Changeset 520

Timestamp:

02/14/07 14:58:00 (6 years ago)

Author:

lincoln

Message:

update sbsnews_website: process subtitle correctly, plus Paul pointed out i botched the previous regex time update!

Files:

• grabbers/sbsnews_website (modified) (6 diffs)
• status (modified) (1 diff)

grabbers/sbsnews_website

@@ -15 +15 @@
 
 my $progname = "sbsnews_website";
-my $version = "0.07";
+my $version = "0.08";
 
 use LWP::UserAgent;
@@ -192 +192 @@
         my @seen_day_tag, my @column_is_day, my @column_maps_to_time;
         my @time_column_wrapped_over_midnight;
+        my @prev_row_prog_tag;
 
         my $seen_time_tag_column = -1;
@@ -254 +255 @@
                                         # soak up the data!
                                         $row[$td_count] = $td1->as_text();
+                                        $row[$td_count] =~ s/(^\s+|\s+$)//g;
                                         &log((sprintf "stored row data %d: '%s'", $td_count, $td1->as_text())) if $opt->{debug};
                                 } else {
@@ -276 +278 @@
                                 # parse time from $row[$tcol] ... format is "13:30-13:50"
                                 my $start_time = -1, my $stop_time = -1;
-                                if ($row[$tcol] =~ /^(\d{1,2}):(\d{1,2})?\- ?(\d{1,2}):(\d{1,2})$/) {
+                                if ($row[$tcol] =~ /^(\d{1,2}):(\d{1,2})\s*\-\s*(\d{1,2}):(\d{1,2})$/) {
                                         $start_time = ($1*60*60)+($2*60);
                                         $stop_time = ($3*60*60)+($4*60);
-                                }
-
-                                # skip sometimes-subtitle-line (but not always)
-                                next if (($start_time == -1) || ($stop_time == -1));
+                                } else {
+                                        &log("couldn't match start/stop time from '".$row[$tcol]."' in column $tcol") if $opt->{debug};
+                                }
+
+                                # sometimes-subtitle-line (but not always)
+                                # if previous line had a programme in this slot and can't match a time,
+                                # use this row data as a sub-title for the programme
+                                if (($start_time == -1) || ($stop_time == -1)) {
+                                        foreach my $col (0..$td_count) {
+                                                if ((defined $column_is_day[$col]) &&
+                                                    (defined $column_maps_to_time[$col]) &&
+                                                    ($column_maps_to_time[$col] == $tcol) &&
+                                                    (defined $prev_row_prog_tag[$col])) {
+                                                        if ($row[$col] ne "") {
+                                                                &log("added subtitle '".$row[$col]."' from col $col to previous-row prog '".
+                                                                  $tv_guide->{$prev_row_prog_tag[$col]}->{title}->[0]->[0]."'") if $opt->{debug};
+
+                                                                $tv_guide->{$prev_row_prog_tag[$col]}->{'sub-title'} = [[ $row[$col], $opt->{lang} ]];
+                                                        }
+                                                        $prev_row_prog_tag[$col] = undef;
+                                                }
+                                        }
+                                        next;
+                                }
 
                                 # apply midnight-wrapping if seen a previous wrap-over-midnight
@@ -303 +325 @@
                                             (defined $column_maps_to_time[$col]) &&
                                             ($column_maps_to_time[$col] == $tcol)) {
-                                                my $progname = $row[$col];
-                                                $progname =~ s/(^\s+|\s+$)//g;
-                                                next if ($progname eq "");
+                                                next if ($row[$col] eq "");
 
                                                 my $prog;
                                                 $prog->{channel} = $channels->{'SBS News'};
-                                                $prog->{title} = [[ $progname, $opt->{lang} ]];
+                                                $prog->{title} = [[ $row[$col], $opt->{lang} ]];
                                                 $prog->{starttime} = $start_time + $day_range[($column_is_day[$col])];
                                                 $prog->{stoptime} = $stop_time + $day_range[($column_is_day[$col])];
@@ -317 +337 @@
 
                                                 $tv_guide->{($prog->{starttime})} = $prog;
+                                                $prev_row_prog_tag[$col] = $prog->{starttime};
 
                                                 &log((sprintf "got prog col %d: start=%d (%s), stop=%d (%s), '%s', day=%d",
-                                                  $col,$start_time,$prog->{starttime},$stop_time,$prog->{stoptime},$progname,$column_is_day[$col]))
+                                                  $col,$start_time,$prog->{starttime},$stop_time,$prog->{stoptime},$row[$col],$column_is_day[$col]))
                                                   if $opt->{debug};
 

status

@@ -4 +4 @@
 grabber         abc_website         2.07
 grabber         abc2_website        2.07
-grabber         sbsnews_website     0.07
+grabber         sbsnews_website     0.08
 grabber         oztivo              0.15
 grabber         jrobbo              0.06