1[![](https://github.com/toddr/Tie-File/workflows/linux/badge.svg)](https://github.com/toddr/Tie-File/actions) [![](https://github.com/toddr/Tie-File/workflows/macos/badge.svg)](https://github.com/toddr/Tie-File/actions) [![](https://github.com/toddr/Tie-File/workflows/windows/badge.svg)](https://github.com/toddr/Tie-File/actions) 2 3# NAME 4 5Tie::File - Access the lines of a disk file via a Perl array 6 7# SYNOPSIS 8 9 # This file documents Tie::File version 0.98 10 use Tie::File; 11 12 tie @array, 'Tie::File', filename or die ...; 13 14 $array[13] = 'blah'; # line 13 of the file is now 'blah' 15 print $array[42]; # display line 42 of the file 16 17 $n_recs = @array; # how many records are in the file? 18 $#array -= 2; # chop two records off the end 19 20 21 for (@array) { 22 s/PERL/Perl/g; # Replace PERL with Perl everywhere in the file 23 } 24 25 # These are just like regular push, pop, unshift, shift, and splice 26 # Except that they modify the file in the way you would expect 27 28 push @array, new recs...; 29 my $r1 = pop @array; 30 unshift @array, new recs...; 31 my $r2 = shift @array; 32 @old_recs = splice @array, 3, 7, new recs...; 33 34 untie @array; # all finished 35 36# DESCRIPTION 37 38`Tie::File` represents a regular text file as a Perl array. Each 39element in the array corresponds to a record in the file. The first 40line of the file is element 0 of the array; the second line is element 411, and so on. 42 43The file is _not_ loaded into memory, so this will work even for 44gigantic files. 45 46Changes to the array are reflected in the file immediately. 47 48Lazy people and beginners may now stop reading the manual. 49 50## `recsep` 51 52What is a 'record'? By default, the meaning is the same as for the 53`<...>` operator: It's a string terminated by `$/`, which is 54probably `"\n"`. (Minor exception: on DOS and Win32 systems, a 55'record' is a string terminated by `"\r\n"`.) You may change the 56definition of "record" by supplying the `recsep` option in the `tie` 57call: 58 59 tie @array, 'Tie::File', $file, recsep => 'es'; 60 61This says that records are delimited by the string `es`. If the file 62contained the following data: 63 64 Curse these pesky flies!\n 65 66then the `@array` would appear to have four elements: 67 68 "Curse th" 69 "e p" 70 "ky fli" 71 "!\n" 72 73An undefined value is not permitted as a record separator. Perl's 74special "paragraph mode" semantics (à la `$/ = ""`) are not 75emulated. 76 77Records read from the tied array do not have the record separator 78string on the end; this is to allow 79 80 $array[17] .= "extra"; 81 82to work as expected. 83 84(See ["autochomp"](#autochomp), below.) Records stored into the array will have 85the record separator string appended before they are written to the 86file, if they don't have one already. For example, if the record 87separator string is `"\n"`, then the following two lines do exactly 88the same thing: 89 90 $array[17] = "Cherry pie"; 91 $array[17] = "Cherry pie\n"; 92 93The result is that the contents of line 17 of the file will be 94replaced with "Cherry pie"; a newline character will separate line 17 95from line 18. This means that this code will do nothing: 96 97 chomp $array[17]; 98 99Because the `chomp`ed value will have the separator reattached when 100it is written back to the file. There is no way to create a file 101whose trailing record separator string is missing. 102 103Inserting records that _contain_ the record separator string is not 104supported by this module. It will probably produce a reasonable 105result, but what this result will be may change in a future version. 106Use 'splice' to insert records or to replace one record with several. 107 108## `autochomp` 109 110Normally, array elements have the record separator removed, so that if 111the file contains the text 112 113 Gold 114 Frankincense 115 Myrrh 116 117the tied array will appear to contain `("Gold", "Frankincense", 118"Myrrh")`. If you set `autochomp` to a false value, the record 119separator will not be removed. If the file above was tied with 120 121 tie @gifts, "Tie::File", $gifts, autochomp => 0; 122 123then the array `@gifts` would appear to contain `("Gold\n", 124"Frankincense\n", "Myrrh\n")`, or (on Win32 systems) `("Gold\r\n", 125"Frankincense\r\n", "Myrrh\r\n")`. 126 127## `mode` 128 129Normally, the specified file will be opened for read and write access, 130and will be created if it does not exist. (That is, the flags 131`O_RDWR | O_CREAT` are supplied in the `open` call.) If you want to 132change this, you may supply alternative flags in the `mode` option. 133See [Fcntl](https://metacpan.org/pod/Fcntl) for a listing of available flags. 134For example: 135 136 # open the file if it exists, but fail if it does not exist 137 use Fcntl 'O_RDWR'; 138 tie @array, 'Tie::File', $file, mode => O_RDWR; 139 140 # create the file if it does not exist 141 use Fcntl 'O_RDWR', 'O_CREAT'; 142 tie @array, 'Tie::File', $file, mode => O_RDWR | O_CREAT; 143 144 # open an existing file in read-only mode 145 use Fcntl 'O_RDONLY'; 146 tie @array, 'Tie::File', $file, mode => O_RDONLY; 147 148Opening the data file in write-only or append mode is not supported. 149 150## `memory` 151 152This is an upper limit on the amount of memory that `Tie::File` will 153consume at any time while managing the file. This is used for two 154things: managing the _read cache_ and managing the _deferred write 155buffer_. 156 157Records read in from the file are cached, to avoid having to re-read 158them repeatedly. If you read the same record twice, the first time it 159will be stored in memory, and the second time it will be fetched from 160the _read cache_. The amount of data in the read cache will not 161exceed the value you specified for `memory`. If `Tie::File` wants 162to cache a new record, but the read cache is full, it will make room 163by expiring the least-recently visited records from the read cache. 164 165The default memory limit is 2Mib. You can adjust the maximum read 166cache size by supplying the `memory` option. The argument is the 167desired cache size, in bytes. 168 169 # I have a lot of memory, so use a large cache to speed up access 170 tie @array, 'Tie::File', $file, memory => 20_000_000; 171 172Setting the memory limit to 0 will inhibit caching; records will be 173fetched from disk every time you examine them. 174 175The `memory` value is not an absolute or exact limit on the memory 176used. `Tie::File` objects contains some structures besides the read 177cache and the deferred write buffer, whose sizes are not charged 178against `memory`. 179 180The cache itself consumes about 310 bytes per cached record, so if 181your file has many short records, you may want to decrease the cache 182memory limit, or else the cache overhead may exceed the size of the 183cached data. 184 185## `dw_size` 186 187(This is an advanced feature. Skip this section on first reading.) 188 189If you use deferred writing (See ["Deferred Writing"](#deferred-writing), below) then 190data you write into the array will not be written directly to the 191file; instead, it will be saved in the _deferred write buffer_ to be 192written out later. Data in the deferred write buffer is also charged 193against the memory limit you set with the `memory` option. 194 195You may set the `dw_size` option to limit the amount of data that can 196be saved in the deferred write buffer. This limit may not exceed the 197total memory limit. For example, if you set `dw_size` to 1000 and 198`memory` to 2500, that means that no more than 1000 bytes of deferred 199writes will be saved up. The space available for the read cache will 200vary, but it will always be at least 1500 bytes (if the deferred write 201buffer is full) and it could grow as large as 2500 bytes (if the 202deferred write buffer is empty.) 203 204If you don't specify a `dw_size`, it defaults to the entire memory 205limit. 206 207## Option Format 208 209`-mode` is a synonym for `mode`. `-recsep` is a synonym for 210`recsep`. `-memory` is a synonym for `memory`. You get the 211idea. 212 213# Public Methods 214 215The `tie` call returns an object, say `$o`. You may call 216 217 $rec = $o->FETCH($n); 218 $o->STORE($n, $rec); 219 220to fetch or store the record at line `$n`, respectively; similarly 221the other tied array methods. (See [perltie](https://metacpan.org/pod/perltie) for details.) You may 222also call the following methods on this object: 223 224## `flock` 225 226 $o->flock(MODE) 227 228will lock the tied file. `MODE` has the same meaning as the second 229argument to the Perl built-in `flock` function; for example 230`LOCK_SH` or `LOCK_EX | LOCK_NB`. (These constants are provided by 231the `use Fcntl ':flock'` declaration.) 232 233`MODE` is optional; the default is `LOCK_EX`. 234 235`Tie::File` maintains an internal table of the byte offset of each 236record it has seen in the file. 237 238When you use `flock` to lock the file, `Tie::File` assumes that the 239read cache is no longer trustworthy, because another process might 240have modified the file since the last time it was read. Therefore, a 241successful call to `flock` discards the contents of the read cache 242and the internal record offset table. 243 244`Tie::File` promises that the following sequence of operations will 245be safe: 246 247 my $o = tie @array, "Tie::File", $filename; 248 $o->flock; 249 250In particular, `Tie::File` will _not_ read or write the file during 251the `tie` call. (Exception: Using `mode => O_TRUNC` will, of 252course, erase the file during the `tie` call. If you want to do this 253safely, then open the file without `O_TRUNC`, lock the file, and use 254`@array = ()`.) 255 256The best way to unlock a file is to discard the object and untie the 257array. It is probably unsafe to unlock the file without also untying 258it, because if you do, changes may remain unwritten inside the object. 259That is why there is no shortcut for unlocking. If you really want to 260unlock the file prematurely, you know what to do; if you don't know 261what to do, then don't do it. 262 263All the usual warnings about file locking apply here. In particular, 264note that file locking in Perl is **advisory**, which means that 265holding a lock will not prevent anyone else from reading, writing, or 266erasing the file; it only prevents them from getting another lock at 267the same time. Locks are analogous to green traffic lights: If you 268have a green light, that does not prevent the idiot coming the other 269way from plowing into you sideways; it merely guarantees to you that 270the idiot does not also have a green light at the same time. 271 272## `autochomp` 273 274 my $old_value = $o->autochomp(0); # disable autochomp option 275 my $old_value = $o->autochomp(1); # enable autochomp option 276 277 my $ac = $o->autochomp(); # recover current value 278 279See ["autochomp"](#autochomp), above. 280 281## `defer`, `flush`, `discard`, and `autodefer` 282 283See ["Deferred Writing"](#deferred-writing), below. 284 285## `offset` 286 287 $off = $o->offset($n); 288 289This method returns the byte offset of the start of the `$n`th record 290in the file. If there is no such record, it returns an undefined 291value. 292 293# Tying to an already-opened filehandle 294 295If `$fh` is a filehandle, such as is returned by `IO::File` or one 296of the other `IO` modules, you may use: 297 298 tie @array, 'Tie::File', $fh, ...; 299 300Similarly if you opened that handle `FH` with regular `open` or 301`sysopen`, you may use: 302 303 tie @array, 'Tie::File', \*FH, ...; 304 305Handles that were opened write-only won't work. Handles that were 306opened read-only will work as long as you don't try to modify the 307array. Handles must be attached to seekable sources of data---that 308means no pipes or sockets. If `Tie::File` can detect that you 309supplied a non-seekable handle, the `tie` call will throw an 310exception. (On Unix systems, it can detect this.) 311 312Note that Tie::File will only close any filehandles that it opened 313internally. If you passed it a filehandle as above, you "own" the 314filehandle, and are responsible for closing it after you have untied 315the @array. 316 317# Deferred Writing 318 319(This is an advanced feature. Skip this section on first reading.) 320 321Normally, modifying a `Tie::File` array writes to the underlying file 322immediately. Every assignment like `$a[3] = ...` rewrites as much of 323the file as is necessary; typically, everything from line 3 through 324the end will need to be rewritten. This is the simplest and most 325transparent behavior. Performance even for large files is reasonably 326good. 327 328However, under some circumstances, this behavior may be excessively 329slow. For example, suppose you have a million-record file, and you 330want to do: 331 332 for (@FILE) { 333 $_ = "> $_"; 334 } 335 336The first time through the loop, you will rewrite the entire file, 337from line 0 through the end. The second time through the loop, you 338will rewrite the entire file from line 1 through the end. The third 339time through the loop, you will rewrite the entire file from line 2 to 340the end. And so on. 341 342If the performance in such cases is unacceptable, you may defer the 343actual writing, and then have it done all at once. The following loop 344will perform much better for large files: 345 346 (tied @a)->defer; 347 for (@a) { 348 $_ = "> $_"; 349 } 350 (tied @a)->flush; 351 352If `Tie::File`'s memory limit is large enough, all the writing will 353done in memory. Then, when you call `->flush`, the entire file 354will be rewritten in a single pass. 355 356(Actually, the preceding discussion is something of a fib. You don't 357need to enable deferred writing to get good performance for this 358common case, because `Tie::File` will do it for you automatically 359unless you specifically tell it not to. See ["Autodeferring"](#autodeferring), 360below.) 361 362Calling `->flush` returns the array to immediate-write mode. If 363you wish to discard the deferred writes, you may call `->discard` 364instead of `->flush`. Note that in some cases, some of the data 365will have been written already, and it will be too late for 366`->discard` to discard all the changes. Support for 367`->discard` may be withdrawn in a future version of `Tie::File`. 368 369Deferred writes are cached in memory up to the limit specified by the 370`dw_size` option (see above). If the deferred-write buffer is full 371and you try to write still more deferred data, the buffer will be 372flushed. All buffered data will be written immediately, the buffer 373will be emptied, and the now-empty space will be used for future 374deferred writes. 375 376If the deferred-write buffer isn't yet full, but the total size of the 377buffer and the read cache would exceed the `memory` limit, the oldest 378records will be expired from the read cache until the total size is 379under the limit. 380 381`push`, `pop`, `shift`, `unshift`, and `splice` cannot be 382deferred. When you perform one of these operations, any deferred data 383is written to the file and the operation is performed immediately. 384This may change in a future version. 385 386If you resize the array with deferred writing enabled, the file will 387be resized immediately, but deferred records will not be written. 388This has a surprising consequence: `@a = (...)` erases the file 389immediately, but the writing of the actual data is deferred. This 390might be a bug. If it is a bug, it will be fixed in a future version. 391 392## Autodeferring 393 394`Tie::File` tries to guess when deferred writing might be helpful, 395and to turn it on and off automatically. 396 397 for (@a) { 398 $_ = "> $_"; 399 } 400 401In this example, only the first two assignments will be done 402immediately; after this, all the changes to the file will be deferred 403up to the user-specified memory limit. 404 405You should usually be able to ignore this and just use the module 406without thinking about deferring. However, special applications may 407require fine control over which writes are deferred, or may require 408that all writes be immediate. To disable the autodeferment feature, 409use 410 411 (tied @o)->autodefer(0); 412 413or 414 415 tie @array, 'Tie::File', $file, autodefer => 0; 416 417Similarly, `->autodefer(1)` re-enables autodeferment, and 418`->autodefer()` recovers the current value of the autodefer setting. 419 420# CONCURRENT ACCESS TO FILES 421 422Caching and deferred writing are inappropriate if you want the same 423file to be accessed simultaneously from more than one process. Other 424optimizations performed internally by this module are also 425incompatible with concurrent access. A future version of this module will 426support a `concurrent => 1` option that enables safe concurrent access. 427 428Previous versions of this documentation suggested using `memory 429=> 0` for safe concurrent access. This was mistaken. Tie::File 430will not support safe concurrent access before version 0.96. 431 432# CAVEATS 433 434(That's Latin for 'warnings'.) 435 436- Reasonable effort was made to make this module efficient. Nevertheless, 437changing the size of a record in the middle of a large file will 438always be fairly slow, because everything after the new record must be 439moved. 440- The behavior of tied arrays is not precisely the same as for regular 441arrays. For example: 442 443 # This DOES print "How unusual!" 444 undef $a[10]; print "How unusual!\n" if defined $a[10]; 445 446 `undef`-ing a `Tie::File` array element just blanks out the 447 corresponding record in the file. When you read it back again, you'll 448 get the empty string, so the supposedly-`undef`'ed value will be 449 defined. Similarly, if you have `autochomp` disabled, then 450 451 # This DOES print "How unusual!" if 'autochomp' is disabled 452 undef $a[10]; 453 print "How unusual!\n" if $a[10]; 454 455 Because when `autochomp` is disabled, `$a[10]` will read back as 456 `"\n"` (or whatever the record separator string is.) 457 458 There are other minor differences, particularly regarding `exists` 459 and `delete`, but in general, the correspondence is extremely close. 460 461- I have supposed that since this module is concerned with file I/O, 462almost all normal use of it will be heavily I/O bound. This means 463that the time to maintain complicated data structures inside the 464module will be dominated by the time to actually perform the I/O. 465When there was an opportunity to spend CPU time to avoid doing I/O, I 466usually tried to take it. 467- You might be tempted to think that deferred writing is like 468transactions, with `flush` as `commit` and `discard` as 469`rollback`, but it isn't, so don't. 470- There is a large memory overhead for each record offset and for each 471cache entry: about 310 bytes per cached data record, and about 21 bytes per offset table entry. 472 473 The per-record overhead will limit the maximum number of records you 474 can access per file. Note that _accessing_ the length of the array 475 via `$x = scalar @tied_file` accesses **all** records and stores their 476 offsets. The same for `foreach (@tied_file)`, even if you exit the 477 loop early. 478 479# SUBCLASSING 480 481This version promises absolutely nothing about the internals, which 482may change without notice. A future version of the module will have a 483well-defined and stable subclassing API. 484 485# WHAT ABOUT `DB_File`? 486 487People sometimes point out that [DB\_File](https://metacpan.org/pod/DB_File) will do something similar, 488and ask why `Tie::File` module is necessary. 489 490There are a number of reasons that you might prefer `Tie::File`. 491A list is available at `http://perl.plover.com/TieFile/why-not-DB_File`. 492 493# AUTHOR 494 495Mark Jason Dominus 496 497To contact the author, send email to: `mjd-perl-tiefile+@plover.com` 498 499To receive an announcement whenever a new version of this module is 500released, send a blank email message to 501`mjd-perl-tiefile-subscribe@plover.com`. 502 503The most recent version of this module, including documentation and 504any news of importance, will be available at 505 506 http://perl.plover.com/TieFile/ 507 508# LICENSE 509 510`Tie::File` version 0.96 is copyright (C) 2003 Mark Jason Dominus. 511 512This library is free software; you may redistribute it and/or modify 513it under the same terms as Perl itself. 514 515These terms are your choice of any of (1) the Perl Artistic Licence, 516or (2) version 2 of the GNU General Public License as published by the 517Free Software Foundation, or (3) any later version of the GNU General 518Public License. 519 520This library is distributed in the hope that it will be useful, 521but WITHOUT ANY WARRANTY; without even the implied warranty of 522MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 523GNU General Public License for more details. 524 525You should have received a copy of the GNU General Public License 526along with this library program; it should be in the file `COPYING`. 527If not, write to the Free Software Foundation, Inc., 51 Franklin Street, 528Fifth Floor, Boston, MA 02110-1301, USA 529 530For licensing inquiries, contact the author at: 531 532 Mark Jason Dominus 533 255 S. Warnock St. 534 Philadelphia, PA 19107 535 536# WARRANTY 537 538`Tie::File` version 0.98 comes with ABSOLUTELY NO WARRANTY. 539For details, see the license. 540 541# THANKS 542 543Gigantic thanks to Jarkko Hietaniemi, for agreeing to put this in the 544core when I hadn't written it yet, and for generally being helpful, 545supportive, and competent. (Usually the rule is "choose any one.") 546Also big thanks to Abhijit Menon-Sen for all of the same things. 547 548Special thanks to Craig Berry and Peter Prymmer (for VMS portability 549help), Randy Kobes (for Win32 portability help), Clinton Pierce and 550Autrijus Tang (for heroic eleventh-hour Win32 testing above and beyond 551the call of duty), Michael G Schwern (for testing advice), and the 552rest of the CPAN testers (for testing generally). 553 554Special thanks to Tels for suggesting several speed and memory 555optimizations. 556 557Additional thanks to: 558Edward Avis / 559Mattia Barbon / 560Tom Christiansen / 561Gerrit Haase / 562Gurusamy Sarathy / 563Jarkko Hietaniemi (again) / 564Nikola Knezevic / 565John Kominetz / 566Nick Ing-Simmons / 567Tassilo von Parseval / 568H. Dieter Pearcey / 569Slaven Rezic / 570Eric Roode / 571Peter Scott / 572Peter Somu / 573Autrijus Tang (again) / 574Tels (again) / 575Juerd Waalboer / 576Todd Rinaldo 577 578# TODO 579 580More tests. (Stuff I didn't think of yet.) 581 582Paragraph mode? 583 584Fixed-length mode. Leave-blanks mode. 585 586Maybe an autolocking mode? 587 588For many common uses of the module, the read cache is a liability. 589For example, a program that inserts a single record, or that scans the 590file once, will have a cache hit rate of zero. This suggests a major 591optimization: The cache should be initially disabled. Here's a hybrid 592approach: Initially, the cache is disabled, but the cache code 593maintains statistics about how high the hit rate would be \*if\* it were 594enabled. When it sees the hit rate get high enough, it enables 595itself. The STAT comments in this code are the beginning of an 596implementation of this. 597 598Record locking with fcntl()? Then the module might support an undo 599log and get real transactions. What a tour de force that would be. 600 601Keeping track of the highest cached record. This would allow reads-in-a-row 602to skip the cache lookup faster (if reading from 1..N with empty cache at 603start, the last cached value will be always N-1). 604 605More tests. 606