From f02e86ff3f0b7ac006a04f9ab506608c4a7b0af4 Mon Sep 17 00:00:00 2001
From: RayPlante <raymond.plante@nist.gov>
Date: Sun, 7 Jul 2024 16:01:04 -0400
Subject: [PATCH] jq/urlencode.jq: update for full UTF-8 support

---
 jq/urldecode.jq | 45 +++++++++++++++++++++++++++++++--------------
 1 file changed, 31 insertions(+), 14 deletions(-)

diff --git a/jq/urldecode.jq b/jq/urldecode.jq
index 6165248..f6c8e97 100644
--- a/jq/urldecode.jq
+++ b/jq/urldecode.jq
@@ -15,34 +15,51 @@ def until(condition; next):
   def u: if condition then . else (next|u) end;
   u;
 
+# interpret a string as a number in some base system
+#
+# Input:     string
+# Output:    number
+#
+def to_i(base):
+  explode
+  | reverse
+  | map(if 65 <= . and . <= 90 then . + 32  else . end)   # downcase
+  | map(if . > 96  then . - 87 else . - 48 end)  # "a" ~ 97 => 10 ~ 87
+  | reduce .[] as $c
+      # base: [power, ans]
+      ([1,0]; (.[0] * base) as $b | [$b, .[1] + (.[0] * $c)]) | .[1];
+
+def hex2utf8(shift; off):
+  [to_i(16)-off+((shift|to_i(16))*64)] | implode;
+
+def hex2utf8(shift):
+  hex2utf8(shift; 128);
+
+def hex2utf8:
+  hex2utf8("0"; 0);
+
 # replace all url-encodings (%XX) in an input string with their unencoded
-# characters.
+# characters.  This decoder should be fully UTF-8 compliant, recognizing
+# the %CX%XX pattern.  
 #
 # Input:  string
 # Output: string
 # 
 def url_decode:
-  # The helper function converts the input string written in the given
-  # "base" to an integer
-  def to_i(base):
-    explode
-    | reverse
-    | map(if 65 <= . and . <= 90 then . + 32  else . end)   # downcase
-    | map(if . > 96  then . - 87 else . - 48 end)  # "a" ~ 97 => 10 ~ 87
-    | reduce .[] as $c
-        # base: [power, ans]
-        ([1,0]; (.[0] * base) as $b | [$b, .[1] + (.[0] * $c)]) | .[1];
- 
   .  as $in
   | length as $length
   | [0, ""]  # i, answer
   | until ( .[0] >= $length;
       .[0] as $i
       |  if $in[$i:$i+1] == "%"
-         then [ $i + 3, .[1] + ([$in[$i+1:$i+3] | to_i(16)] | implode) ]
+         then
+           if $in[$i+1:$i+2] == "C" and $in[$i+3:$i+4] == "%"
+           then [ $i + 6, .[1] + ($in[$i+4:$i+6] | hex2utf8($in[$i+2:$i+3])) ]
+           else [ $i + 3, .[1] + ($in[$i+1:$i+3] | hex2utf8) ]
+           end
          else [ $i + 1, .[1] + $in[$i:$i+1] ]
          end)
-  | .[1];  # answer
+  | .[1];   # answer
 
 # replace url-encodings, including pluses (+), with their corresponding
 # characters.  This is like url_encode, except that it also replaces each