annotate parsedata.pl @ 156:c0c69b3b1f7d

Fix parsing of certain IDs.
author Matti Hamalainen <ccr@tnsp.org>
date Mon, 09 Feb 2015 02:01:44 +0200
parents 1a5994671dcf
children 62687fee6f1c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
1 #!/usr/bin/perl -w
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
2 #
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
3 # Fetch and parse HTML format class timetable into more sane formats
54
0d574e3248b5 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 52
diff changeset
4 # (C) Copyright 2010-2012 Matti Hämäläinen <ccr@tnsp.org>
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
5 #
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
6 use strict;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
7 use Data::Dumper;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
8 use HTML::Entities;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
9
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
10
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
11 sub urlencode($)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
12 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
13 my $value = $_[0];
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
14 $value =~ s/([^a-zA-Z_0-9 ])/"%" . uc(sprintf "%lx" , unpack("C", $1))/eg;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
15 $value =~ tr/ /+/;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
16 return $value;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
17 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
18
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
19
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
20 sub str_trim($)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
21 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
22 my $str = $_[0];
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
23 if (defined($str)) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
24 $str =~ s/^\s*//;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
25 $str =~ s/\s*$//;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
26 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
27 return $str;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
28 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
29
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
30
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
31 sub pop_token_a($)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
32 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
33 my $tokens = $_[0];
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
34 return shift(@$tokens);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
35 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
36
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
37
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
38 sub pop_token($)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
39 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
40 return str_trim(pop_token_a($_[0]));
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
41 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
42
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
43
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
44 sub parse_html_str($)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
45 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
46 my $tokens = $_[0];
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
47 my $token = pop_token($tokens);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
48 my $str = "";
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
49 $token =~ s/^\s*//;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
50
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
51 return undef unless (substr($token, 0, 1) eq '"');
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
52 $token = substr($token, 1);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
53
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
54 while (defined($token)) {
26
eb82a7ddf5bd Oops, 10l.
Matti Hamalainen <ccr@tnsp.org>
parents: 25
diff changeset
55 my $tmp = $token;
25
a076d8d22422 Fix string parsing in the HTML parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
56 $tmp =~ s/\s*$//;
a076d8d22422 Fix string parsing in the HTML parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
57 if (substr($tmp, -1) eq '"') {
a076d8d22422 Fix string parsing in the HTML parser.
Matti Hamalainen <ccr@tnsp.org>
parents: 8
diff changeset
58 $str .= substr($tmp, 0, -1);
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
59 return $str;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
60 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
61 $str .= $token;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
62 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
63 $token = shift(@$tokens);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
64 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
65 return undef;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
66 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
67
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
68
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
69 sub parse_html_tree($$);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
70
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
71 sub parse_html_tree($$)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
72 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
73 my ($tokens, $tree) = @_;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
74
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
75 while (my $token = pop_token($tokens)) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
76 if ($token =~ /^<[!\/]?[a-zA-Z]+/) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
77 $token = lc($token);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
78 if ($token =~ /^<\!.*>$/) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
79 # Ignore comments etc.
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
80 } elsif ($token =~ /^<([a-z]+)(.*)>$/) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
81 my ($name, $args) = ($1, $2);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
82 if ($name eq "meta" || $name eq "img") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
83 my $tmp = {};
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
84 parse_html_tree($tokens, $tree);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
85 return $tree;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
86 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
87 my $tmp = { "name" => $name, "args" => str_trim($args) };
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
88 parse_html_tree($tokens, $tmp);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
89 push(@{$$tree{"nodes"}}, $tmp);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
90 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
91 } elsif ($token =~ /^<\/([a-z]+)>$/) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
92 return $tree;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
93 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
94 die("HORROR TERROR ELITE: $token\n");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
95 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
96 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
97 $token = str_trim(decode_entities($token));
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
98 push(@{$$tree{"nodes"}}, { "name" => "text", "args" => "", "text" => $token }) if length($token) > 0;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
99 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
100 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
101
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
102 return $tree;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
103 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
104
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
105
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
106 sub parse_html($)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
107 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
108 return undef unless defined($_[0]);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
109 my $str = $_[0];
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
110 my $res = { "name" => "", "args" => "" };
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
111 $str =~ tr/\r/ /;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
112 $str =~ tr/\n/ /;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
113 my @tokens = grep { !($_ =~ /^\s*$/) } split(/(<\/?[a-zA-Z]+.*?>)/, $str);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
114 parse_html_tree(\@tokens, $res);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
115 return $res;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
116 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
117
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
118 sub html_find_node($$$);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
119
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
120 sub html_find_node($$$)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
121 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
122 my ($node, $name, $args) = @_;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
123
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
124 if (defined($node)) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
125 if (ref($node) eq "ARRAY") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
126 foreach my $n (@$node) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
127 my $tmp = html_find_node($n, $name, $args);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
128 # Must do it like this, in order not to break the loop
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
129 return $tmp if defined($tmp);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
130 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
131 } elsif (ref($node) eq "HASH") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
132 if (defined($$node{"name"})) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
133 if ($$node{"name"} eq $name) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
134 if ($args ne "") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
135 if (defined($$node{"args"}) && $$node{"args"} =~ /$args/) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
136 } else {
33
ad96ed91de92 Fix recursive parsing with attributes.
Matti Hamalainen <ccr@tnsp.org>
parents: 32
diff changeset
137 return html_find_node($$node{"nodes"}, $name, $args);
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
138 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
139 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
140 return $node;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
141 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
142 return html_find_node($$node{"nodes"}, $name, $args);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
143 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
144 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
145 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
146 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
147 return undef;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
148 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
149
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
150
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
151 sub fnode($$)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
152 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
153 return html_find_node($_[0], $_[1], "");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
154 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
155
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
156
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
157 sub fnodea($$$)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
158 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
159 return html_find_node($_[0], $_[1], $_[2]);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
160 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
161
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
162
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
163 sub escape($)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
164 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
165 my $s = $_[0];
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
166 $s =~ s/(['"])/\\$1/g;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
167 return $s;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
168 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
169
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
170
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
171 sub html_collapse($$);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
172
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
173 sub html_collapse($$)
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
174 {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
175 my ($node, $strip) = @_;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
176 my $str = "";
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
177
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
178 if ($$node{"name"} eq "text") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
179 $str .= $$node{"text"};
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
180 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
181 $str .= "<".$$node{"name"}.">" unless ($strip);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
182 foreach my $n (@{$$node{"nodes"}}) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
183 $str .= html_collapse($n, $strip);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
184 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
185 $str .= "</".$$node{"name"}.">" unless ($strip);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
186 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
187
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
188 return $str;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
189 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
190
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
191
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
192 ###
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
193 ### Main program
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
194 ###
51
0db0d485eb59 Rename some variables, remove "simple" mode.
Matti Hamalainen <ccr@tnsp.org>
parents: 36
diff changeset
195 my $modes = "php|xml";
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
196 my $opt_mode = "php";
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
197 my $opt_dump = 0;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
198 my $opt_filename;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
199 my $opt_outfile;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
200
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
201 while (defined(my $arg = shift)) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
202 if (substr($arg, 0, 1) eq "-") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
203 if ($arg =~ /^-($modes)$/o) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
204 $opt_mode = $1;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
205 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
206 elsif ($arg eq "-dump") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
207 $opt_dump = 1;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
208 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
209 elsif ($arg eq "-o") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
210 $opt_outfile = shift or die("Output filename option -o requires an argument.\n");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
211 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
212 die("Invalid option '$arg'.\n");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
213 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
214 } else {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
215 $opt_filename = $arg;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
216 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
217 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
218
4
dd2bce7ec0c8 Removed http fetching feature, it's not really useful here.
Matti Hamalainen <ccr@tnsp.org>
parents: 3
diff changeset
219 die("Usage: $0 [options] <filename>
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
220
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
221 -php Output a PHP include file with data in arrays (default)
54
0d574e3248b5 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 52
diff changeset
222 -xml Output a simple XML file.
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
223
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
224 -o <filename> Set output filename. Default is to use stdout.
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
225
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
226 -dump Dump HTML tree to stdout and quit.
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
227
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
228 ") unless defined($opt_filename);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
229
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
230
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
231 my $data;
36
bb00a40252e2 Input in ISO-8859-1 and output in UTF-8.
Matti Hamalainen <ccr@tnsp.org>
parents: 33
diff changeset
232 open(my $fh, '<:encoding(iso-8859-1)', $opt_filename) or die("Error opening '$opt_filename': $!\n");
4
dd2bce7ec0c8 Removed http fetching feature, it's not really useful here.
Matti Hamalainen <ccr@tnsp.org>
parents: 3
diff changeset
233 $data = do { local $/; <$fh> };
dd2bce7ec0c8 Removed http fetching feature, it's not really useful here.
Matti Hamalainen <ccr@tnsp.org>
parents: 3
diff changeset
234 close($fh);
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
235
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
236 die("No data in input.\n") unless (defined($data) && $data ne "");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
237
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
238
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
239 # Filter out certain unneeded elements
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
240 $data =~ s/<font[^>]*>//ig;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
241 $data =~ s/<\/font>//ig;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
242 $data =~ s/<\/?center>//ig;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
243 $data =~ s/<br>//ig;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
244 $data =~ s/&nbsp;/ /ig;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
245
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
246 ### Get some general information
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
247 my $otree = parse_html($data);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
248 if ($opt_dump) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
249 print Dumper(fnode($otree, "html"));
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
250 exit;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
251 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
252
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
253 my %class = ();
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
254 my $body = fnode($otree, "body");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
255 if (defined($body) && defined($$body{"nodes"})) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
256 foreach my $n (@{$$body{"nodes"}}) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
257 if ($$n{"name"} eq "text") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
258 push(@{$class{"info"}}, $$n{"text"});
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
259 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
260 elsif ($$n{"name"} eq "b") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
261 push(@{$class{"data"}}, $n);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
262 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
263 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
264 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
265
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
266 # Filter out some more, for easier tree access during table parsing
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
267 $data =~ s/<\/?b>//ig;
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
268 my $tree = parse_html($data);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
269 my $node = fnode(fnode($tree, "body"), "table");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
270 die("No table element found in document. Perhaps the format has changed? :(\n") unless defined($node);
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
271
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
272 ### Parse through the HTML document node tree to find the data we need
51
0db0d485eb59 Rename some variables, remove "simple" mode.
Matti Hamalainen <ccr@tnsp.org>
parents: 36
diff changeset
273 my $cid = 0;
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
274 my $q = $$node{"nodes"};
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
275 my $hourTimes = [];
51
0db0d485eb59 Rename some variables, remove "simple" mode.
Matti Hamalainen <ccr@tnsp.org>
parents: 36
diff changeset
276 my $hourDefs = {};
0db0d485eb59 Rename some variables, remove "simple" mode.
Matti Hamalainen <ccr@tnsp.org>
parents: 36
diff changeset
277 my $hourTable = {};
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
278 my $hourFillTable = {};
53
1c5174dcd92a Clean up maxDays computation.
Matti Hamalainen <ccr@tnsp.org>
parents: 52
diff changeset
279 my $maxDays = 0;
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
280 my $firstHour = 0;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
281 my $lastHour = 0;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
282 my $totalHours = 0;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
283
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
284
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
285 sub parseHourData($$)
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
286 {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
287 my ($l, $rowspan) = @_;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
288 my $chours = $rowspan / 2; # The table is actually in half cells
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
289 my $cdata = [];
145
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
290 my $cturns = 0;
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
291 my $cgrouped = 0;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
292
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
293 # Pull in data for the class/hour cell
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
294 foreach my $h (@{$l}) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
295 if (defined($$h{"nodes"})) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
296 foreach my $b (@{$$h{"nodes"}}) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
297 if (defined($$b{"nodes"})) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
298 my $text = $$b{"nodes"}[0]{"text"};
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
299 $text =~ s/\.$//;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
300
146
1a5994671dcf Oops. Fix a silly 10L.
Matti Hamalainen <ccr@tnsp.org>
parents: 145
diff changeset
301 if ($text =~ /^vuorov/i)
145
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
302 {
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
303 $cturns = 1;
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
304 }
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
305 else
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
306 {
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
307 push(@$cdata, $text);
c028506aa9cc Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents: 80
diff changeset
308 }
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
309 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
310 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
311 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
312 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
313
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
314 # Increased ID if there is data in this class/hour cell
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
315 my $tid;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
316 if (scalar(@$cdata) > 0) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
317 $cid++;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
318 $tid = $cid;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
319 } else {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
320 $tid = 0;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
321 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
322
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
323 # Determine current day
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
324 my $cday = 0;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
325 for (my $x = 0; $x < 7; $x++) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
326 if (!defined($$hourFillTable{$lastHour}{$x})) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
327 $cday = $x;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
328 last;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
329 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
330 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
331 for (my $t = 0; $t < $chours; $t++) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
332 $$hourFillTable{$lastHour + $t}{$cday} = $tid;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
333 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
334
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
335 if ($tid)
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
336 {
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
337 $maxDays = $cday + 1 if ($cday + 1 > $maxDays);
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
338
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
339 # Grouped, if there is another class ID in second slot
156
c0c69b3b1f7d Fix parsing of certain IDs.
Matti Hamalainen <ccr@tnsp.org>
parents: 146
diff changeset
340 $cgrouped = 1 if ($$cdata[1] =~ /^[A-Z]\d{5,6}[A-Z]*$/);
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
341 if ($cgrouped)
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
342 {
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
343 my $cdata1 = [];
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
344 my $cdata2 = [];
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
345 for (my $i = 0; $i < length($cdata); $i += 2)
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
346 {
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
347 push(@$cdata1, $$cdata[$i]) if defined($$cdata[$i]);
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
348 push(@$cdata2, $$cdata[$i+1]) if defined($$cdata[$i+1]);
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
349 }
146
1a5994671dcf Oops. Fix a silly 10L.
Matti Hamalainen <ccr@tnsp.org>
parents: 145
diff changeset
350 $$hourDefs{$cid} = { "turns" => $cturns, "grouped" => $cgrouped, "start" => $lastHour, "hours" => $chours, "data" => [ $cdata1, $cdata2 ] };
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
351 }
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
352 else
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
353 {
146
1a5994671dcf Oops. Fix a silly 10L.
Matti Hamalainen <ccr@tnsp.org>
parents: 145
diff changeset
354 $$hourDefs{$cid} = { "turns" => $cturns, "grouped" => $cgrouped, "start" => $lastHour, "hours" => $chours, "data" => [ $cdata ] };
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
355 }
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
356
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
357 push(@{$$hourTable{$cday}}, $tid);
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
358 $totalHours += $chours;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
359 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
360 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
361
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
362 sub parseHour($)
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
363 {
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
364 if ($_[0] =~ /(\d+):(\d+)/)
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
365 {
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
366 return ((int($1) * 60 + int($2)) * 60);
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
367 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
368 return undef;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
369 }
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
370
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
371
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
372 sub getDataStruct($$);
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
373
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
374 sub getDataStruct($$)
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
375 {
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
376 my @out = ();
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
377 my ($tmp, $first) = @_;
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
378
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
379 if (ref($tmp) eq "ARRAY")
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
380 {
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
381 my @str = ();
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
382 foreach my $item (@{$tmp})
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
383 {
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
384 push(@str, getDataStruct($item, 0));
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
385 }
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
386 if (scalar(@str) > 0)
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
387 {
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
388 push(@out, "array(".join(", ", @str).")") if ($opt_mode eq "php");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
389 push(@out, "<group>".join("", @str)."</group>") if ($opt_mode eq "xml");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
390 }
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
391 }
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
392 elsif (ref($tmp) eq "HASH")
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
393 {
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
394 my @str = ();
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
395 foreach my $key (keys %{$tmp})
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
396 {
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
397 push(@out, "\"".$key."\" => ".getDataStruct($$tmp{$key}, 1)) if ($opt_mode eq "php");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
398 push(@out, "<".$key.">".getDataStruct($$tmp{$key}, 1)."</".$key.">") if ($opt_mode eq "xml");
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
399 }
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
400 if (scalar(@str) > 0)
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
401 {
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
402 push(@out, "array(".join(", ", @str).")") if ($opt_mode eq "php");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
403 push(@out, join("", @str)) if ($opt_mode eq "xml");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
404 }
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
405 }
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
406 elsif ($tmp =~ /^\d+$/)
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
407 {
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
408 if ($opt_mode eq "php")
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
409 {
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
410 push(@out, $tmp);
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
411 }
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
412 else
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
413 {
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
414 push(@out, $first ? $tmp : "<item>".$tmp."</item>");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
415 }
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
416 }
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
417 else
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
418 {
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
419 if ($opt_mode eq "php")
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
420 {
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
421 push(@out, "\"".$tmp."\"");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
422 }
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
423 else
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
424 {
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
425 push(@out, $first ? $tmp : "<item>".$tmp."</item>");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
426 }
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
427 }
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
428
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
429 return join(", ", @out) if ($opt_mode eq "php");
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
430 return join("", @out);
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
431 }
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
432
8
a52a0bdb5ea1 Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
433 # Skip zero position this way (can't use foreach here)
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
434 for (my $i = 1; $i < scalar(@{$q}); $i++) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
435 my $d = $$q[$i]{"nodes"};
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
436 if (defined($d))
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
437 {
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
438 foreach my $n (@{$d})
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
439 {
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
440 my $l = $$n{"nodes"}[0]{"nodes"};
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
441 if (defined($l))
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
442 {
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
443 if ($$n{"args"} =~ /colspan=6\s+rowspan=(\d+)/)
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
444 {
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
445 parseHourData($l, $1);
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
446 }
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
447 elsif ($$n{"args"} =~ /rowspan=2\s+align/)
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
448 {
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
449 my $qstart = parseHour($$l[0]{"nodes"}[0]{"nodes"}[0]{"text"});
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
450 my $qend = parseHour($$l[1]{"nodes"}[0]{"nodes"}[0]{"text"});
72
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
451 if (defined($qstart) && defined($qend))
6fd715063abc Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents: 58
diff changeset
452 {
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
453 push(@$hourTimes, {"start" => $qstart, "end" => $qend});
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
454 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
455 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
456 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
457 }
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
458 $lastHour++;
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
459 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
460 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
461
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
462
8
a52a0bdb5ea1 Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
463 ### Go through hour table, find last day and hour of the week, crop
a52a0bdb5ea1 Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
464 my $flag = 1;
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
465 for (my $y = 0; $y < $lastHour && $flag; $y++) {
51
0db0d485eb59 Rename some variables, remove "simple" mode.
Matti Hamalainen <ccr@tnsp.org>
parents: 36
diff changeset
466 for (my $x = 0; $x < $maxDays && $flag; $x++) {
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
467 $flag = 0 if (defined($$hourFillTable{$y}{$x}) && $$hourFillTable{$y}{$x} != 0);
8
a52a0bdb5ea1 Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
468 }
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
469 $firstHour++ if ($flag);
8
a52a0bdb5ea1 Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
470 }
a52a0bdb5ea1 Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents: 4
diff changeset
471
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
472 $flag = 1;
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
473 for (my $y = $lastHour - 1; $y >= 0 && $flag; $y--) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
474 for (my $x = 0; $x < $maxDays && $flag; $x++) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
475 $flag = 0 if (defined($$hourFillTable{$y}{$x}) && $$hourFillTable{$y}{$x} != 0);
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
476 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
477 $lastHour-- if ($flag);
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
478 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
479
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
480
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
481 ### Open output file, if specified
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
482 if (defined($opt_outfile)) {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
483 open(STDOUT, '>', $opt_outfile) or die("Could not open output file '$opt_outfile'.\n");
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
484 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
485
36
bb00a40252e2 Input in ISO-8859-1 and output in UTF-8.
Matti Hamalainen <ccr@tnsp.org>
parents: 33
diff changeset
486 binmode STDOUT, ':encoding(utf-8)';
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
487
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
488 ### Output data in desired format
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
489 if ($opt_mode eq "php") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
490 print "<?\n".
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
491 "\$classInfo = array(\n".
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
492 " \"general\" => array(".join(", ", map { "\"".escape($_)."\""; } @{$class{"info"}})."),\n".
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
493 " \"info\" => array(".join(", ", map { "\"".escape(html_collapse($_, 1))."\""; } @{$class{"data"}})."),\n".
54
0d574e3248b5 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 52
diff changeset
494 " \"tags\" => array(".join(", ", map { "\"".escape(html_collapse($_, 0))."\""; } @{$class{"data"}})."),\n".
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
495 " \"maxDays\" => $maxDays,\n".
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
496 " \"firstHour\" => $firstHour,\n".
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
497 " \"lastHour\" => $lastHour,\n".
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
498 " \"totalHours\" => $totalHours\n".
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
499 ");\n\n";
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
500
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
501 print "\$classHourTimes = array(\n";
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
502 foreach my $chour (@$hourTimes) {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
503 print " array(\"start\" => ".$$chour{"start"}.", \"end\" => ".$$chour{"end"}."),\n";
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
504 }
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
505 print ");\n\n";
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
506
54
0d574e3248b5 Cleanups.
Matti Hamalainen <ccr@tnsp.org>
parents: 52
diff changeset
507 print "\$classHourDefs = array(\n";
51
0db0d485eb59 Rename some variables, remove "simple" mode.
Matti Hamalainen <ccr@tnsp.org>
parents: 36
diff changeset
508 foreach my $cid (sort { $a <=> $b } keys %{$hourDefs}) {
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
509 print " $cid => array(".getDataStruct($$hourDefs{$cid}, 0)."),\n";
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
510 }
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
511 print ");\n\n";
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
512
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
513 print
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
514 "\$classDayTable = array(\n";
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
515 for (my $y = 0; $y < $maxDays; $y++)
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
516 {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
517 if (defined($$hourTable{$y}))
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
518 {
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
519 print " $y => array(".join(", ", @{$$hourTable{$y}})."),\n";
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
520 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
521 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
522 print ");\n?>\n";
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
523 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
524
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
525 elsif ($opt_mode eq "xml") {
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
526 print "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n".
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
527 "<timetable>\n".
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
528 " <class>\n".
77
49a329e87367 Adjust some tag names to be more sane in the XML output.
Matti Hamalainen <ccr@tnsp.org>
parents: 75
diff changeset
529 " <general>".join("", map { "<item>".$_."</item>"; } @{$class{"info"}})."</general>\n".
49a329e87367 Adjust some tag names to be more sane in the XML output.
Matti Hamalainen <ccr@tnsp.org>
parents: 75
diff changeset
530 " <info>".join("", map { "<item>".html_collapse($_, 1)."</item>"; } @{$class{"data"}})."</info>\n".
52
b2f45dd616bc Oops, the output variable names should not have been changed.
Matti Hamalainen <ccr@tnsp.org>
parents: 51
diff changeset
531 " <maxdays>$maxDays</maxdays>\n".
57
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
532 " <firsthour>$firstHour</firsthour>\n".
93c87f42c803 New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents: 55
diff changeset
533 " <lasthour>$lastHour</lasthour>\n".
58
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
534 " <totalhours>$totalHours</totalhours>\n".
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
535 " </class>\n";
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
536
58
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
537 print " <hours>\n";
75
3d9e42477367 More improvements in the parsing and XML output.
Matti Hamalainen <ccr@tnsp.org>
parents: 74
diff changeset
538 my $cid = 0;
58
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
539 foreach my $chour (@$hourTimes) {
75
3d9e42477367 More improvements in the parsing and XML output.
Matti Hamalainen <ccr@tnsp.org>
parents: 74
diff changeset
540 print " <hour id=\"".$cid."\"><start>".$$chour{"start"}."</start><end>".$$chour{"end"}."</end></hour>\n";
3d9e42477367 More improvements in the parsing and XML output.
Matti Hamalainen <ccr@tnsp.org>
parents: 74
diff changeset
541 $cid++;
58
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
542 }
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
543 print " </hours>\n\n";
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
544
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
545 print " <classes>\n";
75
3d9e42477367 More improvements in the parsing and XML output.
Matti Hamalainen <ccr@tnsp.org>
parents: 74
diff changeset
546 foreach $cid (sort { $a <=> $b } keys %{$hourDefs}) {
74
b51ad733b624 Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents: 72
diff changeset
547 print " <class id=\"$cid\">".getDataStruct($$hourDefs{$cid}, 0)."</class>\n";
58
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
548 }
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
549 print " </classes>\n\n";
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
550
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
551 print " <days>\n";
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
552 for (my $y = 0; $y < $maxDays; $y++)
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
553 {
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
554 if (defined($$hourTable{$y}))
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
555 {
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
556 print " <day id=\"$y\">".join("", map { "<class>".$_."</class>" } @{$$hourTable{$y}})."</day>\n";
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
557 }
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
558 }
119f0cef6498 Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents: 57
diff changeset
559 print " </days>\n";
1
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
560
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
561 print "</timetable>\n";
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
562 }
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
563
21fde93375e9 Add beta code.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
564 close (STDOUT);