From bbb545c4ffa1ac7024555e962d4c37c84023a1c9 Mon Sep 17 00:00:00 2001
From: tagawa <codio@kyokodaniel.com>
Date: Fri, 11 Dec 2015 09:13:42 +0000
Subject: [PATCH] Unicode: Added UTF triggers

---
 lib/DDG/Goodie/Unicode.pm            |  4 ++--
 lib/DDG/Goodie/UnicodeFuzzySearch.pm |  2 +-
 t/Unicode.t                          | 14 ++++++++++++--
 t/UnicodeFuzzySearch.t               | 12 ++++++++++++
 4 files changed, 27 insertions(+), 5 deletions(-)

diff --git a/lib/DDG/Goodie/Unicode.pm b/lib/DDG/Goodie/Unicode.pm
index 6e21198dc..2a2e35b2e 100644
--- a/lib/DDG/Goodie/Unicode.pm
+++ b/lib/DDG/Goodie/Unicode.pm
@@ -10,7 +10,7 @@ use Encode qw/encode_utf8/;
 
 attribution github => ['cosimo', 'Cosimo Streppone'];
 primary_example_queries 'U+590c';
-secondary_example_queries 'unicode white smiling face';
+secondary_example_queries 'unicode white smiling face', 'utf-8 smile';
 description 'get information about a unicode character';
 code_url 'https://github.com/duckduckgo/zeroclickinfo-goodies/blob/master/lib/DDG/Goodie/Unicode.pm';
 name 'Unicode';
@@ -22,7 +22,7 @@ use constant {
     CODEPOINT_RE => qr/^ \s* (?:U \+|\\(?:u|x{(?=.*}))) (?<codepoint> [a-f0-9]{4,6})}? \s* $/xi,
     NAME_RE      => qr/^ (?<name> [A-Z][A-Z\s]+) $/xi,
     CHAR_RE      => qr/^ \s* (?<char> .) \s* $/x,
-    UNICODE_RE   => qr/^ unicode \s+ (.+) $/xi,
+    UNICODE_RE   => qr/^ (?:unicode|utf-(?:8|16|32)) \s+ (.+) $/xi,
     CODEPOINT    => 1,
     NAME         => 2,
     CHAR         => 3,
diff --git a/lib/DDG/Goodie/UnicodeFuzzySearch.pm b/lib/DDG/Goodie/UnicodeFuzzySearch.pm
index 273562163..260e553f4 100644
--- a/lib/DDG/Goodie/UnicodeFuzzySearch.pm
+++ b/lib/DDG/Goodie/UnicodeFuzzySearch.pm
@@ -5,7 +5,7 @@ use strict;
 use DDG::Goodie;
 use URI::Escape::XS;
 
-triggers startend => "unicode", "emoji";
+triggers startend => "unicode", "emoji", "utf-8", "utf-16", "utf-32";
 
 zci is_cached => 1;
 
diff --git a/t/Unicode.t b/t/Unicode.t
index bab277f8a..37e111d86 100644
--- a/t/Unicode.t
+++ b/t/Unicode.t
@@ -20,9 +20,18 @@ ddg_goodie_test(
         # Same should work with the "unicode" start trigger too
         'unicode U+263B' => test_zci("\x{263B} U+263B BLACK SMILING FACE, decimal: 9787, HTML: &#9787;, UTF-8: 0xE2 0x98 0xBB, block: Miscellaneous Symbols"),
 
-        # Lookup by name, "unicode LATIN SMALL LETTER A WITH CIRCUMFLEX")
+        # Lookup by name, "unicode LATIN SMALL LETTER A WITH CIRCUMFLEX"
         "unicode White Smiling Face" => test_zci("\x{263A} U+263A WHITE SMILING FACE, decimal: 9786, HTML: &#9786;, UTF-8: 0xE2 0x98 0xBA, block: Miscellaneous Symbols"),
 
+        # Lookup by name, "utf-8 bullet"
+        "utf-8 bullet" => test_zci("\x{2022} U+2022 BULLET, decimal: 8226, HTML: &#8226;, UTF-8: 0xE2 0x80 0xA2, block: General Punctuation"),
+
+        # Lookup by name, "utf-16 smile"
+        "utf-16 smile" => test_zci("\x{2323} U+2323 SMILE, decimal: 8995, HTML: &#8995;, UTF-8: 0xE2 0x8C 0xA3, block: Miscellaneous Technical"),
+
+        # Lookup by name, "utf-32 custard"
+        "utf-32 custard" => test_zci("\x{1F36E} U+1F36E CUSTARD, decimal: 127854, HTML: &#127854;, UTF-8: 0xF0 0x9F 0x8D 0xAE, block: Miscellaneous Symbols And Pictographs"),
+
         # Lookup by character, "unicode à"
         "unicode \x{263B}" => test_zci("\x{263B} U+263B BLACK SMILING FACE, decimal: 9787, HTML: &#9787;, UTF-8: 0xE2 0x98 0xBB, block: Miscellaneous Symbols"),
 
@@ -31,7 +40,8 @@ ddg_goodie_test(
 
         '\x{2764}' => test_zci("\x{2764} U+2764 HEAVY BLACK HEART, decimal: 10084, HTML: &#10084;, UTF-8: 0xE2 0x9D 0xA4, block: Dingbats"),
 
-        'unicode unknown' => undef
+        'unicode unknown' => undef,
+        'utf-15 bullet' => undef
 );
 
 done_testing;
diff --git a/t/UnicodeFuzzySearch.t b/t/UnicodeFuzzySearch.t
index b58d8c1b8..f778ec541 100644
--- a/t/UnicodeFuzzySearch.t
+++ b/t/UnicodeFuzzySearch.t
@@ -39,6 +39,18 @@ ROTATED HEAVY BLACK HEART BULLET: \x{2765} (U+2765)",
     'unicode 2665' =>
     test_zci("BLACK HEART SUIT: \x{2665} (U+2665)",
             html => "BLACK HEART SUIT: \x{2665} (U+2665)"),
+    # ------
+    'utf-8 2666' =>
+    test_zci("BLACK DIAMOND SUIT: \x{2666} (U+2666)",
+            html => "BLACK DIAMOND SUIT: \x{2666} (U+2666)"),
+    # ------
+    'utf-16 black diamond suit' =>
+    test_zci("BLACK DIAMOND SUIT: \x{2666} (U+2666)",
+            html => "BLACK DIAMOND SUIT: \x{2666} (U+2666)"),
+    # ------
+    'utf-32 black diamond suit' =>
+    test_zci("BLACK DIAMOND SUIT: \x{2666} (U+2666)",
+            html => "BLACK DIAMOND SUIT: \x{2666} (U+2666)"),
     # -- emoji : lower bound
     'unicode cyclone' =>
     test_zci(