Xliff20::updateCounts() - Code Metrics - Inspection of "Fix: When there is no target tag, the check for un..." - matecat/xliff-parser - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#90)

by Domenico

created 2024-08-06 16:47 UTC

Xliff20::updateCounts() A

↳ Parent: Xliff20

Complexity

Conditions	2
Paths	2

Size

Total Lines	5
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	2
eloc	3
nc	2
nop	0
dl	0
loc	5
rs	10
c	1
b	0
f	0

<?php
/**
 * Created by PhpStorm.
 * @author hashashiyyin [email protected] / [email protected]
 * Date: 02/08/24
 * Time: 17:51
 *
 */

namespace Matecat\XliffParser\XliffReplacer;

use Matecat\XliffParser\Utils\Strings;

class Xliff20 extends AbstractXliffReplacer {

    /**
     * @var int
     */
    private int $mdaGroupCounter = 0;
    /**
     * @var bool
     */
    protected bool $unitContainsMda = false;   // check if <unit> already contains a <mda:metadata> (forXliff v 2.*)
    /**
     * @var array
     */
    protected array $nodesToBuffer = [
            'source',
            'mda:metadata',
            'memsource:additionalTagData',
            'originalData',
            'note'
    ];

    /**
     * @inheritDoc
     */
    protected function tagOpen( $parser, string $name, array $attr ) {

        $this->handleOpenUnit( $name, $attr );

        if ( 'mda:metadata' === $name ) {
            $this->unitContainsMda = true;
        }

        $this->checkSetInTarget( $name );

        // open buffer
        $this->setInBuffer( $name );

        // check if we are inside a <target>, obviously this happen only if there are targets inside the trans-unit
        // <target> must be stripped to be replaced, so this check avoids <target> reconstruction
        if ( !$this->inTarget ) {

            $tag = '';

            //
            // ============================================
            // only for Xliff 2.*
            // ============================================
            //
            // In xliff v2 we MUST add <mda:metadata> BEFORE <notes>/<originalData>/<segment>/<ignorable>
            //
            // As documentation says, <unit> contains:
            //
            // - elements from other namespaces, OPTIONAL
            // - Zero or one <notes> elements followed by
            // - Zero or one <originalData> element followed by
            // - One or more <segment> or <ignorable> elements in any order.
            //
            // For more info please refer to:
            //
            // http://docs.oasis-open.org/xliff/xliff-core/v2.0/os/xliff-core-v2.0-os.html#unit
            //
            if ( in_array( $name, [ 'notes', 'originalData', 'segment', 'ignorable' ] ) &&
                    $this->unitContainsMda === false &&
                    !empty( $this->transUnits[ $this->currentTransUnitId ] ) &&
                    !$this->hasWrittenCounts
            ) {
                // we need to update counts here
                $this->updateCounts();
                $this->hasWrittenCounts = true;
                $tag                    .= $this->getWordCountGroupForXliffV2();
                $this->unitContainsMda  = true;
            }

            // construct tag
            $tag .= "<$name ";

            foreach ( $attr as $k => $v ) {
                //normal tag flux, put attributes in it but skip for translation state and set the right value for the attribute
                if ( $k != 'state' ) {
                    $tag .= "$k=\"$v\" ";
                }
            }

            $seg = $this->getCurrentSegment();

            if ( $name === $this->tuTagName && !empty( $seg ) && isset( $seg[ 'sid' ] ) ) {

                // add `mtc:segment-id` to xliff v.2*
                if ( strpos( $tag, 'mtc:segment-id' ) === false ) {
                    $tag .= "mtc:segment-id=\"{$seg[ 'sid' ]}\" ";
                }

            }

            // replace state for xliff v2
            if ( 'segment' === $name ) { // add state to segment in Xliff v2
                [ $stateProp, ] = StatusToStateAttribute::getState( $this->xliffVersion, $seg[ 'status' ] );
                $tag .= $stateProp;
            }

            $tag = $this->handleOpenXliffTag( $name, $attr, $tag );

            $this->checkForSelfClosedTagAndFlush( $parser, $tag );

        }

    }

    /**
     * @param string $name
     * @param array  $attr
     * @param string $tag
     *
     * @return string
     */
    protected function handleOpenXliffTag( string $name, array $attr, string $tag ): string {
        $tag = parent::handleOpenXliffTag( $name, $attr, $tag );
        // add oasis xliff 20 namespace
        if ( $name === 'xliff' && !array_key_exists( 'xmlns:mda', $attr ) ) {
            $tag .= 'xmlns:mda="urn:oasis:names:tc:xliff:metadata:2.0"';
        }

        return $tag;
    }

    /**
     * @inheritDoc
     */
    protected function tagClose( $parser, string $name ) {
        $tag = '';

        /**
         * if is a tag within <target> or
         * if it is an empty tag, do not add closing tag because we have already closed it in
         *
         * self::tagOpen method
         */
        if ( !$this->isEmpty ) {

            if ( !$this->inTarget ) {
                $tag = "</$name>";
            }

            if ( 'target' == $name ) {

                if ( isset( $this->transUnits[ $this->currentTransUnitId ] ) ) {

                    $seg = $this->getCurrentSegment();

                    // update counts
                    if ( !$this->hasWrittenCounts && !empty( $seg ) ) {
                        $this->updateSegmentCounts( $seg );
                    }

                    // delete translations so the prepareSegment
                    // will put source content in target tag
                    if ( $this->sourceInTarget ) {
                        $seg[ 'translation' ] = '';
                        $this->resetCounts();
                    }

                    // append $translation
                    $translation = $this->prepareTranslation( $seg );

                    //append translation
                    $tag = "<target>$translation</target>";

                }

                // signal we are leaving a target
                $this->targetWasWritten = true;
                $this->inTarget         = false;
                $this->postProcAndFlush( $this->outputFP, $tag, true );

            } elseif ( in_array( $name, $this->nodesToBuffer ) ) { // we are closing a critical CDATA section

                $this->bufferIsActive = false;

                // only for Xliff 2.*
                // write here <mda:metaGroup> and <mda:meta> if already present in the <unit>
                if ( 'mda:metadata' === $name && $this->unitContainsMda && !$this->hasWrittenCounts ) {

                    // we need to update counts here
                    $this->updateCounts();
                    $this->hasWrittenCounts = true;

                    $tag = $this->CDATABuffer;
                    $tag .= $this->getWordCountGroupForXliffV2( false );
                    $tag .= "    </mda:metadata>";

                } else {
                    $tag = $this->CDATABuffer . "</$name>";
                }

                $this->CDATABuffer = "";

                //flush to the pointer
                $this->postProcAndFlush( $this->outputFP, $tag );

            } elseif ( 'segment' === $name ) {

                // only for Xliff 2.*
                // if segment has no <target> add it BEFORE </segment>
                if ( !$this->targetWasWritten ) {

                    $seg = $this->getCurrentSegment();

                    if ( isset( $seg[ 'translation' ] ) ) {

                        $translation = $this->prepareTranslation( $seg );
                        // replace the tag
                        $tag = "<target>$translation</target>";

                        $tag .= '</segment>';

                    }

                }

                // update segmentPositionInTu
                $this->segmentInUnitPosition++;

                $this->postProcAndFlush( $this->outputFP, $tag );

                // we are leaving <segment>, reset $segmentHasTarget
                $this->targetWasWritten = false;

            } elseif ( $this->bufferIsActive ) { // this is a tag ( <g | <mrk ) inside a seg or seg-source tag
                $this->CDATABuffer .= "</$name>";
                // Do NOT Flush
            } else { //generic tag closure do Nothing
                // flush to pointer
                $this->postProcAndFlush( $this->outputFP, $tag );
            }
        } else {
            //ok, nothing to be done; reset flag for next coming tag
            $this->isEmpty = false;
        }

        // check if we are leaving a <trans-unit> (xliff v1.*) or <unit> (xliff v2.*)
        if ( $this->tuTagName === $name ) {
            $this->currentTransUnitIsTranslatable = null;
            $this->inTU                           = false;
            $this->unitContainsMda                = false;
            $this->hasWrittenCounts               = false;

            $this->resetCounts();
        }
    }

    /**
     * Update counts
     */
    private function updateCounts() {

        $seg = $this->getCurrentSegment();
        if ( !empty( $seg ) ) {
            $this->updateSegmentCounts( $seg );
        }

    }

    /**
     * @param bool $withMetadataTag
     *
     * @return string
     */
    private function getWordCountGroupForXliffV2( bool $withMetadataTag = true ): string {

        $this->mdaGroupCounter++;
        $segments_count_array = $this->counts[ 'segments_count_array' ];

        $tag = '';

        if ( $withMetadataTag === true ) {
            $tag .= '<mda:metadata>';
        }

        $index = 0;
        foreach ( $segments_count_array as $segments_count_item ) {

            $id = 'word_count_tu[' . $this->currentTransUnitId . '][' . $index . ']';
            $index++;

            $tag .= "    <mda:metaGroup id=\"" . $id . "\" category=\"row_xml_attribute\">
                                <mda:meta type=\"x-matecat-raw\">" . $segments_count_item[ 'raw_word_count' ] . "</mda:meta>
                                <mda:meta type=\"x-matecat-weighted\">" . $segments_count_item[ 'eq_word_count' ] . "</mda:meta>
                            </mda:metaGroup>";
        }

        if ( $withMetadataTag === true ) {
            $tag .= '</mda:metadata>';
        }

        return $tag;

    }

    /**
     * prepare segment tagging for xliff insertion
     *
     * @param array $seg
     *
     * @return string
     */
    protected function prepareTranslation( array $seg ): string {

        $segment     = Strings::removeDangerousChars( $seg [ 'segment' ] );
        $translation = Strings::removeDangerousChars( $seg [ 'translation' ] );
        $dataRefMap  = ( isset( $seg[ 'data_ref_map' ] ) ) ? Strings::jsonToArray( $seg[ 'data_ref_map' ] ) : [];

        if ( $seg [ 'translation' ] == '' ) {
            $translation = $segment;
        } else {
            if ( $this->callback instanceof XliffReplacerCallbackInterface ) {
                $error = ( !empty( $seg[ 'error' ] ) ) ? $seg[ 'error' ] : null;
                if ( $this->callback->thereAreErrors( $seg[ 'sid' ], $segment, $translation, $dataRefMap, $error ) ) {
                    $translation = '|||UNTRANSLATED_CONTENT_START|||' . $segment . '|||UNTRANSLATED_CONTENT_END|||';
                }
            }
        }

        return $translation;

    }

}

1			<?php
2			/**
3			* Created by PhpStorm.
4			* @author hashashiyyin [email protected] / [email protected]
5			* Date: 02/08/24
6			* Time: 17:51
7			*
8			*/
9
10			namespace Matecat\XliffParser\XliffReplacer;
11
12			use Matecat\XliffParser\Utils\Strings;
13
14			class Xliff20 extends AbstractXliffReplacer {
15
16			/**
17			* @var int
18			*/
19			private int $mdaGroupCounter = 0;
20			/**
21			* @var bool
22			*/
23			protected bool $unitContainsMda = false; // check if <unit> already contains a <mda:metadata> (forXliff v 2.*)
24			/**
25			* @var array
26			*/
27			protected array $nodesToBuffer = [
28			'source',
29			'mda:metadata',
30			'memsource:additionalTagData',
31			'originalData',
32			'note'
33			];
34
35			/**
36			* @inheritDoc
37			*/
38			protected function tagOpen( $parser, string $name, array $attr ) {
39
40			$this->handleOpenUnit( $name, $attr );
41
42			if ( 'mda:metadata' === $name ) {
43			$this->unitContainsMda = true;
44			}
45
46			$this->checkSetInTarget( $name );
47
48			// open buffer
49			$this->setInBuffer( $name );
50
51			// check if we are inside a <target>, obviously this happen only if there are targets inside the trans-unit
52			// <target> must be stripped to be replaced, so this check avoids <target> reconstruction
53			if ( !$this->inTarget ) {
54
55			$tag = '';
56
57			//
58			// ============================================
59			// only for Xliff 2.*
60			// ============================================
61			//
62			// In xliff v2 we MUST add <mda:metadata> BEFORE <notes>/<originalData>/<segment>/<ignorable>
63			//
64			// As documentation says, <unit> contains:
65			//
66			// - elements from other namespaces, OPTIONAL
67			// - Zero or one <notes> elements followed by
68			// - Zero or one <originalData> element followed by
69			// - One or more <segment> or <ignorable> elements in any order.
70			//
71			// For more info please refer to:
72			//
73			// http://docs.oasis-open.org/xliff/xliff-core/v2.0/os/xliff-core-v2.0-os.html#unit
74			//
75			if ( in_array( $name, [ 'notes', 'originalData', 'segment', 'ignorable' ] ) &&
76			$this->unitContainsMda === false &&
77			!empty( $this->transUnits[ $this->currentTransUnitId ] ) &&
78			!$this->hasWrittenCounts
79			) {
80			// we need to update counts here
81			$this->updateCounts();
82			$this->hasWrittenCounts = true;
83			$tag .= $this->getWordCountGroupForXliffV2();
84			$this->unitContainsMda = true;
85			}
86
87			// construct tag
88			$tag .= "<$name ";
89
90			foreach ( $attr as $k => $v ) {
91			//normal tag flux, put attributes in it but skip for translation state and set the right value for the attribute
92			if ( $k != 'state' ) {
93			$tag .= "$k=\"$v\" ";
94			}
95			}
96
97			$seg = $this->getCurrentSegment();
98
99			if ( $name === $this->tuTagName && !empty( $seg ) && isset( $seg[ 'sid' ] ) ) {
100
101			// add `mtc:segment-id` to xliff v.2*
102			if ( strpos( $tag, 'mtc:segment-id' ) === false ) {
103			$tag .= "mtc:segment-id=\"{$seg[ 'sid' ]}\" ";
104			}
105
106			}
107
108			// replace state for xliff v2
109			if ( 'segment' === $name ) { // add state to segment in Xliff v2
110			[ $stateProp, ] = StatusToStateAttribute::getState( $this->xliffVersion, $seg[ 'status' ] );
111			$tag .= $stateProp;
112			}
113
114			$tag = $this->handleOpenXliffTag( $name, $attr, $tag );
115
116			$this->checkForSelfClosedTagAndFlush( $parser, $tag );
117
118			}
119
120			}
121
122			/**
123			* @param string $name
124			* @param array $attr
125			* @param string $tag
126			*
127			* @return string
128			*/
129			protected function handleOpenXliffTag( string $name, array $attr, string $tag ): string {
130			$tag = parent::handleOpenXliffTag( $name, $attr, $tag );
131			// add oasis xliff 20 namespace
132			if ( $name === 'xliff' && !array_key_exists( 'xmlns:mda', $attr ) ) {
133			$tag .= 'xmlns:mda="urn:oasis:names:tc:xliff:metadata:2.0"';
134			}
135
136			return $tag;
137			}
138
139			/**
140			* @inheritDoc
141			*/
142			protected function tagClose( $parser, string $name ) {
143			$tag = '';
144
145			/**
146			* if is a tag within <target> or
147			* if it is an empty tag, do not add closing tag because we have already closed it in
148			*
149			* self::tagOpen method
150			*/
151			if ( !$this->isEmpty ) {
152
153			if ( !$this->inTarget ) {
154			$tag = "</$name>";
155			}
156
157			if ( 'target' == $name ) {
158
159			if ( isset( $this->transUnits[ $this->currentTransUnitId ] ) ) {
160
161			$seg = $this->getCurrentSegment();
162
163			// update counts
164			if ( !$this->hasWrittenCounts && !empty( $seg ) ) {
165			$this->updateSegmentCounts( $seg );
166			}
167
168			// delete translations so the prepareSegment
169			// will put source content in target tag
170			if ( $this->sourceInTarget ) {
171			$seg[ 'translation' ] = '';
172			$this->resetCounts();
173			}
174
175			// append $translation
176			$translation = $this->prepareTranslation( $seg );
177
178			//append translation
179			$tag = "<target>$translation</target>";
180
181			}
182
183			// signal we are leaving a target
184			$this->targetWasWritten = true;
185			$this->inTarget = false;
186			$this->postProcAndFlush( $this->outputFP, $tag, true );
187
188			} elseif ( in_array( $name, $this->nodesToBuffer ) ) { // we are closing a critical CDATA section
189
190			$this->bufferIsActive = false;
191
192			// only for Xliff 2.*
193			// write here <mda:metaGroup> and <mda:meta> if already present in the <unit>
194			if ( 'mda:metadata' === $name && $this->unitContainsMda && !$this->hasWrittenCounts ) {
195
196			// we need to update counts here
197			$this->updateCounts();
198			$this->hasWrittenCounts = true;
199
200			$tag = $this->CDATABuffer;
201			$tag .= $this->getWordCountGroupForXliffV2( false );
202			$tag .= " </mda:metadata>";
203
204			} else {
205			$tag = $this->CDATABuffer . "</$name>";
206			}
207
208			$this->CDATABuffer = "";
209
210			//flush to the pointer
211			$this->postProcAndFlush( $this->outputFP, $tag );
212
213			} elseif ( 'segment' === $name ) {
214
215			// only for Xliff 2.*
216			// if segment has no <target> add it BEFORE </segment>
217			if ( !$this->targetWasWritten ) {
218
219			$seg = $this->getCurrentSegment();
220
221			if ( isset( $seg[ 'translation' ] ) ) {
222
223			$translation = $this->prepareTranslation( $seg );
224			// replace the tag
225			$tag = "<target>$translation</target>";
226
227			$tag .= '</segment>';
228
229			}
230
231			}
232
233			// update segmentPositionInTu
234			$this->segmentInUnitPosition++;
235
236			$this->postProcAndFlush( $this->outputFP, $tag );
237
238			// we are leaving <segment>, reset $segmentHasTarget
239			$this->targetWasWritten = false;
240
241			} elseif ( $this->bufferIsActive ) { // this is a tag ( <g \| <mrk ) inside a seg or seg-source tag
242			$this->CDATABuffer .= "</$name>";
243			// Do NOT Flush
244			} else { //generic tag closure do Nothing
245			// flush to pointer
246			$this->postProcAndFlush( $this->outputFP, $tag );
247			}
248			} else {
249			//ok, nothing to be done; reset flag for next coming tag
250			$this->isEmpty = false;
251			}
252
253			// check if we are leaving a <trans-unit> (xliff v1.) or <unit> (xliff v2.)
254			if ( $this->tuTagName === $name ) {
255			$this->currentTransUnitIsTranslatable = null;
256			$this->inTU = false;
257			$this->unitContainsMda = false;
258			$this->hasWrittenCounts = false;
259
260			$this->resetCounts();
261			}
262			}
263
264			/**
265			* Update counts
266			*/
267			private function updateCounts() {
268
269			$seg = $this->getCurrentSegment();
270			if ( !empty( $seg ) ) {
271			$this->updateSegmentCounts( $seg );
272			}
273
274			}
275
276			/**
277			* @param bool $withMetadataTag
278			*
279			* @return string
280			*/
281			private function getWordCountGroupForXliffV2( bool $withMetadataTag = true ): string {
282
283			$this->mdaGroupCounter++;
284			$segments_count_array = $this->counts[ 'segments_count_array' ];
285
286			$tag = '';
287
288			if ( $withMetadataTag === true ) {
289			$tag .= '<mda:metadata>';
290			}
291
292			$index = 0;
293			foreach ( $segments_count_array as $segments_count_item ) {
294
295			$id = 'word_count_tu[' . $this->currentTransUnitId . '][' . $index . ']';
296			$index++;
297
298			$tag .= " <mda:metaGroup id=\"" . $id . "\" category=\"row_xml_attribute\">
299			<mda:meta type=\"x-matecat-raw\">" . $segments_count_item[ 'raw_word_count' ] . "</mda:meta>
300			<mda:meta type=\"x-matecat-weighted\">" . $segments_count_item[ 'eq_word_count' ] . "</mda:meta>
301			</mda:metaGroup>";
302			}
303
304			if ( $withMetadataTag === true ) {
305			$tag .= '</mda:metadata>';
306			}
307
308			return $tag;
309
310			}
311
312			/**
313			* prepare segment tagging for xliff insertion
314			*
315			* @param array $seg
316			*
317			* @return string
318			*/
319			protected function prepareTranslation( array $seg ): string {
320
321			$segment = Strings::removeDangerousChars( $seg [ 'segment' ] );
322			$translation = Strings::removeDangerousChars( $seg [ 'translation' ] );
323			$dataRefMap = ( isset( $seg[ 'data_ref_map' ] ) ) ? Strings::jsonToArray( $seg[ 'data_ref_map' ] ) : [];
324
325			if ( $seg [ 'translation' ] == '' ) {
326			$translation = $segment;
327			} else {
328			if ( $this->callback instanceof XliffReplacerCallbackInterface ) {
329			$error = ( !empty( $seg[ 'error' ] ) ) ? $seg[ 'error' ] : null;
330			if ( $this->callback->thereAreErrors( $seg[ 'sid' ], $segment, $translation, $dataRefMap, $error ) ) {
331			$translation = '\|\|\|UNTRANSLATED_CONTENT_START\|\|\|' . $segment . '\|\|\|UNTRANSLATED_CONTENT_END\|\|\|';
332			}
333			}
334			}
335
336			return $translation;
337
338			}
339
340			}

matecat / xliff-parser

Pull Request — master (#90)

Xliff20::updateCounts() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like