| 1 |
# Dobrica Pavlinusic, <dpavlin@rot13.org> 02/22/07 20:30:00 CET |
| 2 |
|
| 3 |
use strict; |
| 4 |
use warnings; |
| 5 |
|
| 6 |
package Grep::Source; |
| 7 |
|
| 8 |
use Carp qw/verbose/; |
| 9 |
use Module::Pluggable search_path => 'Grep::Source', sub_name => 'sources', require => 1; |
| 10 |
use base qw(Class::Accessor Jifty::Object); |
| 11 |
Grep::Source->mk_accessors( qw(feed uri q new_items collection search_obj tree) ); |
| 12 |
|
| 13 |
use HTML::TreeBuilder; |
| 14 |
use WWW::Mechanize; |
| 15 |
use XML::Feed; |
| 16 |
use URI; |
| 17 |
use HTML::ResolveLink; |
| 18 |
|
| 19 |
use Data::Dump qw/dump/; |
| 20 |
|
| 21 |
=head1 NAME |
| 22 |
|
| 23 |
Grep::Source - base class for implementation of different sources for Grep |
| 24 |
|
| 25 |
=head1 METHODS |
| 26 |
|
| 27 |
This is mostly documentation because most of methods are implemented by plugins. |
| 28 |
|
| 29 |
=head2 sources |
| 30 |
|
| 31 |
my @sources = Grep::Source->sources(); |
| 32 |
|
| 33 |
Returns all available sources. |
| 34 |
|
| 35 |
=cut |
| 36 |
|
| 37 |
Jifty->log->debug("Found source plugins: ", join(", ", __PACKAGE__->sources() ) ); |
| 38 |
|
| 39 |
=head2 new |
| 40 |
|
| 41 |
my $source = Grep::Source->new({ feed => $feed_record }); |
| 42 |
|
| 43 |
This will also setup: |
| 44 |
|
| 45 |
=head2 feed |
| 46 |
|
| 47 |
isa L<Grep::Model::Feed> |
| 48 |
|
| 49 |
=head2 search |
| 50 |
|
| 51 |
my $collection = $source->search( 'query string' ); |
| 52 |
|
| 53 |
It will also setup following accessors: |
| 54 |
|
| 55 |
=head2 q |
| 56 |
|
| 57 |
Search query |
| 58 |
|
| 59 |
=head2 uri |
| 60 |
|
| 61 |
URI of feed with embedded search query |
| 62 |
|
| 63 |
=head2 new_items |
| 64 |
|
| 65 |
Number of new items in result collection |
| 66 |
|
| 67 |
=head2 collection |
| 68 |
|
| 69 |
Actuall results which is L<Grep::Model::ItemCollection>, so following will |
| 70 |
work: |
| 71 |
|
| 72 |
print "and ", $self->collection->count, " total items"; |
| 73 |
|
| 74 |
|
| 75 |
Also setups number of new items |
| 76 |
|
| 77 |
print $source->new_items, " items new"; |
| 78 |
|
| 79 |
=cut |
| 80 |
|
| 81 |
sub search { |
| 82 |
my $self = shift; |
| 83 |
|
| 84 |
my $q = shift; |
| 85 |
|
| 86 |
$q ? $self->q( $q ) : $q = $self->q; |
| 87 |
|
| 88 |
die "no q?" unless ( $self->q ); |
| 89 |
die "no feed?" unless ( $self->feed ); |
| 90 |
die "feed not Grep::Model::Feed" unless ( $self->feed->isa('Grep::Model::Feed') ); |
| 91 |
|
| 92 |
my $message; |
| 93 |
my $uri = $self->feed->uri; |
| 94 |
if ($uri =~ m/%s/) { |
| 95 |
$uri = $self->feed->search_uri( $q ); |
| 96 |
$message = 'Searching'; |
| 97 |
} else { |
| 98 |
$message = 'Fetching'; |
| 99 |
} |
| 100 |
$message .= ' ' . $self->feed->title . " at $uri"; |
| 101 |
|
| 102 |
$self->uri( $uri ); |
| 103 |
|
| 104 |
$self->log->info( $message ); |
| 105 |
|
| 106 |
$self->collection( Grep::Model::ItemCollection->new() ); |
| 107 |
|
| 108 |
my $class = $self->feed->source || 'Grep::Source::Feed'; |
| 109 |
$self->log->debug("using $class"); |
| 110 |
|
| 111 |
$self->search_obj( Grep::Search->new() ); |
| 112 |
$self->log->debug("created " . $self->search_obj); |
| 113 |
|
| 114 |
$class->fetch( $self ); |
| 115 |
|
| 116 |
$self->search_obj->finish; |
| 117 |
|
| 118 |
return $self->collection; |
| 119 |
} |
| 120 |
|
| 121 |
=head2 add_record |
| 122 |
|
| 123 |
Plugins will be called with parametar C<$parent> so they can call this method to add |
| 124 |
record into result collection (and store in cache and index). |
| 125 |
|
| 126 |
$parent->add_record( id => 42, foo => 'bar', ... ); |
| 127 |
|
| 128 |
This will also update L</new_items> |
| 129 |
|
| 130 |
=cut |
| 131 |
|
| 132 |
sub add_record { |
| 133 |
my $self = shift; |
| 134 |
|
| 135 |
$self->log->confess("no search_obj") unless ($self->search_obj); |
| 136 |
|
| 137 |
my $i = Grep::Model::Item->new(); |
| 138 |
|
| 139 |
my $rec = {@_}; |
| 140 |
|
| 141 |
$self->log->debug("resolving links using base ", $rec->{link}); |
| 142 |
my $resolver = HTML::ResolveLink->new( base => $rec->{link} ); |
| 143 |
$rec->{content} = $resolver->resolve( $rec->{content} ); |
| 144 |
|
| 145 |
my ($ok,$msg) = $i->load_or_create( %$rec ); |
| 146 |
|
| 147 |
$msg ||= ''; |
| 148 |
|
| 149 |
if ( $ok ) { |
| 150 |
$self->log->debug("item ", $i->id, ": $msg"); |
| 151 |
$self->collection->add_record( $i ); |
| 152 |
|
| 153 |
# is new record? |
| 154 |
if ( $msg !~ m/^Found/ ) { |
| 155 |
$self->search_obj->add( $i ); |
| 156 |
$self->new_items( ( $self->new_items || 0 ) + 1 ); |
| 157 |
} |
| 158 |
} else { |
| 159 |
warn "can't add entry ", dump( @_ ), "\n"; |
| 160 |
} |
| 161 |
} |
| 162 |
|
| 163 |
=head2 content_class |
| 164 |
|
| 165 |
Return class registred for particular content. |
| 166 |
|
| 167 |
my $class = $source->content_class( $content ); |
| 168 |
|
| 169 |
=cut |
| 170 |
|
| 171 |
sub content_class { |
| 172 |
my $self = shift; |
| 173 |
|
| 174 |
my $content = shift or die "no content?"; |
| 175 |
|
| 176 |
foreach my $s ( $self->sources ) { |
| 177 |
$self->log->debug("testing source class $s"); |
| 178 |
if ( $s->can('content_have') ) { |
| 179 |
my $regex = $s->content_have( $content ) or |
| 180 |
die "${s}->content_have didn't return anything"; |
| 181 |
die "${s}->content_have didn't return regex but ", dump( $regex ), " ref ", ref( $regex ) |
| 182 |
unless ( ref($regex) eq 'Regexp' ); |
| 183 |
if ( $content =~ $regex ) { |
| 184 |
$self->log->debug("${s}->content_have succesful"); |
| 185 |
return $s; |
| 186 |
} |
| 187 |
} |
| 188 |
} |
| 189 |
} |
| 190 |
|
| 191 |
|
| 192 |
=head2 element_by_triplet |
| 193 |
|
| 194 |
Helper method to select element(s) using C<element/attribute/value> triplet using |
| 195 |
L<HTML::TreeBuilder> trees. |
| 196 |
|
| 197 |
my $el = $self->element_by_triplet( |
| 198 |
tree => $tree_or_element, |
| 199 |
triplets => [ qw/ |
| 200 |
div id target |
| 201 |
div class another |
| 202 |
/ ], |
| 203 |
message => 'find search result element', |
| 204 |
fatal => 1, # die instead of warn |
| 205 |
); |
| 206 |
|
| 207 |
=cut |
| 208 |
|
| 209 |
sub element_by_triplet { |
| 210 |
my $self = shift; |
| 211 |
|
| 212 |
my $args = {@_}; |
| 213 |
|
| 214 |
my $tree = $args->{tree} || die "no tree"; |
| 215 |
my $message = $args->{message} || ''; |
| 216 |
my $fatal = $args->{fatal}; |
| 217 |
die "no triplets" unless defined( $args->{triplets} ); |
| 218 |
my @triplets; |
| 219 |
if ( ref( $args->{triplets} ) eq 'ARRAY' ) { |
| 220 |
@triplets = @{ $args->{triplets} }; |
| 221 |
} else { |
| 222 |
@triplets = ( $args->{triplets} ); |
| 223 |
} |
| 224 |
|
| 225 |
push @triplets, ( undef, undef ) if ( $#triplets == 0 ); |
| 226 |
|
| 227 |
die "triplet doesn't have 3 elements but ", $#triplets unless ( |
| 228 |
( $#triplets + 1 ) % 3 == 0 |
| 229 |
); |
| 230 |
|
| 231 |
my ( $el, $attr, $value ); |
| 232 |
|
| 233 |
my @results; |
| 234 |
my @tags; |
| 235 |
|
| 236 |
$self->log->debug("looking for $message ", dump( @triplets )); |
| 237 |
while ( @triplets ) { |
| 238 |
( $el,$attr,$value ) = splice( @triplets, 0, 3 ); |
| 239 |
my $tag = $attr ? "<$el $attr=\"$value\">" : "<$el>"; |
| 240 |
push @tags, $tag; |
| 241 |
@results = $tree->look_down( '_tag', $el, sub { |
| 242 |
return 1 unless ( $attr && $value ); |
| 243 |
( $_[0]->attr( $attr ) || '' ) =~ m/\b\Q$value\E\b/ |
| 244 |
}); |
| 245 |
last if @results; |
| 246 |
} |
| 247 |
|
| 248 |
if ( ! @results ) { |
| 249 |
my $msg = "can't find $message " . join(" ", @tags); |
| 250 |
die $msg if ( $fatal ); |
| 251 |
#warn $msg; |
| 252 |
return; |
| 253 |
} |
| 254 |
|
| 255 |
$self->log->debug("found ", $#results + 1, " elements"); |
| 256 |
#warn dump( map { $_->as_HTML } @results ); |
| 257 |
|
| 258 |
return @results if wantarray; |
| 259 |
return shift @results; |
| 260 |
} |
| 261 |
|
| 262 |
=head2 scrape |
| 263 |
|
| 264 |
Create semi-complex L<WWW::Mechanize> rules to scrape page easily |
| 265 |
|
| 266 |
$parent->scrape( |
| 267 |
# if search string isn't part or URI |
| 268 |
submit_form => { |
| 269 |
fields => { |
| 270 |
value => $parent->q, |
| 271 |
}, |
| 272 |
button => 'fullsearch', |
| 273 |
}, |
| 274 |
# element with search results |
| 275 |
wrapper => [ qw/div class searchresults/ ], |
| 276 |
# element (or tripple) for each result with link |
| 277 |
# <a href=".."> inside it to full-text result |
| 278 |
results => 'dt', |
| 279 |
# collect which element on page linked from results |
| 280 |
scrape => [ qw/div id page/ ], |
| 281 |
# when search returns just single hit, it will redirect to result page |
| 282 |
redirect_single_result => 1, |
| 283 |
); |
| 284 |
|
| 285 |
=cut |
| 286 |
|
| 287 |
sub scrape { |
| 288 |
my $self = shift; |
| 289 |
|
| 290 |
my $args = {@_}; |
| 291 |
|
| 292 |
$self->log->debug("scrape with args ",dump($args)); |
| 293 |
|
| 294 |
my ($feed,$uri,$q) = ($self->feed, $self->uri,$self->q); |
| 295 |
die "no uri" unless ($uri); |
| 296 |
die "feed is not a Grep::Model::Feed but ", ref $feed unless $feed->isa('Grep::Model::Feed'); |
| 297 |
|
| 298 |
sub mech_warn { |
| 299 |
my $m = shift || return; |
| 300 |
warn $m; |
| 301 |
} |
| 302 |
|
| 303 |
my $mech = WWW::Mechanize->new( |
| 304 |
cookie_jar => {}, |
| 305 |
onwarn => \&mech_warn, |
| 306 |
onerror => \&mech_warn, |
| 307 |
); |
| 308 |
|
| 309 |
$mech->get( $uri ); |
| 310 |
|
| 311 |
$self->save( 'get.html', $mech->content ); |
| 312 |
|
| 313 |
if ( my $form = $args->{submit_form} ) { |
| 314 |
$self->log->debug("submit form on $uri with ", dump( $form )); |
| 315 |
$mech->submit_form( %$form ) or die "can't submit form ", dump( $form ); |
| 316 |
$self->save( 'submit.html', $mech->content ); |
| 317 |
} |
| 318 |
|
| 319 |
$self->log->debug("parse result page"); |
| 320 |
|
| 321 |
my $tree = HTML::TreeBuilder->new or die "can't create html tree"; |
| 322 |
$tree->parse( $mech->content ) or die "can't parse fetched content"; |
| 323 |
|
| 324 |
my @wrapper_divs = $self->element_by_triplet( |
| 325 |
tree => $tree, |
| 326 |
triplets => $args->{wrapper}, |
| 327 |
message => 'wrapper for all results', |
| 328 |
# on closer recollection, this shouldn't be ever fatal, because |
| 329 |
# "no results found" page might not contain wrapper |
| 330 |
#fatal => $args->{redirect_single_result} ? 0 : 1, |
| 331 |
); |
| 332 |
|
| 333 |
my $max = 15; |
| 334 |
my $nr = 1; |
| 335 |
|
| 336 |
my $base_uri = $uri; |
| 337 |
$base_uri =~ s!\?.*$!!; |
| 338 |
|
| 339 |
# directly got first result |
| 340 |
if ( $args->{redirect_single_result} && ! @wrapper_divs ) { |
| 341 |
|
| 342 |
my $uri = $mech->uri; $uri->query( undef ); $uri = $uri->canonical; |
| 343 |
|
| 344 |
my $div = $self->element_by_triplet( |
| 345 |
tree => $tree, |
| 346 |
message => "single result - redirect to $uri", |
| 347 |
triplets => $args->{scrape}, |
| 348 |
fatal => 0, |
| 349 |
); |
| 350 |
|
| 351 |
$self->add_record( |
| 352 |
in_feed => $feed, |
| 353 |
title => $mech->title, |
| 354 |
link => $uri, |
| 355 |
content => $div->as_HTML, |
| 356 |
); |
| 357 |
|
| 358 |
$tree->delete; # clear memory! |
| 359 |
return; |
| 360 |
} |
| 361 |
|
| 362 |
my @r; |
| 363 |
|
| 364 |
foreach my $div ( @wrapper_divs ) { |
| 365 |
|
| 366 |
my @r_here = $self->element_by_triplet( |
| 367 |
tree => $div, |
| 368 |
triplets => $args->{results}, |
| 369 |
message => 'result element', |
| 370 |
); |
| 371 |
|
| 372 |
push @r, @r_here if (@r_here); |
| 373 |
} |
| 374 |
|
| 375 |
$self->log->debug("in total, found ", $#r + 1, " results in ", $#wrapper_divs + 1, " result wrapper elements"); |
| 376 |
|
| 377 |
foreach my $dt ( @r ) { |
| 378 |
my $a = $dt->look_down( '_tag', 'a', sub { $_[0]->attr('href') } ); |
| 379 |
if ( $a ) { |
| 380 |
|
| 381 |
my $href = $a->attr('href') or die "can't find href inside <", $args->{results}, ">"; |
| 382 |
|
| 383 |
my $page_uri = URI->new_abs( $href, $base_uri ); |
| 384 |
$page_uri->query( undef ); |
| 385 |
$page_uri = $page_uri->canonical; |
| 386 |
|
| 387 |
if ( my $item = Grep::Model::Item->link_current( $page_uri ) ) { |
| 388 |
Jifty->log->debug("using cached page for $page_uri"); |
| 389 |
$self->collection->add_record( $item ); |
| 390 |
next; |
| 391 |
} |
| 392 |
|
| 393 |
$self->log->debug("fetching page: ",$a->as_text," from $page_uri"); |
| 394 |
if ( $mech->follow_link( url => $href ) ) { |
| 395 |
|
| 396 |
$self->save( "page-${nr}.html", $mech->content ); |
| 397 |
|
| 398 |
my $page_tree = HTML::TreeBuilder->new or die "can't create page tree"; |
| 399 |
$page_tree->parse( $mech->content ) or die "can't parse page at $page_uri"; |
| 400 |
my @divs = $self->element_by_triplet( |
| 401 |
tree => $page_tree, |
| 402 |
message => "result page $nr", |
| 403 |
triplets => $args->{scrape} |
| 404 |
); |
| 405 |
|
| 406 |
if ( @divs ) { |
| 407 |
|
| 408 |
my $html = join("<hr/>\n", map { $_->as_HTML } @divs ); |
| 409 |
$self->log->debug("combined ", $#divs + 1, " elements elements in ", length($html), " bytes"); |
| 410 |
|
| 411 |
$self->add_record( |
| 412 |
in_feed => $feed, |
| 413 |
title => $mech->title, |
| 414 |
link => $page_uri, |
| 415 |
content => $html, |
| 416 |
# summary => |
| 417 |
# category => |
| 418 |
# author => |
| 419 |
# issued => |
| 420 |
# modified => |
| 421 |
); |
| 422 |
|
| 423 |
} else { |
| 424 |
$self->log->debug("NO CONTENT scraped from page $nr"); |
| 425 |
} |
| 426 |
|
| 427 |
$mech->back; |
| 428 |
$page_tree->delete; |
| 429 |
|
| 430 |
} else { |
| 431 |
warn "can't follow uri $page_uri: $!\n"; |
| 432 |
} |
| 433 |
} else { |
| 434 |
$self->log->debug("result $nr doesn't have link inside, ignoring..."); |
| 435 |
} |
| 436 |
|
| 437 |
last if ($nr == $max); |
| 438 |
$nr++; |
| 439 |
} |
| 440 |
|
| 441 |
$tree->delete; # clear memory! |
| 442 |
|
| 443 |
} |
| 444 |
|
| 445 |
=head2 save |
| 446 |
|
| 447 |
save( 'name', $content ); |
| 448 |
|
| 449 |
Save dumps into C</tmp/grep> if writable |
| 450 |
|
| 451 |
=cut |
| 452 |
|
| 453 |
sub save { |
| 454 |
my $self = shift; |
| 455 |
my ( $file, $content ) = @_; |
| 456 |
return unless ( defined($file) && defined($content) ); |
| 457 |
if ( -w '/tmp/grep' ) { |
| 458 |
open(my $f, '>', "/tmp/grep/$file") or die "can't open $file: $!"; |
| 459 |
print $f $content or die "can't write to $file: $!"; |
| 460 |
close $f or die "can't close $file: $!"; |
| 461 |
$self->log->debug("saved $file ",length($content)," bytes"); |
| 462 |
} |
| 463 |
} |
| 464 |
|
| 465 |
1; |
Parent Directory
| 
Revision Log