Merge branch 'master' of https://github.com/clicon/clixon

CHANGELOG.md

@@ -140,8 +140,11 @@
 ### Minor changes
 
 * Regexp improvements
-  * Added check for libxml in configure';
+  * Better compliance with XSD regexps (when transforming to Posix regexps)
+    * Added `\p{L}` and `\p{N}`
+    * Added escaping of `$`
   * Added clixon_util_regexp utility function
+  * Added regexp [test/test_pattern.sh]
 * Yang state get improvements
   * Integrated state and config into same tree on retrieval, not separate trees
   * Added cli functions `cli_show_config_state()` and `cli_show_auto_state()` for showing combined config and state info.

lib/src/clixon_string.c

@@ -709,6 +709,11 @@ clixon_trim(char *str)
  *    "other" characters: #x0000-#x10FFFF]-[\p{P}\p{Z}\p{C}]
  * \i letters + underscore and colon
  * \c XML Namechar, see: https://www.w3.org/TR/2008/REC-xml-20081126/#NT-NameChar
+ *
+ * Not implemented:
+ * \p{X} category escape.  the ones identified in openconfig and yang-models are: 
+ *   \p{L} Letters [ultmo]?
+ *   \p{N} Numbers [dlo]?
  */
 int
 regexp_xsd2posix(char  *xsd,
@@ -718,6 +723,7 @@ regexp_xsd2posix(char  *xsd,
     cbuf *cb = NULL;
     char  x;
     int   i;
+    int   j; /* lookahead */
     int   esc;
     int   minus = 0;
 
@@ -743,12 +749,24 @@ regexp_xsd2posix(char  *xsd,
 	    case 'i': /* initial */
 		cprintf(cb, "[a-zA-Z_:]");
 		break;
-	    case 'w': /* word */
+	    case 'p': /* category escape: \p{IsCategory} */
-		//cprintf(cb, "[0-9a-zA-Z_\\\\-]")
+		j = i+1;
-		cprintf(cb, "[^[:punct:][:space:][:cntrl:]]"); 
+		if (j+2 < strlen(xsd) &&
+		    xsd[j] == '{' &&
+		    xsd[j+2] == '}'){
+		    switch (xsd[j+1]){
+		    case 'L': /* Letters */
+			cprintf(cb, "a-zA-Z"); /* assume in [] */
 			break;
-	    case 'W': /* inverse of \w */
+		    case 'N': /* Numbers */
-		cprintf(cb, "[[:punct:][:space:][:cntrl:]]"); 
+			cprintf(cb, "0-9");
+			break;
+		    default:
+			break;
+		    }
+		    i = j+2;
+		}
+		/* if syntax error, just leave it */
 		break;
 	    case 's':
 		cprintf(cb, "[ \t\r\n]");
@@ -756,6 +774,13 @@ regexp_xsd2posix(char  *xsd,
 	    case 'S':
 		cprintf(cb, "[^ \t\r\n]");
 		break;
+	    case 'w': /* word */
+		//cprintf(cb, "[0-9a-zA-Z_\\\\-]")
+		cprintf(cb, "[^[:punct:][:space:][:cntrl:]]"); 
+		break;
+	    case 'W': /* inverse of \w */
+		cprintf(cb, "[[:punct:][:space:][:cntrl:]]"); 
+		break;
 	    default:
 		cprintf(cb, "\\%c", x);
 		break;
@@ -763,6 +788,8 @@ regexp_xsd2posix(char  *xsd,
 	}
 	else if (x == '\\')
 	    esc++;
+	else if (x == '$')
+	    cprintf(cb, "\\%c", x);
 	else if (x == ']' && minus){
 	    cprintf(cb, "-]");
 	    minus = 0;

lib/src/clixon_yang_parse.l

@@ -96,7 +96,7 @@ identifier      [A-Za-z_][A-Za-z0-9_\-\.]*
 %s STRING
 %s STRINGDQ
 %s STRINGSQ
-%s ESCAPE
+%s DQESC
 %s COMMENT1
 %s COMMENT2
 %s UNKNOWN
@@ -228,7 +228,7 @@ identifier      [A-Za-z_][A-Za-z0-9_\-\.]*
 <STRING>.                 { clixon_yang_parselval.string = strdup(yytext);
                             return CHARS;}
 
-<STRINGDQ>\\              { _YY->yy_lex_state = STRINGDQ; BEGIN(ESCAPE); }
+<STRINGDQ>\\              { _YY->yy_lex_state = STRINGDQ; BEGIN(DQESC); }
 <STRINGDQ>\"              { BEGIN(_YY->yy_lex_string_state); return *yytext; }
 <STRINGDQ>\n              { _YY->yy_linenum++;
                             clixon_yang_parselval.string = strdup(yytext);
@@ -243,7 +243,7 @@ identifier      [A-Za-z_][A-Za-z0-9_\-\.]*
 <STRINGSQ>[^'\n]+       { clixon_yang_parselval.string = strdup(yytext);
                             return CHARS;}
 
-<ESCAPE>.                 { BEGIN(_YY->yy_lex_state); 
+<DQESC>[nt"\\]            { BEGIN(_YY->yy_lex_state); 
                              clixon_yang_parselval.string = strdup(yytext); 
                              return CHARS; }
 <COMMENT1>[^*\n]*        /* eat anything that's not a '*' */

test/lib.sh

@@ -31,6 +31,7 @@ if [ -f ./site.sh ]; then
     # test skiplist.
     for f in $SKIPLIST; do
 	if [ "$testfile" = "$f" ]; then
+	    echo ...skipped
 	    return -1 # skip
 	fi
     done

test/test_yang.sh

@@ -153,7 +153,7 @@ if [ $BE -ne 0 ]; then
     start_backend -s init -f $cfg -y $fyang
 
     new "waiting"
-    sleep $RCWAIT
+    wait_backend
 fi
 
 new "cli defined extension"

util/clixon_util_regexp.c

@@ -33,6 +33,7 @@
 
   * Utility for compiling regexp and checking validity
   *  gcc -I /usr/include/libxml2 regex.c -o regex -lxml2
+  * @see http://www.w3.org/TR/2004/REC-xmlschema-2-20041028
   */
 #ifdef HAVE_CONFIG_H
 #include "clixon_config.h" /* generated by config & autoconf */
@@ -45,8 +46,6 @@
 #include <stdlib.h>
 #include <limits.h>
 
-
-#undef HAVE_LIBXML2
 #ifdef HAVE_LIBXML2 /* Actually it should check for  a header file */
 #include <libxml/xmlregexp.h>
 #endif
@@ -66,7 +65,9 @@
 static int
 regex_libxml2(char *regexp0,
 	      char *content0,
-	      int   nr)
+	      int   nr,
+	      int   debug)
+    
 {
     int        retval = -1;
 #ifdef HAVE_LIBXML2
@@ -92,7 +93,8 @@ regex_libxml2(char *regexp0,
 static int
 regex_posix(char *regexp,
 	    char *content,
-	    int   nr)
+	    int   nr,
+	    int   debug)
 {
     int        retval = -1;
     char       *posix = NULL;
@@ -105,6 +107,7 @@ regex_posix(char *regexp,
     
     if (regexp_xsd2posix(regexp, &posix) < 0)
 	goto done;
+    clicon_debug(1, "posix: %s", posix);
     len0 = strlen(posix);
     if (len0 > sizeof(pattern)-5){
 	fprintf(stderr, "pattern too long\n");
@@ -134,15 +137,15 @@ regex_posix(char *regexp,
 static int
 usage(char *argv0)
 {
-    fprintf(stderr, "usage:%s [options] (either one of -p or -x)\n"
+    fprintf(stderr, "usage:%s [options]\n"
 	    "where options are\n"
             "\t-h \t\tHelp\n"
     	    "\t-D <level>\tDebug\n"
-	    "\t-p          \txsd->posix translation regexp\n"
+	    "\t-p          \txsd->posix translation regexp (default)\n"
-	    "\t-x          \tlibxml2 regexp\n"
+	    "\t-x          \tlibxml2 regexp (alternative to -p)\n"
-	    "\t-n <nr>     \tIterate content match (0 means only compile)\n"
+	    "\t-n <nr>     \tIterate content match (default: 1, 0: no match only compile)\n"
 	    "\t-r <regexp> \tregexp (mandatory)\n"
-	    "\t-c <string> \tValue content string(mandatory)\n",
+	    "\t-c <string> \tValue content string(mandatory if -n > 0)\n",
 	    argv0
 	    );
     exit(0);
@@ -157,10 +160,9 @@ main(int    argc,
     int         c;
     char       *regexp = NULL;
     char       *content = NULL;
-    int         posix = 0;
-    int         libxml2 = 0;
     int         ret;
     int         nr = 1;
+    int         mode = 0; /* 0 is posix, 1 is libxml */
 
     optind = 1;
     opterr = 0;
@@ -174,14 +176,14 @@ main(int    argc,
 		usage(argv0);
 	    break;
 	case 'p': /* xsd->posix */
-	    posix++;
+	    mode = 0;
 	    break;
 	case 'n': /* Number of iterations */
 	    if ((nr = atoi(optarg)) < 0)
 		usage(argv0);
 	    break;
 	case 'x': /* libxml2 */
-	    libxml2++;
+	    mode = 1;
 	    break;
 	case 'r': /* regexp */
 	    regexp = optarg;
@@ -194,22 +196,31 @@ main(int    argc,
 	    break;
 	}
     clicon_log_init(__FILE__, debug?LOG_DEBUG:LOG_INFO, CLICON_LOG_STDERR); 
-    if (regexp == NULL || content == NULL)
+    if (regexp == NULL){
+	fprintf(stderr, "-r mandatory\n");
 	usage(argv0);
-    if (posix == libxml2)
+    }
+    if (nr > 0 && content == NULL){
+	fprintf(stderr, "-c mandatory (if -n > 0)\n");
 	usage(argv0);
+    }
+    if (mode != 0 && mode != 1){
+	fprintf(stderr, "Neither posix or libxml2 set\n");
+	usage(argv0);
+    }
     clicon_debug(1, "regexp:%s", regexp);
     clicon_debug(1, "content:%s", content);
-    if (libxml2){
+    if (mode == 0){
-	if ((ret = regex_libxml2(regexp, content, nr)) < 0)
+	if ((ret = regex_posix(regexp, content, nr, debug)) < 0)
 	    goto done;
+
     }
-    else if (posix){
+    else if (mode == 1){
-	if ((ret = regex_posix(regexp, content, nr)) < 0)
+	if ((ret = regex_libxml2(regexp, content, nr, debug)) < 0)
 	    goto done;
     }
     else
-	goto done;
+	usage(argv0);
     fprintf(stdout, "%d\n", ret);
     exit(ret);
     retval = 0;