17 |
|
|
18 |
=head1 VERSION |
=head1 VERSION |
19 |
|
|
20 |
Version 0.03 |
Version 0.05 |
21 |
|
|
22 |
=cut |
=cut |
23 |
|
|
24 |
our $VERSION = '0.03'; |
our $VERSION = '0.05'; |
25 |
|
|
26 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
27 |
|
|
28 |
|
This module will parse L<WebPAC::Normalize/lookup> directives and generate source |
29 |
|
to produce lookups and normalization. |
30 |
|
|
31 |
|
It's written using L<PPI>, pure-perl parser for perl and heavily influenced by |
32 |
|
reading about LISP. It might be a bit over-the board, but at least it removed |
33 |
|
separate configuration files for lookups. |
34 |
|
|
35 |
|
This is experimental code, but it replaces all older formats which where, |
36 |
|
at one point in time, available in WebPAC. |
37 |
|
|
38 |
FIXME |
FIXME |
39 |
|
|
40 |
=head1 FUNCTIONS |
=head1 FUNCTIONS |
61 |
|
|
62 |
$log->logdie("can't iterate_inputs over this config object") unless ($self->{config}->can('iterate_inputs')); |
$log->logdie("can't iterate_inputs over this config object") unless ($self->{config}->can('iterate_inputs')); |
63 |
|
|
64 |
$self->read_sources; |
$self->_read_sources; |
|
|
|
|
$self->{config}->iterate_inputs( sub { |
|
|
my ($input, $database) = @_; |
|
|
return unless $self->valid_database_input($database, $input->{name}); |
|
|
$self->parse_lookups($database,$input->{name}); |
|
|
} ); |
|
65 |
|
|
66 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
67 |
} |
} |
68 |
|
|
69 |
=head2 read_sources |
=head2 valid_database |
70 |
|
|
71 |
|
my $ok = $parse->valid_database('key'); |
72 |
|
|
73 |
|
=cut |
74 |
|
|
75 |
|
sub valid_database { |
76 |
|
my $self = shift; |
77 |
|
|
78 |
|
my $database = shift || return; |
79 |
|
|
80 |
|
return defined($self->{valid_inputs}->{ _q($database) }); |
81 |
|
} |
82 |
|
|
83 |
|
=head2 valid_database_input |
84 |
|
|
85 |
|
my $ok = $parse->valid_database('database_key','input_name'); |
86 |
|
|
87 |
|
=cut |
88 |
|
|
89 |
|
sub valid_database_input { |
90 |
|
my $self = shift; |
91 |
|
my ($database,$input) = @_; |
92 |
|
$input = _input_name($input); |
93 |
|
return defined($self->{valid_inputs}->{ _q($database) }->{ _q($input) }); |
94 |
|
} |
95 |
|
|
96 |
|
=head2 depends |
97 |
|
|
98 |
|
Return all databases and inputs on which specified one depends |
99 |
|
|
100 |
|
$depends_on = $parser->depends('database','input'); |
101 |
|
|
102 |
|
=cut |
103 |
|
|
104 |
|
sub depends { |
105 |
|
my $self = shift; |
106 |
|
my ($database,$input) = @_; |
107 |
|
$input = _input_name($input); |
108 |
|
$self->_get_logger->debug("depends($database,$input)"); |
109 |
|
return unless ( |
110 |
|
defined( $self->{depends}->{ _q($database) } ) && |
111 |
|
defined( $self->{depends}->{ _q($database) }->{ _q($input) } ) |
112 |
|
); |
113 |
|
return $self->{depends}->{ _q($database) }->{ _q($input) }; |
114 |
|
} |
115 |
|
|
116 |
|
=head2 lookup_create_rules |
117 |
|
|
118 |
my $source_files = $parser->read_sources; |
my $source = $parser->lookup_create_rules($database, $input); |
119 |
|
|
120 |
|
=cut |
121 |
|
|
122 |
|
sub lookup_create_rules { |
123 |
|
my $self = shift; |
124 |
|
my ($database,$input) = @_; |
125 |
|
$input = _input_name($input); |
126 |
|
return unless ( |
127 |
|
defined( $self->{_lookup_create}->{ _q($database) } ) && |
128 |
|
defined( $self->{_lookup_create}->{ _q($database) }->{ _q($input) } ) |
129 |
|
); |
130 |
|
return $self->{_lookup_create}->{ _q($database) }->{ _q($input) }; |
131 |
|
} |
132 |
|
|
133 |
|
=head1 PRIVATE |
134 |
|
|
135 |
|
=head2 _read_sources |
136 |
|
|
137 |
|
my $source_files = $parser->_read_sources; |
138 |
|
|
139 |
Called by L</new>. |
Called by L</new>. |
140 |
|
|
141 |
=cut |
=cut |
142 |
|
|
143 |
sub read_sources { |
sub _read_sources { |
144 |
my $self = shift; |
my $self = shift; |
145 |
|
|
146 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
147 |
|
|
148 |
my $nr = 0; |
my $nr = 0; |
149 |
|
|
150 |
|
my @lookups; |
151 |
|
|
152 |
$self->{config}->iterate_inputs( sub { |
$self->{config}->iterate_inputs( sub { |
153 |
my ($input, $database) = @_; |
my ($input, $database) = @_; |
154 |
|
|
155 |
my $path = $input->{normalize}->{path} || return; |
$log->debug("database: $database input = ", dump($input)); |
|
my $full = $self->{base_path} ? $self->{base_path} . '/' . $path : $path; |
|
156 |
|
|
157 |
$log->logdie("normalization input $full doesn't exist") unless (-e $full); |
foreach my $normalize (@{ $input->{normalize} }) { |
158 |
|
|
159 |
my $s = read_file( $full ) || $log->logdie("can't read $full: $!"); |
my $path = $normalize->{path}; |
160 |
|
return unless($path); |
161 |
|
my $full = $self->{base_path} ? $self->{base_path} . '/' . $path : $path; |
162 |
|
|
163 |
my $input_name = $input->{name} || $log->logdie("can't deduce name of input: ", dump( $input )); |
$log->logdie("normalization input $full doesn't exist") unless (-e $full); |
164 |
|
|
165 |
$log->debug("$database/$input_name: adding $path"); |
my $s = read_file( $full ) || $log->logdie("can't read $full: $!"); |
166 |
|
|
167 |
$self->{valid_inputs}->{$database}->{$input_name} = { |
my $input_name = _input_name($input) || $log->logdie("can't deduce name of input: ", dump( $input )); |
|
source => $s, |
|
|
path => $full, |
|
|
usage => 0, |
|
|
} unless defined($self->{valid_inputs}->{$database}->{$input_name}); |
|
168 |
|
|
169 |
$self->{valid_inputs}->{$database}->{$input_name}->{usage}++; |
$log->debug("$database/$input_name: adding $path"); |
170 |
|
|
171 |
$nr++; |
$self->{valid_inputs}->{$database}->{$input_name}++; |
172 |
|
|
173 |
|
push @lookups, sub { |
174 |
|
$self->_parse_lookups( $database, $input_name, $full, $s ); |
175 |
|
}; |
176 |
|
|
177 |
|
$nr++; |
178 |
|
} |
179 |
} ); |
} ); |
180 |
|
|
181 |
$log->debug("found $nr source files"); |
$log->debug("found $nr source files"); |
182 |
|
|
183 |
|
# parse all lookups |
184 |
|
$_->() foreach (@lookups); |
185 |
|
|
186 |
return $nr; |
return $nr; |
187 |
} |
} |
188 |
|
|
189 |
=head2 parse_lookup |
=head2 _parse_lookups |
190 |
|
|
191 |
|
$parser->_parse_lookups($database,$input,$path,$source); |
192 |
|
|
193 |
|
Called for each normalize source (rules) in each input by L</read_sources> |
194 |
|
|
195 |
$parser->parse_lookups($database,$input); |
It will report invalid databases and inputs in error log after parsing. |
196 |
|
|
197 |
=cut |
=cut |
198 |
|
|
199 |
sub parse_lookups { |
sub _parse_lookups { |
200 |
my $self = shift; |
my $self = shift; |
201 |
my ($database, $input) = @_; |
my ($database, $input, $path, $source) = @_; |
202 |
|
|
203 |
|
$input = _input_name($input); |
204 |
|
|
205 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
206 |
|
|
207 |
$log->logdie("invalid database $database" ) unless $self->valid_database( $database ); |
$log->logdie("invalid database $database" ) unless $self->valid_database( $database ); |
208 |
$log->logdie("invalid input $input of database $database", ) unless $self->valid_database_input( $database, $input ); |
$log->logdie("invalid input $input of database $database", ) unless $self->valid_database_input( $database, $input ); |
209 |
|
|
|
my $source = $self->{valid_inputs}->{$database}->{$input}->{source}; |
|
|
my $path = $self->{valid_inputs}->{$database}->{$input}->{path}; |
|
|
|
|
210 |
$log->logdie("no source found for database $database input $input path $path") unless ($source); |
$log->logdie("no source found for database $database input $input path $path") unless ($source); |
211 |
|
|
212 |
$log->info("parsing lookups for $database/$input in $path"); |
$log->info("parsing lookups for $database/$input from $path (",length($source)," bytes)"); |
213 |
|
|
214 |
my $Document = PPI::Document->new( \$source ) || $log->logdie("can't parse source:\n", $self->{source}); |
my $Document = PPI::Document->new( \$source ) || $log->logdie("can't parse source:\n", $self->{source}); |
215 |
|
|
218 |
|
|
219 |
# Find all the named subroutines |
# Find all the named subroutines |
220 |
|
|
|
my $eval_create; |
|
221 |
$self->{_lookup_errors} = (); |
$self->{_lookup_errors} = (); |
222 |
|
|
223 |
sub _lookup_error { |
sub _lookup_error { |
291 |
return $self->_lookup_error("invalid database $e[3] in $path" ) unless $self->valid_database( $e[3] ); |
return $self->_lookup_error("invalid database $e[3] in $path" ) unless $self->valid_database( $e[3] ); |
292 |
return $self->_lookup_error("invalid input $e[5] of database $e[3] in $path", ) unless $self->valid_database_input( $e[3], $e[5] ); |
return $self->_lookup_error("invalid input $e[5] of database $e[3] in $path", ) unless $self->valid_database_input( $e[3], $e[5] ); |
293 |
|
|
294 |
$eval_create->{ $e[3] }->{ $e[5] } .= $create; |
# save code to create this lookup |
295 |
|
$self->{_lookup_create}->{ _q($e[3]) }->{ _q($e[5]) }->{ _q($key) } .= "# lookup for $e[3]/$e[5]/$key\n\n$create"; |
296 |
|
|
297 |
|
|
298 |
|
if (defined( $self->{depends}->{ $database }->{ $input }->{ _q($e[3]) }->{ _q($e[5]) }->{ $key } )) { |
299 |
|
$log->warn("dependency of $database/$input on $e[3]/$e[5] with $key allready recorded"); |
300 |
|
} |
301 |
|
|
302 |
|
# save this dependency |
303 |
|
$self->{depends}->{ $database }->{ $input }->{ _q($e[3]) }->{ _q($e[5]) }->{$key}++; |
304 |
|
|
305 |
if ($#e < 10) { |
if ($#e < 10) { |
306 |
$e[8]->insert_after( $e[8]->clone ); |
$e[8]->insert_after( $e[8]->clone ); |
316 |
$log->debug(">>> ", $Element->snext_sibling); |
$log->debug(">>> ", $Element->snext_sibling); |
317 |
}); |
}); |
318 |
|
|
319 |
$log->info("create: ", dump($eval_create) ); |
my $normalize_source = $Document->serialize; |
320 |
$log->info("lookup: ", $Document->serialize ); |
$log->debug("create: ", dump($self->{_lookup_create}) ); |
321 |
|
$log->debug("normalize: $normalize_source"); |
322 |
|
|
323 |
|
$self->{_normalize_source}->{$database}->{$input} = $normalize_source; |
324 |
|
|
325 |
if ($self->{debug}) { |
if ($self->{debug}) { |
326 |
my $Dumper = PPI::Dumper->new( $Document ); |
my $Dumper = PPI::Dumper->new( $Document ); |
327 |
$Dumper->print; |
$Dumper->print; |
328 |
} |
} |
329 |
|
|
330 |
$log->error("Parser errors: ", join("\n",@{ $self->{_lookup_errors} }) ) if ($self->{_lookup_errors}); |
$log->error("Parser errors:\n", join("\n",@{ $self->{_lookup_errors} }) ) if ($self->{_lookup_errors}); |
331 |
|
|
332 |
return 1; |
return 1; |
333 |
} |
} |
334 |
|
|
|
=head2 valid_database |
|
335 |
|
|
336 |
my $ok = $parse->valid_database('key'); |
=head2 _q |
337 |
|
|
338 |
=cut |
Strip single or double quotes around value |
339 |
|
|
340 |
sub valid_database { |
_q(qq/'foo'/) -> foo |
|
my $self = shift; |
|
341 |
|
|
342 |
my $database = shift || return; |
=cut |
|
$database =~ s/['"]//g; |
|
343 |
|
|
344 |
return defined($self->{valid_inputs}->{$database}); |
sub _q { |
345 |
|
my $v = shift || return; |
346 |
|
$v =~ s/^['"]*//g; |
347 |
|
$v =~ s/['"]*$//g; |
348 |
|
return $v; |
349 |
} |
} |
350 |
|
|
351 |
=head2 valid_database_input |
=head2 _input_name |
352 |
|
|
353 |
my $ok = $parse->valid_database('database_key','input_name'); |
Return C<name> value if HASH or arg if scalar |
|
|
|
|
=cut |
|
354 |
|
|
355 |
sub valid_database_input { |
_input_name($input) |
|
my $self = shift; |
|
356 |
|
|
357 |
my ($database,$input) = @_; |
=cut |
|
$database =~ s/['"]//g; |
|
|
$input =~ s/['"]//g; |
|
358 |
|
|
359 |
return defined($self->{valid_inputs}->{$database}->{$input}); |
sub _input_name { |
360 |
|
my $input = shift || return; |
361 |
|
if (ref($input) eq 'HASH') { |
362 |
|
die "can't find 'name' value in ", dump($input) unless defined($input->{name}); |
363 |
|
return $input->{name}; |
364 |
|
} else { |
365 |
|
return $input; |
366 |
|
} |
367 |
} |
} |
368 |
|
|
369 |
|
|
370 |
=head1 AUTHOR |
=head1 AUTHOR |
371 |
|
|
372 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |