File::Takeput - Slurp style file IO with locking.
0.10
use File::Takeput;
# Lock some file and read its content.
my @content1 = take('some_file_name.csv');
# Read content of some other file.
# Retry for up to 2.5 seconds if it is already locked.
my @content2 = grab('some_other_file_name.log' , patience => 2.5);
# Append some data to that other file.
append('some_other_file_name.log')->(@some_data);
# Read content of some third file as a single string.
my ($content3) = grab('some_third_file_name.html' , separator => undef);
# Write content back to the first file after editing it.
# The locks will be released right afterwards.
$content1[$_] =~ s/,/;/g for (0..$#content1);
put('some_file_name.csv')->(@content1);Slurp style file IO with locking. The purpose of Takeput is to make it pleasant for you to script file IO. Slurp style is both user friendly and very effective, if you can have your files in memory.
The other major point of Takeput is locking. Takeput is careful to help your script be a good citizen in a busy filesystem. All its file operations respect and set flock locking. If your script misses a lock and does not release it, the lock will be released when your script terminates.
Encoding is often part of file IO operations, but Takeput keeps out of that. It reads and writes file content just as strings of bytes, in a sort of line-based binmode. Use some other module if you need decoding and encoding. For example:
use File::Takeput;
use Encode;
my @article = map {decode('iso-8859-1',$_)} grab 'article.latin-1';Takeput will die on compile-time errors, but not on runtime errors. In case of a runtime error it might issue a warning. But it will always write an error message in $@ and return an error value.
That said, you are able to change how runtime errors are handled, by using the "error" configuration parameter.
Imported by default: append, grab, pass, plunk, put, take
Imported on demand: fgrab, fpass, ftake, reset, set
Appends @data to the $filename file.
Reads and returns the content of the $filename file. Will never change the content of $filename, or create the file.
Reading an empty file will return a list with one element, the empty string. If a false value is returned instead, it is because "grab" could not read the file.
Releases the lock on the $filename file. The content of the file will be the same as when the lock was taken (if everyone respects the locking). This is useful when a lock was taken with the "take" subroutine, but it later turned out that there was nothing to write to the file.
Overwrites the $filename file with @data.
Overwrites the taken $filename file, with @data, and releases the lock on it.
Setting the "create" configuration parameter on this call will not work. Set it on the "take" call instead.
Sets a lock on the $filename file, reads and returns its content. A call to "take" should later on be followed by a call to "put" or "pass".
Reading an empty file will return a list with one element, the empty string. If a false value is returned instead, it is because "take" could not read the file.
A functional version of the "grab" subroutine.
A functional version of the "pass" subroutine.
A functional version of the "take" subroutine.
Note that "take"s twin, "put", also returns a function. With these you can separate the file operations from their definitions. As you can with filehandles. This is true for all the functional subroutines. Here is an example using "ftake" and "put", where they are sent as parameters.
sub changecurr($r,$w,$x) {
$w->( map {s/((\d*\.)?\d+)/$x*$1/ger} $r->() );
};
my $r = ftake('wednesday.csv' , patience => 5);
my $w = put('wednesday.csv');
my $rate = current_rate('GBP');
changecurr($r,$w,$rate);Sets the default configuration parameters back to the Takeput defaults.
Changes the default values to be %settings. Can be reset by calling "reset".
There are six configuration parameters.
A scalar. If true the subroutines that have write intention, will create the file if it does not exist. If false, they will just fail if the file does not exist.
Be careful with this parameter. For example if a process renames the file while another process is waiting for the lock, that other process will open the file with the new name when it gets the lock. If it plunks, it is not to a file with the name it was called with, but to the file with this new name. Maybe not what is wanted...
The "create" parameter is ignored by "put". Use it on "take" instead, if you want this functionality.
A ref to a subroutine that is called if Takeput runs into a runtime error. It will be called without parameters. The $@ variable will be set just prior to the subroutine call, and the subroutines return value will be passed on back to your script. An example:
use Logger::Syslog qw(warning);
use File::Takeput error => sub {warning 'commit.pl: '.$@; die;};
my @data = take('transaction.data' , patience => 10);
do_stuff [@data];
put('transaction.data')->(@data);If you just need non-fatal warnings, here is a simple error handler you can use:
use File::Takeput error => sub {print STDERR "$@\n"; undef;};If the value of "error" is undef, Takeput will not make these calls.
A scalar. If true Takeput will take an exclusive lock on read operations. If false it will just take a shared lock on them, as it normally does.
A string that replaces the "separator" string at the end of each line when reading from a file. When writing to a file the replacement is the other way around. Then "separator" will replace "newline".
If either the "newline" value or the "separator" value is undef, no replacements will be done.
The time in seconds that a call will wait for a lock to be released. The value can be fractional.
If "patience" is set to 0, there will be no waiting.
The string defining the end of a line. It is used in read operations to split the data into lines. Note that the value is read as a bytestring. So take care if you use a special separator in combination with an unusual encoding.
Setting this parameter does not change the value of $/ or vice versa.
The "separator" value cannot de an empty string. If it is undef the data is seen as a single string.
You have a number of options for setting the configuration parameters.
If a parameter is set in more than one way, the most specific setting wins out. In the list above, the item with the lowest number wins out.
All the file operation subroutines can take the configuration parameters as optional named parameters. That means that you can set them per call. The place to write them is after the filename parameter. Example:
my @text = grab 'windows_file.txt' , separator => "\r\n" , newline => "\n";The two subroutines "set" and "reset" is one way to set the default values for the configuration parameters. You use "set" to set custom values, and "reset" to set the values back to the Takeput defaults.
Think of it as assignment statements. If there are multiple calls, the last one is the one that is in effect.
Another way to set the default values is in the use statement that imports Takeput. For example:
use File::Takeput separator => "\n";When you do it like this, the values are set at compile-time. Because of that, Takeput will die on any errors that these settings will give.
The Takeput defaults are:
create: undef (false)
error: undef
exclusive: undef (false)
newline: undef
patience: 0
separator: $/ (at compile time)
Cwd
Fcntl
File::Basename
Time::HiRes
No known issues.
Decide on empty string "separator". It ought to give a list of bytes, but readline gives an unintuitive list. It would be a mess with the line ending transformations.
An empty string will be an invalid value for now.
(c) 2023 Bjørn Hee
Licensed under the Apache License, version 2.0
https://www.apache.org/licenses/LICENSE-2.0.txt