1 |
dpavlin |
686 |
package WebPAC::Parser; |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
|
|
|
7 |
|
|
use PPI; |
8 |
|
|
use PPI::Dumper; |
9 |
|
|
use Data::Dump qw/dump/; |
10 |
|
|
use File::Slurp; |
11 |
|
|
|
12 |
dpavlin |
691 |
use base qw/WebPAC::Common WebPAC::Normalize/; |
13 |
dpavlin |
686 |
|
14 |
|
|
=head1 NAME |
15 |
|
|
|
16 |
|
|
WebPAC::Parser - parse perl normalization configuration files and mungle it |
17 |
|
|
|
18 |
|
|
=head1 VERSION |
19 |
|
|
|
20 |
dpavlin |
691 |
Version 0.03 |
21 |
dpavlin |
686 |
|
22 |
|
|
=cut |
23 |
|
|
|
24 |
dpavlin |
691 |
our $VERSION = '0.03'; |
25 |
dpavlin |
686 |
|
26 |
|
|
=head1 SYNOPSIS |
27 |
|
|
|
28 |
|
|
FIXME |
29 |
|
|
|
30 |
|
|
=head1 FUNCTIONS |
31 |
|
|
|
32 |
|
|
=head2 new |
33 |
|
|
|
34 |
|
|
Create new parser object. |
35 |
|
|
|
36 |
|
|
my $parser = new WebPAC::Parser( |
37 |
|
|
config => new WebPAC::Config(), |
38 |
|
|
base_path => '/optional/path/to/conf', |
39 |
|
|
); |
40 |
|
|
|
41 |
|
|
=cut |
42 |
|
|
|
43 |
|
|
sub new { |
44 |
|
|
my $class = shift; |
45 |
|
|
my $self = {@_}; |
46 |
|
|
bless($self, $class); |
47 |
|
|
|
48 |
|
|
my $log = $self->_get_logger(); |
49 |
|
|
|
50 |
|
|
$log->logdie("config isn't WebPAC::Config") unless ($self->{config} && $self->{config}->isa('WebPAC::Config')); |
51 |
|
|
|
52 |
|
|
$log->logdie("can't iterate_inputs over this config object") unless ($self->{config}->can('iterate_inputs')); |
53 |
|
|
|
54 |
dpavlin |
691 |
$self->read_sources; |
55 |
dpavlin |
686 |
|
56 |
|
|
$self->{config}->iterate_inputs( sub { |
57 |
dpavlin |
689 |
my ($input, $database) = @_; |
58 |
dpavlin |
691 |
return unless $self->valid_database_input($database, $input->{name}); |
59 |
|
|
$self->parse_lookups($database,$input->{name}); |
60 |
|
|
} ); |
61 |
|
|
|
62 |
|
|
$self ? return $self : return undef; |
63 |
|
|
} |
64 |
|
|
|
65 |
|
|
=head2 read_sources |
66 |
|
|
|
67 |
|
|
my $source_files = $parser->read_sources; |
68 |
|
|
|
69 |
|
|
Called by L</new>. |
70 |
|
|
|
71 |
|
|
=cut |
72 |
|
|
|
73 |
|
|
sub read_sources { |
74 |
|
|
my $self = shift; |
75 |
|
|
|
76 |
|
|
my $log = $self->_get_logger(); |
77 |
|
|
|
78 |
|
|
my $nr = 0; |
79 |
|
|
|
80 |
|
|
$self->{config}->iterate_inputs( sub { |
81 |
|
|
my ($input, $database) = @_; |
82 |
|
|
|
83 |
dpavlin |
686 |
my $path = $input->{normalize}->{path} || return; |
84 |
|
|
my $full = $self->{base_path} ? $self->{base_path} . '/' . $path : $path; |
85 |
dpavlin |
691 |
|
86 |
dpavlin |
686 |
$log->logdie("normalization input $full doesn't exist") unless (-e $full); |
87 |
dpavlin |
691 |
|
88 |
dpavlin |
686 |
my $s = read_file( $full ) || $log->logdie("can't read $full: $!"); |
89 |
dpavlin |
691 |
|
90 |
dpavlin |
689 |
my $input_name = $input->{name} || $log->logdie("can't deduce name of input: ", dump( $input )); |
91 |
dpavlin |
691 |
|
92 |
|
|
$log->debug("$database/$input_name: adding $path"); |
93 |
|
|
|
94 |
|
|
$self->{valid_inputs}->{$database}->{$input_name} = { |
95 |
|
|
source => $s, |
96 |
|
|
path => $full, |
97 |
|
|
usage => 0, |
98 |
|
|
} unless defined($self->{valid_inputs}->{$database}->{$input_name}); |
99 |
|
|
|
100 |
|
|
$self->{valid_inputs}->{$database}->{$input_name}->{usage}++; |
101 |
|
|
|
102 |
|
|
$nr++; |
103 |
dpavlin |
686 |
} ); |
104 |
|
|
|
105 |
dpavlin |
691 |
$log->debug("found $nr source files"); |
106 |
dpavlin |
686 |
|
107 |
dpavlin |
691 |
return $nr; |
108 |
dpavlin |
686 |
} |
109 |
|
|
|
110 |
dpavlin |
691 |
=head2 parse_lookup |
111 |
dpavlin |
686 |
|
112 |
dpavlin |
691 |
$parser->parse_lookups($database,$input); |
113 |
|
|
|
114 |
dpavlin |
686 |
=cut |
115 |
|
|
|
116 |
dpavlin |
691 |
sub parse_lookups { |
117 |
dpavlin |
686 |
my $self = shift; |
118 |
dpavlin |
691 |
my ($database, $input) = @_; |
119 |
dpavlin |
686 |
|
120 |
|
|
my $log = $self->_get_logger(); |
121 |
|
|
|
122 |
dpavlin |
691 |
$log->logdie("invalid database $database" ) unless $self->valid_database( $database ); |
123 |
|
|
$log->logdie("invalid input $input of database $database", ) unless $self->valid_database_input( $database, $input ); |
124 |
dpavlin |
686 |
|
125 |
dpavlin |
691 |
my $source = $self->{valid_inputs}->{$database}->{$input}->{source}; |
126 |
|
|
my $path = $self->{valid_inputs}->{$database}->{$input}->{path}; |
127 |
dpavlin |
689 |
|
128 |
dpavlin |
691 |
$log->logdie("no source found for database $database input $input path $path") unless ($source); |
129 |
dpavlin |
686 |
|
130 |
dpavlin |
691 |
$log->info("parsing lookups for $database/$input in $path"); |
131 |
|
|
|
132 |
|
|
my $Document = PPI::Document->new( \$source ) || $log->logdie("can't parse source:\n", $self->{source}); |
133 |
|
|
|
134 |
dpavlin |
686 |
$Document->prune('PPI::Token::Whitespace'); |
135 |
|
|
#$Document->prune('PPI::Token::Operator'); |
136 |
|
|
|
137 |
|
|
# Find all the named subroutines |
138 |
|
|
|
139 |
|
|
my $eval_create; |
140 |
dpavlin |
691 |
$self->{_lookup_errors} = (); |
141 |
dpavlin |
686 |
|
142 |
dpavlin |
691 |
sub _lookup_error { |
143 |
|
|
my $self = shift; |
144 |
|
|
my $msg = shift; |
145 |
|
|
$self->_get_logger->logconfess("error without message?") unless ($msg); |
146 |
|
|
push @{ $self->{_lookup_errors} }, $msg; |
147 |
dpavlin |
690 |
return ''; |
148 |
|
|
} |
149 |
|
|
|
150 |
dpavlin |
686 |
$Document->find( sub { |
151 |
|
|
my ($Document,$Element) = @_; |
152 |
|
|
|
153 |
|
|
$Element->isa('PPI::Token::Word') or return ''; |
154 |
|
|
$Element->content eq 'lookup' or return ''; |
155 |
|
|
|
156 |
dpavlin |
687 |
$log->debug("expansion: ", $Element->snext_sibling); |
157 |
dpavlin |
686 |
|
158 |
|
|
my $args = $Element->snext_sibling; |
159 |
|
|
|
160 |
|
|
my @e = $args->child(0)->elements; |
161 |
dpavlin |
687 |
$log->logdie("hum, expect at least 8 elements, got ", scalar @e, " in $args") if ($#e < 8); |
162 |
dpavlin |
686 |
|
163 |
dpavlin |
687 |
if ($log->is_debug) { |
164 |
|
|
my $report = "found " . scalar @e . " elements:\n"; |
165 |
dpavlin |
686 |
|
166 |
dpavlin |
687 |
foreach my $i ( 0 .. $#e ) { |
167 |
|
|
$report .= sprintf("# %-2d: %-30s %s\n", $i, ( $e[$i] || 'undef' ), $e[$i]->class ); |
168 |
|
|
} |
169 |
|
|
|
170 |
|
|
$log->debug($report); |
171 |
dpavlin |
686 |
} |
172 |
|
|
|
173 |
|
|
my $key_element = $e[8]->clone; |
174 |
|
|
|
175 |
dpavlin |
687 |
$log->logdie("key element must be PPI::Structure::Block") unless $key_element->isa('PPI::Structure::Block'); |
176 |
dpavlin |
686 |
|
177 |
dpavlin |
687 |
$log->debug("key part: ", $key_element); |
178 |
dpavlin |
686 |
|
179 |
|
|
my @key; |
180 |
|
|
|
181 |
|
|
$key_element->find( sub { |
182 |
|
|
my $e = $_[1] || die "no element?"; |
183 |
|
|
$e->isa('PPI::Token::Word') or return ''; |
184 |
|
|
$e->content eq 'rec' or return ''; |
185 |
|
|
|
186 |
|
|
my $kf = $e->snext_sibling; |
187 |
|
|
|
188 |
dpavlin |
687 |
$log->debug("key fragment = $kf"); |
189 |
dpavlin |
686 |
|
190 |
|
|
push @key, eval $kf; |
191 |
dpavlin |
687 |
$log->logdie("can't eval { $kf }: $@") if ($@); |
192 |
dpavlin |
686 |
|
193 |
|
|
return 1; |
194 |
|
|
}); |
195 |
|
|
|
196 |
dpavlin |
687 |
my $key = join('-', @key ) || $log->logdie("no key found!"); |
197 |
dpavlin |
686 |
|
198 |
dpavlin |
687 |
$log->debug("key = $key"); |
199 |
dpavlin |
686 |
|
200 |
|
|
my $create = ' |
201 |
|
|
$coderef = ' . $e[7] . $e[8] . '; |
202 |
|
|
foreach my $v ($coderef->()) { |
203 |
|
|
next unless (defined($v) && $v ne \'\'); |
204 |
|
|
push @{ $lookup->{\'' . $key . '\'}->{$v} }, $mfn; |
205 |
|
|
} |
206 |
|
|
'; |
207 |
|
|
|
208 |
dpavlin |
687 |
$log->debug("create: $create"); |
209 |
dpavlin |
686 |
|
210 |
dpavlin |
691 |
return $self->_lookup_error("invalid database $e[3] in $path" ) unless $self->valid_database( $e[3] ); |
211 |
|
|
return $self->_lookup_error("invalid input $e[5] of database $e[3] in $path", ) unless $self->valid_database_input( $e[3], $e[5] ); |
212 |
dpavlin |
689 |
|
213 |
dpavlin |
686 |
$eval_create->{ $e[3] }->{ $e[5] } .= $create; |
214 |
|
|
|
215 |
|
|
if ($#e < 10) { |
216 |
|
|
$e[8]->insert_after( $e[8]->clone ); |
217 |
|
|
$e[8]->insert_after( $e[7]->clone ); |
218 |
|
|
$e[8]->insert_after( $e[6]->clone ); |
219 |
|
|
} |
220 |
|
|
|
221 |
|
|
$e[7]->remove; |
222 |
|
|
$e[8]->insert_before( new PPI::Token::Quote::Single( "'$key'" ) ); |
223 |
|
|
$e[8]->remove; |
224 |
|
|
|
225 |
|
|
|
226 |
dpavlin |
687 |
$log->debug(">>> ", $Element->snext_sibling); |
227 |
dpavlin |
686 |
}); |
228 |
|
|
|
229 |
dpavlin |
687 |
$log->info("create: ", dump($eval_create) ); |
230 |
|
|
$log->info("lookup: ", $Document->serialize ); |
231 |
dpavlin |
686 |
|
232 |
dpavlin |
687 |
if ($self->{debug}) { |
233 |
|
|
my $Dumper = PPI::Dumper->new( $Document ); |
234 |
|
|
$Dumper->print; |
235 |
|
|
} |
236 |
dpavlin |
686 |
|
237 |
dpavlin |
691 |
$log->error("Parser errors: ", join("\n",@{ $self->{_lookup_errors} }) ) if ($self->{_lookup_errors}); |
238 |
dpavlin |
690 |
|
239 |
dpavlin |
687 |
return 1; |
240 |
dpavlin |
686 |
} |
241 |
|
|
|
242 |
dpavlin |
689 |
=head2 valid_database |
243 |
|
|
|
244 |
|
|
my $ok = $parse->valid_database('key'); |
245 |
|
|
|
246 |
|
|
=cut |
247 |
|
|
|
248 |
|
|
sub valid_database { |
249 |
|
|
my $self = shift; |
250 |
|
|
|
251 |
|
|
my $database = shift || return; |
252 |
|
|
$database =~ s/['"]//g; |
253 |
|
|
|
254 |
|
|
return defined($self->{valid_inputs}->{$database}); |
255 |
|
|
} |
256 |
|
|
|
257 |
|
|
=head2 valid_database_input |
258 |
|
|
|
259 |
|
|
my $ok = $parse->valid_database('database_key','input_name'); |
260 |
|
|
|
261 |
|
|
=cut |
262 |
|
|
|
263 |
|
|
sub valid_database_input { |
264 |
|
|
my $self = shift; |
265 |
|
|
|
266 |
|
|
my ($database,$input) = @_; |
267 |
|
|
$database =~ s/['"]//g; |
268 |
|
|
$input =~ s/['"]//g; |
269 |
|
|
|
270 |
|
|
return defined($self->{valid_inputs}->{$database}->{$input}); |
271 |
|
|
} |
272 |
|
|
|
273 |
dpavlin |
686 |
=head1 AUTHOR |
274 |
|
|
|
275 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
276 |
|
|
|
277 |
|
|
=head1 COPYRIGHT & LICENSE |
278 |
|
|
|
279 |
|
|
Copyright 2006 Dobrica Pavlinusic, All Rights Reserved. |
280 |
|
|
|
281 |
|
|
This program is free software; you can redistribute it and/or modify it |
282 |
|
|
under the same terms as Perl itself. |
283 |
|
|
|
284 |
|
|
=cut |
285 |
|
|
|
286 |
|
|
1; # End of WebPAC::Parser |