improved vcf parser (speed, security)

svn path=/trunk/; revision=139

diff --git a/pages/cards.sh b/pages/cards.sh
index 08b88b94e11b75e8eeeab786df667adb1912af3c..0ead803fb32304a581ba5fcdaae35bcd2e8413de 100755 (executable)
--- a/pages/cards.sh
+++ b/pages/cards.sh
@@ -102,59 +102,63 @@ listcards() {
 }
 
 vcf_parse() {
-  tr -d '\n' <"$1" |sed -r 's:\r ::g;s:\r:\n:g' \
+  sed -r ':X;N;$!bX; s;\r\n[ \t];;g; s;\r\n;\n;g;' "$1" \
   | sed -rn '
-    s:^X-MS-CARDPICTURE:PHOTO:p;
-    s:^X-GENDER:GENDER:p;
-    s:^X-ANNIVERSARY:ANNIVERSARY:p;
-    s:^X-EVOLUTION-ANNIVERSARY:ANNIVERSARY:p;
-    s:^X-KADDRESSBOOK-X-Anniversary:ANNIVERSARY:p;
-    s:^X-AIM(;[^"\:]+|;"[^"]+")*\:(.*)$:IMPP\1\:aim\:\2:p;
-    s:^X-ICQ(;[^"\:]+|;"[^"]+")*\:(.*)$:IMPP\1\:aim\:\2:p;
-    s:^X-GOOGLE-TALK(;[^"\:]+|;"[^"]+")*\:(.*)$:IMPP\1\:xmpp\:\2:p;
-    s:^X-JABBER(;[^"\:]+|;"[^"]+")*\:(.*)$:IMPP\1\:xmpp\:\2:p;
-    s:^X-MSN(;[^"\:]+|;"[^"]+")*\:(.*)$:IMPP\1\:msn\:\2:p;
-    s:^X-YAHOO(;[^"\:]+|;"[^"]+")*\:(.*)$:IMPP\1\:ymsgr\:\2:p;
-    s:^X-SIP(;[^"\:]+|;"[^"]+")*\:(sip\:)?(.*)$:IMPP\1\:sip\:\3:p;
-    s:^LABEL(;[^"\:]+|;"[^"]+")*\:(.*)$:ADR;LABEL="\2"\1\::p;
-    s:^X-EVOLUTION-BLOG-URL:URL:p;
-
-    s:^AGENT:RELATED\;TYPE=agent:p;
-    s:^X-ASSISTANT:RELATED\;TYPE=assistant;VALUE=text:p;
-    s:^X-EVOLUTION-ASSISTANT:RELATED\;TYPE=assistant;VALUE=text:p;
-    s:^X-KADDRESSBOOK-X-AssistantsName:RELATED\;TYPE=assistant;VALUE=text:p;
-    s:^X-MANAGER:RELATED\;TYPE=manager;VALUE=text:p;
-    s:^X-EVOLUTION-MANAGER:RELATED\;TYPE=manager;VALUE=text:p;
-    s:^X-KADDRESSBOOK-X-ManagersName:RELATED\;TYPE=manager;VALUE=text:p;
-    s:^X-SPOUSE:RELATED\;TYPE=spouse;VALUE=text:p;
-    s:^X-EVOLUTION-SPOUSE:RELATED\;TYPE=spouse;VALUE=text:p;
-    s:^X-KADDRESSBOOK-X-SpouseName:RELATED\;TYPE=spouse;VALUE=text:p;
-
-    s:^([A-Z].*)$:\1:p;
-    ' \
-  | sed -r 's:^([^;\:]+)(;[^"\:]+|;"[^"]+")*\:(.*)$:key="\1"\nvalue="\3"\ntag=\2:g' \
-  | while read -r line; do
-    case "$line" in
-      key=*) printf %s\\n "$line"
-       ;;
-      value=*) printf %s\\n "${line}"
-       ;;
-      tag=*) ot=''
-            printf %s "$line" \
-            | sed -r 's:^tag=::;s:\;([A-Z+_-]+="[^"]+"|[A-Z+_-]+=[^\;]+):\n\1:g;' \
-            | sed -rn 's:([A-Z+_-]+)="?(.+)"?:tag\[\1\]="\2":gp' \
-            | sed -r '/^ *$/d' \
-            | sort |while read -r tag; do
-              nt="$(printf %s "$tag" |sed -r 's:^tag\[([A-Z+_-]+)\]="(.*)"$:\1:')"
-              nv="$(printf %s "$tag" |sed -r 's:^tag\[([A-Z+_-]+)\]="(.*)"$:\2:')"
-              [ "$nt" = "$ot" ] && vl="$nv,$vl" || vl="$nv"
-              printf %s\\n "tag[$nt]=\"$vl\""
-              ot="$nt"
-            done
-       ;;
-    esac
-  done \
-  | sed -r 's:[\\$`]:\\&:g'
+    # === turn property names to upper case ===
+    h; s;^([^\;:]+);;;
+    x; s;^([^\;:]+).*$;\1;;
+    y;abcdefghijklmnopqrstuvwxyz;ABCDEFGHIJKLMNOPQRSTUVWXYZ;
+    G; s;\n;;;
+
+    # === Normalise various known vendor properties ===
+                s;^X-MS-CARDPICTURE(\;|:);PHOTO\1;;
+                        s;^X-GENDER(\;|:);GENDER\1;;
+                   s;^X-ANNIVERSARY(\;|:);ANNIVERSARY\1;;
+         s;^X-EVOLUTION-ANNIVERSARY(\;|:);ANNIVERSARY\1;;
+    s;^X-KADDRESSBOOK-X-ANNIVERSARY(\;|:);ANNIVERSARY\1;;
+            s;^X-EVOLUTION-BLOG-URL(\;|:);URL\1;;
+                              s;^AGENT(\;|:);RELATED\;VALUE=text\;TYPE=agent\1;;
+                        s;^X-ASSISTANT(\;|:);RELATED\;VALUE=text\;TYPE=assistant\1;;
+              s;^X-EVOLUTION-ASSISTANT(\;|:);RELATED\;VALUE=text\;TYPE=assistant\1;;
+    s;^X-KADDRESSBOOK-X-ASSISTANTSNAME(\;|:);RELATED\;VALUE=text\;TYPE=assistant\1;;
+                          s;^X-MANAGER(\;|:);RELATED\;VALUE=text\;TYPE=manager\1;;
+                s;^X-EVOLUTION-MANAGER(\;|:);RELATED\;VALUE=text\;TYPE=manager\1;;
+      s;^X-KADDRESSBOOK-X-MANAGERSNAME(\;|:);RELATED\;VALUE=text\;TYPE=manager\1;;
+                           s;^X-SPOUSE(\;|:);RELATED\;VALUE=text\;TYPE=spouse\1;;
+                 s;^X-EVOLUTION-SPOUSE(\;|:);RELATED\;VALUE=text\;TYPE=spouse\1;;
+        s;^X-KADDRESSBOOK-X-SPOUSENAME(\;|:);RELATED\;VALUE=text\;TYPE=spouse\1;;
+
+    # === Normalise obsolete vendor IM properties ===
+            s;^X-AIM((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):;IMPP\1:aim:;;
+            s;^X-ICQ((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):;IMPP\1:aim:;;
+    s;^X-GOOGLE-TALK((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):;IMPP\1:xmpp:;;
+         s;^X-JABBER((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):;IMPP\1:xmpp:;;
+            s;^X-MSN((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):;IMPP\1:msn:;;
+          s;^X-YAHOO((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):;IMPP\1:ymsgr:;;
+            s;^X-SIP((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):(sip:)?;IMPP\1:sip:;;
+
+    # === Update obsolete LABEL property ===
+    s;^LABEL((\;[a-zA-Z0-9-]+|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):(.*)$;ADR\1\;LABEL="\5":;;
+
+    /^([a-zA-Z0-9-]+)((\;[a-zA-Z0-9-]+=?|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):(.*)$/{
+      h;
+      s;^([a-zA-Z0-9-]+)((\;[a-zA-Z0-9-]+=?|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):(.*)$;key='\''\1'\'';;
+      H; g;
+      s;^([a-zA-Z0-9-]+)((\;[a-zA-Z0-9-]+=?|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):([^\n]*)\n.*$;\6;;
+      s;'\'';'\'\\\\\'\'';g
+      s;^.*$;value='\''&'\'';
+      H; g;
+      s;^([a-zA-Z0-9-]+)((\;[a-zA-Z0-9-]+=?|\;[a-zA-Z0-9-]+=([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*)*):([^\n]*)\n.*$;\2;;
+      s;'\'';'\'\\\\\'\'';g
+      s;\;([a-zA-Z0-9-]+)(=|=(([^\;,:"]+|"[^"]+")(,[^\;,:"]+|,"[^"]+")*))?;\ntag[\1]='\''\3'\'';g
+      H; g;
+      s;^[^\n]*[\n ]+;;
+
+      s;\n[\n ]+;\n;g; s;\n$;;;
+
+      p;
+    }
+    '
 }
 
 view_card() {  #Parameter: Cardfile