annotate fetch_weather.pl @ 158:db2b1f74d994

Initial import of weather data fetching backend, written in Perl.
author Matti Hamalainen <ccr@tnsp.org>
date Mon, 02 Jun 2014 12:29:11 +0300
parents
children 4f8a163b2bc1
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
158
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
1 #!/usr/bin/perl -w
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
2 use strict;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
3 use LWP::UserAgent;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
4 use HTML::Entities;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
5 use XML::Simple;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
6 use Text::Iconv;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
7 use Date::Parse;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
8 use Data::Dumper;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
9
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
10 sub str_trim($)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
11 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
12 my $str = $_[0];
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
13 if (defined($str)) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
14 $str =~ s/^\s*//;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
15 $str =~ s/\s*$//;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
16 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
17 return $str;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
18 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
19
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
20
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
21 sub pop_token_a($)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
22 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
23 my $tokens = $_[0];
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
24 return shift(@$tokens);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
25 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
26
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
27
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
28 sub pop_token($)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
29 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
30 return str_trim(pop_token_a($_[0]));
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
31 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
32
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
33
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
34 sub parse_html_str($)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
35 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
36 my $tokens = $_[0];
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
37 my $token = pop_token($tokens);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
38 my $str = "";
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
39 $token =~ s/^\s*//;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
40
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
41 return undef unless (substr($token, 0, 1) eq '"');
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
42 $token = substr($token, 1);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
43
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
44 while (defined($token)) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
45 my $tmp = $token;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
46 $tmp =~ s/\s*$//;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
47 if (substr($tmp, -1) eq '"') {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
48 $str .= substr($tmp, 0, -1);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
49 return $str;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
50 } else {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
51 $str .= $token;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
52 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
53 $token = shift(@$tokens);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
54 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
55 return undef;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
56 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
57
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
58
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
59 sub parse_html_tree($$);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
60
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
61 sub parse_html_tree($$)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
62 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
63 my ($tokens, $tree) = @_;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
64
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
65 while (my $token = pop_token($tokens)) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
66 if ($token =~ /^<[!\/]?[a-zA-Z]+/) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
67 $token = lc($token);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
68 if ($token =~ /^<\!.*>$/) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
69 # Ignore comments etc.
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
70 } elsif ($token =~ /^<([a-z]+[1-9]*)(.*)>$/) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
71 my ($name, $args) = ($1, $2);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
72 if ($name eq "meta" || $name eq "img") {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
73 my $tmp = {};
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
74 parse_html_tree($tokens, $tree);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
75 return $tree;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
76 } else {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
77 my $tmp = { "name" => $name, "args" => str_trim($args) };
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
78 parse_html_tree($tokens, $tmp);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
79 push(@{$$tree{"nodes"}}, $tmp);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
80 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
81 } elsif ($token =~ /^<\/([a-z]+[1-9]*)>$/) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
82 return $tree;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
83 } else {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
84 die("HORROR TERROR ELITE: '$token'\n");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
85 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
86 } else {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
87 $token = str_trim(decode_entities($token));
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
88 push(@{$$tree{"nodes"}}, { "name" => "text", "args" => "", "text" => $token }) if length($token) > 0;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
89 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
90 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
91
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
92 return $tree;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
93 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
94
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
95
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
96 sub parse_html($)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
97 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
98 return undef unless defined($_[0]);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
99 my $str = $_[0];
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
100 my $res = { "name" => "", "args" => "" };
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
101 $str =~ tr/\r/ /;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
102 $str =~ tr/\n/ /;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
103 my @tokens = grep { !($_ =~ /^\s*$/) } split(/(<\/?[a-zA-Z]+.*?>)/, $str);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
104 parse_html_tree(\@tokens, $res);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
105 return $res;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
106 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
107
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
108 sub html_find_node($$$);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
109
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
110 sub html_find_node($$$)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
111 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
112 my ($node, $name, $args) = @_;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
113
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
114 if (defined($node)) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
115 if (ref($node) eq "ARRAY") {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
116 foreach my $n (@$node) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
117 my $tmp = html_find_node($n, $name, $args);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
118 # Must do it like this, in order not to break the loop
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
119 return $tmp if defined($tmp);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
120 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
121 } elsif (ref($node) eq "HASH") {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
122 if (defined($$node{"name"})) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
123 if ($$node{"name"} eq $name) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
124 if ($args ne "") {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
125 if (defined($$node{"args"}) && $$node{"args"} =~ /$args/) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
126 } else {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
127 return html_find_node($$node{"nodes"}, $name, $args);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
128 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
129 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
130 return $node;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
131 } else {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
132 return html_find_node($$node{"nodes"}, $name, $args);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
133 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
134 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
135 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
136 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
137 return undef;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
138 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
139
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
140
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
141 sub fnode($$)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
142 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
143 return html_find_node($_[0], $_[1], "");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
144 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
145
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
146
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
147 sub fnodea($$$)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
148 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
149 return html_find_node($_[0], $_[1], $_[2]);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
150 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
151
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
152
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
153 sub fetch_http($)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
154 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
155 my $tmpAgent = LWP::UserAgent->new;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
156 $tmpAgent->agent("Mozilla/4.0 (compatible; MSIE 6.0; MSIE 5.5; Windows NT 6.0) Opera 10.63 [en]");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
157 $tmpAgent->timeout(10);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
158
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
159 my $tmpRequest = HTTP::Request->new(GET => $_[0]);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
160
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
161 return $tmpAgent->request($tmpRequest);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
162 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
163
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
164
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
165 sub get_node($$$)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
166 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
167 return defined($_[0]->[$_[2]]{"nodes"}[0]{$_[1]}) ? $_[0]->[$_[2]]{"nodes"}[0]{$_[1]} : "";
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
168 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
169
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
170
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
171 sub parse_timestamp($$)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
172 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
173 return $_[1] + str2time($_[0]) - time();
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
174 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
175
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
176
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
177 ###
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
178 ### Main program begins
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
179 ###
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
180 die(
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
181 "Weather Fetch v0.1 by ccr/TNSP <ccr\@tnsp.org>\n".
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
182 "Usage: $0 <output_data_file> <fmi_api_key>\n"
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
183 ) unless scalar(@ARGV) >= 2;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
184
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
185 my $opt_outfile = shift;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
186 my $opt_api_key = shift;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
187 my $weatherdata = {};
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
188
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
189 ### Fetch tiehallinto data
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
190 for (my $i = 1; $i <= 22; $i++)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
191 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
192 my $res = fetch_http("http://alk.tiehallinto.fi/alk/tiesaa/tiesaa_maak_".$i.".html");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
193 if ($res->code >= 200 && $res->code <= 201)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
194 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
195 my $data = $res->decoded_content;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
196
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
197 # Filter out crap tags we don't want or need
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
198 $data =~ s/\n/§/g;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
199 $data =~ s/<!--.*?-->//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
200 $data =~ s/<map[^>]*>.*?<\/map>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
201 $data =~ s/<form[^>]*>.*?<\/form>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
202 $data =~ s/<script[^>]*>.*?<\/script>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
203 $data =~ s/<meta[^>]*>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
204 $data =~ s/<font[^>]*>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
205 $data =~ s/<\/font>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
206 $data =~ s/<span[^>]*>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
207 $data =~ s/<\/span>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
208 $data =~ s/<\/?b>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
209
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
210 $data =~ s/<br>//ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
211 $data =~ s/&nbsp;/ /ig;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
212 $data =~ s/§/\n/g;
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
213
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
214 # Parse the HTML mess
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
215 my $otree = parse_html($data);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
216
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
217 # Find our desired element nodes
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
218 my $odata = fnodea(fnode($otree, "body"), "div", "class=elementc");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
219 my $oupdate = fnode($odata, "p");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
220 my $time_base = time();
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
221 if ($oupdate) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
222 my $tmp = $oupdate->{"nodes"}[0]{"text"};
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
223 if ($tmp =~ /:\s+(\d\d\.\d\d\.\d\d\d\d)\s+(\d\d:\d\d)/) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
224 $time_base = str2time($1);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
225 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
226 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
227
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
228 my $oelems = fnode($odata, "table")->{"nodes"};
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
229 for (my $n = 1; $n < scalar(@$oelems); $n++)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
230 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
231 my $fdata = @$oelems[$n]->{"nodes"};
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
232 $weatherdata->{$fdata->[0]{"nodes"}[0]{"text"}} =
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
233 [
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
234 parse_timestamp(get_node($fdata, "text", 1), $time_base),
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
235 get_node($fdata, "text", 2),
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
236 get_node($fdata, "text", 3),
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
237 get_node($fdata, "text", 4),
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
238 get_node($fdata, "text", 5),
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
239 ];
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
240 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
241 print STDERR ".";
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
242 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
243 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
244
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
245 ### Fetch FMI data
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
246 my $res = fetch_http("http://data.fmi.fi/fmi-apikey/".$opt_api_key.
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
247 "/wfs?request=getFeature&storedquery_id=fmi::observations::weather::cities::multipointcoverage".
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
248 "&parameters=temperature,humidity");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
249
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
250 if ($res->code >= 200 && $res->code <= 201)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
251 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
252 my $xml = XMLin($res->decoded_content,
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
253 KeyAttr => { server => 'name' },
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
254 ForceArray => [ 'server', 'address' ]);
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
255 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
256
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
257 ### Output
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
258 if (defined($opt_outfile)) {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
259 open(STDOUT, '>', $opt_outfile) or die("Could not open output file '$opt_outfile'.\n");
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
260 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
261
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
262 binmode STDOUT, ':encoding(utf-8)';
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
263
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
264 foreach my $key (sort { $a cmp $b } keys %$weatherdata)
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
265 {
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
266 print STDOUT $key."|".join("|", @{$weatherdata->{$key}})."\n";
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
267 }
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
268
db2b1f74d994 Initial import of weather data fetching backend, written in Perl.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
269 close(STDOUT);