| 1 |
dpavlin |
1.1 |
#!/usr/bin/perl -w |
| 2 |
|
|
|
| 3 |
|
|
use LWP::UserAgent; |
| 4 |
|
|
use HTML::TreeBuilder; |
| 5 |
|
|
use strict; |
| 6 |
|
|
require "../common.pm"; |
| 7 |
|
|
|
| 8 |
|
|
my $debug=1; |
| 9 |
|
|
|
| 10 |
|
|
my $dir = open_data_files("sciencedirect"); |
| 11 |
|
|
my $last_tell=0; |
| 12 |
|
|
|
| 13 |
|
|
print MPS "M working...\n"; |
| 14 |
|
|
|
| 15 |
dpavlin |
1.2 |
my $base_url = 'http://www.sciencedirect.com'; |
| 16 |
dpavlin |
1.1 |
my $url = $base_url . '/science?_ob=JournalListURL&_type=subscribed&_stype=title&subjColl=all&_auth=y&_update=y&_frameSeg=M&_title=all&_acct=C000050661&_version=1&_urlVersion=0&_userid=1034703&md5=6d4b6e263318a1d7d2a3b523d861f920'; |
| 17 |
|
|
|
| 18 |
|
|
$debug++ if (lc($ARGV[0]) eq "-d"); |
| 19 |
|
|
|
| 20 |
|
|
sub print_debug { |
| 21 |
|
|
return if (! $debug); |
| 22 |
|
|
open(DEBUG,">> debug") || warn "can't open debug file!"; |
| 23 |
|
|
print DEBUG "###",@_; |
| 24 |
|
|
print @_; |
| 25 |
|
|
close(DEBUG); |
| 26 |
|
|
} |
| 27 |
|
|
|
| 28 |
|
|
print_debug("debug level $debug"); |
| 29 |
|
|
|
| 30 |
|
|
my $ua = new LWP::UserAgent; |
| 31 |
|
|
$ua->agent("Mjesec educational harvester -- contact mglavica\@ffzg.hr 0.0"); |
| 32 |
|
|
$ua->timeout(60); |
| 33 |
|
|
#$ua->env_proxy(); |
| 34 |
|
|
#$ua->proxy(['http', 'ftp'], 'http://proxy.carnet.hr:8001/'); |
| 35 |
|
|
|
| 36 |
|
|
#my $req = HTTP::Request->new(GET => $url); |
| 37 |
|
|
|
| 38 |
|
|
#my $res = $ua->request($req); |
| 39 |
|
|
#if ($res->is_success) { |
| 40 |
|
|
|
| 41 |
|
|
if (1) { |
| 42 |
|
|
|
| 43 |
|
|
my $tree = HTML::TreeBuilder->new; |
| 44 |
|
|
$tree->parse_file("list.html"); # ! |
| 45 |
|
|
|
| 46 |
|
|
foreach my $tr ($tree->look_down('_tag', 'tr')) { |
| 47 |
|
|
my $link; |
| 48 |
|
|
if ($link = $tr->look_down('_tag','a')) { |
| 49 |
|
|
if ($link->attr('href') =~ m{/science\?_ob=JournalURL}) { |
| 50 |
|
|
my $bib = ""; |
| 51 |
|
|
my $mps = ""; |
| 52 |
|
|
|
| 53 |
|
|
$bib .= "%tip Èasopis\n"; |
| 54 |
dpavlin |
1.2 |
$bib .= "%tip on-line\n"; |
| 55 |
dpavlin |
1.1 |
# url |
| 56 |
|
|
$bib .= "%856 $base_url".$link->attr('href')."\n"; |
| 57 |
|
|
# naslov |
| 58 |
dpavlin |
1.2 |
$bib .= "%200+ ".$link->as_text."\n"; |
| 59 |
dpavlin |
1.1 |
$mps .= mps_expand(2,$link->as_text); |
| 60 |
|
|
|
| 61 |
dpavlin |
1.2 |
$mps .= "H ".$link->as_text." <i>(on-line)</i>\n"; |
| 62 |
dpavlin |
1.1 |
|
| 63 |
dpavlin |
1.2 |
print R $bib."\n"; |
| 64 |
dpavlin |
1.1 |
$mps .= "T document text/plain ".(tell(R) - $last_tell)." $dir/bib $last_tell ".tell(R)."\n"; |
| 65 |
|
|
$last_tell=tell(R); |
| 66 |
|
|
|
| 67 |
dpavlin |
1.2 |
print R "\n"; |
| 68 |
dpavlin |
1.1 |
|
| 69 |
|
|
$mps .= "E\n"; |
| 70 |
|
|
|
| 71 |
|
|
print S $mps; |
| 72 |
|
|
print MPS $mps; |
| 73 |
|
|
} |
| 74 |
|
|
} |
| 75 |
|
|
} |
| 76 |
|
|
|
| 77 |
|
|
$tree->delete; # clear memory! |
| 78 |
|
|
|
| 79 |
|
|
} else { |
| 80 |
|
|
warn "can't fetch web page from '$url'"; |
| 81 |
|
|
} |
| 82 |
|
|
|
| 83 |
|
|
print S "M over and out\nX\n"; |
| 84 |
|
|
print MPS "M over and out\nX\n"; |
| 85 |
|
|
close(MPS); |
Parent Directory
| 
Revision Log