Retrieve info about, inject code into and manage Perl subs as objects.
Perl Perl 6
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.



    Devel::Examine::Subs - Get info about, search/replace and inject code
    into Perl files and subs.


        use Devel::Examine::Subs;
        # examine a file
        my $des = Devel::Examine::Subs->new( file => '' );
        # examine all the Perl files in a directory
        my $des = Devel::Examine::Subs->new( file => '/path/to/directory' );
        # load a module by name. Uses %INC to find the path after loading it
        my $des = Devel::Examine::Subs->new( file => 'Some::Module::Name' );

    Get all sub names in a file

        my $aref = $des->all;

    Get all the subs as objects

        $subs = $des->objects;
        for my $sub (@$subs){
            $sub->name;       # name of sub
            $sub->start;      # first line number of sub in file
            $sub->end;        # last line number of sub in file
            $sub->line_count; # number of lines in sub
            $sub->code;       # entire sub code from file
            $sub->lines;      # lines that match search term

    Get the sub objects within a hash

        my $subs = $des->objects( objects_in_hash => 1 );
        for my $sub_name (keys %$subs) {
            print "$sub_name\n";
            my $sub = $subs->{$sub_name};
            print $sub->start . "\n" .
                  $sub->end . "\n";

    Get all subs containing "string" in the body

        my $search = 'string';
        my $aref = $des->has( search => $search );

    Search and replace code in subs

        $des->search_replace( exec => sub { $_[0] =~ s/this/that/g; } );

    Inject code into sub after a search term (preserves previous line's

        my @code = <DATA>;
                        search => 'this',
                        code => \@code,
        # previously uncaught issue
        if ($foo eq "bar"){
            confess 'big bad error';

    Print out all lines in all subs that contain a search term

        my $subs = $des->lines(search => 'this');
        for my $sub (keys %$subs){
            print "\nSub: $sub\n";
            for my $line (@{ $subs->{$sub} }){
                my ($line_num, $text) = each %$line;
                say "Line num: $line_num";
                say "Code: $text\n";

    The structures look a bit differently when 'file' is a directory. You
    need to add one more layer of extraction.

        my $files = $des->objects;
        for my $file (keys %$files){
            for my $sub (@{$files->{$file}}){

    Print all subs within each Perl file under a directory

        my $files = $des->all( file => 'lib/Devel/Examine' );
        for my $file (keys %$files){
            print "$file\n";
            print join('\t', @{$files->{$file}});

    Most methods can include or exclude specific subs

        my $has = $des->has( include => ['dump', 'private'] );
        my $missing = $des->missing( exclude => ['this', 'that'] );
        # note that 'exclude' param renders 'include' invalid


    Gather information about subroutines in Perl files (and in-memory
    modules), with the ability to search/replace code, inject new code, get
    line counts, get start and end line numbers, access the sub's code and
    a myriad of other options. Files are parsed using PPI, not by
    inspecting packages or coderefs.


    See the "PARAMETERS" for the full list of params, and which ones are
    persistent across runs using the same object.


    Mandatory parameters: file => $filename

    Instantiates a new object. If $filename is a directory, we'll iterate
    through it finding all Perl files. If $filename is a module name (eg:
    Data::Dump), we'll attempt to load the module, extract the file for the
    module, and load the file. CAUTION: this will be a production %INC file
    so be careful.

    Only specific params are guaranteed to stay persistent throughout a run
    on the same object, and are best set in new(). These parameters are
    file, extensions, maxdepth, cache, regex, copy, no_indent, backup and

    Note: omit the file parameter if all you'll be using is the module()


    Mandatory parameters: None

    Returns an array reference containing the names of all subroutines
    found in the file, listed in the order they appear in the file.


    Mandatory parameters: search => 'term'

    Returns an array reference containing the names of the subs where the
    subroutine contains the search text.


    Mandatory parameters: search => 'term'

    The exact opposite of has.


    Mandatory parameters: search => 'text'

    Gathers together all line text and line number of all subs where the
    subroutine contains lines matching the search term.

    Returns a hash reference with the subroutine name as the key, the value
    being an array reference which contains a hash reference in the format
    line_number => line_text.


    Mandatory parameters: None

    Optional parameters: objects_in_hash => 1

    Returns an array reference of subroutine objects. If the optional
    objects_in_hash is sent in with a true value, the objects will be
    returned in a hash reference where the key is the sub's name, and the
    value is the sub object.

    See "SYNOPSIS" for the structure of each object.


    Mandatory parameters: 'Module::Name'. Note that this is one public
    method that takes its parameter in string format (as opposed to hash

    Note that this method pulls the subroutine names from the namespace
    (which may include includeed subs. If you only want a list of subs
    within the actual module file, send the module name as the value to the
    file parameter, and use the common methods (all, has, missing etc) to
    extract the names.

    Returns an array reference containing the names of all subs found in
    the module's namespace symbol table.


    After one of the other user methods are called, call this method to get
    returned to you an array of the names of subs you collected, in the
    order that they appear in the file. By default, because we use hashes
    internally, subs aren't ever in proper order.


    Mandatory parameters: exec => $cref

    Core optional parameter: copy => 'filename.txt'

    Coderef should be created in the form sub { $_[0] =~ s/search/replace/;
    };. This allows us to avoid string eval, and allows us to use any regex
    modifiers you choose. The $_[0] element represents each line in the
    file, as we loop over them.


    Parameters: exec => $cref, limit => 1

    This is the entire file brother to the sub-only search_replace(). The
    limit parameter specifies how many successful replacements to do,
    starting at the top of the file. Set to a negative integer for
    unlimited (this is the default).

    The exec parameter is a code reference, eg: my $cref = sub {$_[0] =~
    s/this/that/;}. All standard Perl regular expressions apply, along with
    their modifiers. The $_[0] element represents each line in the file, as
    we loop over them.

    Returns the number of lines changed in file mode, and an empty hashref
    in directory mode.


    Mandatory parameters: search => 'this', code => \@code

    Injects the code in @code into the sub within the file, where the sub
    contains the search term. The same indentation level of the line that
    contains the search term is used for any new code injected. Set
    no_indent parameter to a true value to disable this feature.

    By default, an injection only happens after the first time a search
    term is found. Use the injects parameter (see "PARAMETERS") to change
    this behaviour. Setting to a positive integer beyond 1 will inject
    after that many finds. Set to a negative integer will inject after all

    The code array should contain one line of code (or blank line) per each
    element. (See "SYNOPSIS" for an example). The code is not manipulated
    prior to injection, it is inserted exactly as typed. Best to use a
    heredoc, __DATA__ section or an external text file for the code.

    Optional parameters:


      See search_replace() for a description of how this parameter is used.


      How many injections do you want to do per sub? See "PARAMETERS" for
      more details.


    Parameters (all are mutually exclusive, use only one):

    line_num => 33 with code => \@code or, inject_use => ['use
    Module::Name', 'use Module2::Name'] or, inject_after_sub_def => ['code
    line 1;', 'code line 2;']

    line_num will inject the block of code in the array reference
    immediately after the line number specified.

    inject_use will inject the statements prior to all existing use
    statements that already exist in the file(s). If none are found, will
    inject the statements right after a Package statement if found.

    Technically, you don't have to inject a use statement, but I'd advise

    inject_after_sub_def will inject the lines of code within the array
    reference value immediately following all sub definitions in a file.
    Next line indenting is used, and sub definitions with their opening
    brace on a separate line than the definition itself is caught.


    Parameters: delete => ['string1', 'string2']

    Deletes from the file(s) the entire lines that contain the search

    This method is file based... the work happens prior to digging up subs,
    hence exclude, include and other sub-based parameters have no effect.


    Configure whether to make a filename.bak copy of all files read by DES.
    A true value sent in will enable this feature, a false value will
    disable it. Returns 1 (true) if this feature is enabled, and 0 (false)
    if not.

    Disabled by default.



    Returns a hash where the keys are valid parameter names, and the value
    is a bool where if true, the parameter is persistent (remains between
    calls on the same object) and if false, the param is transient, and
    will be made undef after each method call finishes.


    Parameter format: Hash reference

    All public methods call this method internally. This is the only public
    method that takes its parameters as a single hash reference. The public
    methods set certain variables (filters, engines etc). You can get the
    same effect programatically by using run(). Here's an example that
    performs the same operation as the has() public method:

        my $params = {
                search => 'text',
                post_proc => 'file_lines_contain',
                engine => 'has',
        my $return = $des->run($params);

    This allows for very fine-grained interaction with the application, and
    makes it easy to write new engines and for testing.


    WARNING!: This method is for development of this distribution only!

    While writing new processors, set the processor type to a callback
    within the local working file. When the code performs the actions you
    want it to, put a comment line before the code with #<des> and a line
    following the code with #</des>. DES will slurp in all of that code
    live-time, inject it into the specified processor, and configure it for
    use. See examples/ for an example of creating a new
    'engine' processor.

    Returns 1 on success.



      Informs the system which type of processor to inject and configure.
      Permitted values are 'pre_proc', 'post_proc' and 'engine'.


      Set to a true value, will update the code in the actual installed
      Perl module file, instead of a local copy.

    Optional parameters:


      Set it to a new file name which will be a copy of the specified file,
      and only change the copy. Useful for verifying the changes took

 pre_procs, post_procs, engines

    For development. Returns the list of the respective built-in callbacks.


    There are various parameters that can be used to change the behaviour
    of the application. Some are persistent across calls, and others
    aren't. You can change or null any/all parameters in any call, but some
    should be set in the new() method (set it and forget it).

    The following list are persistent parameters, which need to be manually
    changed or nulled. Consider setting these in new().


      State: Persistent

      Default: None

      The name of a file, directory or module name. Will convert module
      name to a file name if the module is installed on the system. It'll
      require the module temporarily and then 'un'-require it immediately
      after use.

      If set in new(), you can omit it from all subsequent method calls
      until you want it changed. Once changed in a call, the updated value
      will remain persistent until changed again.


      State: Persistent

      Default: Disabled

      Set this to a true value to have a .bak file copy created on all file
      reads. The .bak file will be created in the directory the script is
      run in.


      State: Persistent

      Default: ['*.pm', '*.pl')]

      By default, we load only *.pm and *.pl files. Use this parameter to
      load different files. Only useful when a directory is passed in as
      opposed to a file. This parameter is persistent until manually reset
      and should be set in new().

      Values: Array reference where each element is the names of files to
      find. Any wildcard or regex that are valid in File::Find::Rule's
      <> are valid here. For example, [qw(*.pm *.pl)] is the default.


      When running in directory mode, how many levels deep do you want to
      traverse? Default is unlimited. Set to a positive integer to set.


      State: Persistent

      Default: Undefined

      If multiple calls on the same object are made, caching will save the
      file/directory/sub information, saving tremendous work for subsequent
      calls. This is dependant on certain parameters not changing between

      Set to a true value (1) to enable. Best to call in the new method.


      State: Persistent

      Default: None

      For methods that write to files, you can optionally work on a copy
      that you specify in order to review the changes before modifying a
      production file.

      Set this parameter to the name of an output file. The original file
      will be copied to this name, and we'll work on this copy.


      State: Persistent

      Default: Enabled

      Set to a true value, all values in the 'search' parameter become
      regexes. For example with regex on, /thi?s/ will match "this", but
      without regex, it won't. Without 'regex' enabled, all characters that
      perl treats as special must be escaped. This parameter is persistent;
      it remains until reset manually.


      State: Persistent

      Default: Disabled

      In the processes that write new code to files, the indentation level
      of the line the search term was found on is used for inserting the
      new code by default. Set this parameter to a true value to disable
      this feature and set the new code at the beginning column of the


      State: Persistent

      Not yet implemented.

      Compiles a diff after each edit using the methods that edit files.

    The following parameters are not persistent, ie. they get reset before
    entering the next call on the DES object. They must be passed in to
    each subsequent call if the effect is still desired.


      State: Transient

      Default: None

      An array reference containing the names of subs to include. This (and
      exclude) tell the Processor phase to generate only these subs,
      significantly reducing the work that needs to be done in subsequent
      method calls.


      State: Transient

      Default: None

      An array reference of the names of subs to exclude. See include for
      further details.

      Note that exclude renders include useless.


      State: Transient

      Default: 1

      Informs inject_after() how many injections to perform. For instance,
      if a search term is found five times in a sub, how many of those do
      you want to inject the code after?

      Default is 1. Set to a higher value to achieve more injects. Set to a
      negative integer to inject after all.

    pre_proc_dump, post_proc_dump, engine_dump, cache_dump, core_dump

      State: Transient

      Default: Disabled

      Set to 1 to activate, exit()s after completion.

      Print to STDOUT using Data::Dumper the structure of the data
      following the respective phase. The core_dump will print the state of
      the data, as well as the current state of the entire DES object.

      NOTE: The 'post_proc' phase is run in such a way that the filters can
      be daisy-chained. Due to this reason, the value of post_proc_dump
      works a little differently. For example:

          post_proc => ['one', 'two'];

      ...will execute filter 'one' first, then filter 'two' with the data
      that came out of filter 'one'. Simply set the value to the number
      that coincides with the location of the filter. For instance,
      post_proc_dump => 2; will dump the output from the second filter and
      likewise, 1 will dump after the first.

      For cache_dump, if it is set to one, it'll dump cache but the
      application will continue. Set this parameter to an integer larger
      than one to have the application exit immediately after dumping the
      cache to STDOUT.

    pre_proc_return, post_proc_return, engine_return

      State: Transient

      Default: Disabled

      Returns the structure of data immediately after being processed by
      the respective phase. Useful for writing new 'phases'. (See "SEE
      ALSO" for details).

      NOTE: post_proc_return does not behave like post_proc_dump. It will
      only return after all post_procs have executed.


      State: Transient

      Default: Disabled

      Prints to STDOUT with Data::Dumper the current state of all loaded
      configuration parameters.

    pre_proc, post_proc, engine

      State: Transient

      Default: undef

      These are mainly used to set up the public methods with the proper
      callbacks used by the run() command.

      engine and pre_proc take either a single string that contains a valid
      built-in callback, or a single code reference of a custom callback.

      post_proc works a lot differently. These modules can be
      daisy-chained. Like engine and pre_proc, you can send in a string or
      cref, or to chain, send in an aref where each element is either a
      string or cref. The filters will be executed based on their order in
      the array reference.



    CPAN Testers:


    If Devel::Trace::Subs is installed, you can configure stack tracing.

    In your calling script, set $ENV{DES_TRACE} = 1.

    See perldoc Devel::Trace::Subs for information on how to access the


    perldoc Devel::Examine::Subs::Preprocessor

      Information related to the 'pre_proc' phase core modules.

    perldoc Devel::Examine::Subs::Postprocessor

      Information related to the 'post_proc' phase core modules.

    perldoc Devel::Examine::Subs::Engine

      Information related to the 'engine' phase core modules.


    Steve Bertrand, <steveb at>


    You can find documentation for this module with the perldoc command.

        perldoc Devel::Examine::Subs


    Copyright 2016 Steve Bertrand.

    This program is free software; you can redistribute it and/or modify it
    under the terms of either: the GNU General Public License as published
    by the Free Software Foundation; or the Artistic License.

    See for more information.