File Coverage

blib/lib/App/Greple/jq.pm

Criterion	Covered	Total	%
statement	62	65	95.3
branch	14	20	70.0
condition	1	3	33.3
subroutine	13	15	86.6
pod	0	7	0.0
total	90	110	81.8

=encoding utf-8

=head1 NAME

greple -Mjq - greple module to search JSON data with jq

=head1 SYNOPSIS

greple -Mjq --glob JSON-DATA --IN label pattern

=head1 VERSION

Version 0.06

=head1 DESCRIPTION

This is an experimental module for L to search JSON
formatted text using L as a backend.

Search top level json object which includes both C and
C somewhere in its text representation.

    greple -Mjq 'Marvin Zaphod'

You can search object C<.commit.author.name> includes C like this:

    greple -Mjq --IN .commit.author.name Marvin

Search first C field including C under C<.commit>:

    greple -Mjq --IN .commit..name Marvin

Search any C field including C:

    greple -Mjq --IN author.name Marvin

Search C is C and C is C or C:

    greple -Mjq --IN name Marvin --IN type 'Robot|Android'

Please be aware that this is just a text matching tool for indented
result of L command.  So, for example, C<.commit.author>
includes everything under it and it matches C field name.
Use L filter for more complex and precise operation.

=head1 CAUTION

L commands read entire input before processing.  So it
should not be used for gigantic data or infinite stream.

=head1 INSTALL

=head2 CPANMINUS

    $ cpanm App::Greple::jq

=head1 OPTIONS

=over 7

=item B<--IN> I I

Search I included in I field.

Character C<%> can be used as a wildcard in I string.  So
C<%name> matches labels end with C, and C matches labels
start with C.

If the label is simple string like C, it matches any level of
JSON data.

If the label string contains period (C<.>), it is considered as a
nested labels.  Name C<.name> matches only C label at the top
level.  Name C matches only C entry of some
C hash.

If labels are separated by two or more dots (C<..>), they don't have
to have direct relationship.

=item B<--NOT> I I

Specify negative condition.

=item B<--MUST> I I

Specify required condition.  If there is one or more required
condition, all other positive rules move to optional.  They are not
required but highlighted if exist.

=back

=head1 LABEL SYNTAX

=over 15

=item B<.file>

C at the top level.

=item B<.file.path>

C under C<.file>.

=item B<.file..path>

C in descendants of C<.file>.

=item B

C at any level.

=item B

C at any level.

=item B

Some C in descendants of some C.

=item B<%path>

Any labels end with C.

=item B

Any labels start with C.

=item B<%path%>

Any labels include C.

=back

=head1 EXAMPLES

Search from any C labels.

    greple -Mjq --IN name _mina

Search from C<.process.name> label.

    greple -Mjq --IN .process.name _mina

Object C<.process.name> contains C<_mina> and C<.event> contains
C.

    greple -Mjq --IN .process.name _mina --IN .event EXEC

Object C is 803 and C<.event> contains C or C.

    greple -Mjq --IN ppid 803 --IN event 'FORK|EXEC'

Object C is C<_mina> and C<.event> contains C.

    greple -Mjq --IN name _mina --IN event 'CREATE'

Object C contains C<1132> and C<.event> contains C
with C highlighted.

   greple -Mjq --IN ancestors 1132 --IN event EXEC --IN arguments .

Object C<*pid> label contains 803.

    greple -Mjq --IN %pid 803

Object any  contains C<_mina> under C<.file> and C<.event>
contains C.

    greple -Mjq --IN .file..path _mina --IN .event WRITE

=head1 TIPS

=over 2

=item *

Use C<--all> option to show entire data.

=item *

Use C<--nocolor> option or set C to disable colored
output.

=item *

Use C<-o> option to show only matched part.

=item *

Use C<--blockend=> option to cancel showing block separator.

=item *

Since this module implements original search function, L
B<-i> does not take effect.  Set modifier in regex like C<(?i)pattern>
if you want case-insensitive match.

=item *

Use C<-Mjq::set=debug> to see actual regex.

=item *

Use C<-Mjq::set=noif> if you don't have to use L as an input
filter.  Data have to be well-formatted in that case.

=item *

Use C<--color=always> and set C if you want to
see the output using L.  Put next line in your F<~/.greplerc>
to enable colored output always.

    option default --color=always

=back

=head1 SEE ALSO

L, L

L

=head1 AUTHOR

Kazumasa Utashiro

=head1 LICENSE

Copyright ©︎ 2022-2024 Kazumasa Utashiro

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

=cut

package App::Greple::jq;

use 5.014;
use strict;
use warnings;
use Carp;

our $VERSION = "0.06";

use Exporter 'import';
our @EXPORT = qw(&jq_filter);

use App::Greple::Common;
use App::Greple::Regions qw(match_regions merge_regions);
use Data::Dumper;

my %config;
sub set   { %config = @_ }
sub debug { set debug => 1 }
sub noif  { set noif  => 1 }

my $indent = '  ';
my $indent_re = qr/$indent/;

sub finalize {
    my($mod, $argv) = @_;
    if ($config{noif}) {
	my @default = $mod->default;
	$mod->setopt(default => grep { $_ ne '--jq-filter' } @default);
    }
}

sub re {
    my $pattern = shift;
    my $re = eval { qr/$pattern/ };
    if ($@) {
	die sprintf("$pattern: pattern error - %s\n",
		    $@ =~ /(.*?(?=;|$))/);
    }
    return $re;
}

sub prefix_regex {
    my $path = shift;
    my @prefix_re;
    my $level = '';
    while ($path =~ s/^([^.\n]*?)(\.+)//) {
	my($label, $dot) = ($1, $2);
	$label =~ s/%/.*/g;
	my $label_re = re($label);
	my $start_with = '';
	my $prefix_re = do {
	    if ($label eq '') {
		length($dot) > 1 ? '' : qr{ ^ (?= $indent_re \S) }xm;
	    } else {
		if (length($dot) == 1) {
		    ## using same capture group name is not a good idea
		    ## so make sure to put just for the last one
		    $level      = '?' if $path eq '';
		    $start_with = qr/(?=\S)/;
		}
		qr{
		    ^ (${level} $indent_re*) "$label_re": .* \n
		    (?:
			## single line key-value pair
			\g{-1} $indent_re $start_with .++ \n
		    |
			## indented array/hash
			\g{-1} $indent_re \S .* [\[\{] \n
			(?: \g{-1} $indent_re \s .*+ \n) *+
			\g{-1} $indent_re [\]\}] ,? \n
		    ) *?
		}xm;
	    }
	};
	push @prefix_re, $prefix_re if $prefix_re;
    }
    if ($level eq '') {
	## refering named capture group causes error if it is not used
	## so put dummy expression just to fail
	push @prefix_re, qr/(?(?!))?/;
    }
    @prefix_re
}

sub IN {
    my %opt = @_;
    my $target = delete $opt{&FILELABEL} or die;
    my($label, $pattern) = @opt{qw(label pattern)};
    my @prefix_re = $label =~ s/^((?:.*\.)?)// && prefix_regex($1);
    $label =~ s/%/.*/g;
    my($label_re, $pattern_re) = map re($_), $label, $pattern;
    my $re = qr{
	@prefix_re \K
	^
	(?() (?= \g{level} $indent_re \S ) )	# required level
	(? [ ]*) "$label_re": [ ]*+			# find given label
	(?: . | \n\g{in} \s++ ) *			# and look for ...
	$pattern_re					# pattern
	(?: . | \n\g{in} (?: \s++ | [\]\}] ) ) *	# and take the rest
    }xm;
    warn "$re\n" if $config{debug};
    match_regions pattern => $re;
}

1;

__DATA__