NAME

tidyall - Your all-in-one code tidier and validator

VERSION

version 0.15

SYNOPSIS

# Create a tidyall.ini or .tidyallrc at the top of your project
#
[PerlTidy]
select = **/*.{pl,pm,t}
argv = -noll -it=2

[PerlCritic]
select = lib/**/*.pm
ignore = lib/UtterHack.pm
argv = -severity 3

# Process all files in the current project,
# look upwards from cwd for conf file
#
% tidyall -a

# Process one or more specific files,
# look upwards from the first file for conf file
#
% tidyall file [file...]

# Process a directory recursively
#
% tidyall -r dir

DESCRIPTION

There are a lot of great code tidiers and validators out there. tidyall makes them available from a single unified interface.

You can run tidyall on a single file or on an entire project hierarchy, and configure which tidiers/validators are applied to which files. tidyall will back up files beforehand, and for efficiency will only consider files that have changed since they were last processed.

What's a tidier? What's a validator?

A tidier transforms a file so as to improve its appearance without changing its semantics. Examples include perltidy, podtidy and js-beautify.

A validator analyzes a file for some definition of correctness. Examples include perlcritic, podchecker and jshint.

Many tidiers are also validators, e.g. perltidy will throw an error on badly formed Perl.

To use a tidier or validator with tidyall it must have a corresponding plugin class, usually under the prefix Code::TidyAll::Plugin::. This distribution comes with plugins for:

See Code::TidyAll::Plugin for information about creating your own plugin.

USING TIDYALL

tidyall works on a project basis, where a project is just a directory hierarchy of files. svn or git working directories are typical examples of projects.

The top of the project is called the root directory. In the root directory you'll need a config file named tidyall.ini or .tidyallrc; it defines how various tidiers and validators will be applied to the files in your project.

tidyall will find your root directory and config file automatically depending on how you call it:

tidyall file [file...]

tidyall will search upwards from the first file for the conf file.

tidyall -p/--pipe file

tidyall will search upwards from the specified file for the conf file.

tidyall -a/--all or tidyall -s/--svn or tidyall -g/--git

tidyall will search upwards from the current working directory for the conf file.

tidyall -a --root-dir dir

tidyall will expect to find the conf file in the specified root directory.

You can also pass --conf-name to change the name that is searched for, or --conf-file to specify an explicit path.

CONFIGURATION

The config file (tidyall.ini or .tidyallrc) is in Config::INI format. Here's a sample:

[PerlTidy]
select = **/*.{pl,pm,t}
argv = -noll -it=2

[PerlCritic]
select = lib/**/*.pm
ignore = lib/UtterHack.pm lib/OneTime/*.pm
argv = -severity 3

[PodTidy]
select = lib/**/*.{pm,pod}

In order, the three sections declare:

  • Apply PerlTidy with settings "-noll -it=2" to all *.pl, *.pm, and *.t files.

  • Apply PerlCritic with severity 3 to all Perl modules somewhere underneath "lib/", except for "lib/UtterHack.pm".

  • Apply PodTidy with default settings to all *.pm and *.pod files underneath "lib/".

Standard configuration elements

[class] or [class description]

The header of each section refers to a tidyall plugin. The name is automatically prefixed with Code::TidyAll::Plugin:: unless it begins with a '+', e.g.

; Uses plugin Code::TidyAll::Plugin::PerlTidy
;
[PerlTidy]

; Uses plugin My::TidyAll::Plugin
;
[+My::TidyAll::Plugin]

You can also include an optional description after the class. The description will be ignored and only the first word will be used for the plugin. This allows you to a list a plugin more than once, with different configuration each time. For example, two different PerlCritic configurations:

; Be brtual on libraries
;
[PerlCritic strict]
select = lib/**/*.pm
argv = --brutal

; but gentle on scripts
;
[PerlCritic lenient]
select = bin/**/*.pl
argv = --gentle

Warning: If you simply list the same plugin twice with no description (or the same description), one of them will be silently ignored.

select

One or more File::Zglob patterns, separated by whitespace or on multiple lines, indicating which files to select. At least one is required. e.g.

; All .t and .pl somewhere under bin and t;
; plus all .pm files directly under lib/Foo and lib/Bar
;
select = {bin,t}/**/*.p[lm]
select = lib/{Foo,Bar}/*.pm

; All .txt files anywhere in the project
;
select = **/*.txt

The pattern is relative to the root directory and should have no leading slash. All standard glob characters (*, ?, [], {}) will work; in addition, ** can be used to represent zero or more directories. See File::Zglob documentation for more details.

ignore

One or more File::Zglob patterns, separated by whitespace or on multiple lines, indicating which files to ignore. This is optional and overrides select. e.g.

; All .pl files anywhere under bin, except bin/awful.pl or anywhere
; under bin/tmp
;
select = bin/**/*.pl
ignore = bin/awful.pl bin/tmp/**/*.pl
only_modes

A list of modes, separated by whitespace. e.g.

only_modes = tests cli

The plugin will only run if one of these modes is passed to tidyall via -m or --mode.

except_modes

A list of modes, separated by whitespace. e.g.

except_modes = commit editor

The plugin will not run if one of these modes is passed to tidyall via -m or --mode.

argv

Many plugins (such as perltidy, perlcritic and podtidy) take this option, which specifies arguments to pass to the underlying command-line utility.

PLUGIN ORDER AND ATOMICITY

If multiple plugins match a file, tidiers are applied before validators so that validators are checking the final result. Within those two groups, the plugins are applied in alphabetical order by plugin name/description.

The application of multiple plugins is all-or-nothing. If an error occurs during the application of any plugin, the file is not modified at all.

COMMAND-LINE OPTIONS

-a, --all

Process all files. Does a recursive search for all files in the project hierarchy, starting at the root, and processes any file that matches at least one plugin in the configuration.

-g, --git

Process all added or modified files in the current git working directory.

-h, --help

Print help message

-l, --list

List each file along with the list of plugins it matches (files without any matches are skipped). Does not actually process any files and does not care whether files are cached. Generally used with -a, -g, or -s. e.g.

% tidyall -a -l
lib/CHI.pm (PerlCritic, PerlTidy, PodTidy)
lib/CHI/Benchmarks.pod (PodTidy)
lib/CHI/CacheObject.pm (PerlCritic, PerlTidy, PodTidy)
-m, --mode

Optional mode that can affect which plugins run. Defaults to cli. See "MODES".

-p path, --pipe path

Read content from STDIN and write the resulting content to STDOUT. If successful, tidyall exits with status 0. If an error occurs, tidyall outputs the error message to STDERR, mirrors the input content to STDOUT with no changes, and exits with status 1. The mirroring means that you can safely pipe to your destination regardless of whether an error occurs.

When specifying this option you must specify exactly one filename, relative or absolute, which will be used to determine which plugins to apply and also where the root directory and configuration file are. The file will not actually be read and does need even need to exist.

This option implies --no-backups and --no-cache (since there's no actual file) and --quiet (since we don't want to mix extraneous output with the tidied result).

# Read from STDIN and write to STDOUT, with appropriate plugins
# for some/path.pl (which need not exist)
#
% tidyall --pipe some/path.pl
-r, --recursive

Recursively enter any directories listed on the command-line and process all the files within. By default, directories encountered on the command-line will generate a warning.

-s, --svn

Process all added or modified files in the current svn working directory.

-q, --quiet

Suppress output except for errors.

-v, --verbose

Show extra output.

-I path1,path2,...

Add one or more library paths to @INC, like Perl's -I. Useful if --tidyall-class or plugins are in an alternate lib directory.

--backup-ttl duration

Amount of time before backup files can be purged. Can be a number of seconds or any string recognized by Time::Duration::Parse, e.g. "4h" or "1day". Defaults to "1h".

--check-only

Instead of actually tidying files, check if each file is tidied (i.e. if its tidied version is equal to its current version) and consider it an error if not. This is used by Test::Code::TidyAll and the svn and git pre-commit hooks, for example, to enforce that you've tidied your files.

--conf-file path

Specify relative or absolute path to conf file, instead of searching for it in the usual way.

--conf-name name

Specify a conf file name to search for instead of the defaults (tidyall.ini / .tidyallrc).

--data-dir path

Contains data like backups and cache. Defaults to root_dir/.tidyall.d

--iterations count

Run each tidier transform count times. Default is 1.

In some cases (hopefully rare) the output from a tidier can be different if it is applied multiple times. You may want to perform multiple iterations to make sure the content "settles" into its final tidied form -- especially if the tidiness is being enforced with a version-control hook or a test. Of course, performance will suffer a little. You should rarely need to set this higher than 2.

This only affects tidiers, not validators; e.g. perlcritic and jshint would still only be run once.

--no-backups

Don't backup files before processing.

--no-cache

Don't cache last processed times; process all files every time. See also --refresh-cache.

--output-suffix suffix

Suffix to add to a filename before outputting the modified version, e.g. .tdy. Default is none, which means overwrite the file.

--refresh-cache

Erase any existing cache info before processing each file, then write new cache info. See also --no-cache.

--root-dir

Specify root directory explicitly. Usually this is inferred from the specified files or the current working directory.

--tidyall-class class

Subclass to use instead of Code::TidyAll.

Specifying options in configuration

Almost any command-line option can be specified at the top of the config file, above the plugin sections. Replace dashes with underscores. e.g.

backup_ttl = 4h
iterations = 2
tidyall_class = My::Code::TidyAll

[PerlTidy]
select = **/*.{pl,pm,t}
argv = -noll -it=2

...

If an option is passed in both places, the command-line takes precedence.

EXIT STATUS

tidyall will exit with status 1 if any errors occurred while processing files, and 0 otherwise.

MODES

You can use tidyall in a number of different contexts, and you may not want to run all plugins in all of them.

You can pass a mode to tidyall via -m or --mode, and then specify that certain plugins should only be run in certain modes (via "only_modes") or should be run in all but certain modes (via "except_modes").

Examples of modes:

Now since perlcritic is a bit time-consuming, you might only want to run it during tests and explicit command-line invocation:

[PerlCritic]
select = lib/**/*.pm
only_modes = tests cli
...

Or you could specify that it be run in all modes except the editor:

[PerlCritic]
select = lib/**/*.pm
except_modes = editor
...

If you specify neither only_modes nor except_modes for a plugin, then it will always run.

LAST-PROCESSED CACHE

tidyall keeps track of each file's signature after it was last processed. On subsequent runs, it will only process a file if its signature has changed. The cache is kept in files under the data dir.

You can force a refresh of the cache with --refresh-cache, or turn off the behavior entirely with --no-cache.

BACKUPS

tidyall will backup each file before modifying it. The timestamped backups are kept in a separate directory hierarchy under the data dir.

Old backup files will be purged automatically as part of occasional tidyall runs. The duration specified in --backup-ttl indicates both the minimum amount of time backups should be kept, and the frequency that purges should be run. It may be specified as "30m" or "4 hours" or any string acceptable to Time::Duration::Parse. It defaults to "1h" (1 hour).

You can turn off backups with --no-backups.

RELATED TOOLS

KNOWN BUGS

  • Does not yet work on Windows

ACKNOWLEDGMENTS

Thanks to Jeff Thalhammer for helping me refine this API. Thanks to Jeff for perlcritic, Steve Hancock for perltidy, and all the other authors of great open source tidiers and validators.

SEE ALSO

Code::TidyAll

AUTHOR

Jonathan Swartz <swartz@pobox.com>

COPYRIGHT AND LICENSE

This software is copyright (c) 2011 by Jonathan Swartz.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.