MediaWiki  master
copyFileBackend.php
Go to the documentation of this file.
1 <?php
25 
26 require_once __DIR__ . '/Maintenance.php';
27 
41  protected $statCache = null;
42 
43  public function __construct() {
44  parent::__construct();
45  $this->addDescription( 'Copy files in one backend to another.' );
46  $this->addOption( 'src', 'Backend containing the source files', true, true );
47  $this->addOption( 'dst', 'Backend where files should be copied to', true, true );
48  $this->addOption( 'containers', 'Pipe separated list of containers', true, true );
49  $this->addOption( 'subdir', 'Only do items in this child directory', false, true );
50  $this->addOption( 'ratefile', 'File to check periodically for batch size', false, true );
51  $this->addOption( 'prestat', 'Stat the destination files first (try to use listings)' );
52  $this->addOption( 'skiphash', 'Skip SHA-1 sync checks for files' );
53  $this->addOption( 'missingonly', 'Only copy files missing from destination listing' );
54  $this->addOption( 'syncviadelete', 'Delete destination files missing from source listing' );
55  $this->addOption( 'utf8only', 'Skip source files that do not have valid UTF-8 names' );
56  $this->setBatchSize( 50 );
57  }
58 
59  public function execute() {
60  $backendGroup = $this->getServiceContainer()->getFileBackendGroup();
61  $src = $backendGroup->get( $this->getOption( 'src' ) );
62  $dst = $backendGroup->get( $this->getOption( 'dst' ) );
63  $containers = explode( '|', $this->getOption( 'containers' ) );
64  $subDir = rtrim( $this->getOption( 'subdir', '' ), '/' );
65 
66  $rateFile = $this->getOption( 'ratefile' );
67 
68  foreach ( $containers as $container ) {
69  if ( $subDir != '' ) {
70  $backendRel = "$container/$subDir";
71  $this->output( "Doing container '$container', directory '$subDir'...\n" );
72  } else {
73  $backendRel = $container;
74  $this->output( "Doing container '$container'...\n" );
75  }
76 
77  if ( $this->hasOption( 'missingonly' ) ) {
78  $this->output( "\tBuilding list of missing files..." );
79  $srcPathsRel = $this->getListingDiffRel( $src, $dst, $backendRel );
80  $this->output( count( $srcPathsRel ) . " file(s) need to be copied.\n" );
81  } else {
82  $srcPathsRel = $src->getFileList( [
83  'dir' => $src->getRootStoragePath() . "/$backendRel",
84  'adviseStat' => true // avoid HEADs
85  ] );
86  if ( $srcPathsRel === null ) {
87  $this->fatalError( "Could not list files in $container." );
88  }
89  }
90 
91  if ( $this->getOption( 'prestat' ) && !$this->hasOption( 'missingonly' ) ) {
92  // Build the stat cache for the destination files
93  $this->output( "\tBuilding destination stat cache..." );
94  $dstPathsRel = $dst->getFileList( [
95  'dir' => $dst->getRootStoragePath() . "/$backendRel",
96  'adviseStat' => true // avoid HEADs
97  ] );
98  if ( $dstPathsRel === null ) {
99  $this->fatalError( "Could not list files in $container." );
100  }
101  $this->statCache = [];
102  foreach ( $dstPathsRel as $dstPathRel ) {
103  $path = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
104  $this->statCache[sha1( $path )] = $dst->getFileStat( [ 'src' => $path ] );
105  }
106  $this->output( "done [" . count( $this->statCache ) . " file(s)]\n" );
107  }
108 
109  $this->output( "\tCopying file(s)...\n" );
110  $count = 0;
111  $batchPaths = [];
112  foreach ( $srcPathsRel as $srcPathRel ) {
113  // Check up on the rate file periodically to adjust the concurrency
114  if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
115  $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
116  $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
117  }
118  $batchPaths[$srcPathRel] = 1; // remove duplicates
119  if ( count( $batchPaths ) >= $this->getBatchSize() ) {
120  $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
121  $batchPaths = []; // done
122  }
123  ++$count;
124  }
125  if ( count( $batchPaths ) ) { // left-overs
126  $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
127  }
128  $this->output( "\tCopied $count file(s).\n" );
129 
130  if ( $this->hasOption( 'syncviadelete' ) ) {
131  $this->output( "\tBuilding list of excess destination files..." );
132  $delPathsRel = $this->getListingDiffRel( $dst, $src, $backendRel );
133  $this->output( count( $delPathsRel ) . " file(s) need to be deleted.\n" );
134 
135  $this->output( "\tDeleting file(s)...\n" );
136  $count = 0;
137  $batchPaths = [];
138  foreach ( $delPathsRel as $delPathRel ) {
139  // Check up on the rate file periodically to adjust the concurrency
140  if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
141  $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
142  $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
143  }
144  $batchPaths[$delPathRel] = 1; // remove duplicates
145  if ( count( $batchPaths ) >= $this->getBatchSize() ) {
146  $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
147  $batchPaths = []; // done
148  }
149  ++$count;
150  }
151  if ( count( $batchPaths ) ) { // left-overs
152  $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
153  }
154 
155  $this->output( "\tDeleted $count file(s).\n" );
156  }
157 
158  if ( $subDir != '' ) {
159  $this->output( "Finished container '$container', directory '$subDir'.\n" );
160  } else {
161  $this->output( "Finished container '$container'.\n" );
162  }
163  }
164 
165  $this->output( "Done.\n" );
166  }
167 
174  protected function getListingDiffRel( FileBackend $src, FileBackend $dst, $backendRel ) {
175  $srcPathsRel = $src->getFileList( [
176  'dir' => $src->getRootStoragePath() . "/$backendRel" ] );
177  if ( $srcPathsRel === null ) {
178  $this->fatalError( "Could not list files in source container." );
179  }
180  $dstPathsRel = $dst->getFileList( [
181  'dir' => $dst->getRootStoragePath() . "/$backendRel" ] );
182  if ( $dstPathsRel === null ) {
183  $this->fatalError( "Could not list files in destination container." );
184  }
185  // Get the list of destination files
186  $relFilesDstSha1 = [];
187  foreach ( $dstPathsRel as $dstPathRel ) {
188  $relFilesDstSha1[sha1( $dstPathRel )] = 1;
189  }
190  unset( $dstPathsRel ); // free
191  // Get the list of missing files
192  $missingPathsRel = [];
193  foreach ( $srcPathsRel as $srcPathRel ) {
194  if ( !isset( $relFilesDstSha1[sha1( $srcPathRel )] ) ) {
195  $missingPathsRel[] = $srcPathRel;
196  }
197  }
198  unset( $srcPathsRel ); // free
199 
200  return $missingPathsRel;
201  }
202 
210  protected function copyFileBatch(
211  array $srcPathsRel, $backendRel, FileBackend $src, FileBackend $dst
212  ) {
213  $ops = [];
214  $fsFiles = [];
215  $copiedRel = []; // for output message
216  $domainId = $src->getDomainId();
217 
218  // Download the batch of source files into backend cache...
219  if ( $this->hasOption( 'missingonly' ) ) {
220  $srcPaths = [];
221  foreach ( $srcPathsRel as $srcPathRel ) {
222  $srcPaths[] = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
223  }
224  $t_start = microtime( true );
225  $fsFiles = $src->getLocalReferenceMulti( [ 'srcs' => $srcPaths, 'latest' => 1 ] );
226  $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
227  $this->output( "\n\tDownloaded these file(s) [{$elapsed_ms}ms]:\n\t" .
228  implode( "\n\t", $srcPaths ) . "\n\n" );
229  }
230 
231  // Determine what files need to be copied over...
232  foreach ( $srcPathsRel as $srcPathRel ) {
233  $srcPath = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
234  $dstPath = $dst->getRootStoragePath() . "/$backendRel/$srcPathRel";
235  if ( $this->hasOption( 'utf8only' ) && !mb_check_encoding( $srcPath, 'UTF-8' ) ) {
236  $this->error( "$domainId: Detected illegal (non-UTF8) path for $srcPath." );
237  continue;
238  } elseif ( !$this->hasOption( 'missingonly' )
239  && $this->filesAreSame( $src, $dst, $srcPath, $dstPath )
240  ) {
241  $this->output( "\tAlready have $srcPathRel.\n" );
242  continue; // assume already copied...
243  }
244  $fsFile = array_key_exists( $srcPath, $fsFiles )
245  ? $fsFiles[$srcPath]
246  : $src->getLocalReference( [ 'src' => $srcPath, 'latest' => 1 ] );
247  if ( !$fsFile ) {
248  $src->clearCache( [ $srcPath ] );
249  if ( $src->fileExists( [ 'src' => $srcPath, 'latest' => 1 ] ) === false ) {
250  $this->error( "$domainId: File '$srcPath' was listed but does not exist." );
251  } else {
252  $this->error( "$domainId: Could not get local copy of $srcPath." );
253  }
254  continue;
255  } elseif ( !$fsFile->exists() ) {
256  // FSFileBackends just return the path for getLocalReference() and paths with
257  // illegal slashes may get normalized to a different path. This can cause the
258  // local reference to not exist...skip these broken files.
259  $this->error( "$domainId: Detected possible illegal path for $srcPath." );
260  continue;
261  }
262  $fsFiles[] = $fsFile; // keep TempFSFile objects alive as needed
263  // Note: prepare() is usually fast for key/value backends
264  $status = $dst->prepare( [ 'dir' => dirname( $dstPath ), 'bypassReadOnly' => true ] );
265  if ( !$status->isOK() ) {
266  $this->error( Status::wrap( $status )->getMessage( false, false, 'en' )->text() );
267  $this->fatalError( "$domainId: Could not copy $srcPath to $dstPath." );
268  }
269  $ops[] = [ 'op' => 'store',
270  'src' => $fsFile->getPath(), 'dst' => $dstPath, 'overwrite' => true ];
271  $copiedRel[] = $srcPathRel;
272  }
273 
274  // Copy in the batch of source files...
275  $t_start = microtime( true );
276  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
277  if ( !$status->isOK() ) {
278  sleep( 10 ); // wait and retry copy again
279  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
280  }
281  $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
282  if ( !$status->isOK() ) {
283  $this->error( Status::wrap( $status )->getMessage( false, false, 'en' )->text() );
284  $this->fatalError( "$domainId: Could not copy file batch." );
285  } elseif ( count( $copiedRel ) ) {
286  $this->output( "\n\tCopied these file(s) [{$elapsed_ms}ms]:\n\t" .
287  implode( "\n\t", $copiedRel ) . "\n\n" );
288  }
289  }
290 
297  protected function delFileBatch(
298  array $dstPathsRel, $backendRel, FileBackend $dst
299  ) {
300  $ops = [];
301  $deletedRel = []; // for output message
302  $domainId = $dst->getDomainId();
303 
304  // Determine what files need to be copied over...
305  foreach ( $dstPathsRel as $dstPathRel ) {
306  $dstPath = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
307  $ops[] = [ 'op' => 'delete', 'src' => $dstPath ];
308  $deletedRel[] = $dstPathRel;
309  }
310 
311  // Delete the batch of source files...
312  $t_start = microtime( true );
313  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
314  if ( !$status->isOK() ) {
315  sleep( 10 ); // wait and retry copy again
316  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
317  }
318  $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
319  if ( !$status->isOK() ) {
320  $this->error( Status::wrap( $status )->getMessage( false, false, 'en' )->text() );
321  $this->fatalError( "$domainId: Could not delete file batch." );
322  } elseif ( count( $deletedRel ) ) {
323  $this->output( "\n\tDeleted these file(s) [{$elapsed_ms}ms]:\n\t" .
324  implode( "\n\t", $deletedRel ) . "\n\n" );
325  }
326  }
327 
335  protected function filesAreSame( FileBackend $src, FileBackend $dst, $sPath, $dPath ) {
336  $skipHash = $this->hasOption( 'skiphash' );
337  $srcStat = $src->getFileStat( [ 'src' => $sPath ] );
338  $dPathSha1 = sha1( $dPath );
339  if ( $this->statCache !== null ) {
340  // All dst files are already in stat cache
341  $dstStat = $this->statCache[$dPathSha1] ?? false;
342  } else {
343  $dstStat = $dst->getFileStat( [ 'src' => $dPath ] );
344  }
345  // Initial fast checks to see if files are obviously different
346  $sameFast = (
347  is_array( $srcStat )
348  && is_array( $dstStat ) // dest exists
349  && $srcStat['size'] === $dstStat['size']
350  );
351  // More thorough checks against files
352  if ( !$sameFast ) {
353  $same = false; // no need to look farther
354  } elseif ( isset( $srcStat['md5'] ) && isset( $dstStat['md5'] ) ) {
355  // If MD5 was already in the stat info, just use it.
356  // This is useful as many objects stores can return this in object listing,
357  // so we can use it to avoid slow per-file HEADs.
358  $same = ( $srcStat['md5'] === $dstStat['md5'] );
359  } elseif ( $skipHash ) {
360  // This mode is good for copying to a backup location or resyncing clone
361  // backends in FileBackendMultiWrite (since they get writes second, they have
362  // higher timestamps). However, when copying the other way, this hits loads of
363  // false positives (possibly 100%) and wastes a bunch of time on GETs/PUTs.
364  // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
365  $same = ( $srcStat['mtime'] <= $dstStat['mtime'] );
366  } else {
367  // This is the slowest method which does many per-file HEADs (unless an object
368  // store tracks SHA-1 in listings).
369  $same = ( $src->getFileSha1Base36( [ 'src' => $sPath, 'latest' => 1 ] )
370  === $dst->getFileSha1Base36( [ 'src' => $dPath, 'latest' => 1 ] ) );
371  }
372 
373  return $same;
374  }
375 }
376 
377 $maintClass = CopyFileBackend::class;
378 require_once RUN_MAINTENANCE_IF_MAIN;
Copy all files in one container of one backend to another.
__construct()
Default constructor.
filesAreSame(FileBackend $src, FileBackend $dst, $sPath, $dPath)
array null $statCache
(path sha1 => stat) Pre-computed dst stat entries from listings
delFileBatch(array $dstPathsRel, $backendRel, FileBackend $dst)
execute()
Do the actual work.
getListingDiffRel(FileBackend $src, FileBackend $dst, $backendRel)
copyFileBatch(array $srcPathsRel, $backendRel, FileBackend $src, FileBackend $dst)
Base class for all file backend classes (including multi-write backends).
Definition: FileBackend.php:99
getFileList(array $params)
Get an iterator to list all stored files under a storage directory.
getDomainId()
Get the domain identifier used for this backend (possibly empty).
getFileSha1Base36(array $params)
Get a SHA-1 hash of the content of the file at a storage path in the backend.
clearCache(array $paths=null)
Invalidate any in-process file stat and property cache.
getFileStat(array $params)
Get quick information about a file at a storage path in the backend.
fileExists(array $params)
Check if a file exists at a storage path in the backend.
getRootStoragePath()
Get the root storage path of this backend.
prepare(array $params)
Prepare a storage directory for usage.
doQuickOperations(array $ops, array $opts=[])
Perform a set of independent file operations on some files.
getLocalReferenceMulti(array $params)
Like getLocalReference() except it takes an array of storage paths and yields an order-preserved map ...
getLocalReference(array $params)
Returns a file system file, identical in content to the file at a storage path.
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:66
error( $err, $die=0)
Throw an error to the user.
output( $out, $channel=null)
Throw some output to the user.
hasOption( $name)
Checks to see if a particular option was set.
getServiceContainer()
Returns the main service container.
getBatchSize()
Returns batch size.
addDescription( $text)
Set the description text.
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
getOption( $name, $default=null)
Get an option, or return the default.
setBatchSize( $s=0)
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
Generic operation result class Has warning/error list, boolean status and arbitrary value.
Definition: Status.php:58
$maintClass
return true
Definition: router.php:90