* xsd regular expression support for character classes [https://github.com/clicon/clixon/issues/68]

* added support for \c, \d, \w, \W, \s, \S

CHANGELOG.md

@@ -134,6 +134,8 @@
   * <!DOCTYPE (ie DTD) is not supported.
 
 ### Corrected Bugs
+* xsd regular expression support for character classes [https://github.com/clicon/clixon/issues/68]
+  * added support for \c, \d, \w, \W, \s, \S.
 * Removing newlines from XML data [https://github.com/clicon/clixon/issues/65]
 * [ietf-netconf-notification@2008-07-01.yang validation problem #62](https://github.com/clicon/clixon/issues/62)
 * Ignore CR(\r) in yang files for DOS files

lib/src/clixon_string.c

@@ -641,41 +641,68 @@ clixon_trim(char *str)
  * POSIX ERE regexps according to man regex(3).
  * @param[in]  xsd    Input regex string according XSD
  * @param[out] posix  Output (malloced) string according to POSIX ERE
- * @see http://www.w3.org/TR/2004/REC-xmlschema-2-20041028
+ * @see https://www.w3.org/TR/2004/REC-xmlschema-2-20041028/#regexs
- * @note that the translation is ad-hoc, may need more translations
+ * @see https://www.regular-expressions.info/posixbrackets.html#class translation
+ * Translation is not complete but covers some character sequences:
+ * \d decimal digit
+ * \w alphanum + underscore
  */
 int
 regexp_xsd2posix(char  *xsd,
 		 char **posix)
 {
     int   retval = -1;
-    char *x;
+    cbuf *cb = NULL;
-    char *p = NULL;
+    char  x;
     int   i;
-    int   len;
+    int   esc;
 
-    len = strlen(xsd);
+    if ((cb = cbuf_new()) == NULL){
-    x = xsd;
+	clicon_err(OE_UNIX, errno, "cbuf_new");
-    while ((x = strstr(x, "\\d")) != NULL){
-	len += 3; /* \d --> [0-9] */
-	x += 2;
-    }
-    if ((p = malloc(len+1)) == NULL){
-	clicon_err(OE_UNIX, errno, "malloc");
 	goto done;
     }
-    memset(p, 0, len+1);
+    esc=0;
-    *posix = p;
     for (i=0; i<strlen(xsd); i++){
-	if (strncmp(&xsd[i], "\\d", 2) == 0){
+	x = xsd[i];
-	    strcpy(p, "[0-9]");
+	if (esc){
-	    p += 5; i++;
+	    esc = 0;
+	    switch (x){
+	    case 'c': /* xml namechar */
+		cprintf(cb, "[0-9a-zA-Z\\\\.\\\\-_:]");
+		break;
+	    case 'd':
+		cprintf(cb, "[0-9]");
+		break;
+	    case 'w':
+		cprintf(cb, "[0-9a-zA-Z_\\\\-]");
+		break;
+	    case 'W':
+		cprintf(cb, "[^0-9a-zA-Z_\\\\-]");
+		break;
+	    case 's':
+		cprintf(cb, "[ \t\r\n]");
+		break;
+	    case 'S':
+		cprintf(cb, "[^ \t\r\n]");
+		break;
+	    default:
+		cprintf(cb, "\\%c", x);
+		break;
 	    }
+	}
+	else if (x == '\\')
+	    esc++;
 	else
-	    *p++ = xsd[i];
+	    cprintf(cb, "%c", x);
+    }
+    if ((*posix = strdup(cbuf_get(cb))) == NULL){
+	clicon_err(OE_UNIX, errno, "strdup");
+	goto done;
     }
     retval = 0;
  done:
+    if (cb)
+	cbuf_free(cb);
     return retval;
 }
 

test/test_type.sh

@@ -180,6 +180,16 @@ module example{
     description "transitive type- exists in ex3";
     uses ex2:gr2;
   }
+  leaf digit4{
+     type string {
+         pattern '\d{4}';
+       }
+  }
+  leaf word4{
+     type string {
+         pattern '\w{4}';
+       }
+  }
 }
 EOF
 
@@ -358,6 +368,31 @@ expectfn "$clixon_cli -1f $cfg -l o -y $fyang set len2 ab" 255 "^$"
 new "cli range test len3 42 ok"
 expectfn "$clixon_cli -1f $cfg -l o -y $fyang set len3 hsakjdhkjsahdkjsahdksahdksajdhsakjhd" 0 "^$"
 
+# XSD schema -> POSIX ECE translation
+new "cli yang pattern \d ok"
+expectfn "$clixon_cli -1f $cfg -l o -y $fyang set digit4 0123" 0 "^$"
+
+new "cli yang pattern \d error"
+expectfn "$clixon_cli -1f $cfg -l o -y $fyang set digit4 01b2" 255 "^$"
+
+new "cli yang pattern \w ok"
+expectfn "$clixon_cli -1f $cfg -l o -y $fyang set word4 a2-_" 0 "^$"
+
+new "cli yang pattern \w error"
+expectfn "$clixon_cli -1f $cfg -l o -y $fyang set word4 ab%d3" 255 "^$"
+
+new "netconf pattern \w"
+expecteof "$clixon_netconf -qf $cfg -y $fyang" 0 '<rpc><edit-config><target><candidate/></target><config><word4 xmlns="urn:example:clixon">a-_9</word4></config></edit-config></rpc>]]>]]>' "^<rpc-reply><ok/></rpc-reply>]]>]]>$"
+
+new "netconf pattern \w valid"
+expecteof "$clixon_netconf -qf $cfg -y $fyang" 0 '<rpc><validate><source><candidate/></source></validate></rpc>]]>]]>' "^<rpc-reply><ok/></rpc-reply>]]>]]>$"
+
+new "netconf pattern \w error"
+expecteof "$clixon_netconf -qf $cfg -y $fyang" 0 '<rpc><edit-config><target><candidate/></target><config><word4 xmlns="urn:example:clixon">ab%d3</word4></config></edit-config></rpc>]]>]]>' "^<rpc-reply><ok/></rpc-reply>]]>]]>$"
+
+new "netconf pattern \w valid"
+expecteof "$clixon_netconf -qf $cfg -y $fyang" 0 '<rpc><validate><source><candidate/></source></validate></rpc>]]>]]>' '^<rpc-reply><rpc-error><error-type>application</error-type><error-tag>bad-element</error-tag><error-info><bad-element>word4</bad-element></error-info><error-severity>error</error-severity><error-message>regexp match fail: "ab%d3" does not match \\w{4}</error-message></rpc-error></rpc-reply>]]>]]>$'
+
 if [ $BE -eq 0 ]; then
     exit # BE
 fi