Rewrite juvenes parser for the new site.

juvenes.pl

@@ -1,4 +1,5 @@
 use vars qw(@day_names);
+use JSON;
 
 my $pinni_title = "(TaY) Café Pinni";
 my $bio_title = "(TAYS) Bio";
@@ -6,64 +7,26 @@ my $kliininen_title = "(TAYS) Arvo";
 my $kliininen_fusion_title = "(TAYS) Arvo Fusion Kitchen";
 my $zip_salaattibaari_title = "(TTY) Zip Salaattibaari";
 my @restaurant_info = (
-  [ "(TaY) Yliopiston Ravintola", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TaY__Yliopiston_Ravintola", "M", "left" ],
-  [ "(TaY) Yliopiston Ravintola / Salaattibaari", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TaY__Yliopiston_Ravintola/Salaattibaari", "", "left" ],
-  [ "(TaY) Fusion Kitchen", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TaY__Yliopiston_Ravintola/Fusion_Kitchen", "", "left" ],
-  [ $pinni_title, "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TaY__Caf__Pinni", "M", "middle" ],
-  [ $bio_title, "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TaY_Kauppi__Medica_Bio", "M", "left" ],
-  [ $kliininen_title, "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TaY_Kauppi__Medica_Arvo", "M", "left" ],
-  [ $kliininen_fusion_title, "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TaY_Kauppi__Medica_Arvo/Fusion_Kitchen", "M", "left" ],
-  [ "(TTY) Newton", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TTY__Newton", "", "left" ],
-  [ "(TTY) Zip", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TTY__Zip", "", "right" ],
-  [ "(TTY) Edison", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TTY__Edison", "", "middle" ],
-# There is no Salaattibaari anymore?
-#  [ $zip_salaattibaari_title, "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TTY__Zip/Salaattibaari", "", "right" ],
-  [ "(TTY) Pastabaari", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TTY__Caf____Fast_Voltti/Pastabaari", "", "middle" ],
-  [ "(TTY) Fast Voltti", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TTY__Caf____Fast_Voltti", "", "middle" ],
-  [ "(TTY) Fusion Kitchen", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TTY__Newton/Fusion_Kitchen", "", "left" ],
-  [ "(TAMK) Dot", "http://www.juvenes.fi/Suomeksi/Ravintolat_ja_kahvilat/Opiskelijaravintolat/_TAMK__Dot__Ziberia_", "", "middle" ]
+  [ "(TaY) Yliopiston Ravintola", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/tayp%C3%A4%C3%A4kampus/yliopistonravintola.aspx", "M", "left", 13, 60 ],
+  [ "(TaY) Yliopiston Ravintola / Salaattibaari", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/tayp%C3%A4%C3%A4kampus/yliopistonravintola/salaattibaari.aspx", "", "left", 13, 5 ],
+  [ "(TaY) Fusion Kitchen", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/tayp%C3%A4%C3%A4kampus/yliopistonravintola/fusionkitchen.aspx", "", "left", 13, 3 ],
+  [ $pinni_title, "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/tayp%C3%A4%C3%A4kampus/caf%C3%A9lunchpinni.aspx", "M", "middle", 31, 60 ],
+  [ $bio_title, "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/taykaupinkampus/medicabio.aspx", "M", "left", 5, 60 ],
+  [ $kliininen_title, "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/taykaupinkampus/medicaarvo.aspx", "M", "left", 27, 60 ],
+  [ $kliininen_fusion_title, "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/taykaupinkampus/medicaarvo/fusionkitchen.aspx", "M", "left", 27, 3 ],
+  [ "(TTY) Newton", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/ttykampus/newton.aspx", "", "left", 6, 60],
+  [ "(TTY) Zip", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/ttykampus/zip.aspx", "", "right", 12, 60 ],
+  [ "(TTY) Edison", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/ttykampus/edison.aspx", "", "middle", 2, 60 ],
+  [ "(TTY) Pastabaari", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/ttykampus/caf%C3%A9pastabaarivoltti.aspx", "", "middle", 26, 11 ],
+  [ "(TTY) Fast Voltti", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/ttykampus/caf%C3%A9fastvoltti.aspx", "", "middle", 2, 4 ],
+  [ "(TTY) Fusion Kitchen", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/ttykampus/newton/fusionkitchen.aspx", "", "left", 6, 3 ],
+  [ "(TAMK) Dot", "http://www.juvenes.fi/fi-fi/ravintolatjakahvilat/opiskelijaravintolat/tamkdot.aspx", "", "middle", 15, 1 ]
 );
 
 my @restaurants;
-my ($parse_func, $week, $open_hours, $day_id, $cur_title);
+my ($open_hours, $day_id, $cur_title);
 my ($cur_food, @cur_day_foods, @week_foods);
 
-sub parse_to_eof {
-}
-
-sub parse_skip_to_end_of_div {
-  my $token = shift;
-  
-  if ($token->[0] eq 'E' && $token->[1] eq 'div') {
-    $parse_func = \&parse_more_food;
-  }
-}
-
-sub parse_open_hours_begin {
-  my $token = shift;
-
-  if ($token->[0] eq 'T') {
-    if ($token->[1] =~ /^Aukiolo/) {
-      $parse_func = \&parse_open_hours_end;
-    }
-  }
-}
-
-sub parse_open_hours_end {
-  my $token = shift;
-
-  if ($token->[0] eq 'T') {
-    my $text = $token->[1];
-    if ($text eq 'Erityisruokavaliot') {
-      $parse_func = \&parse_to_eof;
-    } else {
-      $text =~ s/\n//g;
-      $text =~ s/ +$//;
-      $open_hours .= "$text\n" if ($text ne "");
-    }
-  }
-}
-
 sub finish_food {
   chomp $cur_food;
   if ($cur_food =~ /Liha paniini.*tai Kasvis paniini/i && $cur_title eq $pinni_title) {
@@ -80,103 +43,6 @@ sub finish_day {
   $day_id = $day_id + 1;
 }
 
-sub parse_more_food {
-  my $token = shift;
-  
-  if ($token->[0] eq 'S') {
-    my %attrs = %{$token->[2]};
-    if ($token->[1] eq 'div') {
-      if ($attrs{'style'} =~ /display: *none/) {
-	# infobox, skip
-	$parse_func = \&parse_skip_to_end_of_div;
-      } elsif ($attrs{'class'} eq 'Column') {
-	# end of food
-	finish_food();
-	finish_day();
-	$parse_func = \&parse_open_hours_begin;
-      }
-    } elsif ($token->[1] eq 'br') {
-      if ($br_is_new_food) {
-	finish_food();
-      } else {
-	$cur_food .= "\n" if ($cur_food ne "" && substr($cur_food, -1) ne "\n");
-      }
-    }
-  } elsif ($token->[0] eq 'T') {
-    my $text = $token->[1];
-
-    for (my $day = $day_id+1; $day <= 6; $day++) {
-      if ($text eq $day_names[$day]) {
-        while ($day >= $day_id) {
-          finish_food();
-          finish_day();
-          $day = $day - 1;
-        }
-        return;
-      }
-    }
-    if ($text eq "&nbsp;") {
-      # next food
-      finish_food();
-    } else {
-      $text =~ tr/\r\n\t/   /;
-      $text =~ s/ +/ /g;
-      $text =~ s/^ +//;
-      $text =~ s/^\.+//;
-      $text =~ s/ +$//;
-      $text =~ s/sisältää ([^, \)]+)/sis.$1/ig;
-      $cur_food .= $text;
-    }
-  }
-}
-
-sub parse_monday {
-  my $token = shift;
-  
-  for (my $day = 0; $day <= 6; $day++) {
-    if ($token->[0] eq 'T' and $token->[1] eq $day_names[$day]) {
-      while ($day > 0) {
-        finish_day();
-        $day = $day - 1;
-      }
-      $parse_func = \&parse_more_food;
-      break;
-    }
-  }
-}
-
-sub parse_week {
-  my $token = shift;
-  
-  if ($token->[0] eq 'T') {
-    if ($token->[1] =~ /Viikko: (\d+)/) {
-      $week = $1;
-      $parse_func = \&parse_monday;
-    }
-  }
-}
-
-sub parse_juvenes {
-  my ($fname, $info_ref) = @_;
-  my $p = HTML::TokeParser->new($fname) or die("Can't open file $fname");
-
-  my $title = @{$info_ref}[0];
-  $week = "";
-  $open_hours = "";
-  $day_id = 0;
-  $cur_food = "";
-  @cur_day_foods = ();
-  @week_foods = ();
-  $br_is_new_food = $title eq $zip_salaattibaari_title;
-  $cur_title = $title;
-  
-  $parse_func = \&parse_week;
-  while (my $token = $p->get_token) {
-    &$parse_func($token);
-  }
-  push @restaurants, [ $title, $open_hours, $week, [ @week_foods ], $info_ref ];
-}
-
 sub can_merge_bio_kliininen {
   my $day = shift;
 
@@ -215,16 +81,49 @@ sub try_merge_bio_kliininen {
 sub get_juvenes_restaurants {
   my $use_old = shift;
   my $count = 0;
+  # Loops restraurants
   foreach my $i (@restaurant_info) {
     my @info = @{$i};
-    my $temp_fname = "juvenes$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");
-    }
-    if (-f $temp_fname) {
-      parse_juvenes($temp_fname, \@info);
+    my $kitchen = $info[4];
+    my $menutype = $info[5];
+    $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++) {
+      my $temp_fname = "juvenes$count-$weekday.temp.js";
+      my $url = "http://www.juvenes.fi/DesktopModules/Talents.LunchMenu/LunchMenuServices.asmx/GetMenuByWeekday?KitchenId=$kitchen&MenuTypeId=$menutype&Week=$week&Weekday=$weekday&lang='fi'&format=json";
+      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 $jsonp = do { local $/; <FILE> };
+        close(FILE);
+        # the file is encapsulated in ({"d": json}); so we have to double parse it
+        my $data = from_json(substr($jsonp, 1, -2));
+        unless ($data->{'d'} eq 'null') {
+          $data = from_json($data->{'d'});
+          my $mealoptions = $data->{'MealOptions'};
+          # loop different meals
+          foreach my $meal_info (@$mealoptions) {
+            my $menuitems = $meal_info->{'MenuItems'};
+            # loops different foods in a meal
+            foreach my $food_info (@$menuitems) {
+              $cur_food .= "\n" if $cur_food ne ""; 
+              $cur_food .= $food_info->{'Name'};
+              $cur_food .= " (" . $food_info->{'Diets'} . ")" if $food_info->{'Diets'};
+            }
+            finish_food();
+          }
+        }
+        finish_day();
+      }
     }
+    push @restaurants, [ $title, $open_hours, $week, [ @week_foods ], \@info ];
     $count++;
   }
   return @restaurants;