From: Paul Hänsch <paul@plutz.net>
Date: Tue, 9 Jun 2020 18:12:33 +0000 (+0200)
Subject: Squashed 'cgilite/' changes from 5b013b6..b8cec22
X-Git-Url: http://git.plutz.net/?p=serve0;a=commitdiff_plain;h=b982ad7206b71b6a429a68588d71c4e201ec5296

Squashed 'cgilite/' changes from 5b013b6..b8cec22

b8cec22 more escapes for use in html-sh and HTTP headers
4a73c71 limit escaping to necessary characters, more readable output, much faster escaping functions

git-subtree-dir: cgilite
git-subtree-split: b8cec2270b5917e96f57d72c4d1ad96471cad3d3
---

diff --git a/cgilite.sh b/cgilite.sh
index 86ad280..f03f747 100755
--- a/cgilite.sh
+++ b/cgilite.sh
@@ -187,35 +187,48 @@ COOKIE(){
 }
 
 HTML(){
-  # HTML Entity Coding
-  # Prints UTF-8 string as decimal Unicode Code Points
-  # Useful for escaping user input for use in HTML text and attributes
-  { [ $# -eq 0 ] && cat || printf %s "$*"; } \
-  | hexdump -ve '/1 "%03o\n"' \
-  | while read n; do
-    case $n in
-      # bitbanging octal UTF-8 chains into singular 7 digit octal numbers
-      [01]??) printf '0000%s' $n;; # 7 bit ASCII character, nothing to do
-      2??)    printf '%s' ${n#2};; # tail fragment, append 6 bit
-      3[0123]?) printf '000%s' ${n#3};; # 2 octet (11 bit) chain start
-      34?) printf '00%s' ${n#34};; # 3 octet (16 bit) chain start
-      35?) printf '01%s' ${n#35};; # 3 octet chain start, high
-      36?) printf '%s' ${n#36};;   # 4 octet (21 bit) chain start
+  # Escape HTML cahracters
+  # Also escape [, ], and \n for use in html-sh
+  local str out
+  [ $# -eq 0 ] && str="$(cat)" || str="$*"
+  while [ "$str" ]; do
+    case $str in
+      \&*) out="${out}&amp;";;
+      \<*) out="${out}&lt;";;
+      \>*) out="${out}&gt;";;
+      \"*) out="${out}&quot;";;
+      \'*) out="${out}&#x27;";;
+      \[*) out="${out}&#x5B;";;
+      \]*) out="${out}&#x5D;";;
+      "${BR}"*) out="${out}&#x0A;";;
+      *) out="${out}${str%"${str#?}"}";;
     esac
-  done \
-  | sed -E 's;.{7};&\n;g;' \
-  | while read n; do
-    printf '&#%d;' $((0$n))
+    str="${str#?}"
   done
+  printf %s "$out"
 }
 
 URL(){
-  # Code every character in URL escape hex format
-  # except alphanumeric ascii
-
-  { [ $# -eq 0 ] && cat || printf %s "$*"; } \
-  | hexdump -v -e '/1 ",%02X"' \
-  | sed 's;,;%;g; s;%2F;/;g;'
+  # Escape pathes, so they can be used in link tags and HTTP Headers
+  local str out
+  [ $# -eq 0 ] && str="$(cat)" || str="$*"
+  while [ "$str" ]; do
+    case $str in
+      \&*) out="${out}%26";;
+      \"*) out="${out}%22";;
+      \'*) out="${out}%27";;
+      \?*) out="${out}%3F";;
+      \#*) out="${out}%23";;
+      \[*) out="${out}%5B";;
+      \]*) out="${out}%5D";;
+      \ *) out="${out}%20";;
+      "${BR}"*) out="${out}%0A";;
+      %*) out="${out}%25";;
+      *) out="${out}${str%"${str#?}"}";;
+    esac
+    str="${str#?}"
+  done
+  printf %s "$out"
 }
 
 SET_COOKIE(){