MediaWiki  master
copyFileBackend.php
Go to the documentation of this file.
1 <?php
24 require_once __DIR__ . '/Maintenance.php';
25 
39  protected $statCache = null;
40 
41  public function __construct() {
42  parent::__construct();
43  $this->addDescription( 'Copy files in one backend to another.' );
44  $this->addOption( 'src', 'Backend containing the source files', true, true );
45  $this->addOption( 'dst', 'Backend where files should be copied to', true, true );
46  $this->addOption( 'containers', 'Pipe separated list of containers', true, true );
47  $this->addOption( 'subdir', 'Only do items in this child directory', false, true );
48  $this->addOption( 'ratefile', 'File to check periodically for batch size', false, true );
49  $this->addOption( 'prestat', 'Stat the destination files first (try to use listings)' );
50  $this->addOption( 'skiphash', 'Skip SHA-1 sync checks for files' );
51  $this->addOption( 'missingonly', 'Only copy files missing from destination listing' );
52  $this->addOption( 'syncviadelete', 'Delete destination files missing from source listing' );
53  $this->addOption( 'utf8only', 'Skip source files that do not have valid UTF-8 names' );
54  $this->setBatchSize( 50 );
55  }
56 
57  public function execute() {
58  $src = FileBackendGroup::singleton()->get( $this->getOption( 'src' ) );
59  $dst = FileBackendGroup::singleton()->get( $this->getOption( 'dst' ) );
60  $containers = explode( '|', $this->getOption( 'containers' ) );
61  $subDir = rtrim( $this->getOption( 'subdir', '' ), '/' );
62 
63  $rateFile = $this->getOption( 'ratefile' );
64 
65  foreach ( $containers as $container ) {
66  if ( $subDir != '' ) {
67  $backendRel = "$container/$subDir";
68  $this->output( "Doing container '$container', directory '$subDir'...\n" );
69  } else {
70  $backendRel = $container;
71  $this->output( "Doing container '$container'...\n" );
72  }
73 
74  if ( $this->hasOption( 'missingonly' ) ) {
75  $this->output( "\tBuilding list of missing files..." );
76  $srcPathsRel = $this->getListingDiffRel( $src, $dst, $backendRel );
77  $this->output( count( $srcPathsRel ) . " file(s) need to be copied.\n" );
78  } else {
79  $srcPathsRel = $src->getFileList( [
80  'dir' => $src->getRootStoragePath() . "/$backendRel",
81  'adviseStat' => true // avoid HEADs
82  ] );
83  if ( $srcPathsRel === null ) {
84  $this->fatalError( "Could not list files in $container." );
85  }
86  }
87 
88  if ( $this->getOption( 'prestat' ) && !$this->hasOption( 'missingonly' ) ) {
89  // Build the stat cache for the destination files
90  $this->output( "\tBuilding destination stat cache..." );
91  $dstPathsRel = $dst->getFileList( [
92  'dir' => $dst->getRootStoragePath() . "/$backendRel",
93  'adviseStat' => true // avoid HEADs
94  ] );
95  if ( $dstPathsRel === null ) {
96  $this->fatalError( "Could not list files in $container." );
97  }
98  $this->statCache = [];
99  foreach ( $dstPathsRel as $dstPathRel ) {
100  $path = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
101  $this->statCache[sha1( $path )] = $dst->getFileStat( [ 'src' => $path ] );
102  }
103  $this->output( "done [" . count( $this->statCache ) . " file(s)]\n" );
104  }
105 
106  $this->output( "\tCopying file(s)...\n" );
107  $count = 0;
108  $batchPaths = [];
109  foreach ( $srcPathsRel as $srcPathRel ) {
110  // Check up on the rate file periodically to adjust the concurrency
111  if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
112  $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
113  $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
114  }
115  $batchPaths[$srcPathRel] = 1; // remove duplicates
116  if ( count( $batchPaths ) >= $this->getBatchSize() ) {
117  $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
118  $batchPaths = []; // done
119  }
120  ++$count;
121  }
122  if ( count( $batchPaths ) ) { // left-overs
123  $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
124  }
125  $this->output( "\tCopied $count file(s).\n" );
126 
127  if ( $this->hasOption( 'syncviadelete' ) ) {
128  $this->output( "\tBuilding list of excess destination files..." );
129  $delPathsRel = $this->getListingDiffRel( $dst, $src, $backendRel );
130  $this->output( count( $delPathsRel ) . " file(s) need to be deleted.\n" );
131 
132  $this->output( "\tDeleting file(s)...\n" );
133  $count = 0;
134  $batchPaths = [];
135  foreach ( $delPathsRel as $delPathRel ) {
136  // Check up on the rate file periodically to adjust the concurrency
137  if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
138  $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
139  $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
140  }
141  $batchPaths[$delPathRel] = 1; // remove duplicates
142  if ( count( $batchPaths ) >= $this->getBatchSize() ) {
143  $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
144  $batchPaths = []; // done
145  }
146  ++$count;
147  }
148  if ( count( $batchPaths ) ) { // left-overs
149  $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
150  }
151 
152  $this->output( "\tDeleted $count file(s).\n" );
153  }
154 
155  if ( $subDir != '' ) {
156  $this->output( "Finished container '$container', directory '$subDir'.\n" );
157  } else {
158  $this->output( "Finished container '$container'.\n" );
159  }
160  }
161 
162  $this->output( "Done.\n" );
163  }
164 
171  protected function getListingDiffRel( FileBackend $src, FileBackend $dst, $backendRel ) {
172  $srcPathsRel = $src->getFileList( [
173  'dir' => $src->getRootStoragePath() . "/$backendRel" ] );
174  if ( $srcPathsRel === null ) {
175  $this->fatalError( "Could not list files in source container." );
176  }
177  $dstPathsRel = $dst->getFileList( [
178  'dir' => $dst->getRootStoragePath() . "/$backendRel" ] );
179  if ( $dstPathsRel === null ) {
180  $this->fatalError( "Could not list files in destination container." );
181  }
182  // Get the list of destination files
183  $relFilesDstSha1 = [];
184  foreach ( $dstPathsRel as $dstPathRel ) {
185  $relFilesDstSha1[sha1( $dstPathRel )] = 1;
186  }
187  unset( $dstPathsRel ); // free
188  // Get the list of missing files
189  $missingPathsRel = [];
190  foreach ( $srcPathsRel as $srcPathRel ) {
191  if ( !isset( $relFilesDstSha1[sha1( $srcPathRel )] ) ) {
192  $missingPathsRel[] = $srcPathRel;
193  }
194  }
195  unset( $srcPathsRel ); // free
196 
197  return $missingPathsRel;
198  }
199 
207  protected function copyFileBatch(
208  array $srcPathsRel, $backendRel, FileBackend $src, FileBackend $dst
209  ) {
210  $ops = [];
211  $fsFiles = [];
212  $copiedRel = []; // for output message
213  $domainId = $src->getDomainId();
214 
215  // Download the batch of source files into backend cache...
216  if ( $this->hasOption( 'missingonly' ) ) {
217  $srcPaths = [];
218  foreach ( $srcPathsRel as $srcPathRel ) {
219  $srcPaths[] = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
220  }
221  $t_start = microtime( true );
222  $fsFiles = $src->getLocalReferenceMulti( [ 'srcs' => $srcPaths, 'latest' => 1 ] );
223  $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
224  $this->output( "\n\tDownloaded these file(s) [{$elapsed_ms}ms]:\n\t" .
225  implode( "\n\t", $srcPaths ) . "\n\n" );
226  }
227 
228  // Determine what files need to be copied over...
229  foreach ( $srcPathsRel as $srcPathRel ) {
230  $srcPath = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
231  $dstPath = $dst->getRootStoragePath() . "/$backendRel/$srcPathRel";
232  if ( $this->hasOption( 'utf8only' ) && !mb_check_encoding( $srcPath, 'UTF-8' ) ) {
233  $this->error( "$domainId: Detected illegal (non-UTF8) path for $srcPath." );
234  continue;
235  } elseif ( !$this->hasOption( 'missingonly' )
236  && $this->filesAreSame( $src, $dst, $srcPath, $dstPath )
237  ) {
238  $this->output( "\tAlready have $srcPathRel.\n" );
239  continue; // assume already copied...
240  }
241  $fsFile = array_key_exists( $srcPath, $fsFiles )
242  ? $fsFiles[$srcPath]
243  : $src->getLocalReference( [ 'src' => $srcPath, 'latest' => 1 ] );
244  if ( !$fsFile ) {
245  $src->clearCache( [ $srcPath ] );
246  if ( $src->fileExists( [ 'src' => $srcPath, 'latest' => 1 ] ) === false ) {
247  $this->error( "$domainId: File '$srcPath' was listed but does not exist." );
248  } else {
249  $this->error( "$domainId: Could not get local copy of $srcPath." );
250  }
251  continue;
252  } elseif ( !$fsFile->exists() ) {
253  // FSFileBackends just return the path for getLocalReference() and paths with
254  // illegal slashes may get normalized to a different path. This can cause the
255  // local reference to not exist...skip these broken files.
256  $this->error( "$domainId: Detected possible illegal path for $srcPath." );
257  continue;
258  }
259  $fsFiles[] = $fsFile; // keep TempFSFile objects alive as needed
260  // Note: prepare() is usually fast for key/value backends
261  $status = $dst->prepare( [ 'dir' => dirname( $dstPath ), 'bypassReadOnly' => 1 ] );
262  if ( !$status->isOK() ) {
263  $this->error( Status::wrap( $status )->getMessage( false, false, 'en' )->text() );
264  $this->fatalError( "$domainId: Could not copy $srcPath to $dstPath." );
265  }
266  $ops[] = [ 'op' => 'store',
267  'src' => $fsFile->getPath(), 'dst' => $dstPath, 'overwrite' => 1 ];
268  $copiedRel[] = $srcPathRel;
269  }
270 
271  // Copy in the batch of source files...
272  $t_start = microtime( true );
273  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
274  if ( !$status->isOK() ) {
275  sleep( 10 ); // wait and retry copy again
276  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
277  }
278  $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
279  if ( !$status->isOK() ) {
280  $this->error( Status::wrap( $status )->getMessage( false, false, 'en' )->text() );
281  $this->fatalError( "$domainId: Could not copy file batch." );
282  } elseif ( count( $copiedRel ) ) {
283  $this->output( "\n\tCopied these file(s) [{$elapsed_ms}ms]:\n\t" .
284  implode( "\n\t", $copiedRel ) . "\n\n" );
285  }
286  }
287 
294  protected function delFileBatch(
295  array $dstPathsRel, $backendRel, FileBackend $dst
296  ) {
297  $ops = [];
298  $deletedRel = []; // for output message
299  $domainId = $dst->getDomainId();
300 
301  // Determine what files need to be copied over...
302  foreach ( $dstPathsRel as $dstPathRel ) {
303  $dstPath = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
304  $ops[] = [ 'op' => 'delete', 'src' => $dstPath ];
305  $deletedRel[] = $dstPathRel;
306  }
307 
308  // Delete the batch of source files...
309  $t_start = microtime( true );
310  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
311  if ( !$status->isOK() ) {
312  sleep( 10 ); // wait and retry copy again
313  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
314  }
315  $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
316  if ( !$status->isOK() ) {
317  $this->error( Status::wrap( $status )->getMessage( false, false, 'en' )->text() );
318  $this->fatalError( "$domainId: Could not delete file batch." );
319  } elseif ( count( $deletedRel ) ) {
320  $this->output( "\n\tDeleted these file(s) [{$elapsed_ms}ms]:\n\t" .
321  implode( "\n\t", $deletedRel ) . "\n\n" );
322  }
323  }
324 
332  protected function filesAreSame( FileBackend $src, FileBackend $dst, $sPath, $dPath ) {
333  $skipHash = $this->hasOption( 'skiphash' );
334  $srcStat = $src->getFileStat( [ 'src' => $sPath ] );
335  $dPathSha1 = sha1( $dPath );
336  if ( $this->statCache !== null ) {
337  // All dst files are already in stat cache
338  $dstStat = $this->statCache[$dPathSha1] ?? false;
339  } else {
340  $dstStat = $dst->getFileStat( [ 'src' => $dPath ] );
341  }
342  // Initial fast checks to see if files are obviously different
343  $sameFast = (
344  is_array( $srcStat ) // sanity check that source exists
345  && is_array( $dstStat ) // dest exists
346  && $srcStat['size'] === $dstStat['size']
347  );
348  // More thorough checks against files
349  if ( !$sameFast ) {
350  $same = false; // no need to look farther
351  } elseif ( isset( $srcStat['md5'] ) && isset( $dstStat['md5'] ) ) {
352  // If MD5 was already in the stat info, just use it.
353  // This is useful as many objects stores can return this in object listing,
354  // so we can use it to avoid slow per-file HEADs.
355  $same = ( $srcStat['md5'] === $dstStat['md5'] );
356  } elseif ( $skipHash ) {
357  // This mode is good for copying to a backup location or resyncing clone
358  // backends in FileBackendMultiWrite (since they get writes second, they have
359  // higher timestamps). However, when copying the other way, this hits loads of
360  // false positives (possibly 100%) and wastes a bunch of time on GETs/PUTs.
361  // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
362  $same = ( $srcStat['mtime'] <= $dstStat['mtime'] );
363  } else {
364  // This is the slowest method which does many per-file HEADs (unless an object
365  // store tracks SHA-1 in listings).
366  $same = ( $src->getFileSha1Base36( [ 'src' => $sPath, 'latest' => 1 ] )
367  === $dst->getFileSha1Base36( [ 'src' => $dPath, 'latest' => 1 ] ) );
368  }
369 
370  return $same;
371  }
372 }
373 
374 $maintClass = CopyFileBackend::class;
375 require_once RUN_MAINTENANCE_IF_MAIN;
Copy all files in one container of one backend to another.
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:39
getFileStat(array $params)
Get quick information about a file at a storage path in the backend.
error( $err, $die=0)
Throw an error to the user.
getOption( $name, $default=null)
Get an option, or return the default.
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:86
setBatchSize( $s=0)
Set the batch size.
getDomainId()
Get the domain identifier used for this backend (possibly empty).
hasOption( $name)
Checks to see if a particular option exists.
getLocalReferenceMulti(array $params)
Like getLocalReference() except it takes an array of storage paths and yields an order-preserved map ...
addDescription( $text)
Set the description text.
fileExists(array $params)
Check if a file exists at a storage path in the backend.
array null $statCache
(path sha1 => stat) Pre-computed dst stat entries from listings
getFileSha1Base36(array $params)
Get a SHA-1 hash of the content of the file at a storage path in the backend.
output( $out, $channel=null)
Throw some output to the user.
doQuickOperations(array $ops, array $opts=[])
Perform a set of independent file operations on some files.
prepare(array $params)
Prepare a storage directory for usage.
getListingDiffRel(FileBackend $src, FileBackend $dst, $backendRel)
static wrap( $sv)
Succinct helper method to wrap a StatusValue.
Definition: Status.php:55
getRootStoragePath()
Get the root storage path of this backend.
getBatchSize()
Returns batch size.
Base class for all file backend classes (including multi-write backends).
Definition: FileBackend.php:94
getFileList(array $params)
Get an iterator to list all stored files under a storage directory.
filesAreSame(FileBackend $src, FileBackend $dst, $sPath, $dPath)
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
$maintClass
clearCache(array $paths=null)
Invalidate any in-process file stat and property cache.
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
copyFileBatch(array $srcPathsRel, $backendRel, FileBackend $src, FileBackend $dst)
delFileBatch(array $dstPathsRel, $backendRel, FileBackend $dst)
getLocalReference(array $params)
Returns a file system file, identical in content to the file at a storage path.