Mercurial > hg > lukkari
annotate parsedata.pl @ 181:9c3100ab29cc
Cosmetics.
| author | Matti Hamalainen <ccr@tnsp.org> |
|---|---|
| date | Tue, 08 Sep 2015 10:48:48 +0300 |
| parents | eaac0a84a7d2 |
| children | 9f7eb4db99b4 |
| rev | line source |
|---|---|
| 1 | 1 #!/usr/bin/perl -w |
| 2 # | |
| 3 # Fetch and parse HTML format class timetable into more sane formats | |
| 170 | 4 # (C) Copyright 2010-2015 Matti Hämäläinen <ccr@tnsp.org> |
| 1 | 5 # |
| 6 use strict; | |
| 7 use Data::Dumper; | |
| 8 use HTML::Entities; | |
| 9 | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
10 ### |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
11 ### Some globals |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
12 ### |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
13 my $modes = "php|xml"; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
14 my $opt_mode = "php"; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
15 my $opt_dump = 0; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
16 my $opt_filename; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
17 my $opt_outfile; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
18 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
19 my $cid = 0; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
20 my $hourTimes = []; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
21 my $hourDefs = {}; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
22 my $hourTable = {}; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
23 my $hourFillTable = {}; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
24 my $maxDays = 0; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
25 my $firstHour = 0; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
26 my $lastHour = 0; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
27 my $totalHours = 0; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
28 |
| 1 | 29 |
| 30 sub str_trim($) | |
| 31 { | |
| 32 my $str = $_[0]; | |
| 170 | 33 if (defined($str)) |
| 34 { | |
| 1 | 35 $str =~ s/^\s*//; |
| 36 $str =~ s/\s*$//; | |
| 37 } | |
| 38 return $str; | |
| 39 } | |
| 40 | |
| 41 | |
| 42 sub pop_token_a($) | |
| 43 { | |
| 44 my $tokens = $_[0]; | |
| 45 return shift(@$tokens); | |
| 46 } | |
| 47 | |
| 48 | |
| 49 sub pop_token($) | |
| 50 { | |
| 51 return str_trim(pop_token_a($_[0])); | |
| 52 } | |
| 53 | |
| 54 | |
| 55 sub parse_html_str($) | |
| 56 { | |
| 57 my $tokens = $_[0]; | |
| 58 my $token = pop_token($tokens); | |
| 59 my $str = ""; | |
| 60 $token =~ s/^\s*//; | |
| 61 | |
| 62 return undef unless (substr($token, 0, 1) eq '"'); | |
| 63 $token = substr($token, 1); | |
| 64 | |
| 170 | 65 while (defined($token)) |
| 66 { | |
| 26 | 67 my $tmp = $token; |
|
25
a076d8d22422
Fix string parsing in the HTML parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
8
diff
changeset
|
68 $tmp =~ s/\s*$//; |
| 170 | 69 if (substr($tmp, -1) eq '"') |
| 70 { | |
|
25
a076d8d22422
Fix string parsing in the HTML parser.
Matti Hamalainen <ccr@tnsp.org>
parents:
8
diff
changeset
|
71 $str .= substr($tmp, 0, -1); |
| 1 | 72 return $str; |
| 170 | 73 } |
| 74 else | |
| 75 { | |
| 1 | 76 $str .= $token; |
| 77 } | |
| 78 $token = shift(@$tokens); | |
| 79 } | |
| 80 return undef; | |
| 81 } | |
| 82 | |
| 83 | |
| 84 sub parse_html_tree($$); | |
| 85 | |
| 86 sub parse_html_tree($$) | |
| 87 { | |
| 88 my ($tokens, $tree) = @_; | |
| 89 | |
| 170 | 90 while (my $token = pop_token($tokens)) |
| 91 { | |
| 92 if ($token =~ /^<[!\/]?[a-zA-Z]+/) | |
| 93 { | |
| 1 | 94 $token = lc($token); |
| 170 | 95 if ($token =~ /^<\!.*>$/) |
| 96 { | |
| 1 | 97 # Ignore comments etc. |
| 170 | 98 } |
| 99 elsif ($token =~ /^<([a-z]+)(.*)>$/) | |
| 100 { | |
| 1 | 101 my ($name, $args) = ($1, $2); |
| 170 | 102 if ($name eq "meta" || $name eq "img") |
| 103 { | |
| 1 | 104 my $tmp = {}; |
| 105 parse_html_tree($tokens, $tree); | |
| 106 return $tree; | |
| 170 | 107 } |
| 108 else | |
| 109 { | |
| 1 | 110 my $tmp = { "name" => $name, "args" => str_trim($args) }; |
| 111 parse_html_tree($tokens, $tmp); | |
| 112 push(@{$$tree{"nodes"}}, $tmp); | |
| 113 } | |
| 170 | 114 } |
| 115 elsif ($token =~ /^<\/([a-z]+)>$/) | |
| 116 { | |
| 1 | 117 return $tree; |
| 170 | 118 } |
| 119 else | |
| 120 { | |
| 1 | 121 die("HORROR TERROR ELITE: $token\n"); |
| 122 } | |
| 170 | 123 } |
| 124 else | |
| 125 { | |
| 1 | 126 $token = str_trim(decode_entities($token)); |
| 127 push(@{$$tree{"nodes"}}, { "name" => "text", "args" => "", "text" => $token }) if length($token) > 0; | |
| 128 } | |
| 129 } | |
| 130 | |
| 131 return $tree; | |
| 132 } | |
| 133 | |
| 134 | |
| 135 sub parse_html($) | |
| 136 { | |
| 137 return undef unless defined($_[0]); | |
| 138 my $str = $_[0]; | |
| 139 my $res = { "name" => "", "args" => "" }; | |
| 140 $str =~ tr/\r/ /; | |
| 141 $str =~ tr/\n/ /; | |
| 142 my @tokens = grep { !($_ =~ /^\s*$/) } split(/(<\/?[a-zA-Z]+.*?>)/, $str); | |
| 143 parse_html_tree(\@tokens, $res); | |
| 144 return $res; | |
| 145 } | |
| 146 | |
| 147 sub html_find_node($$$); | |
| 148 | |
| 149 sub html_find_node($$$) | |
| 150 { | |
| 151 my ($node, $name, $args) = @_; | |
| 152 | |
| 170 | 153 if (defined($node)) |
| 154 { | |
| 155 if (ref($node) eq "ARRAY") | |
| 156 { | |
| 157 foreach my $n (@$node) | |
| 158 { | |
| 1 | 159 my $tmp = html_find_node($n, $name, $args); |
| 160 # Must do it like this, in order not to break the loop | |
| 161 return $tmp if defined($tmp); | |
| 162 } | |
| 170 | 163 } |
| 164 elsif (ref($node) eq "HASH") | |
| 165 { | |
| 166 if (defined($$node{"name"})) | |
| 167 { | |
| 168 if ($$node{"name"} eq $name) | |
| 169 { | |
| 1 | 170 if ($args ne "") { |
| 170 | 171 if (defined($$node{"args"}) && $$node{"args"} =~ /$args/) |
| 172 { | |
| 173 } | |
| 174 else | |
| 175 { | |
|
33
ad96ed91de92
Fix recursive parsing with attributes.
Matti Hamalainen <ccr@tnsp.org>
parents:
32
diff
changeset
|
176 return html_find_node($$node{"nodes"}, $name, $args); |
| 1 | 177 } |
| 178 } | |
| 179 return $node; | |
| 170 | 180 } |
| 181 else | |
| 182 { | |
| 1 | 183 return html_find_node($$node{"nodes"}, $name, $args); |
| 184 } | |
| 185 } | |
| 186 } | |
| 187 } | |
| 188 return undef; | |
| 189 } | |
| 190 | |
| 191 | |
| 192 sub fnode($$) | |
| 193 { | |
| 194 return html_find_node($_[0], $_[1], ""); | |
| 195 } | |
| 196 | |
| 197 | |
| 198 sub fnodea($$$) | |
| 199 { | |
| 200 return html_find_node($_[0], $_[1], $_[2]); | |
| 201 } | |
| 202 | |
| 203 | |
| 204 sub escape($) | |
| 205 { | |
| 206 my $s = $_[0]; | |
| 207 $s =~ s/(['"])/\\$1/g; | |
| 208 return $s; | |
| 209 } | |
| 210 | |
| 211 | |
| 212 sub html_collapse($$); | |
| 213 | |
| 214 sub html_collapse($$) | |
| 215 { | |
| 216 my ($node, $strip) = @_; | |
| 217 my $str = ""; | |
| 218 | |
| 170 | 219 if ($$node{"name"} eq "text") |
| 220 { | |
| 1 | 221 $str .= $$node{"text"}; |
| 170 | 222 } |
| 223 else | |
| 224 { | |
| 1 | 225 $str .= "<".$$node{"name"}.">" unless ($strip); |
| 170 | 226 foreach my $n (@{$$node{"nodes"}}) |
| 227 { | |
| 1 | 228 $str .= html_collapse($n, $strip); |
| 229 } | |
| 230 $str .= "</".$$node{"name"}.">" unless ($strip); | |
| 231 } | |
| 232 | |
| 233 return $str; | |
| 234 } | |
| 235 | |
| 236 | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
237 sub parse_hour_data($$) |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
238 { |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
239 my ($l, $rowspan) = @_; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
240 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
|
241 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
|
242 my $cturns = 0; |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
243 my $cgrouped = 0; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
244 |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
245 # Pull in data for the class/hour cell |
| 170 | 246 foreach my $h (@{$l}) |
| 247 { | |
| 248 if (defined($$h{"nodes"})) | |
| 249 { | |
| 250 foreach my $b (@{$$h{"nodes"}}) | |
| 251 { | |
| 252 if (defined($$b{"nodes"})) | |
| 253 { | |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
254 my $text = $$b{"nodes"}[0]{"text"}; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
255 $text =~ s/\.$//; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
256 |
| 146 | 257 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
|
258 { |
|
c028506aa9cc
Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents:
80
diff
changeset
|
259 $cturns = 1; |
|
c028506aa9cc
Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents:
80
diff
changeset
|
260 } |
|
c028506aa9cc
Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents:
80
diff
changeset
|
261 else |
|
c028506aa9cc
Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents:
80
diff
changeset
|
262 { |
|
c028506aa9cc
Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents:
80
diff
changeset
|
263 push(@$cdata, $text); |
|
c028506aa9cc
Separate the concepts of "every other week" classes and split group classes.
Matti Hamalainen <ccr@tnsp.org>
parents:
80
diff
changeset
|
264 } |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
265 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
266 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
267 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
268 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
269 |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
270 # 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
|
271 my $tid; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
272 if (scalar(@$cdata) > 0) { |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
273 $cid++; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
274 $tid = $cid; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
275 } else { |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
276 $tid = 0; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
277 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
278 |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
279 # Determine current day |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
280 my $cday = 0; |
| 170 | 281 for (my $x = 0; $x < 7; $x++) |
| 282 { | |
| 283 if (!defined($$hourFillTable{$lastHour}{$x})) | |
| 284 { | |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
285 $cday = $x; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
286 last; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
287 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
288 } |
| 170 | 289 for (my $t = 0; $t < $chours; $t++) |
| 290 { | |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
291 $$hourFillTable{$lastHour + $t}{$cday} = $tid; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
292 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
293 |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
294 if ($tid) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
295 { |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
296 $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
|
297 |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
298 # Grouped, if there is another class ID in second slot |
|
169
04c4f1a95dbd
Fix parsing of some course codes.
Matti Hamalainen <ccr@tnsp.org>
parents:
168
diff
changeset
|
299 $cgrouped = 1 if ($$cdata[1] =~ /^([A-Z]\d{5,6}[A-Z]*|[A-Z0-9]{6,8})$/); |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
300 if ($cgrouped) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
301 { |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
302 my $cdata1 = []; |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
303 my $cdata2 = []; |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
304 for (my $i = 0; $i < length($cdata); $i += 2) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
305 { |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
306 push(@$cdata1, $$cdata[$i]) if defined($$cdata[$i]); |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
307 push(@$cdata2, $$cdata[$i+1]) if defined($$cdata[$i+1]); |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
308 } |
| 146 | 309 $$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
|
310 } |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
311 else |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
312 { |
| 146 | 313 $$hourDefs{$cid} = { "turns" => $cturns, "grouped" => $cgrouped, "start" => $lastHour, "hours" => $chours, "data" => [ $cdata ] }; |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
314 } |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
315 |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
316 push(@{$$hourTable{$cday}}, $tid); |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
317 $totalHours += $chours; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
318 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
319 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
320 |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
321 sub parse_hour_header($) |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
322 { |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
323 if ($_[0] =~ /(\d+):(\d+)/) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
324 { |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
325 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
|
326 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
327 return undef; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
328 } |
| 1 | 329 |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
330 |
| 174 | 331 sub get_hour_data_struct($$); |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
332 |
| 174 | 333 sub get_hour_data_struct($$) |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
334 { |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
335 my @out = (); |
|
74
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
336 my ($tmp, $first) = @_; |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
337 |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
338 if (ref($tmp) eq "ARRAY") |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
339 { |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
340 my @str = (); |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
341 foreach my $item (@{$tmp}) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
342 { |
| 174 | 343 push(@str, get_hour_data_struct($item, 0)); |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
344 } |
|
74
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
345 if (scalar(@str) > 0) |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
346 { |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
347 push(@out, "array(".join(", ", @str).")") if ($opt_mode eq "php"); |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
348 push(@out, "<group>".join("", @str)."</group>") if ($opt_mode eq "xml"); |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
349 } |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
350 } |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
351 elsif (ref($tmp) eq "HASH") |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
352 { |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
353 my @str = (); |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
354 foreach my $key (keys %{$tmp}) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
355 { |
| 174 | 356 push(@out, "\"".$key."\" => ".get_hour_data_struct($$tmp{$key}, 1)) if ($opt_mode eq "php"); |
| 357 push(@out, "<".$key.">".get_hour_data_struct($$tmp{$key}, 1)."</".$key.">") if ($opt_mode eq "xml"); | |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
358 } |
|
74
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
359 if (scalar(@str) > 0) |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
360 { |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
361 push(@out, "array(".join(", ", @str).")") if ($opt_mode eq "php"); |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
362 push(@out, join("", @str)) if ($opt_mode eq "xml"); |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
363 } |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
364 } |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
365 elsif ($tmp =~ /^\d+$/) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
366 { |
|
74
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
367 if ($opt_mode eq "php") |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
368 { |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
369 push(@out, $tmp); |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
370 } |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
371 else |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
372 { |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
373 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
|
374 } |
|
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 else |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
377 { |
|
74
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
378 if ($opt_mode eq "php") |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
379 { |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
380 push(@out, "\"".$tmp."\""); |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
381 } |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
382 else |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
383 { |
|
b51ad733b624
Improvements in the parser, now also produces "XML" output.
Matti Hamalainen <ccr@tnsp.org>
parents:
72
diff
changeset
|
384 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
|
385 } |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
386 } |
|
74
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 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
|
389 return join("", @out); |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
390 } |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
391 |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
392 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
393 ### |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
394 ### Main program |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
395 ### |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
396 |
| 181 | 397 while (defined(my $arg = shift)) |
| 398 { | |
| 399 if (substr($arg, 0, 1) eq "-") | |
| 400 { | |
| 401 if ($arg =~ /^-($modes)$/o) | |
| 402 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
403 $opt_mode = $1; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
404 } |
| 181 | 405 elsif ($arg eq "-dump") |
| 406 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
407 $opt_dump = 1; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
408 } |
| 181 | 409 elsif ($arg eq "-o") |
| 410 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
411 $opt_outfile = shift or die("Output filename option -o requires an argument.\n"); |
| 181 | 412 } |
| 413 else | |
| 414 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
415 die("Invalid option '$arg'.\n"); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
416 } |
| 181 | 417 } |
| 418 else | |
| 419 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
420 $opt_filename = $arg; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
421 } |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
422 } |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
423 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
424 die("Usage: $0 [options] <filename> |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
425 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
426 -php Output a PHP include file with data in arrays (default) |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
427 -xml Output a simple XML file. |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
428 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
429 -o <filename> Set output filename. Default is to use stdout. |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
430 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
431 -dump Dump HTML tree to stdout and quit. |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
432 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
433 ") unless defined($opt_filename); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
434 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
435 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
436 my $data; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
437 open(my $fh, '<:encoding(iso-8859-1)', $opt_filename) or die("Error opening '$opt_filename': $!\n"); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
438 $data = do { local $/; <$fh> }; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
439 close($fh); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
440 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
441 die("No data in input.\n") unless (defined($data) && $data ne ""); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
442 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
443 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
444 # Filter out certain unneeded elements |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
445 $data =~ s/<font[^>]*>//ig; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
446 $data =~ s/<\/font>//ig; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
447 $data =~ s/<\/?center>//ig; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
448 $data =~ s/<br>//ig; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
449 $data =~ s/ / /ig; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
450 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
451 ### Get some general information |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
452 my $otree = parse_html($data); |
| 163 | 453 if ($opt_dump) |
| 454 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
455 print Dumper(fnode($otree, "html")); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
456 exit; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
457 } |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
458 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
459 my %class = (); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
460 my $body = fnode($otree, "body"); |
| 181 | 461 if (defined($body) && defined($$body{"nodes"})) |
| 462 { | |
| 463 foreach my $n (@{$$body{"nodes"}}) | |
| 464 { | |
| 465 if ($$n{"name"} eq "text") | |
| 466 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
467 push(@{$class{"info"}}, $$n{"text"}); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
468 } |
| 181 | 469 elsif ($$n{"name"} eq "b") |
| 470 { | |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
471 push(@{$class{"data"}}, $n); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
472 } |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
473 } |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
474 } |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
475 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
476 # Filter out some more, for easier tree access during table parsing |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
477 $data =~ s/<\/?b>//ig; |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
478 my $tree = parse_html($data); |
|
165
1d17dbb864ff
Fix parsing of new format of UNTIS html output.
Matti Hamalainen <ccr@tnsp.org>
parents:
164
diff
changeset
|
479 my $node = fnodea(fnode($tree, "body"), "table", "border=\"3\" rules=\"all\" cellpadding=\"1\" cellspacing=\"1\""); |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
480 die("No table element found in document. Perhaps the format has changed? :(\n") unless defined($node); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
481 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
482 |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
483 ### Parse through the HTML document node tree to find the data we need |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
484 |
|
8
a52a0bdb5ea1
Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents:
4
diff
changeset
|
485 # Skip zero position this way (can't use foreach here) |
| 164 | 486 my $q = $$node{"nodes"}; |
| 161 | 487 for (my $i = 1; $i < scalar(@{$q}); $i++) |
| 488 { | |
| 1 | 489 my $d = $$q[$i]{"nodes"}; |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
490 if (defined($d)) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
491 { |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
492 foreach my $n (@{$d}) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
493 { |
| 1 | 494 my $l = $$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
|
495 if (defined($l)) |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
496 { |
|
165
1d17dbb864ff
Fix parsing of new format of UNTIS html output.
Matti Hamalainen <ccr@tnsp.org>
parents:
164
diff
changeset
|
497 if ($$n{"args"} =~ /colspan=12\s+rowspan=(\d+)/) |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
498 { |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
499 parse_hour_data($l, $1); |
| 1 | 500 } |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
501 elsif ($$n{"args"} =~ /rowspan=2\s+align/) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
502 { |
|
162
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
503 my $qstart = parse_hour_header($$l[0]{"nodes"}[0]{"nodes"}[0]{"text"}); |
|
70f432e3d1dc
Some remodeling here and there.
Matti Hamalainen <ccr@tnsp.org>
parents:
161
diff
changeset
|
504 my $qend = parse_hour_header($$l[1]{"nodes"}[0]{"nodes"}[0]{"text"}); |
|
72
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
505 if (defined($qstart) && defined($qend)) |
|
6fd715063abc
Clean up some parsing operations.
Matti Hamalainen <ccr@tnsp.org>
parents:
58
diff
changeset
|
506 { |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
507 push(@$hourTimes, {"start" => $qstart, "end" => $qend}); |
| 1 | 508 } |
| 509 } | |
| 510 } | |
| 511 } | |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
512 $lastHour++; |
| 1 | 513 } |
| 514 } | |
| 515 | |
| 516 | |
|
8
a52a0bdb5ea1
Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents:
4
diff
changeset
|
517 ### 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
|
518 my $flag = 1; |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
519 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
|
520 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
|
521 $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
|
522 } |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
523 $firstHour++ if ($flag); |
|
8
a52a0bdb5ea1
Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents:
4
diff
changeset
|
524 } |
|
a52a0bdb5ea1
Crop off empty hours (time slots) and days.
Matti Hamalainen <ccr@tnsp.org>
parents:
4
diff
changeset
|
525 |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
526 $flag = 1; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
527 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
|
528 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
|
529 $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
|
530 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
531 $lastHour-- if ($flag); |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
532 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
533 |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
534 |
| 1 | 535 ### Open output file, if specified |
| 536 if (defined($opt_outfile)) { | |
| 537 open(STDOUT, '>', $opt_outfile) or die("Could not open output file '$opt_outfile'.\n"); | |
| 538 } | |
| 539 | |
|
36
bb00a40252e2
Input in ISO-8859-1 and output in UTF-8.
Matti Hamalainen <ccr@tnsp.org>
parents:
33
diff
changeset
|
540 binmode STDOUT, ':encoding(utf-8)'; |
| 1 | 541 |
| 542 ### Output data in desired format | |
| 174 | 543 if ($opt_mode eq "php") |
| 544 { | |
| 1 | 545 print "<?\n". |
| 546 "\$classInfo = array(\n". | |
| 547 " \"general\" => array(".join(", ", map { "\"".escape($_)."\""; } @{$class{"info"}})."),\n". | |
| 548 " \"info\" => array(".join(", ", map { "\"".escape(html_collapse($_, 1))."\""; } @{$class{"data"}})."),\n". | |
| 54 | 549 " \"tags\" => array(".join(", ", map { "\"".escape(html_collapse($_, 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
|
550 " \"maxDays\" => $maxDays,\n". |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
551 " \"firstHour\" => $firstHour,\n". |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
552 " \"lastHour\" => $lastHour,\n". |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
553 " \"totalHours\" => $totalHours\n". |
| 1 | 554 ");\n\n"; |
| 555 | |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
556 print "\$classHourTimes = array(\n"; |
| 174 | 557 foreach my $chour (@$hourTimes) |
| 558 { | |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
559 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
|
560 } |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
561 print ");\n\n"; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
562 |
| 54 | 563 print "\$classHourDefs = array(\n"; |
| 174 | 564 foreach my $cid (sort { $a <=> $b } keys %{$hourDefs}) |
| 565 { | |
| 566 print " $cid => array(".get_hour_data_struct($$hourDefs{$cid}, 0)."),\n"; | |
| 1 | 567 } |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
568 print ");\n\n"; |
| 1 | 569 |
|
57
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
570 print |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
571 "\$classDayTable = array(\n"; |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
572 for (my $y = 0; $y < $maxDays; $y++) |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
573 { |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
574 if (defined($$hourTable{$y})) |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
575 { |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
576 print " $y => array(".join(", ", @{$$hourTable{$y}})."),\n"; |
| 1 | 577 } |
| 578 } | |
| 579 print ");\n?>\n"; | |
| 580 } | |
| 581 | |
| 174 | 582 elsif ($opt_mode eq "xml") |
| 583 { | |
| 1 | 584 print "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n". |
| 585 "<timetable>\n". | |
| 586 " <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
|
587 " <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
|
588 " <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
|
589 " <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
|
590 " <firsthour>$firstHour</firsthour>\n". |
|
93c87f42c803
New parser output format, parser logic cleaned up a bit.
Matti Hamalainen <ccr@tnsp.org>
parents:
55
diff
changeset
|
591 " <lasthour>$lastHour</lasthour>\n". |
|
58
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
592 " <totalhours>$totalHours</totalhours>\n". |
| 1 | 593 " </class>\n"; |
| 594 | |
|
58
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
595 print " <hours>\n"; |
|
75
3d9e42477367
More improvements in the parsing and XML output.
Matti Hamalainen <ccr@tnsp.org>
parents:
74
diff
changeset
|
596 my $cid = 0; |
| 174 | 597 foreach my $chour (@$hourTimes) |
| 598 { | |
|
75
3d9e42477367
More improvements in the parsing and XML output.
Matti Hamalainen <ccr@tnsp.org>
parents:
74
diff
changeset
|
599 print " <hour id=\"".$cid."\"><start>".$$chour{"start"}."</start><end>".$$chour{"end"}."</end></hour>\n"; |
|
3d9e42477367
More improvements in the parsing and XML output.
Matti Hamalainen <ccr@tnsp.org>
parents:
74
diff
changeset
|
600 $cid++; |
|
58
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
601 } |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
602 print " </hours>\n\n"; |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
603 |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
604 print " <classes>\n"; |
| 174 | 605 foreach $cid (sort { $a <=> $b } keys %{$hourDefs}) |
| 606 { | |
| 607 print " <class id=\"$cid\">".get_hour_data_struct($$hourDefs{$cid}, 0)."</class>\n"; | |
|
58
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
608 } |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
609 print " </classes>\n\n"; |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
610 |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
611 print " <days>\n"; |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
612 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
|
613 { |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
614 if (defined($$hourTable{$y})) |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
615 { |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
616 print " <day id=\"$y\">".join("", map { "<class>".$_."</class>" } @{$$hourTable{$y}})."</day>\n"; |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
617 } |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
618 } |
|
119f0cef6498
Work on XML output support (not finished yet.)
Matti Hamalainen <ccr@tnsp.org>
parents:
57
diff
changeset
|
619 print " </days>\n"; |
| 1 | 620 |
| 621 print "</timetable>\n"; | |
| 622 } | |
| 623 | |
| 624 close (STDOUT); |