7 |
|
|
8 |
use WWW::Mechanize; |
use WWW::Mechanize; |
9 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
|
use File::Slurp; |
|
10 |
|
|
11 |
my $mech = WWW::Mechanize->new( |
my $mech = WWW::Mechanize->new( |
12 |
autocheck => 1, |
autocheck => 1, |
13 |
cookie_jar => undef, |
cookie_jar => undef, |
14 |
); |
); |
15 |
|
|
16 |
my $step = 1; |
our $step = 0; |
17 |
|
|
18 |
sub save_mech { |
sub save_mech { |
19 |
my $mech = shift; |
my ( $mech, $path ) = @_; |
|
my $path = "/tmp/isi.$step.html"; |
|
|
write_file $path, $mech->content; |
|
|
warn "# step $step ", -s $path; |
|
|
$mech->dump_all; |
|
20 |
$step++; |
$step++; |
21 |
|
$path ||= sprintf('/tmp/isi.%02d.%s', $step, $mech->{ct} =~ m{html}i ? 'html' : 'txt' ); |
22 |
|
$mech->save_content( $path ); |
23 |
|
warn "# [$step] $path ", -s $path, " ", $mech->ct; |
24 |
|
$mech->dump_all; |
25 |
} |
} |
26 |
|
|
27 |
warn "# get session"; |
warn "# get session"; |
46 |
save_mech $mech; |
save_mech $mech; |
47 |
|
|
48 |
my $from = 1; |
my $from = 1; |
49 |
my $range_size = 10; |
my $range_size = 50; |
50 |
|
|
51 |
$mech->submit_form( |
while ( $from ) { |
|
form_name => 'summary_output_form', |
|
|
fields => { |
|
|
record_select_type => 'range', |
|
|
mark_from => $from, |
|
|
mark_to => $from += $range_size, |
|
|
mark_id => 'WOS', |
|
|
|
|
|
qo_fields => 'fullrecord', |
|
|
citedref => 'citedref', |
|
|
|
|
|
save_options => 'plain_text', |
|
|
|
|
|
fields => 'Full', |
|
|
format => 'save', |
|
|
}, |
|
|
button => 'save', |
|
|
); |
|
|
save_mech $mech; |
|
52 |
|
|
53 |
warn "# save_file"; |
my $to = $from + $range_size; |
54 |
$mech->follow_link( url_regex => qr/save_file/ ); |
|
55 |
save_mech $mech; |
$mech->submit_form( |
56 |
|
form_name => 'summary_output_form', |
57 |
|
fields => { |
58 |
|
record_select_type => 'range', |
59 |
|
mark_from => $from, |
60 |
|
mark_to => $to, |
61 |
|
mark_id => 'WOS', |
62 |
|
|
63 |
|
qo_fields => 'fullrecord', |
64 |
|
citedref => 'citedref', |
65 |
|
|
66 |
|
save_options => 'plain_text', |
67 |
|
|
68 |
|
fields => 'Full', |
69 |
|
format => 'save', |
70 |
|
}, |
71 |
|
button => 'save', |
72 |
|
); |
73 |
|
save_mech $mech; |
74 |
|
|
75 |
|
last if $mech->content =~ m{invalid API call}; |
76 |
|
|
77 |
|
warn "# save_file"; |
78 |
|
$mech->follow_link( url_regex => qr/save_file/ ); |
79 |
|
save_mech $mech => "/tmp/isi.$q.$from-$to.txt"; |
80 |
|
|
81 |
|
$from += $range_size; |
82 |
|
|
83 |
|
$mech->back; |
84 |
|
$mech->back; |
85 |
|
#save_mech $mech; |
86 |
|
|
87 |
|
} |