From f7af5bbcd8bb8a650cd8957664a8f3759eaa1623 Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 11:56:32 -0600
Subject: [PATCH 01/10] Handle control characters

---
 lib/DDG/Goodie/Bin2Unicode.pm | 82 +++++++++++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index f742d6b8d58..f944c9bd2a6 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -14,6 +14,77 @@ triggers query => qr{^([01\s]{8,})(?:\s+(?:to\s+)?(?:unicode|text|ascii))?$};
 
 my $MAX_CODE_PT = 1114111;
 
+my %ctrl_chars = (
+    0 => 'Null character (NUL)',
+    1 => 'Start of Heading (SOH)',
+    2 => 'Start of Text (STX)',
+    3 => 'End-of-text character (ETX)',
+    4 => 'End-of-transmission character (EOT)',
+    5 => 'Enquiry character (ENQ)',
+    6 => 'Acknowledge character (ACK)',
+    7 => 'Bell character (BEL)',
+    8 => 'Backspace (BS)',
+    9 => 'Horizontal tab (HT)',
+    10 => 'Line feed (LF)',
+    11 => 'Vertical tab (VT)',
+    12 => 'Form feed (FF)',
+    13 => 'Carriage return (CR)',
+    14 => 'Shift Out (SO)',
+    15 => 'Shift In (SI)',
+    16 => 'Data Link Escape (DLE)',
+    17 => 'Device Control 1 (DC1)',
+    18 => 'Device Control 2 (DC2)',
+    19 => 'Device Control 3 (DC3)',
+    20 => 'Device Control 4 (DC4)',
+    21 => 'Negative-acknowledge character (NAK)',
+    22 => 'Synchronous Idle (SYN)',
+    23 => 'End of Transmission Block (ETB)',
+    24 => 'Cancel character (CAN)',
+    25 => 'End of Medium (EM)',
+    26 => 'Substitute character (SUB)',
+    27 => 'Escape character (ESC)',
+    28 => 'File Separator (FS)',
+    29 => 'Group Separator (GS)',
+    30 => 'Record Separator (RS)',
+    31 => 'Unit Separator (US)',
+    32 => 'Space (SP)',
+    127 => 'Delete (DEL)',
+    128 => 'Padding Character (PAD)',
+    129 => 'High Octet Preset (HOP)',
+    130 => 'Break Permitted Here (BPH)',
+    131 => 'No Break Here (NBH)',
+    132 => 'Index (IND)',
+    133 => 'Next Line (NEL)',
+    134 => 'Start of Selected Area (SSA)',
+    135 => 'End of Selected Area (ESA)',
+    136 => 'Character Tabulation Set (HTS)',
+    137 => 'Character Tabulation with Justification (HTJ)',
+    138 => 'Line Tabulation Set (VTS)',
+    139 => 'Partial Line Forward (PLD)',
+    140 => 'Partial Line Backward (PLU)',
+    141 => 'Reverse Line Feed (RI)',
+    142 => 'Single-Shift Two (SS2)',
+    143 => 'Single-Shift Three (SS3)',
+    144 => 'Device Control String (DCS)',
+    145 => 'Private Use 1 (PU1)',
+    146 => 'Private Use 2 (PU2)',
+    147 => 'Set Transmit State (STS)',
+    148 => 'Cancel character (CCH)',
+    149 => 'Message Waiting (MW)',
+    150 => 'Start of Protected Area (SPA)',
+    151 => 'End of Protected Area (EPA)',
+    152 => 'Start of String (SOS)',
+    153 => 'Single Graphic Character Introducer (SGCI)',
+    154 => 'Single Character Intro Introducer (SCI)',
+    155 => 'Control Sequence Introducer (CSI)',
+    156 => 'String Terminator (ST)',
+    157 => 'Operating System Command (OSC)',
+    158 => 'Private Message (PM)',
+    159 => 'Application Program Command (APC)'
+);
+
+my $zaahirs_hideout = '0' x 48;
+
 handle matches => sub {
     my $q = $_; # orginal query
     my $bin_string = shift @_; # captured binary string
@@ -27,6 +98,17 @@ handle matches => sub {
             return if length($b) % 8;
             # Overflow/non-portable warnings expected
             my $i = oct("0b$b");
+            if($bin_string eq $zaahirs_hideout){
+                $str = q{Congratulations, you've discovered Zaahir's hideout!};
+                last;
+            }
+            elsif(exists $ctrl_chars{$i}){
+                if(@bins == 1){
+                    $str = "Control character: $ctrl_chars{$i}";
+                    last;
+                }
+                else{ return } # only allow a single binary to unicode char
+            }
             # Assume ascii if out of range or explicitly requested.
             # This will work for characters all in the same string
             # but will not print the right non-ascii characters *if*

From 0a602ad68f71c213ef22571780bff8603e6d78bc Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 12:15:00 -0600
Subject: [PATCH 02/10] Alter logic.  Translate control characters if more than
 one and check for word characters in end string.  This allow for spaces as we
 used to do

---
 lib/DDG/Goodie/Bin2Unicode.pm | 21 +++++++++++----------
 1 file changed, 11 insertions(+), 10 deletions(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index f944c9bd2a6..e87f34dd19b 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -89,26 +89,23 @@ handle matches => sub {
     my $q = $_; # orginal query
     my $bin_string = shift @_; # captured binary string
 
+    my $str;
+    if($bin_string eq $zaahirs_hideout){
+        $str = q{Congratulations, you've discovered Zaahir's hideout!};
+        goto DONE;
+    }
     my $want_ascii = $q =~ /\bascii\b/;
 
     my @bins = $bin_string =~ /([01]+|\s+)/g;
-    my $str;
     for my $b (@bins){
         if($b =~ /^[01]+$/){
             return if length($b) % 8;
             # Overflow/non-portable warnings expected
             my $i = oct("0b$b");
-            if($bin_string eq $zaahirs_hideout){
-                $str = q{Congratulations, you've discovered Zaahir's hideout!};
+            if((exists $ctrl_chars{$i}) && (@bins == 1)){
+                $str = "Control character: $ctrl_chars{$i}";
                 last;
             }
-            elsif(exists $ctrl_chars{$i}){
-                if(@bins == 1){
-                    $str = "Control character: $ctrl_chars{$i}";
-                    last;
-                }
-                else{ return } # only allow a single binary to unicode char
-            }
             # Assume ascii if out of range or explicitly requested.
             # This will work for characters all in the same string
             # but will not print the right non-ascii characters *if*
@@ -122,6 +119,10 @@ handle matches => sub {
         }
     }
 
+    DONE:
+
+    return unless $str =~ /\w+/;
+
     return "Binary '$bin_string' converted to " . $want_ascii ? 'ascii' : 'unicode' . " is '$str'",
         structured_answer => {
             id => 'bin2unicode',

From 45acabc018514302f11918ec3410bef7b7aa696d Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 12:39:17 -0600
Subject: [PATCH 03/10] Check for all control/space characters

---
 lib/DDG/Goodie/Bin2Unicode.pm | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index e87f34dd19b..01e9d2097ec 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -121,7 +121,7 @@ handle matches => sub {
 
     DONE:
 
-    return unless $str =~ /\w+/;
+    return if $str =~ /^[\x{0}-\x{20}\x{7F}-\x{9F}]+$/; # all control/space
 
     return "Binary '$bin_string' converted to " . $want_ascii ? 'ascii' : 'unicode' . " is '$str'",
         structured_answer => {

From 07f2d3a08e6d7cecfd8d7e74ca77504cbab119c3 Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 12:41:21 -0600
Subject: [PATCH 04/10] Add link

---
 lib/DDG/Goodie/Bin2Unicode.pm | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index 01e9d2097ec..5adb41f71ae 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -121,7 +121,8 @@ handle matches => sub {
 
     DONE:
 
-    return if $str =~ /^[\x{0}-\x{20}\x{7F}-\x{9F}]+$/; # all control/space
+    # return if all control/space (https://en.wikipedia.org/wiki/List_of_Unicode_characters#Control_codes)
+    return if $str =~ /^[\x{0}-\x{20}\x{7F}-\x{9F}]+$/;
 
     return "Binary '$bin_string' converted to " . $want_ascii ? 'ascii' : 'unicode' . " is '$str'",
         structured_answer => {

From 77f4032b5512810b0046ebc8dcb5af615f76a890 Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 12:58:50 -0600
Subject: [PATCH 05/10] Update regex to be easier to understand

---
 lib/DDG/Goodie/Bin2Unicode.pm | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index 5adb41f71ae..794fa42b389 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -122,7 +122,7 @@ handle matches => sub {
     DONE:
 
     # return if all control/space (https://en.wikipedia.org/wiki/List_of_Unicode_characters#Control_codes)
-    return if $str =~ /^[\x{0}-\x{20}\x{7F}-\x{9F}]+$/;
+    return if $str =~ /^[\p{Control}]\s]+$/;
 
     return "Binary '$bin_string' converted to " . $want_ascii ? 'ascii' : 'unicode' . " is '$str'",
         structured_answer => {

From e479164c70370e9befd2948b4b7cf778c6973083 Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 13:08:44 -0600
Subject: [PATCH 06/10] Only need space, not class of spaces

---
 lib/DDG/Goodie/Bin2Unicode.pm | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index 794fa42b389..c8c51457dcd 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -122,7 +122,7 @@ handle matches => sub {
     DONE:
 
     # return if all control/space (https://en.wikipedia.org/wiki/List_of_Unicode_characters#Control_codes)
-    return if $str =~ /^[\p{Control}]\s]+$/;
+    return if $str =~ /^[\p{Control}] ]+$/;
 
     return "Binary '$bin_string' converted to " . $want_ascii ? 'ascii' : 'unicode' . " is '$str'",
         structured_answer => {

From 6a73236464b01083f4f2d8b1afed951a6c4abaf1 Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 13:20:50 -0600
Subject: [PATCH 07/10] space/delete not strictly control characters

---
 lib/DDG/Goodie/Bin2Unicode.pm | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index c8c51457dcd..07c0ad5c784 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -103,7 +103,8 @@ handle matches => sub {
             # Overflow/non-portable warnings expected
             my $i = oct("0b$b");
             if((exists $ctrl_chars{$i}) && (@bins == 1)){
-                $str = "Control character: $ctrl_chars{$i}";
+                $str = $ctrl_chars{$i};
+				$str = "Control character: $str" unless ($i == 32) || ($i == 127);
                 last;
             }
             # Assume ascii if out of range or explicitly requested.

From e3fa7ad6e160ab40660635ab948a89c0ebbfa19c Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 15:06:24 -0600
Subject: [PATCH 08/10] Remove extra bracket from regex Add test for control
 characters

---
 lib/DDG/Goodie/Bin2Unicode.pm | 2 +-
 t/Bin2Unicode.t               | 4 +++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index 07c0ad5c784..130193a724f 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -123,7 +123,7 @@ handle matches => sub {
     DONE:
 
     # return if all control/space (https://en.wikipedia.org/wiki/List_of_Unicode_characters#Control_codes)
-    return if $str =~ /^[\p{Control}] ]+$/;
+    return if $str =~ /^[\p{Control} ]+$/;
 
     return "Binary '$bin_string' converted to " . $want_ascii ? 'ascii' : 'unicode' . " is '$str'",
         structured_answer => {
diff --git a/t/Bin2Unicode.t b/t/Bin2Unicode.t
index c04f9f9a8d6..5975a801585 100644
--- a/t/Bin2Unicode.t
+++ b/t/Bin2Unicode.t
@@ -59,7 +59,9 @@ ddg_goodie_test(
         'Muchas Gracias CompasMð',
         1)),
     '010101' => undef,
-    '201 to text' => undef
+    '201 to text' => undef,
+    '00000000000000001' => undef,
+    '00000001 00000010' => undef
 );
 
 done_testing;

From 77e72484421750b2bdb0c7e297318a76bf7ced16 Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 15:29:03 -0600
Subject: [PATCH 09/10] Test for all multiple control character sequences
 failing test for single control character conversions returning

---
 t/Bin2Unicode.t | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/t/Bin2Unicode.t b/t/Bin2Unicode.t
index 5975a801585..76ba4815568 100644
--- a/t/Bin2Unicode.t
+++ b/t/Bin2Unicode.t
@@ -29,6 +29,12 @@ sub gen_struc_ans {
         };
 }
 
+my %ctrl_tests;
+for (1..32, 127..159){
+    my $b = sprintf '%08b', $_;
+	$ctrl_tests{join ' ', $b, $b} = undef;
+}
+
 ddg_goodie_test(
     [qw( DDG::Goodie::Bin2Unicode )],
     '0110100001100101011011000110110001101111 to text' => test_zci(gen_struc_ans(
@@ -58,10 +64,19 @@ ddg_goodie_test(
         '0100110101110101011000110110100001100001011100110010000001000111011100100110000101100011011010010110000101110011001000000100001101101111011011010111000001100001011100110100110111110000',
         'Muchas Gracias CompasMð',
         1)),
+    '00000000' => test_zci(gen_struc_ans(
+        '00000000',
+        '00000000',
+        'Control character: Null character (NUL)',
+        0)),
+    '000000000000000000100000' => test_zci(gen_struc_ans(
+        '000000000000000000100000',
+        '000000000000000000100000',
+        'Space (SP)',
+        0)),
     '010101' => undef,
     '201 to text' => undef,
-    '00000000000000001' => undef,
-    '00000001 00000010' => undef
+    %ctrl_tests
 );
 
 done_testing;

From f21a1dd1f9e65dd834d605b94b46a83006a921d0 Mon Sep 17 00:00:00 2001
From: Zach Thompson <zach@duckduckgo.com>
Date: Sun, 20 Mar 2016 15:59:33 -0600
Subject: [PATCH 10/10] Put label below regex

---
 lib/DDG/Goodie/Bin2Unicode.pm | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/lib/DDG/Goodie/Bin2Unicode.pm b/lib/DDG/Goodie/Bin2Unicode.pm
index 130193a724f..914c827c7e5 100644
--- a/lib/DDG/Goodie/Bin2Unicode.pm
+++ b/lib/DDG/Goodie/Bin2Unicode.pm
@@ -120,11 +120,10 @@ handle matches => sub {
         }
     }
 
-    DONE:
-
     # return if all control/space (https://en.wikipedia.org/wiki/List_of_Unicode_characters#Control_codes)
     return if $str =~ /^[\p{Control} ]+$/;
 
+    DONE:
     return "Binary '$bin_string' converted to " . $want_ascii ? 'ascii' : 'unicode' . " is '$str'",
         structured_answer => {
             id => 'bin2unicode',