Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 146 |
|
0.00% |
0 / 13 |
CRAP | |
0.00% |
0 / 1 |
| DiffHistoryBlob | |
0.00% |
0 / 146 |
|
0.00% |
0 / 13 |
1980 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| addItem | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| getItem | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| setText | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| getText | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| compress | |
0.00% |
0 / 49 |
|
0.00% |
0 / 1 |
132 | |||
| diff | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| patch | |
0.00% |
0 / 33 |
|
0.00% |
0 / 1 |
90 | |||
| xdiffAdler32 | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| uncompress | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
12 | |||
| __sleep | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
30 | |||
| __wakeup | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
30 | |||
| isHappy | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Efficient concatenated text storage. |
| 4 | * |
| 5 | * @license GPL-2.0-or-later |
| 6 | * @file |
| 7 | */ |
| 8 | |
| 9 | // NO_NAMESPACE Compatibility with serialized objects in permanent storage |
| 10 | |
| 11 | /** |
| 12 | * Diff-based history compression |
| 13 | * Requires xdiff and zlib |
| 14 | * |
| 15 | * WARNING: Objects of this class are serialized and permanently stored in the DB. |
| 16 | * Do not change the name or visibility of any property! |
| 17 | */ |
| 18 | class DiffHistoryBlob implements HistoryBlob { |
| 19 | /** @var string[] Uncompressed item cache */ |
| 20 | public $mItems = []; |
| 21 | |
| 22 | /** @var int Total uncompressed size */ |
| 23 | public $mSize = 0; |
| 24 | |
| 25 | /** |
| 26 | * @var array|null Array of diffs. If a diff D from A to B is notated D = B - A, |
| 27 | * and Z is an empty string: |
| 28 | * |
| 29 | * { item[map[i]] - item[map[i-1]] where i > 0 |
| 30 | * diff[i] = { |
| 31 | * { item[map[i]] - Z where i = 0 |
| 32 | */ |
| 33 | public $mDiffs; |
| 34 | |
| 35 | /** @var array The diff map, see above */ |
| 36 | public $mDiffMap; |
| 37 | |
| 38 | /** @var string|null The key for getText() |
| 39 | */ |
| 40 | public $mDefaultKey; |
| 41 | |
| 42 | /** @var string|null Compressed storage */ |
| 43 | public $mCompressed; |
| 44 | |
| 45 | /** @var bool True if the object is locked against further writes */ |
| 46 | public $mFrozen = false; |
| 47 | |
| 48 | /** |
| 49 | * @var int The maximum uncompressed size before the object becomes sad |
| 50 | * Should be less than max_allowed_packet |
| 51 | */ |
| 52 | public $mMaxSize = 10_000_000; |
| 53 | |
| 54 | /** @var int The maximum number of text items before the object becomes sad */ |
| 55 | public $mMaxCount = 100; |
| 56 | |
| 57 | /** Constants from xdiff.h */ |
| 58 | private const XDL_BDOP_INS = 1; |
| 59 | private const XDL_BDOP_CPY = 2; |
| 60 | private const XDL_BDOP_INSB = 3; |
| 61 | |
| 62 | public function __construct() { |
| 63 | if ( !function_exists( 'gzdeflate' ) ) { |
| 64 | throw new RuntimeException( "Need zlib support to read or write DiffHistoryBlob\n" ); |
| 65 | } |
| 66 | } |
| 67 | |
| 68 | /** |
| 69 | * @param string $text |
| 70 | * @return string |
| 71 | */ |
| 72 | public function addItem( $text ) { |
| 73 | if ( $this->mFrozen ) { |
| 74 | throw new BadMethodCallException( __METHOD__ . ": Cannot add more items after sleep/wakeup" ); |
| 75 | } |
| 76 | |
| 77 | $this->mItems[] = $text; |
| 78 | $this->mSize += strlen( $text ); |
| 79 | $this->mDiffs = null; // later |
| 80 | return (string)( count( $this->mItems ) - 1 ); |
| 81 | } |
| 82 | |
| 83 | /** |
| 84 | * @param string $key |
| 85 | * @return string |
| 86 | */ |
| 87 | public function getItem( $key ) { |
| 88 | return $this->mItems[(int)$key]; |
| 89 | } |
| 90 | |
| 91 | /** |
| 92 | * @param string $text |
| 93 | */ |
| 94 | public function setText( $text ) { |
| 95 | $this->mDefaultKey = $this->addItem( $text ); |
| 96 | } |
| 97 | |
| 98 | /** |
| 99 | * @return string |
| 100 | */ |
| 101 | public function getText() { |
| 102 | return $this->getItem( $this->mDefaultKey ); |
| 103 | } |
| 104 | |
| 105 | private function compress() { |
| 106 | if ( !function_exists( 'xdiff_string_rabdiff' ) ) { |
| 107 | throw new RuntimeException( "Need xdiff support to write DiffHistoryBlob\n" ); |
| 108 | } |
| 109 | if ( $this->mDiffs !== null ) { |
| 110 | // Already compressed |
| 111 | return; |
| 112 | } |
| 113 | if ( $this->mItems === [] ) { |
| 114 | return; |
| 115 | } |
| 116 | |
| 117 | // Create two diff sequences: one for main text and one for small text |
| 118 | $sequences = [ |
| 119 | 'small' => [ |
| 120 | 'tail' => '', |
| 121 | 'diffs' => [], |
| 122 | 'map' => [], |
| 123 | ], |
| 124 | 'main' => [ |
| 125 | 'tail' => '', |
| 126 | 'diffs' => [], |
| 127 | 'map' => [], |
| 128 | ], |
| 129 | ]; |
| 130 | $smallFactor = 0.5; |
| 131 | |
| 132 | $mItemsCount = count( $this->mItems ); |
| 133 | for ( $i = 0; $i < $mItemsCount; $i++ ) { |
| 134 | $text = $this->mItems[$i]; |
| 135 | if ( $i == 0 ) { |
| 136 | $seqName = 'main'; |
| 137 | } else { |
| 138 | $mainTail = $sequences['main']['tail']; |
| 139 | if ( strlen( $text ) < strlen( $mainTail ) * $smallFactor ) { |
| 140 | $seqName = 'small'; |
| 141 | } else { |
| 142 | $seqName = 'main'; |
| 143 | } |
| 144 | } |
| 145 | |
| 146 | $tail = $sequences[$seqName]['tail']; |
| 147 | $diff = $this->diff( $tail, $text ); |
| 148 | $sequences[$seqName]['diffs'][] = $diff; |
| 149 | $sequences[$seqName]['map'][] = $i; |
| 150 | $sequences[$seqName]['tail'] = $text; |
| 151 | } |
| 152 | |
| 153 | // Knit the sequences together |
| 154 | $tail = ''; |
| 155 | $this->mDiffs = []; |
| 156 | $this->mDiffMap = []; |
| 157 | foreach ( $sequences as $seq ) { |
| 158 | if ( $seq['diffs'] === [] ) { |
| 159 | continue; |
| 160 | } |
| 161 | if ( $tail === '' ) { |
| 162 | $this->mDiffs[] = $seq['diffs'][0]; |
| 163 | } else { |
| 164 | $head = $this->patch( '', $seq['diffs'][0] ); |
| 165 | $this->mDiffs[] = $this->diff( $tail, $head ); |
| 166 | } |
| 167 | $this->mDiffMap[] = $seq['map'][0]; |
| 168 | $diffsCount = count( $seq['diffs'] ); |
| 169 | for ( $i = 1; $i < $diffsCount; $i++ ) { |
| 170 | $this->mDiffs[] = $seq['diffs'][$i]; |
| 171 | $this->mDiffMap[] = $seq['map'][$i]; |
| 172 | } |
| 173 | $tail = $seq['tail']; |
| 174 | } |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * @param string $t1 |
| 179 | * @param string $t2 |
| 180 | * @return string |
| 181 | */ |
| 182 | private function diff( $t1, $t2 ) { |
| 183 | return xdiff_string_rabdiff( $t1, $t2 ); |
| 184 | } |
| 185 | |
| 186 | /** |
| 187 | * @param string $base |
| 188 | * @param string $diff |
| 189 | * @return bool|string |
| 190 | */ |
| 191 | private function patch( $base, $diff ) { |
| 192 | if ( function_exists( 'xdiff_string_bpatch' ) ) { |
| 193 | return xdiff_string_bpatch( $base, $diff ); |
| 194 | } |
| 195 | |
| 196 | # Pure PHP implementation |
| 197 | |
| 198 | $header = unpack( 'Vofp/Vcsize', substr( $diff, 0, 8 ) ); |
| 199 | |
| 200 | # Check the checksum |
| 201 | $ofp = $this->xdiffAdler32( $base ); |
| 202 | if ( $ofp !== substr( $diff, 0, 4 ) ) { |
| 203 | wfDebug( __METHOD__ . ": incorrect base checksum" ); |
| 204 | return false; |
| 205 | } |
| 206 | if ( $header['csize'] != strlen( $base ) ) { |
| 207 | wfDebug( __METHOD__ . ": incorrect base length" ); |
| 208 | return false; |
| 209 | } |
| 210 | |
| 211 | $p = 8; |
| 212 | $out = ''; |
| 213 | while ( $p < strlen( $diff ) ) { |
| 214 | $x = unpack( 'Cop', substr( $diff, $p, 1 ) ); |
| 215 | $op = $x['op']; |
| 216 | ++$p; |
| 217 | switch ( $op ) { |
| 218 | case self::XDL_BDOP_INS: |
| 219 | $x = unpack( 'Csize', substr( $diff, $p, 1 ) ); |
| 220 | $p++; |
| 221 | $out .= substr( $diff, $p, $x['size'] ); |
| 222 | $p += $x['size']; |
| 223 | break; |
| 224 | case self::XDL_BDOP_INSB: |
| 225 | $x = unpack( 'Vcsize', substr( $diff, $p, 4 ) ); |
| 226 | $p += 4; |
| 227 | $out .= substr( $diff, $p, $x['csize'] ); |
| 228 | $p += $x['csize']; |
| 229 | break; |
| 230 | case self::XDL_BDOP_CPY: |
| 231 | $x = unpack( 'Voff/Vcsize', substr( $diff, $p, 8 ) ); |
| 232 | $p += 8; |
| 233 | $out .= substr( $base, $x['off'], $x['csize'] ); |
| 234 | break; |
| 235 | default: |
| 236 | wfDebug( __METHOD__ . ": invalid op" ); |
| 237 | return false; |
| 238 | } |
| 239 | } |
| 240 | return $out; |
| 241 | } |
| 242 | |
| 243 | /** |
| 244 | * Compute a binary "Adler-32" checksum as defined by LibXDiff, i.e. with |
| 245 | * the bytes backwards and initialised with 0 instead of 1. See T36428. |
| 246 | * |
| 247 | * @param string $s |
| 248 | * @return string |
| 249 | */ |
| 250 | public function xdiffAdler32( $s ) { |
| 251 | static $init = null; |
| 252 | $init ??= str_repeat( "\xf0", 205 ) . "\xee" . str_repeat( "\xf0", 67 ) . "\x02"; |
| 253 | |
| 254 | // The real Adler-32 checksum of $init is zero, so it initialises the |
| 255 | // state to zero, as it is at the start of LibXDiff's checksum |
| 256 | // algorithm. Appending the subject string then simulates LibXDiff. |
| 257 | return strrev( hash( 'adler32', $init . $s, true ) ); |
| 258 | } |
| 259 | |
| 260 | private function uncompress() { |
| 261 | if ( !$this->mDiffs ) { |
| 262 | return; |
| 263 | } |
| 264 | $tail = ''; |
| 265 | $mDiffsCount = count( $this->mDiffs ); |
| 266 | for ( $diffKey = 0; $diffKey < $mDiffsCount; $diffKey++ ) { |
| 267 | $textKey = $this->mDiffMap[$diffKey]; |
| 268 | $text = $this->patch( $tail, $this->mDiffs[$diffKey] ); |
| 269 | $this->mItems[$textKey] = $text; |
| 270 | $tail = $text; |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | /** |
| 275 | * @return array |
| 276 | */ |
| 277 | public function __sleep() { |
| 278 | $this->compress(); |
| 279 | if ( $this->mItems === [] ) { |
| 280 | $info = false; |
| 281 | } else { |
| 282 | // Take forward differences to improve the compression ratio for sequences |
| 283 | $map = ''; |
| 284 | $prev = 0; |
| 285 | foreach ( $this->mDiffMap as $i ) { |
| 286 | if ( $map !== '' ) { |
| 287 | $map .= ','; |
| 288 | } |
| 289 | $map .= $i - $prev; |
| 290 | $prev = $i; |
| 291 | } |
| 292 | $info = [ |
| 293 | 'diffs' => $this->mDiffs, |
| 294 | 'map' => $map |
| 295 | ]; |
| 296 | } |
| 297 | if ( $this->mDefaultKey !== null ) { |
| 298 | $info['default'] = $this->mDefaultKey; |
| 299 | } |
| 300 | $this->mCompressed = gzdeflate( serialize( $info ) ); |
| 301 | return [ 'mCompressed' ]; |
| 302 | } |
| 303 | |
| 304 | public function __wakeup() { |
| 305 | // addItem() doesn't work if mItems is partially filled from mDiffs |
| 306 | $this->mFrozen = true; |
| 307 | $info = HistoryBlobUtils::unserializeArray( gzinflate( $this->mCompressed ) ); |
| 308 | $this->mCompressed = null; |
| 309 | |
| 310 | if ( !$info ) { |
| 311 | // Empty object |
| 312 | return; |
| 313 | } |
| 314 | |
| 315 | if ( isset( $info['default'] ) ) { |
| 316 | $this->mDefaultKey = $info['default']; |
| 317 | } |
| 318 | $this->mDiffs = $info['diffs']; |
| 319 | if ( isset( $info['base'] ) ) { |
| 320 | // Old format |
| 321 | $this->mDiffMap = range( 0, count( $this->mDiffs ) - 1 ); |
| 322 | array_unshift( $this->mDiffs, |
| 323 | pack( 'VVCV', 0, 0, self::XDL_BDOP_INSB, strlen( $info['base'] ) ) . |
| 324 | $info['base'] ); |
| 325 | } else { |
| 326 | // New format |
| 327 | $map = explode( ',', $info['map'] ); |
| 328 | $cur = 0; |
| 329 | $this->mDiffMap = []; |
| 330 | foreach ( $map as $i ) { |
| 331 | $cur += $i; |
| 332 | $this->mDiffMap[] = $cur; |
| 333 | } |
| 334 | } |
| 335 | $this->uncompress(); |
| 336 | } |
| 337 | |
| 338 | /** @inheritDoc */ |
| 339 | public function isHappy(): bool { |
| 340 | return $this->mSize < $this->mMaxSize |
| 341 | && count( $this->mItems ) < $this->mMaxCount; |
| 342 | } |
| 343 | |
| 344 | } |