File::ByLine - Line-by-line file access loops
version 1.181981
use File::ByLine; # # Procedural Interface (Simple!) # # Execute a routine for each line of a file dolines { say "Line: $_" } "file.txt"; forlines "file.txt", sub { say "Line: $_" }; # Grep (match) lines of a file my (@result) = greplines { m/foo/ } "file.txt"; # Apply a function to each line and return result my (@result) = maplines { lc($_) } "file.txt"; # Parallelized forlines/dolines routines # (Note: Requires Parallel::WorkUnit to be installed) parallel_dolines { foo($_) } "file.txt", 10; parallel_forlines "file.txt", 10, sub { foo($_); }; # Parallelized maplines and greplines my (@result) = parallel_greplines { m/foo/ } "file.txt", 10; my (@result) = parallel_maplines { lc($_) } "file.txt", 10; # Read an entire file, split into lines my (@result) = readlines "file.txt"; # # Functional Interface # # Execute a routine for each line of a file my $byline = File::ByLine->new(); $byline->do( sub { say "Line: $_" }, "file.txt"); # Grep (match) lines of a file my $byline = File::ByLine->new(); my (@result) = $byline->grep( sub { m/foo/ }, "file.txt"); # Apply a function to each line and return result my $byline = File::ByLine->new(); my (@result) = $byline->map( sub { lc($_) }, "file.txt"); # Parallelized routines # (Note: Requires Parallel::WorkUnit to be installed) my $byline = File::ByLine->new(); $byline->processes(10); $byline->do( sub { foo($_) }, "file.txt"); my (@grep_result) = $byline->grep( sub { m/foo/ }, "file.txt"); my (@map_result) = $byline->map( sub { lc($_) }, "file.txt"); # Skip the header line my $byline = File::ByLine->new(); $byline->skip_header(1); $byline->do( sub { foo($_) }, "file.txt"); my (@grep_result) = $byline->grep( sub { m/foo/ }, "file.txt"); my (@map_result) = $byline->map( sub { lc($_) }, "file.txt"); # Process the header line my $byline = File::ByLine->new(); $byline->header_handler( sub { say $_; } ); $byline->do( sub { foo($_) }, "file.txt"); my (@grep_result) = $byline->grep( sub { m/foo/ }, "file.txt"); my (@map_result) = $byline->map( sub { lc($_) }, "file.txt"); # Read an entire file, split into lines my (@result) = readlines "file.txt"; # Alternative way of specifying filenames my $byline = File::ByLine->new(); $byline->file("file.txt") $byline->do( sub { foo($_) } ); my (@grep_result) = $byline->grep( sub { m/foo/ } ); my (@map_result) = $byline->map( sub { lc($_) } );
Finding myself writing the same trivial loops to read files, or relying on modules like Perl6::Slurp that didn't quite do what I needed (abstracting the loop), it was clear something easy, simple, and sufficiently Perl-ish was needed.
Perl6::Slurp
dolines { say "Line: $_" } "file.txt"; dolines \&func, "file.txt";
This function calls a coderef once for each line in the file. The file is read line-by-line, removes the newline character(s), and then executes the coderef.
Each line (without newline) is passed to the coderef as the only parameter to the coderef. It is also placed into $_.
$_
This function returns the number of lines in the file.
This is similar to forlines(), except for order of arguments. The author recommends this form for short code blocks - I.E. a coderef that fits on one line. For longer, multi-line code blocks, the author recommends the forlines() syntax.
forlines()
Instead of a single filename, an arrayref can be passed in, in which case the files are read in turn as if they are all one file. Note that if the file doesn't end in a newline, a newline is inserted before processing the next file.
forlines "file.txt", sub { say "Line: $_" }; forlines "file.txt", \&func;
This is similar to dolines(), except for order of arguments. The author recommends this when using longer, multi-line code blocks, even though it is not orthogonal with the maplines()/greplines() routines.
dolines()
maplines()
greplines()
my (@result) = parallel_dolines { foo($_) } "file.txt", 10;
Requires Parallel::WorkUnit to be installed.
Three parameters are requied: a codref, a filename, and number of simultanious child threads to use.
This function performs similar to dolines(), except that it does its' operations in parallel using fork() and Parallel::WorkUnit. Because the code in the coderef is executed in a child process, any changes it makes to variables in high scopes will not be visible outside that single child. In general, it will be safest to not modify anything that belongs outside this scope.
fork()
Note that the file will be read in several chunks, with each chunk being processed in a different thread. This means that the child threads may be operating on very different sections of the file simultaniously and no specific order of execution of the coderef should be expected!
Because of the mechanism used to split the file into chunks for processing, each thread may process a somewhat different number of lines. This is particularly true if there are a mix of very long and very short lines. The splitting routine splits the file into roughly equal size chunks by byte count, not line count.
Otherwise, this function is identical to dolines(). See the documentation for dolines() or forlines() for information about how this might differ from parallel_forlines().
parallel_forlines()
my (@result) = parallel_forlines "file.txt", 10, sub { foo($_) };
Three parameters are requied: a filename, a codref, and number of simultanious child threads to use.
This function performs similar to forlines(), except that it does its' operations in parallel using fork() and Parallel::WorkUnit. Because the code in the coderef is executed in a child process, any changes it makes to variables in high scopes will not be visible outside that single child. In general, it will be safest to not modify anything that belongs outside this scope.
Otherwise, this function is identical to forlines(). See the documentation for forlines() or dolines() for information about how this might differ from parallel_dolines().
parallel_dolines()
my (@result) = greplines { m/foo/ } "file.txt";
This function calls a coderef once for each line in the file, and, based on the return value of that coderef, returns only the lines where the coderef evaluates to true. This is similar to the grep built-in function, except operating on file input rather than array input.
grep
This function returns the lines for which the coderef evaluates as true.
my (@result) = parallel_greplines { m/foo/ } "file.txt", 10;
Three parameters are requied: a coderef, filename, and number of simultanious child threads to use.
This function performs similar to greplines(), except that it does its' operations in parallel using fork() and Parallel::WorkUnit. Because the code in the coderef is executed in a child process, any changes it makes to variables in high scopes will not be visible outside that single child. In general, it will be safest to not modify anything that belongs outside this scope.
If a large amount of data is returned, the overhead of passing the data from child to parents may exceed the benefit of parallelization. However, if there is substantial line-by-line processing, there likely will be a speedup, but trivial loops will not speed up.
Note that the file will be read in several chunks, with each chunk being processed in a different thread. This means that the child threads may be operating on very different sections of the file simultaniously and no specific order of execution of the coderef should be expected! However, the results will be returned in the same order as greplines() would return them.
Otherwise, this function is identical to greplines().
my (@result) = maplines { lc($_) } "file.txt";
This function calls a coderef once for each line in the file, and, returns an array of return values from those calls. This follows normal Perl rules - basically if the coderef returns a list, all elements of that list are added as distinct elements to the return value array. If the coderef returns an empty list, no elements are added.
This is meant to be similar to the built-in map function.
map
my (@result) = parallel_maplines { lc($_) } "file.txt", 10;
This function performs similar to maplines(), except that it does its' operations in parallel using fork() and Parallel::WorkUnit. Because the code in the coderef is executed in a child process, any changes it makes to variables in high scopes will not be visible outside that single child. In general, it will be safest to not modify anything that belongs outside this scope.
Note that the file will be read in several chunks, with each chunk being processed in a different thread. This means that the child threads may be operating on very different sections of the file simultaniously and no specific order of execution of the coderef should be expected! However, the results will be returned in the same order as maplines() would return them.
Otherwise, this function is identical to maplines().
my (@result) = readlines "file.txt";
This function simply returns an array of lines (without newlines) read from a file.
The object oriented interface was implemented in version 1.181860.
my $byline = File::ByLine->new();
Constructs a new object, suitable for the object oriented calls below.
$extended = $byline->extended_info(); $byline->extended_info(1);
This was added in version 1.181951.
Gets and sets the "extended information" flag. This defaults to false, but if set to a true value this will pass a second parameter to all user-defined code (such as the per-line code function in dolines and do and the header_handler function.
dolines
do
header_handler
For all code, this information will be passed as the second argument to the user defined code. It will be a hashref with the following keys defined:
filename
object
File::ByLine
process_number
This object should not be modified by user code. In addition, no attributes of the explict or implicit File::ByLine object passed as part of this hashref should be modified within user code.
my $file = $byline->file(); $byline->file("abc.txt"); $byline->file( [ "abc.txt", "def.txt" ] ); $byline->file( "$abc.txt", "def.txt" );
Gets and sets the default filename used by the methods in the object oriented interface. The default value is undef which indicates that no default filename is provided.
undef
Instead of a single filename, a list or arrayref can be passed in, in which case the files are read in turn as if they are all one file. Note that if the file doesn't end in a newline, a newline is inserted before processing the next file.
my $all_files = $byline->header_all_files(); $byline->header_all_files(1);
Gets and sets whether the object oriented methods will call header_handler for every file if multiple files are passed into the file attribute.
file
The anticipated usage of this would be with extended_info set to true, with the header_handler function examining the filename attribute of the extended info hashref. Note that all headers may be read before any line in any file is read, to better accommodate parallel code execution. I.E. the headers of all files may be read at once before any data line is read.
extended_info
my $handler = $byline->header_handler(); $byline->header_handler( sub { ... } );
Specifies code that should be executed on the header row of the input file. This defaults to undef, which indicates no header handler is specified. When a header handler is specified, the first row of the file is sent to this handler, and is not sent to the code provided to the various do/grep/map/lines methods in the object oriented interface.
The code is called with one or two parameters, the header line, and, if the extended_info attribute is set, the extended information hashref. The header line is also stored in $_.
When set, this is always executed in the parent process, not in the child processes that are spawned (in the case of processes being greater than one).
processes
You cannot set this to true while a header_skip value is set.
header_skip
my $procs = $byline->processes(); $byline->processes(10);
This gets and sets the degree of parallelism most methods will use. The default degree is 1, which indicates all tasks should only use a single process. Specifying 2 or greater will use multiple processes to operate on the file (see documentation for the parallel_* functions described above for more details).
1
2
my $unreadable = $byline->skip_unreadable(); $byline->skip_unreadable(10);
This was added in version 1.181980.
If this attribute is true, unreadable files are treated as empty files during processing. The default is false, in which case an exception is thrown when an access attempt is made to an unreadable file.
$byline->f(); # Alias for file $byline->ei(); # Alias for extended_info $byline->haf(); # Alias for header_all_files $byline->hh(); # Alias for header_handler $byline->hs(); # Alias for header_skip $byline->p(); # Alias for processes $byline->su(); # Alias for skip_unreadable
Short name aliases were added in version 1.181980.
Each attribute listed above has a corresponding short name. This short name can also be used as a constructor argument.
$byline->do( sub { ... }, "file.txt" );
This performs the dolines functionality, calling the code provided. If the filename is not provided, the file attribute is used for this. See the dolines and parallel_dolines functions for more information on how this functions.
parallel_dolines
Each line (without newline) is passed to the coderef as the first parameter to the coderef. It is also placed into $_. If the extended_info attribute is true, the extended information hashref will be passed as the second parameter.
my (@output) = $byline->grep( sub { ... }, "file.txt" );
This performs the greplines functionality, calling the code provided. If the filename is not provided, the file attribute is used for this. See the greplines and parallel_greplines functions for more information on how this functions.
greplines
parallel_greplines
The output is a list of all input lines where the code reference produces a true result.
my (@output) = $byline->map( sub { ... }, "file.txt" );
This performs the maplines functionality, calling the code provided. If the filename is not provided, the file attribute is used for this. See the maplines and parallel_maplines functions for more information on how this functions.
maplines
parallel_maplines
The output is the list produced by calling the passed-in code repeatively for each line of input.
my (@output) = $byline->lines( "file.txt" );
This performs the readlines functionality. If the filename is not provided, the file attribute is used for this. See the readlines function for more information on how this functions.
readlines
The output is a list of all input lines.
Note that this function is unaffected by the value of the processes attribute - it always executes in the parent process.
The Parallel::WorkUnit module is a recommended dependency. It is required to use the parallel_* functions - all other functionality works fine without it.
parallel_*
Some CPAN clients will automatically try to install recommended dependency, but others won't (cpan often, but not always, will; cpanm will not by default). In the cases where it is not automatically installed, you need to install Parallel::WorkUnit to get this functionality.
Joelle Maslak <jmaslak@antelope.net>
This software is copyright (c) 2018 by Joelle Maslak.
This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.
To install File::ByLine, copy and paste the appropriate command in to your terminal.
cpanm
cpanm File::ByLine
CPAN shell
perl -MCPAN -e shell install File::ByLine
For more information on module installation, please visit the detailed CPAN module installation guide.