much faster hex decode function

[cgilite] / cgilite.sh

diff --git a/cgilite.sh b/cgilite.sh
index 2e538324c0a8b5ec45f6fb2235f254fcf359db22..7f828ddd6c8e2234c7cb3e2625505b5796e8eb77 100755 (executable)
--- a/cgilite.sh
+++ b/cgilite.sh
@@ -1,6 +1,6 @@
 #!/bin/sh
 
-# Copyright 2017 - 2020 Paul Hänsch
+# Copyright 2017 - 2021 Paul Hänsch
 #
 # This is CGIlite.
 # A collection of posix shell functions for writing CGI scripts.
@@ -18,6 +18,9 @@
 # You should have received a copy of the GNU Affero General Public License
 # along with CGIlite.  If not, see <http://www.gnu.org/licenses/>. 
 
+[ -n "$include_cgilite" ] && return 0
+# guard set after webserver part
+
 # ksh and zsh workaround
 # set -o posix # ksh, not portable
 setopt -o OCTAL_ZEROES 2>&-
@@ -27,10 +30,11 @@ BR='
 '
 cgilite_timeout=2
 
-debug(){ [ $# -gt 0 ] && printf '%s\n' "$@" >&2 || tee -a /dev/stderr; }
-
 PATH(){ 
   local str seg out
+  # normalize path
+  # read from stdin if no arguments are provided
+
   [ $# -eq 0 ] && str="$(cat)" || str="$*"
   while [ "$str" ]; do
     seg=${str%%/*}; str="${str#*/}"
@@ -44,18 +48,58 @@ PATH(){
   [ "${str}" -a "${out}" ] && printf %s "$out" || printf %s/ "${out%/}"
 }
 
-HEX_DECODE='
-  s;\\;\\\\;g; :HEXDECODE_X; s;%([^0-9A-F]);\\045\1;g; tHEXDECODE_X;
-  # Hexadecimal { %00 - %FF } will be transformed to octal { \000 - \377 } for posix printf
-  s;%[0123].;&\\0;g; s;%[4567].;&\\1;g; s;%[89AB].;&\\2;g; s;%[CDEF].;&\\3;g;
-  s;%[048C][0-7]\\.;&0;g; s;%[048C][89A-F]\\.;&1;g; s;%[159D][0-7]\\.;&2;g; s;%[159D][89A-F]\\.;&3;g;
-  s;%[26AE][0-7]\\.;&4;g; s;%[26AE][89A-F]\\.;&5;g; s;%[37BF][0-7]\\.;&6;g; s;%[37BF][89A-F]\\.;&7;g;
-  s;%.[08](\\..);\10;g; s;%.[19](\\..);\11;g; s;%.[2A](\\..);\12;g; s;%.[3B](\\..);\13;g;
-  s;%.[4C](\\..);\14;g; s;%.[5D](\\..);\15;g; s;%.[6E](\\..);\16;g; s;%.[7F](\\..);\17;g;
-'
-
 HEX_DECODE(){
-  printf -- "$(printf %s "$1" |sed -E "$HEX_DECODE")"
+  local pfx="$1" in="$2" out
+  # Print out Data encoded as Hex
+  #
+  # Arguments:
+  # pfx - required, prefix for a hex tupel, e.g. "\x", "%" "\", may be empty
+  # in - required, string to be decoded
+  #
+  # anything that does not constitute a tupel of valid Hex numerals
+  # will be copied to the output literally
+
+  while [ "$in" ]; do
+    case $in in
+      "$pfx"[0-9a-fA-F][0-9a-fA-F]*) in="${in#${pfx}}";;
+      \\*) in="${in#?}"; out="${out}\\\\"; continue;;
+       %*) in="${in#?}"; out="${out}%%";  continue;;
+        *) out="${out}${in%"${in#?}"}"; in="${in#?}"; continue;;
+    esac;
+
+    # Hex escaes for printf (e.g. \x41) are not portable 
+    # The portable way for Hex output is transforming Hex to Octal
+    # (e.g. \x41 = \101)
+    case $in in
+        [0123]?*) out="${out}\\0";;
+        [4567]?*) out="${out}\\1";;
+      [89aAbB]?*) out="${out}\\2";;
+      [c-fC-F]?*) out="${out}\\3";;
+    esac
+    case $in in
+            [048cC][0-7]*) out="${out}0";;
+       [048cC][89a-fA-F]*) out="${out}1";;
+            [159dD][0-7]*) out="${out}2";;
+       [159dD][89a-fA-F]*) out="${out}3";;
+           [26aAeE][0-7]*) out="${out}4";;
+      [26aAeE][89a-fA-F]*) out="${out}5";;
+           [37bBfF][0-7]*) out="${out}6";;
+      [37bBfF][89a-fA-F]*) out="${out}7";;
+    esac
+    case $in in
+       ?[08]*) out="${out}0";;
+       ?[19]*) out="${out}1";;
+      ?[2aA]*) out="${out}2";;
+      ?[3bB]*) out="${out}3";;
+      ?[4cC]*) out="${out}4";;
+      ?[5dD]*) out="${out}5";;
+      ?[6eE]*) out="${out}6";;
+      ?[7fF]*) out="${out}7";;
+    esac
+    in="${in#?}"
+    in="${in#?}"
+  done
+  printf -- "$out"
 }
 
 if [ -z "$REQUEST_METHOD" ]; then
@@ -78,7 +122,7 @@ if [ -z "$REQUEST_METHOD" ]; then
     kill $cgilite_watchdog
 
     SERVER_PROTOCOL="${SERVER_PROTOCOL%${CR}}"
-    PATH_INFO="$(HEX_DECODE "${REQUEST_URI%\?*}" |PATH)"
+    PATH_INFO="$(HEX_DECODE % "${REQUEST_URI%\?*}" |PATH)"
     [ "${REQUEST_URI}" = "${REQUEST_URI#*\?}" ] \
     && QUERY_STRING='' \
     || QUERY_STRING="${REQUEST_URI#*\?}"
@@ -121,11 +165,14 @@ if [ -z "$REQUEST_METHOD" ]; then
   exit 0
 fi
 
+include_cgilite="$0"
+
 if [ "${REQUEST_METHOD}" = POST -a "${CONTENT_LENGTH:-0}" -gt 0 -a \
      "${CONTENT_TYPE}" = "application/x-www-form-urlencoded" ]; then
   cgilite_post="$(head -c "$CONTENT_LENGTH")"
 fi
 
+debug(){ [ $# -gt 0 ] && printf '%s\n' "$@" >&2 || tee -a /dev/stderr; }
 [ "${DEBUG+x}" ] && env >&2
 
 cgilite_count(){
@@ -141,7 +188,7 @@ cgilite_value(){
     str="${str#*&${name}=}"
     cnt=$((cnt - 1))
   done
-  printf -- "$(printf %s "${str%%&*}" |sed -E 's;\+; ;g;'"$HEX_DECODE")"
+  HEX_DECODE % "$(printf %s "${str%%&*}" |tr + \  )"
 }
 
 cgilite_keys(){
@@ -153,15 +200,15 @@ cgilite_keys(){
   | sort -u
 }
 
-GET(){ cgilite_value "${QUERY_STRING}" $@; }
+GET(){ cgilite_value "${QUERY_STRING}" "$@"; }
 GET_COUNT(){ cgilite_count "${QUERY_STRING}" $1; }
 GET_KEYS(){ cgilite_keys "${QUERY_STRING}"; }
 
-POST(){ cgilite_value "${cgilite_post}" $@; }
+POST(){ cgilite_value "${cgilite_post}" "$@"; }
 POST_COUNT(){ cgilite_count "${cgilite_post}" $1; }
 POST_KEYS(){ cgilite_keys "${cgilite_post}"; }
 
-REF(){ cgilite_value "${HTTP_REFERER#*\?}" $@; }
+REF(){ cgilite_value "${HTTP_REFERER#*\?}" "$@"; }
 REF_COUNT(){ cgilite_count "${HTTP_REFERER#*\?}" $1; }
 REF_KEYS(){ cgilite_keys "${HTTP_REFERER#*\?}"; }
 
@@ -181,7 +228,7 @@ HEADER(){
 }
 
 COOKIE(){
-  HEX_DECODE "$(
+  HEX_DECODE % "$(
     HEADER Cookie \
     | grep -oE '(^|; ?)'"$1"'=[^;]*' \
     | sed -En "${2:-1}"'{s;^[^=]+=;;; s;\+; ;g; p;}'
@@ -202,6 +249,7 @@ HTML(){
       \'*) out="${out}&#x27;";;
       \[*) out="${out}&#x5B;";;
       \]*) out="${out}&#x5D;";;
+      "${CR}"*) out="${out}&#x0D;";;
       "${BR}"*) out="${out}&#x0A;";;
       *) out="${out}${str%"${str#?}"}";;
     esac
@@ -224,6 +272,8 @@ URL(){
       \[*) out="${out}%5B";;
       \]*) out="${out}%5D";;
       \ *) out="${out}%20";;
+      "        "*) out="${out}%09";;
+      "${CR}"*) out="${out}%0D";;
       "${BR}"*) out="${out}%0A";;
       %*) out="${out}%25";;
       *) out="${out}${str%"${str#?}"}";;