| 1 |
#!/usr/local/bin/perl -w |
| 2 |
|
| 3 |
use LWP::UserAgent; |
| 4 |
use strict; |
| 5 |
use DBI; |
| 6 |
use HTML::TreeBuilder; |
| 7 |
|
| 8 |
my $debug=1; |
| 9 |
|
| 10 |
my $url = 'http://custom.marketwatch.com/custom/alliance/ftmw/invrel.asp?siteid=!plivadd-0773-4F6D-DHID-QN1112NFTD0X&symb=PLVD'; |
| 11 |
|
| 12 |
$debug++ if (lc($ARGV[0]) eq "-d"); |
| 13 |
|
| 14 |
my %val; |
| 15 |
|
| 16 |
sub print_debug { |
| 17 |
return if (! $debug); |
| 18 |
open(DEBUG,">> debug") || warn "can't open debug file!"; |
| 19 |
print DEBUG "###",@_,"\n"; |
| 20 |
print @_,"\n"; |
| 21 |
close(DEBUG); |
| 22 |
} |
| 23 |
|
| 24 |
sub print_val { |
| 25 |
return if (! $debug); |
| 26 |
foreach (keys %val) { |
| 27 |
print "$_: $val{$_}\n"; |
| 28 |
} |
| 29 |
} |
| 30 |
|
| 31 |
print_debug("debug level $debug"); |
| 32 |
|
| 33 |
my $dbh = DBI->connect("DBI:Pg:dbname=corp","","") || die $DBI::errstr; |
| 34 |
|
| 35 |
my $ua = new LWP::UserAgent; |
| 36 |
$ua->agent("pliva_harvester 0.0"); |
| 37 |
$ua->timeout(60); |
| 38 |
$ua->env_proxy(); |
| 39 |
$ua->proxy(['http', 'ftp'], 'http://proxy.pliva.hr:8080/'); |
| 40 |
|
| 41 |
|
| 42 |
my $close_time='21:21:21 CET'; |
| 43 |
|
| 44 |
sub insert { |
| 45 |
my $sth = $dbh->prepare("select count(date) from stocks where date='".$val{date}."'"); |
| 46 |
$sth->execute(); |
| 47 |
my ($nr) = $sth->fetchrow_array; |
| 48 |
if ($nr == 0) { |
| 49 |
my $sql="insert into stocks values ('$val{date}','LSE',$val{curr},$val{change},$val{high},$val{low},$val{open})"; |
| 50 |
$sql=~s/,\+(\d)/,$1/g; # nuke + which pgsql doesn't like |
| 51 |
if ($sql =~ m,n/a,i) { |
| 52 |
print_val(); |
| 53 |
print_debug("undefined values found. not inserting in db"); |
| 54 |
} else { |
| 55 |
$dbh->do("$sql") || die "$sql\n".$DBI::errstr; |
| 56 |
} |
| 57 |
print_debug("sql: $sql\n"); |
| 58 |
} else { |
| 59 |
print_debug("skip: $val{date}\n"); |
| 60 |
} |
| 61 |
} |
| 62 |
|
| 63 |
my $req = HTTP::Request->new(GET => $url); |
| 64 |
my $tree = HTML::TreeBuilder->new; |
| 65 |
|
| 66 |
#my $res = $ua->request($req); |
| 67 |
#if ($res->is_success) { |
| 68 |
# print_debug("html: ".$res->content."\n"); |
| 69 |
# $tree->parse($res->content) || die "can't parse html file!"; |
| 70 |
|
| 71 |
if (1) { |
| 72 |
$tree->parse_file("out-without_proxy") || die "can't parse html file!"; |
| 73 |
|
| 74 |
# find date |
| 75 |
my $t = $tree; |
| 76 |
# $t = $tree->look_down('_tag', 'td'); |
| 77 |
# print "##td: ",$t->as_text,"\n"; |
| 78 |
if ($t = $tree->look_down('class', 'ft-quotedate')) { |
| 79 |
if ($t->as_text =~ m,(\d+):(\d+) (\d+)/(\d+)/(\d+),) { |
| 80 |
my ($h,$m,$dd,$mm,$yy) = ($1,$2,$3,$4,$5); |
| 81 |
my $date=($yy+2000)."-$mm-$dd"; |
| 82 |
my (undef,undef,$local_h) = localtime(time); |
| 83 |
$h += 12 if ($local_h - $h > 12); |
| 84 |
my $time="$h:$m"; |
| 85 |
print_debug("time: $time date: $date"); |
| 86 |
$val{date}="$date $time"; |
| 87 |
} else { |
| 88 |
die "can't recognise date format ".$t->as_text; |
| 89 |
} |
| 90 |
} else { |
| 91 |
die "can't find date in html"; |
| 92 |
} |
| 93 |
|
| 94 |
# last, change |
| 95 |
if (my @q = $tree->look_down('class', 'ft-quoteLG')) { |
| 96 |
if ($#q+1 == 2) { |
| 97 |
$val{curr} = $q[0]->as_text; |
| 98 |
$val{change} = $q[1]->as_text; |
| 99 |
} else { |
| 100 |
die "can't find 2 classes ft-quoteLG"; |
| 101 |
} |
| 102 |
} else { |
| 103 |
die "can't find class ft-quoteLG (last value and change)"; |
| 104 |
} |
| 105 |
|
| 106 |
# high, low, open |
| 107 |
if (my @q = $tree->look_down('class', 'ft-quoteMd')) { |
| 108 |
if ($#q+1 == 11) { |
| 109 |
$val{high} = $q[1]->as_text; |
| 110 |
$val{low} = $q[2]->as_text; |
| 111 |
$val{open} = $q[3]->as_text; |
| 112 |
} else { |
| 113 |
die "can't find 11 classes ft-quoteMd"; |
| 114 |
} |
| 115 |
} else { |
| 116 |
die "can't find class ft-quoteMd"; |
| 117 |
} |
| 118 |
|
| 119 |
insert(); |
| 120 |
} else { |
| 121 |
warn "can't fetch stock data"; |
| 122 |
} |
| 123 |
|
| 124 |
print_val(); |
| 125 |
|
| 126 |
$dbh->disconnect; |
| 127 |
|
Parent Directory
| 
Revision Log