NAME File::Data - interface to file data DESCRIPTION Wraps all the accessing of a file into a convenient set of calls for reading and writing data, including a simple regex interface. SYNOPSIS use strict; use File::Data; my $o_dat = File::Data->new('./jabber'); $o_dat->write(" Bewxre the Jabberwock my son,\n"); $o_dat->prepend("The Jxbberwock by Lewis Cxrroll:\n"); $o_dat->append(" the claws thxt snxtch,\n ...\n"); $o_dat->insert(2, " the jaws which bite.\n"); $o_dat->replace('x', 'a'); print $o_dat->SEARCH('The.+\n')->REPLACE("The.+\n", '')->return('search'); print $o_dat->read(); Or, perhaps more seriously :-} my $o_sgm = File::Data->new('./sgmlfile'); print "new SGML data: ".$o_sgm->replace( '\<\s*((?i)tag)\s*\>\s*((?s).*)\s*\<\s*((?i)\s*\/\s*tag)\s*\>', qq|key="val"|, ) if $o_sgm; See the METHODS manpage and the EXAMPLES manpage. EXPLANATION The idea is to standardise accessing of files for repetitive and straight forward tasks, and remove the repeated and therefore error prone file access I have seen in many sites, where varying, (with equivalently varying success), methods are used to achieve essentially the same result - a simple search and replace and/or a regex match. Approaches to opening and working with files vary so much, where one person may wish to know if a file exists, another wishes to know whether the target is a file, or if it is readable, or writable and so on. Sometimes, in production code even (horror), file's are opened without any checks of whether the open was succesful. Then there's a loop through each line to find the first or many patterns to read and/or replace. With a failure, normally the only message is 'permission denied', is that read or write access, does the file even exist? etc. This module attempts to provide a plain/generic interface to accessing a file's data. This will not suit every situation, but I have included some examples which will hopefully demonstrate that it may be used in situations where people would normally go through the same procedure for the umpteenth time to get at the same data. Theoretically you can mix and match your read and writes so long as you don't open read-only. my $o_dat = File::Data->new($file); my @partial = $o_dat->search($pattern); my $i_cnt = $o_dat->replace($search, $replace); One last thing - I'm sure this could be made much more efficient, and I'll be very interested to try and incorporate any suggestions to that effect. Note though that the intention has been to create a simple moderately consistent interface, rather than a complicated one. Sometimes it's better to roll your own, and sometimes you don't have to reinvent the wheel - TMTOWTDI. METHODS new Create a new File::Data object (default read-write). my $o_rw = File::Data->new($filename); # read-write my $o_ro = File::Data->new($filename, 'ro'); # read-only Note that if you open a file read-only and then attempt to write to it, that will be regarded as an error, even if you change the permissions in the meantime. Each file should have it's own discrete object. Look in the EXAMPLES manpage for a more complete explanation of possible arguments to the new() method read Read all data from file my @data = $o_dat->read; write Write data to file my @written = $o_dat->write; prepend Prepend to file my @prepended = $o_dat->prepend(\@lines); insert Insert data at line number, starting from '0' my @inserted = $o_dat->insert($i_lineno, \@lines); append Append to file my @appended = $o_dat->append(\@lines); search Retrieve data out of a file, simple list of all matches found are returned. Note - you must use capturing parentheses for this to work! my @addrs = $o_dat->search('/^(.*\@.*)$/'); my @names = $o_dat->search('/^(?:[^:]:){4}([^:]+):/'); replace Replace data in a 'search and replace' manner, returns the final data. my @data = $o_dat->replace($search, $replace); my @data = $o_dat->replace( q|\ q|'my.sales.com'|, ); This is simple, in that you can do almost anything in the search side, but the replace side is a bit more restricted, as we can't effect the replacement modifiers on the fly. If you really need this, perhaps (?{}) can help? do Simple wrapper for method calls, returning the object, so that you can chain them. my $o_dat = $o_dat->do('insert', @insertargs)->do(\'append', @appendargs)->do('read'); An addendum to this method, and to make life generally easier, is that you can also call any of the above methods in uppercase, to call via do() eg; my @data = $o_dat->WRITE($this)->APPEND->($that)->read; First argument is the method to call, followed by the arguments that method expects. perl -MFile::Data -e "print File::Data->new($file)->INSERT(3, \"third line\n\")->read"; If you want to get at the output of a particular called method see the return() entry elsewhere in this document return Returns the product of the given (or last) do(), undef on failure. my @prepended = $o_dat->PREPEND($a)->APPEND($b)->return('prepend'); my @appended = $o_dat->PREPEND($a)->APPEND($b)->return; # like read() create placeholder - unsupported delete placeholder - unsupported info placeholder - unsupported VARIABLES Various variables may be set affecting the behaviour of the module. $File::Data::DEBUG Set to 0 (default) or 1 for debugging information to be printed on STDOUT. $File::Data::DEBUG = 1; Alternatively set to a regex of any of the prime methods to debug them individually. $File::Data::DEBUG = '(ap|pre)pend'; $File::Data::FATAL Will die if there is any failure in accessing the file, or reading the data. Default = 0 (don't die - just warn); $File::Data::FATAL = 1; # die $File::Data::REFERENCE Will return a reference, not a list, useful with large files. Default is 0, ie; methods normally returns a list. Hopefully future versions of perl may return a reference if you request one, but as this is not supported generically yet, nor do we, so we require the variable to be set. There may be an argument to make this a reference by default, feedback will decide. $File::Data::REFERENCE = 1; my $a_ref = $o_dat->search('.*'); print "The log: \n".@{ $a_ref }; $File::Data::SILENT Set to something other than zero if you don't want error messages ?-\ $File::Data::SILENT = 0; # per line $File::Data::STRING Where regex's are used, default behaviour is to treate the entire file as a single scalar string, so that, for example, (?ms:...) matches are effective. Unset if you don't want this behaviour. $File::Data::STRING = 0; # per line $File::Data::PERMISSIONS File will be opened read-write (insert() compatible) unless this variable is set explicitly or given via new(). In either case, unless it is one of our keys declared below, it will be passed on to FileHandle and otherwise not modified. We don't support fancy permission sets. Read-only permissions may be explicitly set using one of the following keys: $File::Data::PERMISSIONS = 'ro'; # or readonly or < Or, equivalently, for read-write (default): $File::Data::PERMISSIONS = 'rw'; # or readwrite or +< # ================================================================ SPECIAL ... AUTOLOAD Any unrecognised function will be passed to the FileHandle object for final consideration, behaviour is then effectively 'o_dat ISA FileHandle'. $o_dat->truncate; EXAMPLES Typical construction examples: my $o_rw = File::Data->new($filename, 'rw'); my $o_ro = File::Data->new($filename, 'ro'); error Failure is indicated by an error routine being called, this will print out any error to STDERR, unless warnings are declared fatal, in which case we croak. You can register your own error handlers for any method mentioned in the the METHOD manpage section of this document, in addition is a special init call for initial file opening and general setting up. Create a read-write object with a callback for all errors: my $o_rw = File::Data->new($filename, 'ro', { 'error' => \&myerror, }); Create a read-only object with a separate object handler for each error type: my $o_rw = File::Data->new($filename, 'rw', { 'error' => $o_generic->error_handler, 'insert' => $o_handler->insert_error, 'open' => $o_open_handler, 'read' => \&carp, 'write' => \&write_error, }); commandline From the command line: Cnew('./test.txt')->write('some stuff')"> And (very non-obfuscated) C< perl -MFile::Data -e "@x=sort qw(perl another hacker just); print \ map {split(\"\n\", ucfirst(\$_).' ')} File::Data->new('./japh')-> \ WRITE(shift(@x).\"\n\")-> \ APPEND(shift(@x).\"\n\")-> \ PREPEND(shift(@x).\"\n\")-> \ INSERT(2, shift(@x).\"\n\")->read" > If you still have problems, mail me the output of make test TEST_VERBOSE=1 PRIVATE Private methods not expected to be called by anybody, and completely unsupported. Expected to metamorphose regularly - do not call these - you have been warned! _var Variable get/set method my $get = $o_dat->_var($key); # get my $set = $o_dat->_var($key, $val); # set _debug Print given args on STDOUT $o_dat->_debug($msg) if $File::Data::DEBUG; _vars Return dumped env and object key and values print $o_dat->_vars; _err Get/set error handling methods/objects my $c_sub = $o_dat->_err('insert'); # or default _error By default prints error to STDERR, will croak if File::Data::FATAL set. See the EXAMPLES manpage for info on how to pass your own error handlers in. _mapfile Maps file my $file = $o_dat->_mapfile($filename); _mapperms Maps given permissions to appropriate form for FileHandle my $perms = $o_dat->_mapperms('+<'); _maperrs Map error handlers, if given my $h_errs = $o_dat->_maperrs(\%error_handlers); _enter Mark the entering of a special section, or state my $entered = $o_dat->enter('search'); _leave Mark the leaving of a special section, or state my $left = $o_dat->_leave('search'); _fh Get and set FileHandle. Returns undef otherwise. my $FH = $o_dat->_fh($FH); UTILITY Private methods not expected to be called by anybody, and completely unsupported. Expected to metamorphose regularly - do not call these - you have been warned! The following utility methods return integer values 1 = success 0 = failure _init Setup object, open a file, with permissions. my $i_ok = $o_date->_init($file, $perm, $h_errs); _check_access Checks the args for existence and appropriate permissions etc. my $i_isok = $o_dat->_check_access($filename, $permissions); _open Open the file my $i_ok = $o_dat->_open; _lock Lock the file my $i_ok = $o_dat->_lock; _unlock Unlock the file my $i_ok = $o_dat->unlock; _close Close the filehandle my $i_ok = $o_dat->_close; AUTHOR Richard Foley richard.foley@rfi.net 2001 For those that are interested, the docs and tests were (mostly) written before the code.