Code Coverage for /workspace/src/extensions/Wikispeech/includes/Segment/Segmenter.php

1	<?php
2
3	namespace MediaWiki\Wikispeech\Segment;
4
5	/**
6	* @file
7	* @ingroup Extensions
8	* @license GPL-2.0-or-later
9	*/
10
11	/**
12	* Used for dividing text into segments, that can then be sent to the
13	* Speechoid service. Also calculates values for variables that are needed
14	* for highlighting.
15	*
16	* @since 0.1.10
17	*/
18	abstract class Segmenter {
19
20	/**
21	* Divide a cleaned content array into segments, one for each sentence.
22	*
23	* @since 0.1.10
24	* @param SegmentContent[] $cleanedContent An array of items returned by `Cleaner::cleanHtml()`.
25	* @return Segment[] An array of segments, each containing the `CleanedText's in that segment.
26	*/
27	abstract public function segmentSentences( array $cleanedContent ): array;
28
29	/**
30	* Used to evaluate hash of segments, the primary key for stored utterances.
31	*
32	* @since 0.1.10
33	* @param Segment $segment The segment to be evaluated.
34	* @return string SHA256 message digest
35	*/
36	public function evaluateHash( Segment $segment ): string {
37	$context = hash_init( 'sha256' );
38	foreach ( $segment->getContent() as $part ) {
39	hash_update( $context, $part->getString() );
40	hash_update( $context, "\n" );
41	}
42	return hash_final( $context );
43	// Uncommenting below block can be useful during creation of
44	// new test cases as you might need to figure out hashes.
45	//LoggerFactory::getInstance( 'Segmenter' )
46	// ->info( __METHOD__ . ': {segement} : {hash}', [
47	// 'segment' => $segment,
48	// 'hash' => $hash
49	// ] );
50	}
51
52	}