regexp, regsub: add support for -expanded

Fixes #311

Signed-off-by: Steve Bennett <steveb@workware.net.au>

6 files changed, 94 insertions, 21 deletions

diff --git a/jim-regexp.c b/jim-regexp.c
index 8516c80..28cede4 100644
--- a/jim-regexp.c
+++ b/jim-regexp.c
@@ -137,10 +137,10 @@ int Jim_RegexpCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
     int eflags = 0;
     int option;
     enum {
-        OPT_INDICES,  OPT_NOCASE, OPT_LINE, OPT_ALL, OPT_INLINE, OPT_START, OPT_END
+        OPT_INDICES,  OPT_NOCASE, OPT_LINE, OPT_ALL, OPT_INLINE, OPT_START, OPT_EXPANDED, OPT_END
     };
     static const char * const options[] = {
-        "-indices", "-nocase", "-line", "-all", "-inline", "-start", "--", NULL
+        "-indices", "-nocase", "-line", "-all", "-inline", "-start", "-expanded", "--", NULL
     };
 
     for (i = 1; i < argc; i++) {
@@ -185,6 +185,15 @@ int Jim_RegexpCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
                     return JIM_ERR;
                 }
                 break;
+
+            case OPT_EXPANDED:
+#ifdef REG_EXPANDED
+                regcomp_flags |= REG_EXPANDED;
+                break;
+#else
+                Jim_SetResultFormatted(interp, "not supported: %#s", argv[i]);
+                return JIM_ERR;
+#endif
         }
     }
     if (argc - i < 2) {
@@ -361,10 +370,10 @@ int Jim_RegsubCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
     const char *pattern;
     int option;
     enum {
-        OPT_NOCASE, OPT_LINE, OPT_ALL, OPT_START, OPT_COMMAND, OPT_END
+        OPT_NOCASE, OPT_LINE, OPT_ALL, OPT_START, OPT_COMMAND, OPT_EXPANDED, OPT_END
     };
     static const char * const options[] = {
-        "-nocase", "-line", "-all", "-start", "-command", "--", NULL
+        "-nocase", "-line", "-all", "-start", "-command", "-expanded", "--", NULL
     };
 
     for (i = 1; i < argc; i++) {
@@ -405,6 +414,15 @@ int Jim_RegsubCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
             case OPT_COMMAND:
                 opt_command = 1;
                 break;
+
+            case OPT_EXPANDED:
+#ifdef REG_EXPANDED
+                regcomp_flags |= REG_EXPANDED;
+                break;
+#else
+                Jim_SetResultFormatted(interp, "not supported: %#s", argv[i]);
+                return JIM_ERR;
+#endif
         }
     }
     if (argc - i != 3 && argc - i != 4) {
diff --git a/jim_tcl.txt b/jim_tcl.txt
index fd58d50..6975401 100644
--- a/jim_tcl.txt
+++ b/jim_tcl.txt
@@ -3901,7 +3901,7 @@ See `aio read`
 
 regexp
 ~~~~~~
-+*regexp ?-nocase? ?-line? ?-indices? ?-start* 'offset'? *?-all? ?-inline? ?--?* 'exp string ?matchVar? ?subMatchVar subMatchVar \...?'+
++*regexp ?-nocase? ?-line? ?-indices? ?-start* 'offset'? *?-all? ?-inline? ?-expanded? ?--?* 'exp string ?matchVar? ?subMatchVar subMatchVar \...?'+
 
 Determines whether the regular expression +'exp'+ matches part or
 all of +'string'+ and returns 1 if it does, 0 if it doesn't.
@@ -3974,13 +3974,17 @@ The following switches modify the behaviour of +'regexp'+
     data, plus one element for each subexpression in the regular
     expression.
 
++*-expanded*+::
+    Enables use of the expanded regular expression syntax where whitespace
+    and comments are ignored.
+
 +*--*+::
     Marks the end of switches. The argument following this one will be
     treated as +'exp'+ even if it starts with a +-+.
 
 regsub
 ~~~~~~
-+*regsub ?-nocase? ?-all? ?-line? ?-command? ?-start* 'offset'? ?*--*? 'exp string subSpec ?varName?'+
++*regsub ?-nocase? ?-all? ?-line? ?-command? ?-expanded? ?-start* 'offset'? ?*--*? 'exp string subSpec ?varName?'+
 
 This command matches the regular expression +'exp'+ against
 +'string'+ using the rules described in REGULAR EXPRESSIONS
@@ -4062,6 +4066,10 @@ The following switches modify the behaviour of +'regsub'+
     start matching the regular expression. +'offset'+ will be
     constrained to the bounds of the input string.
 
++*-expanded*+::
+    Enables use of the expanded regular expression syntax where whitespace
+    and comments are ignored.
+
 +*--*+::
     Marks the end of switches. The argument following this one will be
     treated as +'exp'+ even if it starts with a +-+.
diff --git a/jimregexp.c b/jimregexp.c
index 136b0c0..1fe6d8b 100644
--- a/jimregexp.c
+++ b/jimregexp.c
@@ -1,5 +1,5 @@
 /*
- * vi:se ts=8:
+ * vi:se ts=8 sw=8:
  *
  * regcomp and regexec -- regsub and regerror are elsewhere
  *
@@ -216,6 +216,41 @@ static int str_int_len(const int *seq)
 	return n;
 }
 
+/* skips preg->regparse past white space and comments to end of line if REG_EXPANDED */
+static char *reg_expanded_new_pattern(const char *exp)
+{
+	/* Make a copy and do removal in place as the final will always be no longer than the original */
+	char *newexp = strdup(exp);
+	char *d = newexp;
+	const char *s = exp;
+	int escape = 0;
+
+	while (*s) {
+		if (escape) {
+			escape = 0;
+			continue;
+		}
+		else if (*s == '\\') {
+			escape = 1;
+		}
+		else if (strchr(" \t\r\n\f\v", *s)) {
+			s++;
+			continue;
+		}
+		else if (*s == '#') {
+			/* skip comments to end of line */
+			s++;
+			while (*s && *s != '\n') {
+				s++;
+			}
+			continue;
+		}
+		*d++ = *s++;
+	}
+	*d++ = '\0';
+	return newexp;
+}
+
 /*
  - regcomp - compile a regular expression into internal code
  *
@@ -246,6 +281,11 @@ int jim_regcomp(regex_t *preg, const char *exp, int cflags)
 	if (exp == NULL)
 		FAIL(preg, REG_ERR_NULL_ARGUMENT);
 
+	if (cflags & REG_EXPANDED) {
+		preg->exp = reg_expanded_new_pattern(exp);
+		exp = preg->exp;
+	}
+
 	/* First pass: determine size, legality. */
 	preg->cflags = cflags;
 	preg->regparse = exp;
@@ -1911,6 +1951,7 @@ size_t jim_regerror(int errcode, const regex_t *preg, char *errbuf,  size_t errb
 
 void jim_regfree(regex_t *preg)
 {
+	free(preg->exp);
 	free(preg->program);
 }
 
diff --git a/jimregexp.h b/jimregexp.h
index e18178e..86a94f6 100644
--- a/jimregexp.h
+++ b/jimregexp.h
@@ -49,6 +49,7 @@ typedef struct regexp {
 	int regmust;		/* Internal use only. */
 	int regmlen;		/* Internal use only. */
 	int *program;		/* Allocated */
+	char *exp;			/* NULL or allocated version of regcomp expression (for REG_EXPANDED) */
 
 	/* working state - compile */
 	const char *regparse;		/* Input-scan pointer. */
@@ -73,6 +74,7 @@ typedef regexp regex_t;
 #define REG_ICASE 2
 
 #define REG_NOTBOL 16
+#define REG_EXPANDED 32
 
 enum {
 	REG_NOERROR,      /* Success.  */
diff --git a/tests/regexp.test b/tests/regexp.test
index 7aeb72e..c7f949f 100644
--- a/tests/regexp.test
+++ b/tests/regexp.test
@@ -199,9 +199,9 @@ test regexp-6.1 {regexp errors} {
 test regexp-6.2 {regexp errors} {
     list [catch {regexp -nocase a} msg] $msg
 } {1 {wrong # args: should be "regexp ?-switch ...? exp string ?matchVar? ?subMatchVar ...?"}}
-test regexp-6.3 {regexp errors} jim {
+test regexp-6.3 {regexp errors} -constraints jim -body {
     list [catch {regexp -gorp a} msg] $msg
-} {1 {bad switch "-gorp": must be --, -all, -indices, -inline, -line, -nocase, or -start}}
+} -result {1 {bad switch "-gorp": must be --, -all, -expanded, -indices, -inline, -line, -nocase, or -start}}
 test regexp-6.4 {regexp errors} {
     catch {regexp a( b} msg
 } 1
@@ -367,7 +367,7 @@ test regexp-11.4 {regsub errors} {
 } {1 {wrong # args: should be "regsub ?-switch ...? exp string subSpec ?varName?"}}
 test regexp-11.5 {regsub errors} -constraints jim -body {
     list [catch {regsub -gorp a b c} msg] $msg
-} -result {1 {bad switch "-gorp": must be --, -all, -command, -line, -nocase, or -start}}
+} -result {1 {bad switch "-gorp": must be --, -all, -command, -expanded, -line, -nocase, or -start}}
 test regexp-11.6 {regsub errors} {
     catch {regsub -nocase a( b c d} msg
 } 1
diff --git a/tests/regexp2.test b/tests/regexp2.test
index c3db17d..ebcf01b 100644
--- a/tests/regexp2.test
+++ b/tests/regexp2.test
@@ -463,12 +463,12 @@ test regexpComp-9.6 {-all option to regsub} {
     }
 } {1 123xxx}
 
-#test regexpComp-10.1 {expanded syntax in regsub} {
-#    evalInProc {
-#	set foo xxx
-#	list [regsub -expanded ". \#comment\n  . \#comment2" abc def foo] $foo
-#    }
-#} {1 defc}
+test regexpComp-10.1 {expanded syntax in regsub} {
+    evalInProc {
+	set foo xxx
+	list [regsub -expanded ". \#comment\n  . \#comment2" abc def foo] $foo
+    }
+} {1 defc}
 test regexpComp-10.2 {newline sensitivity in regsub} {
     evalInProc {
 	set foo xxx
@@ -523,11 +523,11 @@ test regexpComp-11.4 {regsub errors} {
 	list [catch {regsub a b c d e f} msg] $msg
     }
 } {1 {wrong # args: should be "regsub ?-switch ...? exp string subSpec ?varName?"}}
-#test regexpComp-11.5 {regsub errors} {
-#    evalInProc {
-#	list [catch {regsub -gorp a b c} msg] $msg
-#    }
-#} {1 {bad switch "-gorp": must be -all, -nocase, -expanded, -line, -linestop, -lineanchor, -start, or --}}
+test regexpComp-11.5 {regsub errors} {
+    evalInProc {
+	list [catch {regsub -gorp a b c} msg] $msg
+    }
+} {1 {bad switch "-gorp": must be --, -all, -command, -expanded, -line, -nocase, or -start}}
 test regexpComp-11.6 {regsub errors} {
     evalInProc {
 	list [catch {regsub -nocase a( b c d} msg] $msg
@@ -959,4 +959,8 @@ test regexp-26.3 {regexp operator =~ invalid regexp} -body {
     expr {"abc" =~ {[}}
 } -returnCodes error -result {couldn't compile regular expression pattern: brackets [] not balanced}
 
+test regexp-27.1 {regexp expanded} -body {
+    regexp -expanded -all -inline { a ( b b ) + } {abbbbbbcde}
+} -returnCodes ok -result {abbbbbb bb}
+
 testreport