1<?php 2 3/** 4 * @copyright 2021 Greg Roach <greg@subaqua.co.uk> 5 * @license GPLv3+ 6 */ 7 8declare(strict_types=1); 9 10namespace Fisharebest\Webtrees\GedcomFilters; 11 12use Fisharebest\Webtrees\Encodings\EncodingInterface; 13use Fisharebest\Webtrees\Registry; 14use php_user_filter; 15 16use function stream_bucket_append; 17use function stream_bucket_make_writeable; 18use function stream_bucket_new; 19use function substr; 20 21use const PSFS_FEED_ME; 22use const PSFS_PASS_ON; 23 24/** 25 * Filter a GEDCOM data stream, converting to UTF-8. 26 * 27 * These properties are created after the class is instantiated. 28 * 29 * @property string $filtername 30 * @property array<string,string> $params 31 * @property resource $stream 32 */ 33class GedcomEncodingFilter extends php_user_filter 34{ 35 private string $data = ''; 36 37 private EncodingInterface|null $src_encoding = null; 38 39 private EncodingInterface|null $dst_encoding = null; 40 41 /** 42 * Initialization 43 */ 44 public function onCreate(): bool 45 { 46 parent::onCreate(); 47 48 $src_encoding = $this->params['src_encoding'] ?? ''; 49 $dst_encoding = $this->params['dst_encoding'] ?? 'UTF-8'; 50 51 if ($src_encoding !== '') { 52 $this->src_encoding = Registry::encodingFactory()->make($src_encoding); 53 } 54 55 $this->dst_encoding = Registry::encodingFactory()->make($dst_encoding); 56 57 return true; 58 } 59 60 /** 61 * Filter some data. 62 * 63 * @param resource $in Read from this input stream 64 * @param resource $out Write to this output stream 65 * @param int $consumed Count of bytes processed 66 * @param bool $closing Is the input about to end? 67 * 68 * @return int PSFS_PASS_ON / PSFS_FEED_ME / PSFS_ERR_FATAL 69 */ 70 public function filter($in, $out, &$consumed, $closing): int 71 { 72 $return = PSFS_FEED_ME; 73 74 // While input data is available, continue to read it. 75 while ($bucket_in = stream_bucket_make_writeable($in)) { 76 $this->data .= $bucket_in->data; 77 $consumed += $bucket_in->datalen; 78 79 $this->src_encoding ??= Registry::encodingFactory()->detect($this->data); 80 81 if ($this->src_encoding instanceof EncodingInterface) { 82 $bytes = $this->src_encoding->convertibleBytes($this->data); 83 $data_in = substr($this->data, 0, $bytes); 84 $data_out = $this->dst_encoding->fromUtf8($this->src_encoding->toUtf8($data_in)); 85 $bucket_out = stream_bucket_new($this->stream, $data_out); 86 $this->data = substr($this->data, $bytes); 87 $return = PSFS_PASS_ON; 88 stream_bucket_append($out, $bucket_out); 89 } 90 } 91 92 // Process the final record. 93 if ($closing && $this->data !== '') { 94 $this->src_encoding ??= Registry::encodingFactory()->make('UTF-8'); 95 $data_out = $this->dst_encoding->fromUtf8($this->src_encoding->toUtf8($this->data)); 96 $bucket_out = stream_bucket_new($this->stream, $data_out); 97 $return = PSFS_PASS_ON; 98 stream_bucket_append($out, $bucket_out); 99 } 100 101 return $return; 102 } 103} 104