Strings::toASCII() - Code Metrics - Inspection of "Merge pull request #234 from robmachado/master" - nfephp-org/sped-common - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 2938c8...c15d53 )

by Roberto

created 2020-11-14 20:07 UTC

Strings::toASCII() A

↳ Parent: Strings

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	2

Importance

Changes

Metric	Value
dl	0
loc	6
ccs	0
cts	4
cp	0
rs	10
c	0
b	0
f	0
cc	1
nc	1
nop	1
crap	2

<?php

namespace NFePHP\Common;

/**
 * Classe auxiliar para o tratamento de strings
 * @category   NFePHP
 * @package    NFePHP\Common\Strings
 * @copyright  Copyright (c) 2008-2017
 * @license    http://www.gnu.org/licenses/lesser.html LGPL v3
 * @author     Roberto L. Machado <linux dot rlm at gmail dot com>
 * @link       http://github.com/nfephp-org/nfephp for the canonical source repository
 */

use ForceUTF8\Encoding;

class Strings
{
    
    /**
     * Includes missing or unsupported properties in stdClass inputs
     * and Replace all unsuported chars
     *
     * @param \stdClass $std
     * @param array $possible
     * @return \stdClass
     */
    public static function equilizeParameters(
        \stdClass $std,
        $possible,
        $replaceAccentedChars = false
    ) {
        $arr = get_object_vars($std);
        foreach ($possible as $key) {
            if (!array_key_exists($key, $arr)) {
                $std->$key = null;
            } else {
                if (is_string($std->$key)) {
                    $std->$key = trim(self::replaceUnacceptableCharacters($std->$key));
                    if ($replaceAccentedChars) {
                        $std->$key = self::toASCII($std->$key);
                    }
                }
            }
        }
        return $std;
    }

    /**
     * Replace all specials characters from string and retuns only 128 basics
     * NOTE: only for UTF-8
     * @param string $string
     * @return  string
     */
    public static function replaceSpecialsChars($string)
    {
        $string = self::squashCharacters($string);
        $string = str_replace('&', 'e', $string);
        $string = preg_replace("/[^a-zA-Z0-9 @#,-_.;:$%\/]/", "", $string);
        return preg_replace("/[<>]/", "", $string);
    }
    
    /**
     * Clear inputs for build in XML
     * Only UTF-8 characters is acceptable
     * & isolated, less than, greater than, quotation marks and apostrophes
     * should be replaced by their html equivalent
     * Carriage Return, Tab and Line Feed is not acceptable in strings
     * Multiple spaces is not acceptable in strings
     * And no other control character is acceptable either
     * @param string|null $input
     * @return string|null
     */
    public static function replaceUnacceptableCharacters($input)
    {
        if (empty($input)) {
            return $input;
        }
        $input = str_replace('&', ' & ', $input);
        $input = preg_replace('/(?:\s\s+)/', ' ', $input);
        //& isolated, less than, greater than, quotation marks and apostrophes
        //should be replaced by their html equivalent
        $input = str_replace(
            ['& ','<','>','"',"'"],
            ['&amp; ','&lt;','&gt;','&quot;','&#39;'],
            $input
        );
        $input = self::normalize($input);
        return trim($input);
    }
    
    /**
     * Converts all UTF-8 remains in ASCII
     * @param string $input
     * @return string
     */
    public static function toASCII($input)
    {
        $input = self::normalize($input);
        $input = self::squashCharacters($input);
        return mb_convert_encoding($input, 'ascii');
    }
    
    /**
     * Replaces all accented characters of their ASCII equivalents
     * @param string $input
     * @return string
     */
    public static function squashCharacters($input)
    {
        $input = trim($input);
        $aFind = ['á','à','ã','â','é','ê','í','ó','ô','õ','ö','ú','ü',
            'ç','Á','À','Ã','Â','É','Ê','Í','Ó','Ô','Õ','Ö','Ú','Ü','Ç'];
        $aSubs = ['a','a','a','a','e','e','i','o','o','o','o','u','u',
            'c','A','A','A','A','E','E','I','O','O','O','O','U','U','C'];
        return str_replace($aFind, $aSubs, $input);
    }
    
    /**
     * Replace all non-UTF-8 chars to UTF-8
     * Remove all control chars
     * Remove all multiple spaces
     * @param string $input
     * @return string
     */
    public static function normalize($input)
    {
        //Carriage Return, Tab and Line Feed is not acceptable in strings
        $input = str_replace(["\r","\t","\n"], "", $input);
        //Multiple spaces is not acceptable in strings
        $input = preg_replace('/(?:\s\s+)/', ' ', $input);
        //Only UTF-8 characters is acceptable
        $input = Encoding::fixUTF8($input);
        $input = preg_replace(
            '/[\x00-\x08\x10\x0B\x0C\x0E-\x19\x7F]'.
            '|[\x00-\x7F][\x80-\xBF]+'.
            '|([\xC0\xC1]|[\xF0-\xFF])[\x80-\xBF]*'.
            '|[\xC2-\xDF]((?![\x80-\xBF])|[\x80-\xBF]{2,})'.
            '|[\xE0-\xEF](([\x80-\xBF](?![\x80-\xBF]))|(?![\x80-\xBF]{2})|[\x80-\xBF]{3,})/S',
            '',
            $input
        );
        $input = preg_replace('/\xE0[\x80-\x9F][\x80-\xBF]'.
            '|\xED[\xA0-\xBF][\x80-\xBF]/S', '', $input);
        //And no other control character is acceptable either
        return preg_replace('/[[:cntrl:]]/', '', $input);
    }

    /**
     * Remove all non numeric characters from string
     * @param string $string
     * @return string
     */
    public static function onlyNumbers($string)
    {
        return preg_replace("/[^0-9]/", "", $string);
    }
    
    /**
     * Remove unwanted attributes, prefixes, sulfixes and other control
     * characters like \r \n \s \t
     * @param string $string
     * @param boolean $removeEncodingTag remove encoding tag from a xml
     * @return string
     */
    public static function clearXmlString($string, $removeEncodingTag = false)
    {
        $aFind = array(
            'xmlns:default="http://www.w3.org/2000/09/xmldsig#"',
            ' standalone="no"',
            'default:',
            ':default',
            "\n",
            "\r",
            "\t"
        );
        $retXml = str_replace($aFind, "", $string);
        $retXml = preg_replace('/(\>)\s*(\<)/m', '$1$2', $retXml);
        if ($removeEncodingTag) {
            $retXml = self::deleteAllBetween($retXml, '<?xml', '?>');
        }
        return $retXml;
    }
    
    /**
     * Remove all characters between markers
     * @param string $string
     * @param string $beginning
     * @param string $end
     * @return string
     */
    public static function deleteAllBetween($string, $beginning, $end)
    {
        $beginningPos = strpos($string, $beginning);
        $endPos = strpos($string, $end);
        if ($beginningPos === false || $endPos === false) {
            return $string;
        }
        $textToDelete = substr($string, $beginningPos, ($endPos + strlen($end)) - $beginningPos);
        return str_replace($textToDelete, '', $string);
    }
    
    /**
     * Clears the xml after adding the protocol, removing repeated namespaces
     * @param string $string
     * @return string
     */
    public static function clearProtocoledXML($string)
    {
        $procXML = self::clearXmlString($string);
        $aApp = array('nfe','cte','mdfe');
        foreach ($aApp as $app) {
            $procXML = str_replace(
                'xmlns="http://www.portalfiscal.inf.br/'.$app.'" xmlns="http://www.w3.org/2000/09/xmldsig#"',
                'xmlns="http://www.portalfiscal.inf.br/'.$app.'"',
                $procXML
            );
        }
        return $procXML;
    }
    
    /**
     * Remove some alien chars from txt
     * @param string $txt
     * @return string
     */
    public static function removeSomeAlienCharsfromTxt($txt)
    {
        //remove CRs and TABs
        $txt = str_replace(["\r","\t"], "", $txt);
        //remove multiple spaces
        $txt = preg_replace('/(?:\s\s+)/', ' ', $txt);
        //remove spaces at begin and end of fields
        $txt = str_replace(["| "," |"], "|", $txt);
        return $txt;
    }
    
    /**
     * Creates a string ramdomically with the specified length
     * @param int $length
     * @return string
     */
    public static function randomString($length)
    {
        $keyspace = '0123456789abcdefghijklmnopqr'
            . 'stuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ';
        $str = '';
        $max = mb_strlen($keyspace, '8bit') - 1;
        for ($i = 0; $i < $length; ++$i) {
            $str .= $keyspace[rand(0, $max)];
        }
        return $str;
    }
}


1		<?php
2
3		namespace NFePHP\Common;
4
5		/**
6		* Classe auxiliar para o tratamento de strings
7		* @category NFePHP
8		* @package NFePHP\Common\Strings
9		* @copyright Copyright (c) 2008-2017
10		* @license http://www.gnu.org/licenses/lesser.html LGPL v3
11		* @author Roberto L. Machado <linux dot rlm at gmail dot com>
12		* @link http://github.com/nfephp-org/nfephp for the canonical source repository
13		*/
14
15		use ForceUTF8\Encoding;
16
17		class Strings
18		{
19
20		/**
21		* Includes missing or unsupported properties in stdClass inputs
22		* and Replace all unsuported chars
23		*
24		* @param \stdClass $std
25		* @param array $possible
26		* @return \stdClass
27		*/
28		public static function equilizeParameters(
29		\stdClass $std,
30		$possible,
31		$replaceAccentedChars = false
32		) {
33		$arr = get_object_vars($std);
34		foreach ($possible as $key) {
35		if (!array_key_exists($key, $arr)) {
36		$std->$key = null;
37		} else {
38		if (is_string($std->$key)) {
39		$std->$key = trim(self::replaceUnacceptableCharacters($std->$key));
40		if ($replaceAccentedChars) {
41		$std->$key = self::toASCII($std->$key);
42		}
43		}
44		}
45		}
46		return $std;
47		}
48
49		/**
50		* Replace all specials characters from string and retuns only 128 basics
51		* NOTE: only for UTF-8
52		* @param string $string
53		* @return string
54		*/
55	1	public static function replaceSpecialsChars($string)
56		{
57	1	$string = self::squashCharacters($string);
58	1	$string = str_replace('&', 'e', $string);
59	1	$string = preg_replace("/[^a-zA-Z0-9 @#,-_.;:$%\/]/", "", $string);
60	1	return preg_replace("/[<>]/", "", $string);
61		}
62
63		/**
64		* Clear inputs for build in XML
65		* Only UTF-8 characters is acceptable
66		* & isolated, less than, greater than, quotation marks and apostrophes
67		* should be replaced by their html equivalent
68		* Carriage Return, Tab and Line Feed is not acceptable in strings
69		* Multiple spaces is not acceptable in strings
70		* And no other control character is acceptable either
71		* @param string\|null $input
72		* @return string\|null
73		*/
74	1	public static function replaceUnacceptableCharacters($input)
75		{
76	1	if (empty($input)) {
77		return $input;
78		}
79	1	$input = str_replace('&', ' & ', $input);
80	1	$input = preg_replace('/(?:\s\s+)/', ' ', $input);
81		//& isolated, less than, greater than, quotation marks and apostrophes
82		//should be replaced by their html equivalent
83	1	$input = str_replace(
84	1	['& ','<','>','"',"'"],
85	1	['& ','<','>','"','''],
86		$input
87		);
88	1	$input = self::normalize($input);
89	1	return trim($input);
90		}
91
92		/**
93		* Converts all UTF-8 remains in ASCII
94		* @param string $input
95		* @return string
96		*/
97		public static function toASCII($input)
98		{
99		$input = self::normalize($input);
100		$input = self::squashCharacters($input);
101		return mb_convert_encoding($input, 'ascii');
102		}
103
104		/**
105		* Replaces all accented characters of their ASCII equivalents
106		* @param string $input
107		* @return string
108		*/
109	1	public static function squashCharacters($input)
110		{
111	1	$input = trim($input);
112	1	$aFind = ['á','à','ã','â','é','ê','í','ó','ô','õ','ö','ú','ü',
113		'ç','Á','À','Ã','Â','É','Ê','Í','Ó','Ô','Õ','Ö','Ú','Ü','Ç'];
114	1	$aSubs = ['a','a','a','a','e','e','i','o','o','o','o','u','u',
115		'c','A','A','A','A','E','E','I','O','O','O','O','U','U','C'];
116	1	return str_replace($aFind, $aSubs, $input);
117		}
118
119		/**
120		* Replace all non-UTF-8 chars to UTF-8
121		* Remove all control chars
122		* Remove all multiple spaces
123		* @param string $input
124		* @return string
125		*/
126	1	public static function normalize($input)
127		{
128		//Carriage Return, Tab and Line Feed is not acceptable in strings
129	1	$input = str_replace(["\r","\t","\n"], "", $input);
130		//Multiple spaces is not acceptable in strings
131	1	$input = preg_replace('/(?:\s\s+)/', ' ', $input);
132		//Only UTF-8 characters is acceptable
133	1	$input = Encoding::fixUTF8($input);
134	1	$input = preg_replace(
135		'/[\x00-\x08\x10\x0B\x0C\x0E-\x19\x7F]'.
136		'\|[\x00-\x7F][\x80-\xBF]+'.
137		'\|([\xC0\xC1]\|[\xF0-\xFF])[\x80-\xBF]*'.
138		'\|[\xC2-\xDF]((?![\x80-\xBF])\|[\x80-\xBF]{2,})'.
139	1	'\|[\xE0-\xEF](([\x80-\xBF](?![\x80-\xBF]))\|(?![\x80-\xBF]{2})\|[\x80-\xBF]{3,})/S',
140	1	'',
141		$input
142		);
143	1	$input = preg_replace('/\xE0[\x80-\x9F][\x80-\xBF]'.
144	1	'\|\xED[\xA0-\xBF][\x80-\xBF]/S', '', $input);
145		//And no other control character is acceptable either
146	1	return preg_replace('/[[:cntrl:]]/', '', $input);
147		}
148
149		/**
150		* Remove all non numeric characters from string
151		* @param string $string
152		* @return string
153		*/
154	1	public static function onlyNumbers($string)
155		{
156	1	return preg_replace("/[^0-9]/", "", $string);
157		}
158
159		/**
160		* Remove unwanted attributes, prefixes, sulfixes and other control
161		* characters like \r \n \s \t
162		* @param string $string
163		* @param boolean $removeEncodingTag remove encoding tag from a xml
164		* @return string
165		*/
166	2	public static function clearXmlString($string, $removeEncodingTag = false)
167		{
168		$aFind = array(
169	2	'xmlns:default="http://www.w3.org/2000/09/xmldsig#"',
170		' standalone="no"',
171		'default:',
172		':default',
173		"\n",
174		"\r",
175		"\t"
176		);
177	2	$retXml = str_replace($aFind, "", $string);
178	2	$retXml = preg_replace('/(\>)\s*(\<)/m', '$1$2', $retXml);
179	2	if ($removeEncodingTag) {
180	1	$retXml = self::deleteAllBetween($retXml, '<?xml', '?>');
181		}
182	2	return $retXml;
183		}
184
185		/**
186		* Remove all characters between markers
187		* @param string $string
188		* @param string $beginning
189		* @param string $end
190		* @return string
191		*/
192	2	public static function deleteAllBetween($string, $beginning, $end)
193		{
194	2	$beginningPos = strpos($string, $beginning);
195	2	$endPos = strpos($string, $end);
196	2	if ($beginningPos === false \|\| $endPos === false) {
197		return $string;
198		}
199	2	$textToDelete = substr($string, $beginningPos, ($endPos + strlen($end)) - $beginningPos);
200	2	return str_replace($textToDelete, '', $string);
201		}
202
203		/**
204		* Clears the xml after adding the protocol, removing repeated namespaces
205		* @param string $string
206		* @return string
207		*/
208	1	public static function clearProtocoledXML($string)
209		{
210	1	$procXML = self::clearXmlString($string);
211	1	$aApp = array('nfe','cte','mdfe');
212	1	foreach ($aApp as $app) {
213	1	$procXML = str_replace(
214	1	'xmlns="http://www.portalfiscal.inf.br/'.$app.'" xmlns="http://www.w3.org/2000/09/xmldsig#"',
215	1	'xmlns="http://www.portalfiscal.inf.br/'.$app.'"',
216		$procXML
217		);
218		}
219	1	return $procXML;
220		}
221
222		/**
223		* Remove some alien chars from txt
224		* @param string $txt
225		* @return string
226		*/
227	1	public static function removeSomeAlienCharsfromTxt($txt)
228		{
229		//remove CRs and TABs
230	1	$txt = str_replace(["\r","\t"], "", $txt);
231		//remove multiple spaces
232	1	$txt = preg_replace('/(?:\s\s+)/', ' ', $txt);
233		//remove spaces at begin and end of fields
234	1	$txt = str_replace(["\| "," \|"], "\|", $txt);
235	1	return $txt;
236		}
237
238		/**
239		* Creates a string ramdomically with the specified length
240		* @param int $length
241		* @return string
242		*/
243	1	public static function randomString($length)
244		{
245		$keyspace = '0123456789abcdefghijklmnopqr'
246	1	. 'stuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ';
247	1	$str = '';
248	1	$max = mb_strlen($keyspace, '8bit') - 1;
249	1	for ($i = 0; $i < $length; ++$i) {
250	1	$str .= $keyspace[rand(0, $max)];
251		}
252	1	return $str;
253		}
254		}
255

nfephp-org / sped-common

Push — master ( 2938c8...c15d53 )

Strings::toASCII() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like