194 |
die "no uri" unless ($uri); |
die "no uri" unless ($uri); |
195 |
die "feed is not a Grep::Model::Feed but ", ref $feed unless $feed->isa('Grep::Model::Feed'); |
die "feed is not a Grep::Model::Feed but ", ref $feed unless $feed->isa('Grep::Model::Feed'); |
196 |
|
|
197 |
sub save_html { |
sub mech_warn { |
198 |
my ( $file, $content ) = @_; |
my $m = shift || return; |
199 |
if ( -w '/tmp/grep' ) { |
warn $m; |
|
open(my $f, '>', "/tmp/grep/${file}.html") or die "can't open $file: $!"; |
|
|
print $f $content or die "can't write to $file: $!"; |
|
|
close $f or die "can't close $file: $!"; |
|
|
} |
|
200 |
} |
} |
201 |
|
|
202 |
my $mech = WWW::Mechanize->new(); |
my $mech = WWW::Mechanize->new( |
203 |
|
cookie_jar => {}, |
204 |
|
onwarn => \&mech_warn, |
205 |
|
onerror => \&mech_warn, |
206 |
|
); |
207 |
|
|
208 |
$mech->get( $uri ); |
$mech->get( $uri ); |
209 |
|
|
210 |
save_html( 'get', $mech->content ); |
$self->save( 'get.html', $mech->content ); |
211 |
|
|
212 |
if ( $args->{submit_form} ) { |
if ( my $form = $args->{submit_form} ) { |
213 |
warn "submit form on $uri\n"; |
warn "submit form on $uri with ", dump( $form ),"\n"; |
214 |
$mech->submit_form( %{ $args->{submit_form} } ) or die "can't submit form"; |
$mech->submit_form( %$form ) or die "can't submit form ", dump( $form ); |
215 |
save_html( 'submit', $mech->content ); |
$self->save( 'submit.html', $mech->content ); |
216 |
} |
} |
217 |
|
|
218 |
warn "parse result page\n"; |
warn "parse result page\n"; |
230 |
( $_[0]->attr( $attr ) || '' ) eq $value; |
( $_[0]->attr( $attr ) || '' ) eq $value; |
231 |
}); |
}); |
232 |
|
|
233 |
die "can't find results wrapper <$el $attr=\"$value\">" unless ( $div ); |
if ( ! $div ) { |
234 |
|
warn "can't find results wrapper <$el $attr=\"$value\">"; |
235 |
|
return; |
236 |
|
} |
237 |
|
|
238 |
my $max = 5; |
my $max = 5; |
239 |
my $nr = 1; |
my $nr = 1; |
253 |
warn "fetching page: ",$a->as_text," from $page_uri\n"; |
warn "fetching page: ",$a->as_text," from $page_uri\n"; |
254 |
if ( $mech->follow_link( url => $a->attr('href') ) ) { |
if ( $mech->follow_link( url => $a->attr('href') ) ) { |
255 |
|
|
256 |
save_html( "page-${nr}", $mech->content ); |
$self->save( "page-${nr}.html", $mech->content ); |
257 |
|
|
258 |
my $page_tree = HTML::TreeBuilder->new or die "can't create page tree"; |
my $page_tree = HTML::TreeBuilder->new or die "can't create page tree"; |
259 |
$page_tree->parse( $mech->content ) or die "can't parse page at $page_uri"; |
$page_tree->parse( $mech->content ) or die "can't parse page at $page_uri"; |
260 |
|
|
261 |
my ( $el,$attr,$value ) = @{ $args->{scrape} }; |
( $el,$attr,$value ) = @{ $args->{scrape} }; |
262 |
my $div = $page_tree->look_down( '_tag', $el, sub { ( $_[0]->attr( $attr ) || '' ) eq $value } ); |
$div = $page_tree->look_down( '_tag', $el, sub { ( $_[0]->attr( $attr ) || '' ) eq $value } ); |
263 |
|
|
264 |
die "can't find <$el $attr=\"$value\">" unless ($div); |
die "can't find <$el $attr=\"$value\">" unless ($div); |
265 |
|
|
291 |
|
|
292 |
} |
} |
293 |
|
|
294 |
|
=head2 save |
295 |
|
|
296 |
|
save( 'name', $content ); |
297 |
|
|
298 |
|
Save dumps into C</tmp/grep> if writable |
299 |
|
|
300 |
|
=cut |
301 |
|
|
302 |
|
sub save { |
303 |
|
my $self = shift; |
304 |
|
my ( $file, $content ) = @_; |
305 |
|
if ( -w '/tmp/grep' ) { |
306 |
|
open(my $f, '>', "/tmp/grep/$file") or die "can't open $file: $!"; |
307 |
|
print $f $content or die "can't write to $file: $!"; |
308 |
|
close $f or die "can't close $file: $!"; |
309 |
|
Jifty->log->debug("saved $file ",length($content)," bytes"); |
310 |
|
} |
311 |
|
} |
312 |
|
|
313 |
1; |
1; |