cleanup

- accented_chars changes for the future
- substitutions in the Encode Goodie are now after the ‘else’ block
- regex change to one liners
- tests

lib/DDG/Goodie/HTMLEntitiesDecode.pm

@@ -31,9 +31,9 @@ sub append_css {
 };
 
 handle remainder => sub {
-    $_ =~ s/^\s*//g; # remove front whitespace
+    $_ =~ s/^\s+|\s+$//g; # remove front and back whitespace
     $_ =~ s/^(for|of)\s+//g; # remove filler words at the start
-    $_ =~ s/\s*$//g; # remove back whitespace
+    $_ =~ s/^\s+|\s+$//g; # remove front and back whitespace that existed in between that may show up after removing the filler words
     return unless ((/^(&?#(?:[0-9]+(?!_))+;?)$/) || (/^(&(?:[a-zA-Z]+(?!_))+;?)$/) || (/^(&?#[xX](?:[0-9A-Fa-f]+(?!_))+;?)$/)); # decimal (') || text with no underscores (¢) || hex (')
                                                                                                                                 # "&" optional for all
                                                                                                                                 # ";" optional except in text type

lib/DDG/Goodie/HTMLEntitiesEncode.pm

@@ -256,39 +256,36 @@ attribution web     =>      ["http://nishanths.github.io", "Nishanth Shanmugham"
 handle remainder => sub {
 
     # General query cleanup
-    $_ =~ s/^\s*//g; # remove front whitespace
+    $_ =~ s/^\s+|\s+$//g; # remove front and back whitespace
-    $_ =~ s/\s*$//g; # remove back whitespace
     $_ =~ s/^(for|of)\s+//g; # remove filler words at the start (note: this will remove 'for' in "for euro sign", but not 'for' in "formula sign")
     $_ =~ s/(symbol|sign)//g; # remove 'symbol' and 'sign'
-    $_ =~ s/^\s*//g; # remove front whitespace again that may show up after removing the words above
+    $_ =~ s/^\s+|\s+$//g; # remove front and back whitespace that existed in between that may show up after removing the words above
-    $_ =~ s/\s*$//g; # remove back whitespace again that may show up after removing the words above
 
     # Hash-specific query cleanup for better hits
     my $hashes_query = $_;
     $hashes_query =~ s/\-/ /g; # change '-' to ' '
-    $hashes_query =~ s/"//g; # remove double quote
+    $hashes_query =~ s/"|'//g; # remove double and single quotes
-    $hashes_query =~ s/'//g; # remove single quote
 
     # Hashes lookup
     if ($hashes_query) {
         my $key;
         my $value;
         # Query is for accented character
-        if ($hashes_query =~ /^(a|A|e|E|i|I|o|O|u|U)\s*(grave|acute)$/) {
+        if ($hashes_query =~ /^([a-zA-Z])\s*(grave|acute)$/i) {
-            $hashes_query =~ s/\s*//g; # remove in-between spaces
+            $hashes_query = $1 . lc $2; # $1's capitalization matters for accented characters lookup, lc $2 allows for more freedom in queries
-            $key = $hashes_query; # capitalization matters for accented characters lookup
+            $key = $hashes_query;
             $value = $accented_chars{$key};
         # Not an accented character -- lookup the $codes hash instead
         } else { 
             $key = lc $hashes_query;
             $value = $codes{$key};
+        }
         # Try again after substitutions if there is no hit
         unless (defined $value) {
             $key =~ s/brackets/bracket/g;
             $key =~ s/quotes/quote/g;
             $value = $codes{$key};
         }
-        }
         # Make final answer
         if (defined $value) {
             my $text = make_text($value);

share/goodie/htmlentities_encode/style.css

@@ -1,12 +1,6 @@
 .zci--answer .zci--htmlentitiesencode {
     padding-top: 0.25em;
     padding-bottom: 0.25em;
-}
-.zci--answer .zci--htmlentitiesencode .line-single {
-    font-weight: 300;
-    font-size: 1.5em;
-}
-.zci--answer .zci--htmlentitiesencode .line-multiple {
     font-weight: 300;
     font-size: 1.5em;
 }

t/HTMLEntitiesEncode.t

@@ -23,7 +23,7 @@ ddg_goodie_test(
     'html entity A-acute' => test_zci("Encoded HTML Entity: Á",html => qr/Aacute/),
 
     # Variety in querying accented chars #2
-    'html entity for E grave' => test_zci("Encoded HTML Entity: È", html => qr/Egrave/),
+    'html entity for E Grave' => test_zci("Encoded HTML Entity: È", html => qr/Egrave/),
 
     # Query is a single typed-in character to encode
     'html escape &' => test_zci("Encoded HTML Entity: &", html => qr/amp/),

t/URLEncode.t

@@ -35,6 +35,9 @@ ddg_goodie_test(
 
     'www.heroku.com/{rawwr!@#$%^&*()+=__} escapeurl' => test_zci("Percent-encoded URL: www.heroku.com%2F%7Brawwr!%40%23%24%25%5E%26*()%2B%3D__%7D",
     html => qr/www.heroku.com%2F\%7Brawwr\!%40%23%24%25%5E%26\*\(\)%2B%3D__\%7D/),
+
+    'hello there escapeurl' => test_zci("Percent-encoded URL: hello%20there",
+    html => qr/hello%20there/),
 );
 
 done_testing;