DataRefReplace::transform() - Code Metrics - matecat/subfiltering - Measure and Improve Code Quality continuously with Scrutinizer

DataRefReplace::transform() A
last analyzed 2025-10-27 13:05 UTC

↳ Parent: DataRefReplace

Complexity

Conditions	3
Paths	4

Size

Total Lines	18
Code Lines	9

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	10
CRAP Score	3

Importance

Changes	2
Bugs	0	Features	0

Metric	Value
cc	3
eloc	9
c	2
b	0
f	0
nc	4
nop	1
dl	0
loc	18
ccs	10
cts	10
cp	1
crap	3
rs	9.9666

<?php

namespace Matecat\SubFiltering\Filters;

use Exception;
use Matecat\SubFiltering\Commons\AbstractHandler;
use Matecat\SubFiltering\Enum\CTypeEnum;
use Matecat\SubFiltering\Utils\DataRefReplacer;
use Matecat\XmlParser\XmlParser;

class DataRefReplace extends AbstractHandler {

    /**
     * @var array
     */
    private array $dataRefMap = [];

    /**
     * DataRefReplace constructor.
     */
    public function __construct() {
        parent::__construct();
    }

    /**
     * @inheritDoc
     */
    public function transform( string $segment ): string {

        if ( empty( $this->dataRefMap ) ) {
            $this->dataRefMap = $this->pipeline->getDataRefMap();
        }

        // dataRefMap is present only in xliff 2.0 files
        if ( empty( $this->dataRefMap ) ) {
            $segment = $this->replace_Ph_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );

            return $this->replace_Pc_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );
        }

        $dataRefReplacer = new DataRefReplacer( $this->dataRefMap );
        $segment         = $dataRefReplacer->replace( $segment );
        $segment         = $this->replace_Ph_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );

        return $this->replace_Pc_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );
    }

    /**
     * This function replace encoded ph tags (from Xliff 2.0) without any dataRef correspondence
     * to regular Matecat <ph> tag for UI presentation
     *
     * Example:
     *
     * We can control who sees content when with <ph id="source1" dataRef="source1"/>Visibility Constraints.
     *
     * Is transformed to:
     *
     * We can control who sees content when with &lt;ph id="mtc_ph_u_1" equiv-text="base64:PHBoIGlkPSJzb3VyY2UxIiBkYXRhUmVmPSJzb3VyY2UxIi8+"/&gt;Visibility Constraints.
     *
     * @param $segment
     *
     * @return string
     */
    private function replace_Ph_TagsWithoutDataRefCorrespondenceToMatecatPhTags( string $segment ): string {

        preg_match_all( '/<(ph .*?)>/iu', $segment, $phTags );

        if ( count( $phTags[ 0 ] ) === 0 ) {
            return $segment;
        }

        foreach ( $phTags[ 0 ] as $phTag ) {

            // check if phTag has not any correspondence on dataRef map
            if ( $this->isAValidPhTag( $phTag ) ) {
                $segment = preg_replace(
                        '/' . preg_quote( $phTag, '/' ) . '/',
                        '<ph id="' . $this->getPipeline()->getNextId() .
                        '" ctype="' . CTypeEnum::ORIGINAL_PH_OR_NOT_DATA_REF .
                        '" equiv-text="base64:' . base64_encode( $phTag ) .
                        '"/>',
                        $segment,
                        1 // replace ONLY ONE occurrence
                );
            }
        }

        return $segment;
    }

    /**
     * This function checks if a ph tag with dataRef attribute
     * and without equiv-text
     * a correspondence on dataRef map
     *
     * @param string $phTag
     *
     * @return bool
     */
    private function isAValidPhTag( string $phTag ): bool {

        // try not to throw exception for wrong segments with opening tags and no closing
        try {
            $parsed = XmlParser::parse( $phTag, true );
        } catch ( Exception $e ) {
            return false;
        }

        // check for matecat ctype
        $cType = isset( $parsed[ 0 ]->attributes[ 'ctype' ] ) ? $parsed[ 0 ]->attributes[ 'ctype' ] : 'not-found';
        if ( CTypeEnum::isMatecatCType( $cType ) ) {
            return false;
        }

        // if has equiv-text don't touch
        if ( isset( $parsed[ 0 ]->attributes[ 'equiv-text' ] ) ) {
            return false;
        }

        if ( isset( $parsed[ 0 ]->attributes[ 'dataRef' ] ) ) {
            return !array_key_exists( $parsed[ 0 ]->attributes[ 'dataRef' ], $this->dataRefMap );
        }

        return true;

    }

    /**
     * This function replace encoded pc tags (from Xliff 2.0) without any dataRef correspondence
     * to regular Matecat <ph> tag for UI presentation
     *
     * Example:
     *
     * Text &lt;ph id="source1_1" dataType="pcStart" originalData="Jmx0O3BjIGlkPSJzb3VyY2UxIiBkYXRhUmVmU3RhcnQ9InNvdXJjZTEiIGRhdGFSZWZFbmQ9InNvdXJjZTEiJmd0Ow==" dataRef="source1" equiv-text="base64:eA=="/&gt;&lt;pc id="1u" type="fmt" subType="m:u"&gt;link&lt;/pc&gt;&lt;ph id="source1_2" dataType="pcEnd" originalData="Jmx0Oy9wYyZndDs=" dataRef="source1" equiv-text="base64:eA=="/&gt;.
     *
     * is transformed to:
     *
     * Text &lt;ph id="source1_1" dataType="pcStart" originalData="Jmx0O3BjIGlkPSJzb3VyY2UxIiBkYXRhUmVmU3RhcnQ9InNvdXJjZTEiIGRhdGFSZWZFbmQ9InNvdXJjZTEiJmd0Ow==" dataRef="source1" equiv-text="base64:eA=="/&gt;&lt;ph id="mtc_u_1" equiv-text="base64:Jmx0O3BjIGlkPSIxdSIgdHlwZT0iZm10IiBzdWJUeXBlPSJtOnUiJmd0Ow=="/&gt;link&lt;ph id="mtc_u_2" equiv-text="base64:Jmx0Oy9wYyZndDs="/&gt;&lt;ph id="source1_2" dataType="pcEnd" originalData="Jmx0Oy9wYyZndDs=" dataRef="source1" equiv-text="base64:eA=="/&gt;.
     *
     * @param $segment
     *
     * @return string
     */
    private function replace_Pc_TagsWithoutDataRefCorrespondenceToMatecatPhTags( string $segment ): string {

        preg_match_all( '/<(pc .*?)>/iu', $segment, $openingPcTags );
        preg_match_all( '|<(/pc)>|iu', $segment, $closingPcTags );

        if ( count( $openingPcTags[ 0 ] ) === 0 ) {
            return $segment;
        }

        foreach ( $openingPcTags[ 0 ] as $openingPcTag ) {
            $segment = preg_replace(
                    '/' . preg_quote( $openingPcTag, '/' ) . '/',
                    '<ph id="' . $this->getPipeline()->getNextId() .
                    '" ctype="' . CTypeEnum::ORIGINAL_PC_OPEN_NO_DATA_REF .
                    '" equiv-text="base64:' . base64_encode( $openingPcTag ) .
                    '"/>',
                    $segment,
                    1
            );
        }

        foreach ( $closingPcTags[ 0 ] as $closingPcTag ) {
            $segment = preg_replace(
                    '/' . preg_quote( $closingPcTag, '/' ) . '/',
                    '<ph id="' . $this->getPipeline()->getNextId() .
                    '" ctype="' . CTypeEnum::ORIGINAL_PC_CLOSE_NO_DATA_REF .
                    '" equiv-text="base64:' . base64_encode( $closingPcTag ) .
                    '"/>',
                    $segment,
                    1
            );
        }

        return $segment;
    }
}

1		<?php
2
3		namespace Matecat\SubFiltering\Filters;
4
5		use Exception;
6		use Matecat\SubFiltering\Commons\AbstractHandler;
7		use Matecat\SubFiltering\Enum\CTypeEnum;
8		use Matecat\SubFiltering\Utils\DataRefReplacer;
9		use Matecat\XmlParser\XmlParser;
10
11		class DataRefReplace extends AbstractHandler {
12
13		/**
14		* @var array
15		*/
16		private array $dataRefMap = [];
17
18		/**
19		* DataRefReplace constructor.
20		*/
21	57	public function __construct() {
22	57	parent::__construct();
23		}
24
25		/**
26		* @inheritDoc
27		*/
28	57	public function transform( string $segment ): string {
29
30	57	if ( empty( $this->dataRefMap ) ) {
31	57	$this->dataRefMap = $this->pipeline->getDataRefMap();
32		}
33
34		// dataRefMap is present only in xliff 2.0 files
35	57	if ( empty( $this->dataRefMap ) ) {
36	45	$segment = $this->replace_Ph_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );
37
38	45	return $this->replace_Pc_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );
39		}
40
41	12	$dataRefReplacer = new DataRefReplacer( $this->dataRefMap );
42	12	$segment = $dataRefReplacer->replace( $segment );
43	12	$segment = $this->replace_Ph_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );
44
45	12	return $this->replace_Pc_TagsWithoutDataRefCorrespondenceToMatecatPhTags( $segment );
46		}
47
48		/**
49		* This function replace encoded ph tags (from Xliff 2.0) without any dataRef correspondence
50		* to regular Matecat <ph> tag for UI presentation
51		*
52		* Example:
53		*
54		* We can control who sees content when with <ph id="source1" dataRef="source1"/>Visibility Constraints.
55		*
56		* Is transformed to:
57		*
58		* We can control who sees content when with <ph id="mtc_ph_u_1" equiv-text="base64:PHBoIGlkPSJzb3VyY2UxIiBkYXRhUmVmPSJzb3VyY2UxIi8+"/>Visibility Constraints.
59		*
60		* @param $segment
61		*
62		* @return string
63		*/
64	57	private function replace_Ph_TagsWithoutDataRefCorrespondenceToMatecatPhTags( string $segment ): string {
65
66	57	preg_match_all( '/<(ph .*?)>/iu', $segment, $phTags );
67
68	57	if ( count( $phTags[ 0 ] ) === 0 ) {
69	20	return $segment;
70		}
71
72	37	foreach ( $phTags[ 0 ] as $phTag ) {
73
74		// check if phTag has not any correspondence on dataRef map
75	37	if ( $this->isAValidPhTag( $phTag ) ) {
76	5	$segment = preg_replace(
77	5	'/' . preg_quote( $phTag, '/' ) . '/',
78	5	'<ph id="' . $this->getPipeline()->getNextId() .
79	5	'" ctype="' . CTypeEnum::ORIGINAL_PH_OR_NOT_DATA_REF .
80	5	'" equiv-text="base64:' . base64_encode( $phTag ) .
81	5	'"/>',
82	5	$segment,
83	5	1 // replace ONLY ONE occurrence
84	5	);
85		}
86		}
87
88	37	return $segment;
89		}
90
91		/**
92		* This function checks if a ph tag with dataRef attribute
93		* and without equiv-text
94		* a correspondence on dataRef map
95		*
96		* @param string $phTag
97		*
98		* @return bool
99		*/
100	37	private function isAValidPhTag( string $phTag ): bool {
101
102		// try not to throw exception for wrong segments with opening tags and no closing
103		try {
104	37	$parsed = XmlParser::parse( $phTag, true );
105	1	} catch ( Exception $e ) {
106	1	return false;
107		}
108
109		// check for matecat ctype
110	36	$cType = isset( $parsed[ 0 ]->attributes[ 'ctype' ] ) ? $parsed[ 0 ]->attributes[ 'ctype' ] : 'not-found';
111	36	if ( CTypeEnum::isMatecatCType( $cType ) ) {
112	34	return false;
113		}
114
115		// if has equiv-text don't touch
116	6	if ( isset( $parsed[ 0 ]->attributes[ 'equiv-text' ] ) ) {
117	1	return false;
118		}
119
120	5	if ( isset( $parsed[ 0 ]->attributes[ 'dataRef' ] ) ) {
121	2	return !array_key_exists( $parsed[ 0 ]->attributes[ 'dataRef' ], $this->dataRefMap );
122		}
123
124	3	return true;
125
126		}
127
128		/**
129		* This function replace encoded pc tags (from Xliff 2.0) without any dataRef correspondence
130		* to regular Matecat <ph> tag for UI presentation
131		*
132		* Example:
133		*
134		* Text <ph id="source1_1" dataType="pcStart" originalData="Jmx0O3BjIGlkPSJzb3VyY2UxIiBkYXRhUmVmU3RhcnQ9InNvdXJjZTEiIGRhdGFSZWZFbmQ9InNvdXJjZTEiJmd0Ow==" dataRef="source1" equiv-text="base64:eA=="/><pc id="1u" type="fmt" subType="m:u">link</pc><ph id="source1_2" dataType="pcEnd" originalData="Jmx0Oy9wYyZndDs=" dataRef="source1" equiv-text="base64:eA=="/>.
135		*
136		* is transformed to:
137		*
138		* Text <ph id="source1_1" dataType="pcStart" originalData="Jmx0O3BjIGlkPSJzb3VyY2UxIiBkYXRhUmVmU3RhcnQ9InNvdXJjZTEiIGRhdGFSZWZFbmQ9InNvdXJjZTEiJmd0Ow==" dataRef="source1" equiv-text="base64:eA=="/><ph id="mtc_u_1" equiv-text="base64:Jmx0O3BjIGlkPSIxdSIgdHlwZT0iZm10IiBzdWJUeXBlPSJtOnUiJmd0Ow=="/>link<ph id="mtc_u_2" equiv-text="base64:Jmx0Oy9wYyZndDs="/><ph id="source1_2" dataType="pcEnd" originalData="Jmx0Oy9wYyZndDs=" dataRef="source1" equiv-text="base64:eA=="/>.
139		*
140		* @param $segment
141		*
142		* @return string
143		*/
144	57	private function replace_Pc_TagsWithoutDataRefCorrespondenceToMatecatPhTags( string $segment ): string {
145
146	57	preg_match_all( '/<(pc .*?)>/iu', $segment, $openingPcTags );
147	57	preg_match_all( '\|<(/pc)>\|iu', $segment, $closingPcTags );
148
149	57	if ( count( $openingPcTags[ 0 ] ) === 0 ) {
150	53	return $segment;
151		}
152
153	4	foreach ( $openingPcTags[ 0 ] as $openingPcTag ) {
154	4	$segment = preg_replace(
155	4	'/' . preg_quote( $openingPcTag, '/' ) . '/',
156	4	'<ph id="' . $this->getPipeline()->getNextId() .
157	4	'" ctype="' . CTypeEnum::ORIGINAL_PC_OPEN_NO_DATA_REF .
158	4	'" equiv-text="base64:' . base64_encode( $openingPcTag ) .
159	4	'"/>',
160	4	$segment,
161	4	1
162	4	);
163		}
164
165	4	foreach ( $closingPcTags[ 0 ] as $closingPcTag ) {
166	4	$segment = preg_replace(
167	4	'/' . preg_quote( $closingPcTag, '/' ) . '/',
168	4	'<ph id="' . $this->getPipeline()->getNextId() .
169	4	'" ctype="' . CTypeEnum::ORIGINAL_PC_CLOSE_NO_DATA_REF .
170	4	'" equiv-text="base64:' . base64_encode( $closingPcTag ) .
171	4	'"/>',
172	4	$segment,
173	4	1
174	4	);
175		}
176
177	4	return $segment;
178		}
179		}

matecat / subfiltering

DataRefReplace::transform() A last analyzed 2025-10-27 13:05 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

DataRefReplace::transform() A
last analyzed 2025-10-27 13:05 UTC