| 1 | #!/usr/bin/perl |
|---|
| 2 | # |
|---|
| 3 | # citysearch TV guide grabber |
|---|
| 4 | # |
|---|
| 5 | |
|---|
| 6 | my $version = '4.0.0'; |
|---|
| 7 | |
|---|
| 8 | use strict; |
|---|
| 9 | use Getopt::Long; |
|---|
| 10 | use POSIX; |
|---|
| 11 | use Data::Dumper; |
|---|
| 12 | use IO::File; |
|---|
| 13 | use XMLTV; |
|---|
| 14 | use HTML::TreeBuilder; |
|---|
| 15 | use Shepherd::Common; |
|---|
| 16 | |
|---|
| 17 | # --------------------------------------------------------------------------- |
|---|
| 18 | # --- Global Variables |
|---|
| 19 | |
|---|
| 20 | my $progname = "citysearch"; |
|---|
| 21 | |
|---|
| 22 | my $DATASOURCE = "http://citysearch.com.au"; |
|---|
| 23 | |
|---|
| 24 | my $lang = 'en'; |
|---|
| 25 | my $debug = 0; |
|---|
| 26 | my $channels; |
|---|
| 27 | my $opt_channels; |
|---|
| 28 | my $opt = { }; |
|---|
| 29 | my $gaps; |
|---|
| 30 | my %stats; |
|---|
| 31 | my $shows; |
|---|
| 32 | my $cache; |
|---|
| 33 | my $runtime = time; |
|---|
| 34 | my $zerohr; |
|---|
| 35 | my @skipped_channels; |
|---|
| 36 | |
|---|
| 37 | # --------------------------------------------------------------------------- |
|---|
| 38 | # --- Setup |
|---|
| 39 | |
|---|
| 40 | print "$progname $version\n"; |
|---|
| 41 | |
|---|
| 42 | $| = 1; |
|---|
| 43 | |
|---|
| 44 | &get_command_line_options; |
|---|
| 45 | |
|---|
| 46 | exit 0 if ($opt->{version}); |
|---|
| 47 | |
|---|
| 48 | &help if ($opt->{help}); |
|---|
| 49 | |
|---|
| 50 | &set_defaults; |
|---|
| 51 | |
|---|
| 52 | &read_channels_file; |
|---|
| 53 | |
|---|
| 54 | unless ($channels) |
|---|
| 55 | { |
|---|
| 56 | print "ERROR: No channels requested. Please use --channels_file.\n"; |
|---|
| 57 | exit 33; |
|---|
| 58 | } |
|---|
| 59 | |
|---|
| 60 | &read_gaps_file; |
|---|
| 61 | |
|---|
| 62 | &read_cache; |
|---|
| 63 | |
|---|
| 64 | &set_region; |
|---|
| 65 | |
|---|
| 66 | &get_guide_data; |
|---|
| 67 | |
|---|
| 68 | &calculate_stop_times; |
|---|
| 69 | |
|---|
| 70 | &details; |
|---|
| 71 | |
|---|
| 72 | &write_cache; |
|---|
| 73 | |
|---|
| 74 | &write_xml; |
|---|
| 75 | |
|---|
| 76 | &Shepherd::Common::print_stats($progname, $version, $runtime, %stats); |
|---|
| 77 | |
|---|
| 78 | &log("Done."); |
|---|
| 79 | exit; |
|---|
| 80 | |
|---|
| 81 | |
|---|
| 82 | # --------------------------------------------------------------------------- |
|---|
| 83 | # --- Subs |
|---|
| 84 | |
|---|
| 85 | sub get_guide_data |
|---|
| 86 | { |
|---|
| 87 | &log("Grabbing data for days " . $opt->{offset} . |
|---|
| 88 | " - " . ($opt->{days} - 1) . |
|---|
| 89 | ($opt->{output} ? " into " . $opt->{output} : '') . |
|---|
| 90 | "."); |
|---|
| 91 | |
|---|
| 92 | # Calculate midnight on day zero in epoch time |
|---|
| 93 | my @today = localtime($runtime); # 0=sec,1=min,2=hour,3=day,4=month,5=year,6=wday,7=yday,8=isdst |
|---|
| 94 | $zerohr = $runtime - (($today[0]) + ($today[1]*60) + ($today[2]*60*60)); |
|---|
| 95 | |
|---|
| 96 | for my $day ($opt->{offset} .. ($opt->{days} - 1)) |
|---|
| 97 | { |
|---|
| 98 | my $dow = &POSIX::strftime("%A", localtime($runtime + ($day * 86400))); |
|---|
| 99 | |
|---|
| 100 | &log("Day $day ($dow)"); |
|---|
| 101 | |
|---|
| 102 | my $start_hr = 0; |
|---|
| 103 | if (!$day) |
|---|
| 104 | { |
|---|
| 105 | $start_hr = int($today[2] / 3) * 3; |
|---|
| 106 | } |
|---|
| 107 | for (my $hr = $start_hr; $hr < 24; $hr += 3) |
|---|
| 108 | { |
|---|
| 109 | |
|---|
| 110 | &log("Time window $hr:00 - " . ($hr+3) . ":00"); |
|---|
| 111 | |
|---|
| 112 | my $url = "$DATASOURCE/tvguide/$day/$hr:00?city=" . $opt->{'rname'}; |
|---|
| 113 | my $guidedata = &Shepherd::Common::get_url($url); |
|---|
| 114 | exit 11 unless ($guidedata); |
|---|
| 115 | |
|---|
| 116 | # Verify that the guide page really is for the day we want. |
|---|
| 117 | my $daystr = &POSIX::strftime("%A, %d %B", localtime($runtime + ($day * 86400))); |
|---|
| 118 | $daystr =~ s/, 0(\d)/, $1/; |
|---|
| 119 | |
|---|
| 120 | unless ($guidedata =~ /<p class="date">$daystr<\/p>/) |
|---|
| 121 | { |
|---|
| 122 | &log("Exiting: couldn't locate daystring \"$daystr\" in guide page for $dow."); |
|---|
| 123 | exit 21; |
|---|
| 124 | } |
|---|
| 125 | |
|---|
| 126 | &parse_guide($guidedata, $day, $hr); |
|---|
| 127 | } |
|---|
| 128 | } |
|---|
| 129 | &log("Found " . &num_items($shows) . " shows on " . scalar(keys %$shows) . " channels."); |
|---|
| 130 | } |
|---|
| 131 | |
|---|
| 132 | sub parse_guide |
|---|
| 133 | { |
|---|
| 134 | my ($guidedata, $day, $window) = @_; |
|---|
| 135 | |
|---|
| 136 | &log("Parsing guide page (Day $day hr $window).") if ($debug); |
|---|
| 137 | my $tree = HTML::TreeBuilder->new; |
|---|
| 138 | $tree->no_space_compacting(1); |
|---|
| 139 | $tree->parse($guidedata); |
|---|
| 140 | $tree->eof; |
|---|
| 141 | |
|---|
| 142 | foreach my $table ($tree->look_down(_tag => 'table', class => 'tvGuideTable')) |
|---|
| 143 | { |
|---|
| 144 | &log("Found table.") if ($debug); |
|---|
| 145 | foreach my $tr ($table->look_down(_tag => 'tr')) |
|---|
| 146 | { |
|---|
| 147 | my $ctag = $tr->look_down(_tag => 'td', class => 'channel'); |
|---|
| 148 | next unless ($ctag); |
|---|
| 149 | |
|---|
| 150 | $ctag = $ctag->look_down(_tag => 'p', class => 'channelName'); |
|---|
| 151 | my $channame = $ctag->as_text; |
|---|
| 152 | my $chanid = $channels->{$channame}; |
|---|
| 153 | unless ($chanid) |
|---|
| 154 | { |
|---|
| 155 | unless (grep $_ eq $channame, @skipped_channels) |
|---|
| 156 | { |
|---|
| 157 | &log("Skipping unsubscribed channel \"$channame\"."); |
|---|
| 158 | push @skipped_channels, $channame; |
|---|
| 159 | } |
|---|
| 160 | next; |
|---|
| 161 | } |
|---|
| 162 | |
|---|
| 163 | &log("Channel $channame") if ($debug); |
|---|
| 164 | |
|---|
| 165 | # When we hit a "Continue Before" block, it means we're missing |
|---|
| 166 | # a show's start time. Skip next show in this case. |
|---|
| 167 | my $continue_before = 0; |
|---|
| 168 | |
|---|
| 169 | foreach my $td ($tr->look_down(_tag => 'td')) |
|---|
| 170 | { |
|---|
| 171 | my $td_class = $td->attr('class'); |
|---|
| 172 | next unless ($td_class and $td_class =~ /(\d\d)(\d\d)/); |
|---|
| 173 | my $block_start = ($1 * 3600) + ($2 * 60); |
|---|
| 174 | |
|---|
| 175 | $continue_before = 1 if ($td_class =~ /continueBefore/); |
|---|
| 176 | |
|---|
| 177 | foreach my $showblock ($td->look_down(_tag => 'div', class => 'programWrapper')) |
|---|
| 178 | { |
|---|
| 179 | |
|---|
| 180 | if ($continue_before) |
|---|
| 181 | { |
|---|
| 182 | $continue_before = 0; |
|---|
| 183 | next; |
|---|
| 184 | } |
|---|
| 185 | |
|---|
| 186 | my $show; |
|---|
| 187 | |
|---|
| 188 | $show->{channel} = $chanid; |
|---|
| 189 | |
|---|
| 190 | my $start = $block_start; |
|---|
| 191 | if ($showblock->as_HTML =~ /<span class="oddStartTime">(\d+)\.(\d+)<\/span> ([ap]m)/) |
|---|
| 192 | { |
|---|
| 193 | $start = ($1 * 3600) + ($2 * 60); |
|---|
| 194 | $start += (12 * 3600) if ($3 eq 'pm'); |
|---|
| 195 | $start -= (12 * 3600) if ($3 eq 'am' and $1 == 12); |
|---|
| 196 | } |
|---|
| 197 | $show->{start} = $zerohr + (86400 * $day) + $start; |
|---|
| 198 | |
|---|
| 199 | my $atag = $showblock->look_down(_tag => 'a'); |
|---|
| 200 | unless ($atag) |
|---|
| 201 | { |
|---|
| 202 | # Caused by "No information available" entries |
|---|
| 203 | &log("Empty show block: day $day hr $window chan $channame") if ($debug); |
|---|
| 204 | next; |
|---|
| 205 | } |
|---|
| 206 | $show->{title} = $atag->as_text(); |
|---|
| 207 | |
|---|
| 208 | die "Missing pid: day $day hr $window chan $channame title $show->{title}" unless ($atag->attr('href') =~ m"/tvguide/viewTvProgram/tvReviews-(.*)"); |
|---|
| 209 | $show->{pid} = $1; |
|---|
| 210 | |
|---|
| 211 | #if ($showblock->as_HTML =~ /<span class="accessLink">(\w+)<\/span>/) |
|---|
| 212 | #{ |
|---|
| 213 | # push @{$show->{category}}, $1; |
|---|
| 214 | #} |
|---|
| 215 | |
|---|
| 216 | if ($showblock->as_HTML =~ /Rpt/) |
|---|
| 217 | { |
|---|
| 218 | $show->{'previously-shown'} = { }; |
|---|
| 219 | } |
|---|
| 220 | &log("- $show->{title}") if ($debug); |
|---|
| 221 | $shows->{$chanid}->{$show->{start}} = $show; |
|---|
| 222 | } |
|---|
| 223 | } |
|---|
| 224 | } |
|---|
| 225 | } |
|---|
| 226 | $tree->delete; |
|---|
| 227 | } |
|---|
| 228 | |
|---|
| 229 | sub details |
|---|
| 230 | { |
|---|
| 231 | # iterate through our list, compare to cache, lookup if necessary |
|---|
| 232 | my $count = 0; |
|---|
| 233 | my $num_shows = &num_items($shows); |
|---|
| 234 | foreach my $ch (keys %$shows) |
|---|
| 235 | { |
|---|
| 236 | foreach my $s (sort keys %{$shows->{$ch}}) |
|---|
| 237 | { |
|---|
| 238 | my $show = $shows->{$ch}->{$s}; |
|---|
| 239 | if ($show->{start} > $zerohr + (86400 * $opt->{days})) |
|---|
| 240 | { |
|---|
| 241 | &log("Late : " . $show->{title}) if ($debug); |
|---|
| 242 | delete $shows->{$ch}->{$s}; |
|---|
| 243 | } |
|---|
| 244 | elsif ($show->{stop} and $show->{stop} < $zerohr + (86400 * $opt->{offset})) |
|---|
| 245 | { |
|---|
| 246 | &log("Early : " . $show->{title}) if ($debug); |
|---|
| 247 | delete $shows->{$ch}->{$s}; |
|---|
| 248 | } |
|---|
| 249 | elsif ($gaps and &is_outside_gaps($show->{channel}, $show->{start}, $show->{stop})) |
|---|
| 250 | { |
|---|
| 251 | &log("Nongap: " . $show->{title}) if ($debug); |
|---|
| 252 | delete $shows->{$ch}->{$s}; |
|---|
| 253 | } |
|---|
| 254 | elsif ( |
|---|
| 255 | $cache |
|---|
| 256 | and |
|---|
| 257 | $cache->{$ch} |
|---|
| 258 | and |
|---|
| 259 | $cache->{$ch}->{$s} |
|---|
| 260 | and |
|---|
| 261 | $cache->{$ch}->{$s}->{details} |
|---|
| 262 | and |
|---|
| 263 | $cache->{$ch}->{$s}->{stop} eq $show->{stop} |
|---|
| 264 | and |
|---|
| 265 | $cache->{$ch}->{$s}->{title} eq $show->{title}) |
|---|
| 266 | { |
|---|
| 267 | &log("Cached: ". $show->{title}) if ($debug); |
|---|
| 268 | $shows->{$ch}->{$s} = $cache->{$ch}->{$s}; |
|---|
| 269 | $stats{cache_hits}++; |
|---|
| 270 | $stats{shows}++; |
|---|
| 271 | } |
|---|
| 272 | else |
|---|
| 273 | { |
|---|
| 274 | &log("New : " . $show->{title}) if ($debug); |
|---|
| 275 | my $html = &fetch_details($show->{pid}); |
|---|
| 276 | if ($html) |
|---|
| 277 | { |
|---|
| 278 | &parse_details($html, $show); |
|---|
| 279 | $show->{details} = 1; |
|---|
| 280 | $cache->{$ch}->{$s} = $show; |
|---|
| 281 | $stats{shows}++; |
|---|
| 282 | } |
|---|
| 283 | else |
|---|
| 284 | { |
|---|
| 285 | &log("Couldn't fetch " . $show->{title} . |
|---|
| 286 | " (pid " . $show->{pid} . ")!"); |
|---|
| 287 | } |
|---|
| 288 | } |
|---|
| 289 | $count++; |
|---|
| 290 | if ($count % 25 == 0) |
|---|
| 291 | { |
|---|
| 292 | &log(sprintf " ...processed %d of %d shows [%s elapsed, %d new, %d cached, %d unwanted]", |
|---|
| 293 | $count, $num_shows, |
|---|
| 294 | &Shepherd::Common::pretty_duration(time - $runtime), |
|---|
| 295 | $stats{shows} - $stats{cache_hits}, |
|---|
| 296 | $stats{cache_hits}, |
|---|
| 297 | $count - $stats{shows}); |
|---|
| 298 | } |
|---|
| 299 | } |
|---|
| 300 | } |
|---|
| 301 | } |
|---|
| 302 | |
|---|
| 303 | sub fetch_details |
|---|
| 304 | { |
|---|
| 305 | my $pid = shift; |
|---|
| 306 | |
|---|
| 307 | my $url = "$DATASOURCE/tvpopup/tvReviews-$pid?city=".$opt->{'rname'}; |
|---|
| 308 | my $html = &Shepherd::Common::get_url($url); |
|---|
| 309 | |
|---|
| 310 | # my $html = &Shepherd::Common::get_url(url => $url, referer => 'http://citysearch.com.au/tvguide/viewTvProgram/tvReviews-39204745-94-1'); |
|---|
| 311 | return $html; |
|---|
| 312 | } |
|---|
| 313 | |
|---|
| 314 | sub parse_details |
|---|
| 315 | { |
|---|
| 316 | my ($html, $show) = @_; |
|---|
| 317 | |
|---|
| 318 | &log("Parsing \"$show->{title}\"") if ($debug); |
|---|
| 319 | my $tree = HTML::TreeBuilder->new_from_content($html); |
|---|
| 320 | |
|---|
| 321 | my $block = $tree->look_down(_tag => 'div', id => 'popupContent'); |
|---|
| 322 | unless ($block) |
|---|
| 323 | { |
|---|
| 324 | print "Dumping bad HTML:\n$html\n" if ($debug); |
|---|
| 325 | print "Can't parse details page! Title: $show->{title}\n"; |
|---|
| 326 | exit 22; |
|---|
| 327 | } |
|---|
| 328 | |
|---|
| 329 | my $desc = $block->right(); |
|---|
| 330 | if ($desc) |
|---|
| 331 | { |
|---|
| 332 | $desc = &strip_whitespace($desc->as_text); |
|---|
| 333 | $show->{desc} = $desc if ($desc); |
|---|
| 334 | } else { &log("Missing desc???\n"); } |
|---|
| 335 | |
|---|
| 336 | my (%video, $category, %type); |
|---|
| 337 | foreach my $label ($block->look_down(_tag => 'label')) |
|---|
| 338 | { |
|---|
| 339 | if ($label->as_text =~ /(.*?):/) |
|---|
| 340 | { |
|---|
| 341 | my $value = $label->right()->as_text; |
|---|
| 342 | if ($1 eq 'Type') |
|---|
| 343 | { |
|---|
| 344 | $category = &strip_whitespace($value); |
|---|
| 345 | } |
|---|
| 346 | elsif ($1 eq 'Country') |
|---|
| 347 | { |
|---|
| 348 | $show->{country} = $value; |
|---|
| 349 | } |
|---|
| 350 | elsif ($1 eq 'Language') |
|---|
| 351 | { |
|---|
| 352 | $show->{language} = $value; |
|---|
| 353 | } |
|---|
| 354 | elsif ($1 eq 'Cast') |
|---|
| 355 | { |
|---|
| 356 | foreach (split /, /, $value) |
|---|
| 357 | { |
|---|
| 358 | push @{$show->{credits}{actor}}, &strip_whitespace($_); |
|---|
| 359 | } |
|---|
| 360 | } |
|---|
| 361 | elsif ($1 eq 'Director') |
|---|
| 362 | { |
|---|
| 363 | foreach (split /, /, $value) |
|---|
| 364 | { |
|---|
| 365 | push @{$show->{credits}{director}}, &strip_whitespace($_); |
|---|
| 366 | } |
|---|
| 367 | } |
|---|
| 368 | elsif ($1 eq 'Writer') # unseen |
|---|
| 369 | { |
|---|
| 370 | foreach (split /, /, $value) |
|---|
| 371 | { |
|---|
| 372 | push @{$show->{credits}{writer}}, &strip_whitespace($_); |
|---|
| 373 | } |
|---|
| 374 | } |
|---|
| 375 | elsif ($1 eq 'Duration') |
|---|
| 376 | { |
|---|
| 377 | if ($2 =~ /(\d+) min/) |
|---|
| 378 | { |
|---|
| 379 | $show->{length} = $1 * 60; |
|---|
| 380 | if (!$show->{stop}) { |
|---|
| 381 | $show->{stop} = $show->{start} + ($1 * 60); |
|---|
| 382 | &log("Filled in stop time! $1 minutes.") if ($debug); |
|---|
| 383 | } |
|---|
| 384 | } |
|---|
| 385 | } |
|---|
| 386 | elsif ($1 eq 'Format') |
|---|
| 387 | { |
|---|
| 388 | foreach my $info (split /, /, $value) |
|---|
| 389 | { |
|---|
| 390 | $info = &strip_whitespace($info); |
|---|
| 391 | if ($info eq 'Closed Captions') |
|---|
| 392 | { |
|---|
| 393 | push @{$show->{'subtitles'}}, 'teletext'; |
|---|
| 394 | } |
|---|
| 395 | elsif ($info eq 'Subtitles') |
|---|
| 396 | { |
|---|
| 397 | push @{$show->{'subtitles'}}, 'onscreen'; |
|---|
| 398 | } |
|---|
| 399 | elsif ($info eq 'Widescreen') |
|---|
| 400 | { |
|---|
| 401 | $video{aspect} = '16:9'; |
|---|
| 402 | } |
|---|
| 403 | elsif ($info eq 'High Definition') |
|---|
| 404 | { |
|---|
| 405 | $video{'quality'} = 'HDTV'; |
|---|
| 406 | } |
|---|
| 407 | elsif ($info eq 'Premiere') |
|---|
| 408 | { |
|---|
| 409 | $show->{'premiere'} = [ $info ]; |
|---|
| 410 | $type{premiere} = 1; |
|---|
| 411 | } |
|---|
| 412 | elsif ($info eq 'Live') |
|---|
| 413 | { |
|---|
| 414 | $type{live} = 1; |
|---|
| 415 | } |
|---|
| 416 | elsif ($info eq 'Final' || $info eq 'Finale') # unseen |
|---|
| 417 | { |
|---|
| 418 | $type{final} = 1; |
|---|
| 419 | } |
|---|
| 420 | elsif ($info eq 'Return') # unseen |
|---|
| 421 | { |
|---|
| 422 | $type{return} = 1; |
|---|
| 423 | } |
|---|
| 424 | elsif ($info eq 'Repeat') |
|---|
| 425 | { |
|---|
| 426 | $show->{'previously-shown'} = { }; |
|---|
| 427 | } |
|---|
| 428 | elsif ($info eq 'Movie') |
|---|
| 429 | { |
|---|
| 430 | $type{movie} = 1; |
|---|
| 431 | } |
|---|
| 432 | else |
|---|
| 433 | { |
|---|
| 434 | &log("Unknown info field: \"$info\""); |
|---|
| 435 | } |
|---|
| 436 | } |
|---|
| 437 | } |
|---|
| 438 | elsif ($1 eq 'Rating') |
|---|
| 439 | { |
|---|
| 440 | $show->{rating} = $value; |
|---|
| 441 | } |
|---|
| 442 | elsif ($1 eq 'Year') |
|---|
| 443 | { |
|---|
| 444 | $show->{date} = $value; |
|---|
| 445 | } |
|---|
| 446 | elsif ($1 eq 'Channel' or $1 eq 'Time') |
|---|
| 447 | { |
|---|
| 448 | # ignore: handled elsewhere |
|---|
| 449 | } |
|---|
| 450 | else |
|---|
| 451 | { |
|---|
| 452 | &log("Ignoring $1: $value") if ($debug); |
|---|
| 453 | } |
|---|
| 454 | } |
|---|
| 455 | else |
|---|
| 456 | { |
|---|
| 457 | &log("Unknown label: " .$label->as_text); |
|---|
| 458 | } |
|---|
| 459 | } |
|---|
| 460 | $show->{video} = { %video } if (%video); |
|---|
| 461 | $show->{category} = [ &Shepherd::Common::generate_category( |
|---|
| 462 | $show->{title}, $category, %type) ]; |
|---|
| 463 | |
|---|
| 464 | $tree->delete; |
|---|
| 465 | |
|---|
| 466 | print "Parsed: " . Dumper($show) if ($debug); |
|---|
| 467 | } |
|---|
| 468 | |
|---|
| 469 | sub calculate_stop_times |
|---|
| 470 | { |
|---|
| 471 | foreach my $ch (keys %$shows) |
|---|
| 472 | { |
|---|
| 473 | my $last_start_time; |
|---|
| 474 | foreach my $s (reverse sort keys %{$shows->{$ch}}) |
|---|
| 475 | { |
|---|
| 476 | $shows->{$ch}->{$s}->{stop} = $last_start_time if ($last_start_time); |
|---|
| 477 | $last_start_time = $shows->{$ch}->{$s}->{start}; |
|---|
| 478 | } |
|---|
| 479 | } |
|---|
| 480 | } |
|---|
| 481 | |
|---|
| 482 | sub write_xml |
|---|
| 483 | { |
|---|
| 484 | my %writer_args = ( encoding => 'ISO-8859-1' ); |
|---|
| 485 | |
|---|
| 486 | &log("Writing " . &num_items($shows) . " shows to XML."); |
|---|
| 487 | |
|---|
| 488 | if ($opt->{output}) |
|---|
| 489 | { |
|---|
| 490 | my $fh = new IO::File(">" . $opt->{output}) |
|---|
| 491 | or die "Can't open " . $opt->{output} . ": $!"; |
|---|
| 492 | $writer_args{OUTPUT} = $fh; |
|---|
| 493 | } |
|---|
| 494 | |
|---|
| 495 | my $writer = new XMLTV::Writer(%writer_args); |
|---|
| 496 | |
|---|
| 497 | $writer->start |
|---|
| 498 | ( { 'source-info-url' => $DATASOURCE, |
|---|
| 499 | 'source-info-name' => "Citysearch", |
|---|
| 500 | 'generator-info-name' => "$progname $version"} ); |
|---|
| 501 | |
|---|
| 502 | for my $channel (sort keys %$channels) |
|---|
| 503 | { |
|---|
| 504 | $writer->write_channel( { |
|---|
| 505 | 'display-name' => [ [ $channel, $lang ] ], |
|---|
| 506 | 'id' => $channels->{$channel} } ); |
|---|
| 507 | } |
|---|
| 508 | |
|---|
| 509 | foreach my $ch (sort keys %$shows) |
|---|
| 510 | { |
|---|
| 511 | foreach my $s (sort keys %{$shows->{$ch}}) |
|---|
| 512 | { |
|---|
| 513 | # Don't return shows with no stop time |
|---|
| 514 | next unless ($shows->{$ch}->{$s}->{stop}); |
|---|
| 515 | |
|---|
| 516 | # Format for XMLTV-compliance |
|---|
| 517 | my %p = %{$shows->{$ch}->{$s}}; |
|---|
| 518 | foreach my $field ('title', 'sub-title', 'desc', 'country') |
|---|
| 519 | { |
|---|
| 520 | $p{$field} = [[ $p{$field}, $lang ]] if ($p{$field}); |
|---|
| 521 | } |
|---|
| 522 | $p{language} = [ $p{language}, $lang ] if ($p{language}); |
|---|
| 523 | $p{start} = &POSIX::strftime("%Y%m%d%H%M", localtime($p{start})); |
|---|
| 524 | $p{stop} = &POSIX::strftime("%Y%m%d%H%M", localtime($p{stop})); |
|---|
| 525 | $p{rating} = [[ $p{rating}, 'ABA', undef ]] if ($p{rating}); |
|---|
| 526 | if ($p{category} && ref($p{category}) eq "ARRAY" |
|---|
| 527 | && $p{category}[0] && ref($p{category}[0]) ne "ARRAY") # obsolete 14/10/2007 |
|---|
| 528 | { |
|---|
| 529 | foreach (@{$p{category}}) |
|---|
| 530 | { |
|---|
| 531 | $_ = [ &Shepherd::Common::translate_category($_), $lang ]; |
|---|
| 532 | } |
|---|
| 533 | } |
|---|
| 534 | if ($p{subtitles}) |
|---|
| 535 | { |
|---|
| 536 | my @s; |
|---|
| 537 | foreach (@{$p{subtitles}}) |
|---|
| 538 | { |
|---|
| 539 | push @s, { type => $_ }; |
|---|
| 540 | } |
|---|
| 541 | $p{subtitles} = [ @s ]; |
|---|
| 542 | } |
|---|
| 543 | delete $p{pid}; |
|---|
| 544 | delete $p{details}; |
|---|
| 545 | |
|---|
| 546 | &log("-> " . $shows->{$ch}->{$s}->{title}) if ($debug); |
|---|
| 547 | # print Dumper(\%p); |
|---|
| 548 | $shows->{$ch}->{$s}->{start} = &POSIX::strftime("%Y%m%d%H%M", localtime($s)); |
|---|
| 549 | $writer->write_programme(\%p); |
|---|
| 550 | } |
|---|
| 551 | } |
|---|
| 552 | |
|---|
| 553 | $writer->end(); |
|---|
| 554 | } |
|---|
| 555 | |
|---|
| 556 | # --------------------------------------------------------------------- |
|---|
| 557 | # Helper subs |
|---|
| 558 | |
|---|
| 559 | sub num_items |
|---|
| 560 | { |
|---|
| 561 | my $hash = shift; |
|---|
| 562 | my $count = 0; |
|---|
| 563 | foreach my $ch (keys %$hash) |
|---|
| 564 | { |
|---|
| 565 | $count += scalar keys %{$hash->{$ch}}; |
|---|
| 566 | } |
|---|
| 567 | return $count; |
|---|
| 568 | } |
|---|
| 569 | |
|---|
| 570 | sub is_outside_gaps |
|---|
| 571 | { |
|---|
| 572 | my ($ch, $start, $stop) = @_; |
|---|
| 573 | |
|---|
| 574 | foreach my $gap (@{$gaps->{$ch}}) |
|---|
| 575 | { |
|---|
| 576 | if ($gap =~ /(\d+)-(\d+)/) |
|---|
| 577 | { |
|---|
| 578 | return 0 if ($stop > $1 and $start < $2); |
|---|
| 579 | } |
|---|
| 580 | } |
|---|
| 581 | return 1; |
|---|
| 582 | } |
|---|
| 583 | |
|---|
| 584 | sub strip_whitespace |
|---|
| 585 | { |
|---|
| 586 | $_[0] =~ /^\s*(.*?)\s*$/ ? $1 : $_[0]; |
|---|
| 587 | } |
|---|
| 588 | |
|---|
| 589 | # --------------------------------------------------------------------- |
|---|
| 590 | # Setup subs |
|---|
| 591 | |
|---|
| 592 | |
|---|
| 593 | sub read_cache |
|---|
| 594 | { |
|---|
| 595 | $cache = Shepherd::Common::read_cache(\$opt->{'cache-file'}); |
|---|
| 596 | if ($cache) |
|---|
| 597 | { |
|---|
| 598 | &log("Retrieved " . &num_items($cache) . " cached items from file."); |
|---|
| 599 | &clean_cache; |
|---|
| 600 | } |
|---|
| 601 | else |
|---|
| 602 | { |
|---|
| 603 | $cache = { }; |
|---|
| 604 | &log("Not using cache."); |
|---|
| 605 | } |
|---|
| 606 | if ($opt->{'dump-cache'}) |
|---|
| 607 | { |
|---|
| 608 | &log("Dumping cache."); |
|---|
| 609 | print Dumper($cache); |
|---|
| 610 | exit 0; |
|---|
| 611 | } |
|---|
| 612 | } |
|---|
| 613 | |
|---|
| 614 | sub clean_cache |
|---|
| 615 | { |
|---|
| 616 | my $cutoff = $runtime - 86400; |
|---|
| 617 | &log("Removing cached shows that finish earlier than " . localtime($cutoff) . ".") if ($debug); |
|---|
| 618 | my $count = 0; |
|---|
| 619 | |
|---|
| 620 | foreach my $ch (keys %$cache) |
|---|
| 621 | { |
|---|
| 622 | foreach my $s (keys %{$cache->{$ch}}) |
|---|
| 623 | { |
|---|
| 624 | if ($cache->{$ch}->{$s}->{stop} < $cutoff) |
|---|
| 625 | { |
|---|
| 626 | &log("Removing $cache->{$ch}->{$s}->{title}.") if ($debug); |
|---|
| 627 | delete $cache->{$ch}->{$s}; |
|---|
| 628 | $count++; |
|---|
| 629 | } |
|---|
| 630 | } |
|---|
| 631 | } |
|---|
| 632 | &log("Removed $count stale items from cache.") if ($count); |
|---|
| 633 | } |
|---|
| 634 | |
|---|
| 635 | sub write_cache |
|---|
| 636 | { |
|---|
| 637 | my $n = &num_items($cache); |
|---|
| 638 | return unless ($n); |
|---|
| 639 | &log("Writing $n shows to cache."); |
|---|
| 640 | Shepherd::Common::write_cache($opt->{'cache-file'}, $cache); |
|---|
| 641 | } |
|---|
| 642 | |
|---|
| 643 | sub set_region |
|---|
| 644 | { |
|---|
| 645 | my %regions = ( 81 => 'adelaide', 75 => 'brisbane', 126 => 'canberra', |
|---|
| 646 | 74 => 'darwin', 88 => 'hobart', 94 => 'melbourne', |
|---|
| 647 | 101 => 'perth', 73 => 'sydney' ); |
|---|
| 648 | unless ($regions{$opt->{region}}) |
|---|
| 649 | { |
|---|
| 650 | &log("ERROR: unsupported region \"$opt->{region}\"."); |
|---|
| 651 | exit 32; |
|---|
| 652 | } |
|---|
| 653 | $opt->{rname} = $regions{$opt->{region}}; |
|---|
| 654 | &log("Datasource: $DATASOURCE (". $opt->{'rname'} . ')') if ($debug); |
|---|
| 655 | } |
|---|
| 656 | |
|---|
| 657 | |
|---|
| 658 | sub get_command_line_options |
|---|
| 659 | { |
|---|
| 660 | &Getopt::Long::Configure('pass_through'); |
|---|
| 661 | &GetOptions($opt, qw( |
|---|
| 662 | help |
|---|
| 663 | debug |
|---|
| 664 | output=s |
|---|
| 665 | days=i |
|---|
| 666 | offset=i |
|---|
| 667 | region=i |
|---|
| 668 | dump-cache |
|---|
| 669 | cache-file=s |
|---|
| 670 | channels_file=s |
|---|
| 671 | gaps_file=s |
|---|
| 672 | version |
|---|
| 673 | warper |
|---|
| 674 | )); |
|---|
| 675 | $debug = $opt->{debug}; |
|---|
| 676 | |
|---|
| 677 | if (@ARGV) |
|---|
| 678 | { |
|---|
| 679 | &log("\nUnknown option(s): @ARGV\n"); |
|---|
| 680 | } |
|---|
| 681 | } |
|---|
| 682 | |
|---|
| 683 | sub set_defaults |
|---|
| 684 | { |
|---|
| 685 | my $defaults = { |
|---|
| 686 | 'days' => 7, |
|---|
| 687 | 'offset' => 0, |
|---|
| 688 | 'region' => 94, |
|---|
| 689 | 'output' => &getcwd . '/output.xmltv', |
|---|
| 690 | 'cache-file' => &getcwd . '/' . $progname . '.cache', |
|---|
| 691 | 'channels_file' => &getcwd . '/channels.conf' |
|---|
| 692 | }; |
|---|
| 693 | |
|---|
| 694 | foreach (keys %$defaults) |
|---|
| 695 | { |
|---|
| 696 | unless (defined $opt->{$_}) |
|---|
| 697 | { |
|---|
| 698 | $opt->{$_} = $defaults->{$_}; |
|---|
| 699 | } |
|---|
| 700 | } |
|---|
| 701 | |
|---|
| 702 | $opt->{'days'} = 7 if ($opt->{'days'} > 7); |
|---|
| 703 | |
|---|
| 704 | &Shepherd::Common::set_defaults( |
|---|
| 705 | stats => \%stats, |
|---|
| 706 | delay => "1-5", |
|---|
| 707 | debug => $debug, |
|---|
| 708 | webwarper => $opt->{warper} |
|---|
| 709 | ); |
|---|
| 710 | |
|---|
| 711 | &Shepherd::Common::setup_ua( cookie_jar => 1 ); |
|---|
| 712 | |
|---|
| 713 | # Initialize stats |
|---|
| 714 | %stats = ( ); |
|---|
| 715 | foreach (qw( cache_hits shows )) |
|---|
| 716 | { |
|---|
| 717 | $stats{$_} = 0; |
|---|
| 718 | } |
|---|
| 719 | } |
|---|
| 720 | |
|---|
| 721 | sub read_channels_file |
|---|
| 722 | { |
|---|
| 723 | &read_config_file('channels', 'channels_file'); |
|---|
| 724 | } |
|---|
| 725 | |
|---|
| 726 | sub read_gaps_file |
|---|
| 727 | { |
|---|
| 728 | &read_config_file('gaps', 'gaps_file'); |
|---|
| 729 | if ($gaps) |
|---|
| 730 | { |
|---|
| 731 | foreach (keys %$gaps) |
|---|
| 732 | { |
|---|
| 733 | $gaps->{$channels->{$_}} = $gaps->{$_}; |
|---|
| 734 | delete $gaps->{$_}; |
|---|
| 735 | } |
|---|
| 736 | } |
|---|
| 737 | } |
|---|
| 738 | |
|---|
| 739 | sub read_config_file |
|---|
| 740 | { |
|---|
| 741 | my ($name, $arg) = @_; |
|---|
| 742 | |
|---|
| 743 | return unless ($opt->{$arg}); |
|---|
| 744 | &log("Reading $name file: $opt->{$arg}"); |
|---|
| 745 | if (-r $opt->{$arg}) |
|---|
| 746 | { |
|---|
| 747 | local (@ARGV, $/) = ($opt->{$arg}); |
|---|
| 748 | no warnings 'all'; |
|---|
| 749 | eval <>; |
|---|
| 750 | die "Can't parse $name file: $@" if ($@); |
|---|
| 751 | } |
|---|
| 752 | else |
|---|
| 753 | { |
|---|
| 754 | &log("Unable to read $name file."); |
|---|
| 755 | } |
|---|
| 756 | } |
|---|
| 757 | |
|---|
| 758 | sub log |
|---|
| 759 | { |
|---|
| 760 | &Shepherd::Common::log(@_); |
|---|
| 761 | } |
|---|
| 762 | |
|---|
| 763 | sub help |
|---|
| 764 | { |
|---|
| 765 | print q{ |
|---|
| 766 | Command-line options: |
|---|
| 767 | --help Print this message |
|---|
| 768 | --version Show current version |
|---|
| 769 | |
|---|
| 770 | --output <file> Write XML into the specified file |
|---|
| 771 | --channels_file <file> Read channel subscriptions from file |
|---|
| 772 | |
|---|
| 773 | --region <n> Grab data for region code <n> |
|---|
| 774 | --days <n> Grab <n> days of data (today being day 1) |
|---|
| 775 | --offset <n> Skip the first <n> days |
|---|
| 776 | |
|---|
| 777 | --debug Print lots of debugging output |
|---|
| 778 | }; |
|---|
| 779 | exit 0; |
|---|
| 780 | } |
|---|
