Fix Sodexo parsing.

sodexo.pl

@@ -1,21 +1,30 @@
 use vars qw(@day_names);
+use POSIX qw(strftime);
+
 
 my @restaurant_info = (
   #[ "(TaY) Sodexo Linna", "http://www.sodexo.fi/fi-FI/linna/lounas/", "right" ],
   #[ "(TTY) Sodexo Erkkeri", "http://www.sodexo.fi/fi-FI/erkkeri/lounas/", "left" ]
-  [ "(TaY) Sodexo Linna", "http://www.sodexo.fi/linna", "right" ],
-  [ "(TTY) Sodexo Erkkeri", "http://www.sodexo.fi/erkkeri", "left" ]
+  [ "(TaY) Sodexo Linna", "http://www.sodexo.fi/linna", "right", 92],
+  [ "(TTY) Sodexo Erkkeri", "http://www.sodexo.fi/erkkeri", "left", 100]
 );
 
 my ($cur_text, $cur_title, $parse_func, $day_id, $week);
 my (@cur_day_foods, @week_foods);
 
+sub sodexo_finish_food {
+  chomp $cur_food;
+  push @cur_day_foods, utf8_to_8859($cur_food) if ($cur_food ne "");
+  $cur_food = "";
+}
+
 sub sodexo_finish_day {
   push @week_foods, [@cur_day_foods];
   @cur_day_foods = ();
   $day_id = $day_id + 1;
 }
 
+
 sub utf8_to_8859 {
   $_ = shift;
 
@@ -26,163 +35,54 @@ sub utf8_to_8859 {
   return $_;
 }
 
-sub sodexo_parse_finish {
-}
-
-sub sodexo_parse_after_food {
-  my $token = shift;
-
-  if ($token->[0] eq 'E' && $token->[1] eq 'tbody') {
-    sodexo_finish_day();
-    $parse_func = \&sodexo_parse_finish;
-  } elsif ($token->[0] eq 'S' && $token->[1] eq 'td') {
-    $parse_func = \&sodexo_parse_to_food;
-    sodexo_parse_to_food($token);
-  } elsif ($token->[0] eq 'S' && $token->[1] eq 'th') {
-    sodexo_finish_day();
-    $parse_func = \&sodexo_parse_to_food;
-  }
-}
-
-sub sodexo_finish_food {
-  if ($cur_text ne "") {
-    push @cur_day_foods, "$cur_title ($cur_text)";
-  } else {
-    push @cur_day_foods, "$cur_title";
-  }
-  $parse_func = \&sodexo_parse_after_food;
-}
-
-sub sodexo_parse_allergy {
-  my $token = shift;
-
-  if ($token->[0] eq 'T' && !($token->[1] =~ /\s*\/\s*/)) {
-    my $text = $token->[1];
-    $text =~ s/^\s+//;
-    $text =~ s/\s+$//;
-    $cur_text .= ", " if $cur_text ne "";
-    $cur_text .= $text;
-  } elsif ($token->[0] eq 'E' && $token->[1] eq 'td') {
-    sodexo_finish_food();
-  }
-}
-
-sub sodexo_parse_to_allergy {
-  my $token = shift;
-
-  if ($token->[0] eq 'S' && $token->[1] eq 'td') {
-    my %attrs = %{$token->[2]};
-    if ($attrs{'class'} eq 'food-properties') {
-      $parse_func = \&sodexo_parse_allergy;
-    }
-  }
-}
-
-sub sodexo_parse_food {
-  my $token = shift;
-
-  if ($token->[0] eq 'T' && $token->[1] =~ /\S/) {
-     $cur_title .= utf8_to_8859($token->[1]);
-     $parse_func = \&sodexo_parse_to_allergy;
-  }
-}
-
-sub sodexo_parse_type {
-  my $token = shift;
-
-  if ($token->[0] eq 'T' && $token->[1] =~ /\S/) {
-    if ($token->[1] !~ /^ $/) {
-        $cur_title = utf8_to_8859($token->[1]) . ": ";
-    }
-    $parse_func = \&sodexo_parse_food; 
-  }
-}
-
-sub sodexo_parse_to_food {
-  my $token = shift;
-  # Allow list to skip days, ie. Helatorstai
-  if ($token->[0] eq 'T') {
-    for (my $day = $day_id; $day <= 6; ++$day) {
-      if ($token->[1] eq $day_names[$day]) {
-        while ($day > $day_id) {
-          sodexo_finish_day();
-          $day = $day - 1;
-        }
-        break;
-      }
-    }
-  }
-  if ($token->[0] eq 'S' && $token->[1] eq "td") {
-    my %attrs = %{$token->[2]};
-    if ($attrs{'class'} eq "food-type") {
-      $cur_title = '';
-      $cur_text = '';
-      $parse_func = \&sodexo_parse_type;
-    }
-    if ($attrs{'class'} eq "food-desc") {
-      $cur_title = '';
-      $cur_text = '';
-      $parse_func = \&sodexo_parse_food;
-    }
-  }
-}
-
-sub sodexo_parse_week {
-  my $token = shift;
-  if ($token->[0] eq 'T' && $token->[1] =~ /Viikko (\d+)/) {
-    $week = $1;
-    $parse_func = \&sodexo_parse_to_food;
-  }
-}
-
-sub sodexo_parse_to_week {
-  my $token = shift;
-  if ($token->[0] eq 'S' && $token->[1] eq 'div') {
-    my %attrs = %{$token->[2]};
-    if ($attrs{'class'} eq "list-date") {
-      $parse_func = \&sodexo_parse_week;
-    }
-  }
-}
-
-sub parse_sodexo {
-  my ($fname, $info_ref) = @_;
-  my $title = @{$info_ref}[0];
-  my $url = @{$info_ref}[1];
-  my $align = @{$info_ref}[2];
-  my $p = HTML::TokeParser->new($fname) or die("Can't open file $fname");
-
-  $week = "";
-  $day_id = 0;
-  @cur_day_foods = ();
-  @week_foods = ();
-
-  my $week_day = `date +%w`;
-  for (my $i = 1; $i < $week_day; ++$i) {
-    sodexo_finish_day();
-  }
-
-  $parse_func = \&sodexo_parse_to_week;
-  while (my $token = $p->get_token) {
-    &$parse_func($token);
-  }
-  return [ $title, "", $week, [ @week_foods ], [ $title, $url, "M", $align ] ];
-}
-
 sub get_sodexo_restaurants {
   my $use_old = shift;
   my $count = 0;
-  my @restaurants = ();
+  # Loops restraurants
   foreach my $i (@restaurant_info) {
     my @info = @{$i};
-    my $temp_fname = "sodexo$count.temp.html";
-    my $url = $info[1];
-    if (!-f $temp_fname || !$use_old) {
-      system("wget -q --timeout=10 -O $temp_fname.tmp '$url' && mv $temp_fname.tmp $temp_fname");
-      return undef if (!-f $temp_fname);
+    my $kitchen = $info[3];
+    $title = $info[0];
+    $cur_title = $title;
+    $open_hours = "";
+    @week_foods = ();
+    my $week = `date +%V`;
+    chomp($week);
+    # Loop weekdays
+    for (my $weekday = 1; $weekday < 7; $weekday++) {
+
+      # Get current unix timestamp and week
+      $s = strftime "%s", localtime;
+      $v = strftime "%w", localtime;
+      # Calculate current weekday 
+      $s = $s - ($v - 2 + $weekday)  * 84600;
+      $timestr = strftime "%Y/%m/%d", localtime($s);
+
+      my $url = "http://www.sodexo.fi/ruokalistat/output/daily_json/$kitchen/$timestr/fi";
+      my $temp_fname = "sodexo$count-$weekday.temp.js";  
+      if (!-f $temp_fname || !$use_old) {
+        system("wget -q --timeout=10 -O $temp_fname.tmp \"$url\" && mv $temp_fname.tmp $temp_fname");
+      }
+      if (-f $temp_fname) {
+        open(FILE, $temp_fname);
+        my $json = do { local $/; <FILE> };
+        close(FILE);
+        # the file is encapsulated in ({"d": json}); so we have to double parse it
+        my $data = from_json($json);
+        my $courses = $data->{'courses'};
+        # loop different meals
+        foreach my $course_info (@$courses) {
+          if ($course_info->{'category'} ne 'Aamupala') {
+            $cur_food = $course_info->{'title_fi'};
+            $cur_food .= " (" . $course_info->{'properties'} . ")" if $course_info->{'properties'};
+          }
+          sodexo_finish_food();
+        }
+        sodexo_finish_day();
+      }
     }
-    push @restaurants, parse_sodexo($temp_fname, \@info);
-    ++$count;
+    push @restaurants, [ $title, $open_hours, $week, [ @week_foods ], \@info ];
+    $count++;
   }
   return @restaurants;
 }