23 |
sub save_mech { |
sub save_mech { |
24 |
my ( $mech, $path ) = @_; |
my ( $mech, $path ) = @_; |
25 |
$step++; |
$step++; |
26 |
$path ||= sprintf('/tmp/isi.%02d.%s', $step, $mech->{ct} =~ m{html}i ? 'html' : 'txt' ); |
mkdir '/tmp/isi/' unless -e '/tmp/isi'; |
27 |
|
my $base_path = sprintf('/tmp/isi/%04d', $step); |
28 |
|
$path ||= $base_path . ( $mech->{ct} =~ m{html}i ? '.html' : '.txt' ); |
29 |
$mech->save_content( $path ); |
$mech->save_content( $path ); |
30 |
warn "# [$step] $path ", -s $path, " ", $mech->ct; |
warn "# [$step] $path ", -s $path, " ", $mech->ct, "\n"; |
31 |
$mech->dump_all if $dump; |
open(my $dump, '>', "$base_path.dump.txt"); |
32 |
|
$mech->dump_all($dump); |
33 |
} |
} |
34 |
|
|
35 |
warn "# get session"; |
warn "# get session"; |
36 |
$mech->get( 'http://isiknowledge.com/?DestApp=WOS' ); |
$mech->get( 'http://isiknowledge.com/?DestApp=WOS' ); |
37 |
save_mech $mech; |
save_mech $mech; |
38 |
|
|
39 |
warn "# advanced serach"; |
sub search { |
40 |
$mech->follow_link( url_regex => qr/AdvancedSearch/ ); |
my $q = shift; |
|
save_mech $mech; |
|
|
|
|
|
warn "# cookie_jar ", dump $mech->cookie_jar; |
|
41 |
|
|
42 |
$mech->submit_form( |
warn "# advanced serach"; |
43 |
fields => { |
$mech->follow_link( url_regex => qr/AdvancedSearch/ ); |
44 |
'value(input1)' => $q, |
save_mech $mech; |
45 |
} |
|
46 |
); |
warn "# cookie_jar ", dump $mech->cookie_jar; |
47 |
save_mech $mech; |
|
48 |
|
$mech->submit_form( |
49 |
|
fields => { |
50 |
|
'value(input1)' => $q, |
51 |
|
} |
52 |
|
); |
53 |
|
save_mech $mech; |
54 |
|
|
55 |
warn "# summary"; |
warn "# summary"; |
56 |
$mech->follow_link( url_regex => qr/summary/ ); |
$mech->follow_link( url_regex => qr/summary/ ); |
57 |
save_mech $mech; |
save_mech $mech; |
58 |
|
} |
59 |
|
|
60 |
sub get_results { |
sub get_results { |
61 |
my $q = shift; |
my $q = shift; |
87 |
|
|
88 |
if ( $mech->content =~ m{invalid API call} ) { |
if ( $mech->content =~ m{invalid API call} ) { |
89 |
$mech->back; |
$mech->back; |
90 |
return; |
last; |
91 |
} |
} |
92 |
|
|
93 |
warn "# save_file $from - $to [$q]"; |
warn "range $from - $to [$q]\n"; |
94 |
$mech->follow_link( url_regex => qr/save_file/ ); |
$mech->follow_link( url_regex => qr/save_file/ ); |
95 |
save_mech $mech => "/tmp/isi.$q.$from-$to.txt"; |
save_mech $mech => "/tmp/isi.$q.$from-$to.txt"; |
96 |
|
|
100 |
$mech->back; |
$mech->back; |
101 |
#save_mech $mech; |
#save_mech $mech; |
102 |
|
|
103 |
} # while |
} |
104 |
|
|
105 |
} |
} |
106 |
|
|
|
get_results $q; |
|
107 |
|
|
108 |
save_mech $mech; |
sub citations { |
109 |
warn "# citations"; |
save_mech $mech; |
110 |
$mech->follow_link( url_regex => qr/search_mode=CitationReport/ ); |
warn "# citation report"; |
111 |
save_mech $mech; |
$mech->follow_link( url_regex => qr/search_mode=CitationReport/ ); |
112 |
|
save_mech $mech; |
113 |
|
|
114 |
|
warn "view citing articles"; |
115 |
|
$mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ ); |
116 |
|
save_mech $mech; |
117 |
|
} |
118 |
|
|
119 |
$mech->follow_link( url_regex => qr/search_mode=TotalCitingArticles/ ); |
sub years { |
120 |
save_mech $mech; |
my $years_url = $mech->find_link( text_regex => qr/more options/ )->url_abs; |
121 |
|
warn "## $years_url"; |
122 |
|
$years_url =~ s{ra_name=\w+}{ra_name=PublicationYear} || die "ra_name"; |
123 |
|
warn "# refine years (hidden by javascript)"; |
124 |
|
warn "http://apps.isiknowledge.com/RAMore.do?product=WOS&search_mode=TotalCitingArticles&SID=T1o6bChdN9PGP1LN1Nh&qid=3&ra_mode=more&ra_name=PublicationYear&db_id=WOS&viewType=raMore\n$years_url\n"; |
125 |
|
$mech->get( $years_url ); |
126 |
|
save_mech $mech; |
127 |
|
|
128 |
|
my $html = $mech->content; |
129 |
|
my @years; |
130 |
|
while ( $html =~ s{>(\d\d\d\d)\s\((\d+)\)</label.+?value="PublicationYear_}{} ) { |
131 |
|
push @years, [ $1 => $2 ]; |
132 |
|
} |
133 |
|
warn "# years ",dump @years; |
134 |
|
$mech->back; |
135 |
|
return @years; |
136 |
|
} |
137 |
|
|
138 |
|
search $q; |
139 |
|
years; |
140 |
|
get_results $q; |
141 |
|
|
142 |
|
citations; |
143 |
|
years; |
144 |
get_results $q . '.citing'; |
get_results $q . '.citing'; |
145 |
|
|