135 |
|
|
136 |
my $rec = {@_}; |
my $rec = {@_}; |
137 |
|
|
138 |
warn "resolving links"; |
$self->log->debug("resolving links using base ", $rec->{link}); |
139 |
my $resolver = HTML::ResolveLink->new( base => $rec->{link} ); |
my $resolver = HTML::ResolveLink->new( base => $rec->{link} ); |
140 |
$rec->{content} = $resolver->resolve( $rec->{content} ); |
$rec->{content} = $resolver->resolve( $rec->{content} ); |
141 |
|
|
197 |
|
|
198 |
my $args = {@_}; |
my $args = {@_}; |
199 |
|
|
200 |
warn "scrape got args ",dump($args); |
$self->log->debug("scrape with args ",dump($args)); |
201 |
|
|
202 |
my ($feed,$uri,$q) = ($self->feed, $self->uri,$self->q); |
my ($feed,$uri,$q) = ($self->feed, $self->uri,$self->q); |
203 |
die "no uri" unless ($uri); |
die "no uri" unless ($uri); |
219 |
$self->save( 'get.html', $mech->content ); |
$self->save( 'get.html', $mech->content ); |
220 |
|
|
221 |
if ( my $form = $args->{submit_form} ) { |
if ( my $form = $args->{submit_form} ) { |
222 |
warn "submit form on $uri with ", dump( $form ),"\n"; |
$self->log->debug("submit form on $uri with ", dump( $form )); |
223 |
$mech->submit_form( %$form ) or die "can't submit form ", dump( $form ); |
$mech->submit_form( %$form ) or die "can't submit form ", dump( $form ); |
224 |
$self->save( 'submit.html', $mech->content ); |
$self->save( 'submit.html', $mech->content ); |
225 |
} |
} |
226 |
|
|
227 |
warn "parse result page\n"; |
$self->log->debug("parse result page"); |
228 |
|
|
229 |
my $tree = HTML::TreeBuilder->new or die "can't create html tree"; |
my $tree = HTML::TreeBuilder->new or die "can't create html tree"; |
230 |
$tree->parse( $mech->content ) or die "can't parse fetched content"; |
$tree->parse( $mech->content ) or die "can't parse fetched content"; |
232 |
die "wrapper doesn't have 3 elements but ", $#{ $args->{wrapper} } unless ( $#{ $args->{wrapper} } == 2 ); |
die "wrapper doesn't have 3 elements but ", $#{ $args->{wrapper} } unless ( $#{ $args->{wrapper} } == 2 ); |
233 |
my ( $el,$attr,$value ) = @{ $args->{wrapper} }; |
my ( $el,$attr,$value ) = @{ $args->{wrapper} }; |
234 |
|
|
235 |
warn "looking for <$el $attr=\"$value\">"; |
$self->log->debug("looking for <$el $attr=\"$value\">"); |
236 |
|
|
237 |
my $div = $tree->look_down( '_tag', $el, sub { |
my $div = $tree->look_down( '_tag', $el, sub { |
|
warn dump( $_[0]->attr( $attr ) ),$/; |
|
238 |
( $_[0]->attr( $attr ) || '' ) eq $value; |
( $_[0]->attr( $attr ) || '' ) eq $value; |
239 |
}); |
}); |
240 |
|
|
258 |
$page_uri->query( undef ); |
$page_uri->query( undef ); |
259 |
$page_uri = $page_uri->canonical; |
$page_uri = $page_uri->canonical; |
260 |
|
|
261 |
warn "fetching page: ",$a->as_text," from $page_uri\n"; |
$self->log->debug("fetching page: ",$a->as_text," from $page_uri"); |
262 |
if ( $mech->follow_link( url => $a->attr('href') ) ) { |
if ( $mech->follow_link( url => $a->attr('href') ) ) { |
263 |
|
|
264 |
$self->save( "page-${nr}.html", $mech->content ); |
$self->save( "page-${nr}.html", $mech->content ); |