error $e) {
            $e->log(MIDCOM_LOG_WARN);
            return $stringid;
        }
    }

    /**
     * This is a shortcut for echo $this->get_string(...);.
     *
     * To keep the naming stable with the actual l10n class, this is not called
     * echo_string (Zend won't allow $l10n->echo().)
     *
     * @see midcom_services_i18n_l10n::get()
     * @see get_string()
     */
    public function show_string(string $stringid, string $component)
    {
        echo $this->get_string($stringid, $component);
    }

    /**
     * This is a calling wrapper to the iconv library.
     *
     * See the PHP iconv() function for the exact parameter definitions.
     *
     * @return mixed The converted string or false on any error.
     */
    private function iconv(string $source_charset, string $destination_charset, string $string)
    {
        $result = @iconv($source_charset, $destination_charset, $string);
        if ($result === false && !empty($string)) {
            debug_add("Iconv returned failed to convert a string, returning an empty string.", MIDCOM_LOG_WARN);
            debug_print_r("Tried to convert this string from {$source_charset} to {$destination_charset}:", $string);
            midcom::get()->debug->log_php_error(MIDCOM_LOG_WARN);
            return false;
        }
        return $result;
    }

    /**
     * Convert a string assumed to be in the currently active charset to UTF8.
     *
     * @return string The string converted to UTF-8
     */
    public function convert_to_utf8(string $string)
    {
        if ($this->_current_charset == 'utf-8') {
            return $string;
        }
        return $this->iconv($this->_current_charset, 'utf-8', $string);
    }

    /**
     * Convert a string assumed to be in UTF-8 to the currently active charset.
     *
     * @return string The string converted to the current charset
     */
    public function convert_from_utf8(string $string)
    {
        if ($this->_current_charset == 'utf-8') {
            return $string;
        }
        return $this->iconv('utf-8', $this->_current_charset, $string);
    }

    /**
     * Converts the given string to the current site charset.
     *
     * @return string The converted string.
     */
    public function convert_to_current_charset(string $string)
    {
        $charset = mb_detect_encoding($string, "UTF-8, UTF-7, ASCII, ISO-8859-15");
        debug_add("mb_detect_encoding got {$charset}");
        return $this->iconv($charset, $this->_current_charset, $string);
    }

    /**
     * Wrapped html_entity_decode call
     */
    public function html_entity_decode(string $text) : string
    {
        return html_entity_decode($text, ENT_COMPAT, $this->_current_charset);
    }
}


1		<?php
2		/**
3		* @package midcom.services
4		* @author The Midgard Project, http://www.midgard-project.org
5		* @copyright The Midgard Project, http://www.midgard-project.org
6		* @license http://www.gnu.org/licenses/lgpl.html GNU Lesser General Public License
7		*/
8
9		use Symfony\Component\Intl\Intl;
10		use Symfony\Component\Intl\Locales;
11		use Symfony\Component\HttpFoundation\RequestStack;
12		use Symfony\Component\HttpFoundation\Request;
13		use Symfony\Component\Translation\Translator;
14
15		/**
16		* This is a basic MidCOM Service which provides an interfaces to the
17		* various I18n facilities of MidCOM.
18		*
19		* The I18n service serves as a central access point for all aspects
20		* around internationalization and localization. It provides auto-detection
21		* of language data using HTTP Content-Negotiation along with a cookie-based
22		* fallback.
23		*
24		* This class is able to run independently from midcom_application
25		* due to the fact that it is used in the cache_hit code.
26		*
27		* Use this class to set the language preferences (charset and locale) and to gain
28		* access to the l10n string databases. A few helpers which can be used to ease
29		* translation work (like charset conversion) are in here as well.
30		*
31		* All language codes used here are ISO 639-1 two-letter codes.
32		*
33		* @package midcom.services
34		*/
35		class midcom_services_i18n
36		{
37		/**
38		* @var string
39		*/
40		private $_fallback_language;
41
42		/**
43		* @var midcom_services_i18n_l10n[]
44		*/
45		private $_obj_l10n = [];
46
47		/**
48		* @var string
49		*/
50		private $_current_language;
51
52		/**
53		* @var string
54		*/
55		private $_current_charset = 'utf-8';
56
57		/**
58		* Initialize the available i18n framework by determining the desired language
59		* from these different sources: HTTP Content Negotiation, Client side language cookie.
60		*
61		* Its two parameters set the default language in case that none is supplied
62		* via HTTP Content Negotiation or through Cookies.
63		*
64		* The default language set on startup is currently hardcoded to 'en',
65		* you should override it after initialization, if you want something
66		* else using the setter methods below.
67		*
68		* The fallback language is read from the MidCOM configuration directive
69		* <i>i18n_fallback_language</i>.
70		*/
71	3	public function __construct(RequestStack $request_stack, string $fallback_language)
72		{
73	3	$this->_fallback_language = $fallback_language;
74
75	3	$found = false;
76	3	if ($request = $request_stack->getCurrentRequest()) {
77	2	$found = $this->_read_cookie($request) \|\| $this->_read_http_negotiation($request);
78		}
79	3	if (!$found) {
80	1	$this->set_language($fallback_language);
81		}
82		}
83
84		/**
85		* Try to pull the user's preferred language and
86		* character set out of a cookie named "midcom_services_i18n".
87		*/
88	2	private function _read_cookie(Request $request) : bool
89		{
90	2	if (!$request->cookies->has('midcom_services_i18n')) {
91	2	return false;
92		}
93
94		$rawdata = base64_decode($request->cookies->get('midcom_services_i18n'));
95		$array = unserialize($rawdata);
96
97		if ( !array_key_exists('language', $array)
98		\|\| !array_key_exists('charset', $array)) {
99		debug_add("Rejecting cookie, it seems invalid.");
100		return false;
101		}
102		$this->set_charset($array['charset']);
103		return $this->set_language($array['language']);
104		}
105
106		/**
107		* Pull available language out of the HTTP Headers
108		*
109		* q-parameters for prioritization are supported.
110		*/
111	2	private function _read_http_negotiation(Request $request) : bool
112		{
113	2	foreach ($request->getLanguages() as $lang) {
114		// we can't use strings like en_US, so we only use the first two characters
115	2	$lang = substr($lang, 0, 2);
116	2	if ($this->set_language($lang)) {
117	2	return true;
118		}
119		}
120		return false;
121		}
122
123	1	public function get_translator(array $prefixes) : Translator
124		{
125	1	$locale = $this->get_current_language();
126	1	$translator = new Translator($locale);
127	1	foreach ($prefixes as $prefix) {
128	1	$translator->addResource('xlf', $prefix . $locale . '.xlf', $locale);
129		}
130	1	return $translator;
131		}
132
133		/**
134		* Set output character set.
135		*/
136		public function set_charset(string $charset)
137		{
138		$this->_current_charset = strtolower($charset);
139		}
140
141		/**
142		* Set output language.
143		*
144		* This will set the character encoding to the language's default
145		* encoding and will also set the system locale to the one
146		* specified in the language database.
147		*
148		* If you want another character encoding as the default one, you
149		* have to override it manually using midcom_services_i18n::set_charset()
150		* after calling this method.
151		*
152		* @param string $lang Language ISO 639-1 code
153		*/
154	3	public function set_language(string $lang) : bool
155		{
156		try {
157	3	Locales::getName($lang);
158		} catch (Exception $e) {
159		debug_add("Language {$lang} not found:" . $e->getMessage(), MIDCOM_LOG_ERROR);
160		return false;
161		}
162
163	3	$this->_current_language = $lang;
164
165	3	setlocale(LC_ALL, $lang);
166	3	if (Intl::isExtensionLoaded()) {
167	3	Locale::setDefault($lang);
168		}
169
170	3	foreach ($this->_obj_l10n as $object) {
171		$object->set_language($lang);
172		}
173	3	return true;
174		}
175
176		/**
177		* Set the fallback language.
178		*/
179		public function set_fallback_language(string $lang)
180		{
181		$this->_fallback_language = $lang;
182		foreach ($this->_obj_l10n as $object) {
183		$object->set_fallback_language($lang);
184		}
185		}
186
187		/**
188		* Returns the current language code
189		*/
190	398	public function get_current_language() : string
191		{
192	398	return $this->_current_language;
193		}
194
195		/**
196		* Returns the current fallback language code
197		*/
198	51	public function get_fallback_language() : string
199		{
200	51	return $this->_fallback_language;
201		}
202
203		/**
204		* Returns the current character set
205		*/
206	108	public function get_current_charset() : string
207		{
208	108	return $this->_current_charset;
209		}
210
211		/**
212		* Returns a l10n class instance which can be used to
213		* access the localization data of the current component.
214		*
215		* If loading failed, midcom_error is thrown, otherwise the l10n
216		* db cache is populated accordingly.
217		*
218		* @see midcom_services_i18n_l10n
219		*/
220	406	public function get_l10n(string $component = 'midcom') : midcom_services_i18n_l10n
221		{
222	406	if (!array_key_exists($component, $this->_obj_l10n)) {
223	5	$this->_obj_l10n[$component] = new midcom_services_i18n_l10n($component, $this->_current_language, $this->_fallback_language);
224		}
225
226	406	return $this->_obj_l10n[$component];
227		}
228
229		/**
230		* Returns a translated string using the l10n database specified in the function
231		* arguments.
232		*
233		* @see midcom_services_i18n_l10n::get()
234		*/
235	370	public function get_string(string $stringid, string $component) : string
236		{
237		try {
238	370	return $this->get_l10n($component)->get($stringid);
239		} catch (midcom_error $e) {
240		$e->log(MIDCOM_LOG_WARN);
241		return $stringid;
242		}
243		}
244
245		/**
246		* This is a shortcut for echo $this->get_string(...);.
247		*
248		* To keep the naming stable with the actual l10n class, this is not called
249		* echo_string (Zend won't allow $l10n->echo().)
250		*
251		* @see midcom_services_i18n_l10n::get()
252		* @see get_string()
253		*/
254		public function show_string(string $stringid, string $component)
255		{
256		echo $this->get_string($stringid, $component);
257		}
258
259		/**
260		* This is a calling wrapper to the iconv library.
261		*
262		* See the PHP iconv() function for the exact parameter definitions.
263		*
264		* @return mixed The converted string or false on any error.
265		*/
266		private function iconv(string $source_charset, string $destination_charset, string $string)
267		{
268		$result = @iconv($source_charset, $destination_charset, $string);
269		if ($result === false && !empty($string)) {
270		debug_add("Iconv returned failed to convert a string, returning an empty string.", MIDCOM_LOG_WARN);
271		debug_print_r("Tried to convert this string from {$source_charset} to {$destination_charset}:", $string);
272		midcom::get()->debug->log_php_error(MIDCOM_LOG_WARN);
273		return false;
274		}
275		return $result;
276		}
277
278		/**
279		* Convert a string assumed to be in the currently active charset to UTF8.
280		*
281		* @return string The string converted to UTF-8
282		*/
283		public function convert_to_utf8(string $string)
284		{
285		if ($this->_current_charset == 'utf-8') {
286		return $string;
287		}
288		return $this->iconv($this->_current_charset, 'utf-8', $string);
289		}
290
291		/**
292		* Convert a string assumed to be in UTF-8 to the currently active charset.
293		*
294		* @return string The string converted to the current charset
295		*/
296	381	public function convert_from_utf8(string $string)
297		{
298	381	if ($this->_current_charset == 'utf-8') {
299	381	return $string;
300		}
301		return $this->iconv('utf-8', $this->_current_charset, $string);
302		}
303
304		/**
305		* Converts the given string to the current site charset.
306		*
307		* @return string The converted string.
308		*/
309		public function convert_to_current_charset(string $string)
310		{
311		$charset = mb_detect_encoding($string, "UTF-8, UTF-7, ASCII, ISO-8859-15");
312		debug_add("mb_detect_encoding got {$charset}");
313		return $this->iconv($charset, $this->_current_charset, $string);
314		}
315
316		/**
317		* Wrapped html_entity_decode call
318		*/
319	14	public function html_entity_decode(string $text) : string
320		{
321	14	return html_entity_decode($text, ENT_COMPAT, $this->_current_charset);
322		}
323		}
324

flack / openpsa

Push — master ( 07213d...3b1a54 )

midcom_services_i18n::html_entity_decode() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like