PATCH: =~ regex match

[Phil Pennock <zsh-workers+phil.pennock@xxxxxxxxxxxx>]

My mental processes push me towards regexps perhaps more than is
healthy.  zsh/pcre's -pcre-match conditional operator is firmly in my
toolkit.

Earlier today I discovered that bash has a =~ operator for doing
extended regexp comparisons.  I became envious, so I did something about
it.  Which didn't involve giving up all the conveniences of zsh which
have spoiled me.

The below patch makes =~ an operator which silently auto-loads zsh/pcre
and uses -pcre-match from there.  This is not bash-compatible in that:

 1: it's PCRE -- far more to my liking :^)
 2: if the regexp pattern is bad, it returns false/1 not false/2.

Further, I became aware of =~ because someone was ranting about bash3.2
breaking their scripts by insisting that the regexp must not be quoted.
zsh's =~ is more compatible with older bash scripts by accepting either.

The Src/cond.c @@ -139,7 +153,8 @@ patch covers a case where a bug in my
developing exposed a bug in zsh's internal-protection error handling,
whereby the else condition of an "if non-null and other condition"
assumed the non-null aspect.

My uncovering that bug does more to demonstrate how little I understand
things, so if there are other issues, please do let me know.

This =~ will return an error cleanly if zsh/pcre is not available for
whatever reason.

Index: Src/cond.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/cond.c,v
retrieving revision 1.8
diff -p -u -r1.8 cond.c
--- Src/cond.c	30 May 2006 22:35:03 -0000	1.8
+++ Src/cond.c	26 Apr 2007 04:05:18 -0000
@@ -34,7 +34,7 @@ int tracingcond;
 
 static char *condstr[COND_MOD] = {
     "!", "&&", "||", "==", "!=", "<", ">", "-nt", "-ot", "-ef", "-eq",
-    "-ne", "-lt", "-gt", "-le", "-ge"
+    "-ne", "-lt", "-gt", "-le", "-ge", "=~"
 };
 
 /*
@@ -53,14 +53,14 @@ int
 evalcond(Estate state, char *fromtest)
 {
     struct stat *st;
-    char *left, *right;
+    char *left, *right, *overridename;
     Wordcode pcode;
     wordcode code;
     int ctype, htok = 0, ret;
 
  rec:
 
-    left = right = NULL;
+    left = right = overridename = NULL;
     pcode = state->pc++;
     code = *pcode;
     ctype = WC_COND_TYPE(code);
@@ -92,13 +92,27 @@ evalcond(Estate state, char *fromtest)
 	    state->pc = pcode + (WC_COND_SKIP(code) + 1);
 	    return ret;
 	}
+    case COND_REGEX:
+	{
+	    int loaded;
+	    loaded = load_module_silence("zsh/pcre", 1);
+	    if (!loaded) {
+		zwarnnam(fromtest, "zsh/pcre not available for regex");
+		return 2;
+	    }
+	    ctype = COND_MODI;
+	    overridename = "-pcre-match";
+	}
     case COND_MOD:
     case COND_MODI:
 	{
 	    Conddef cd;
-	    char *name = ecgetstr(state, EC_NODUP, NULL), **strs;
+	    char *name = overridename;
+	    char **strs;
 	    int l = WC_COND_SKIP(code);
 
+	    if (name == NULL)
+		name = ecgetstr(state, EC_NODUP, NULL);
 	    if (ctype == COND_MOD)
 		strs = ecgetarr(state, l, EC_DUP, NULL);
 	    else {
@@ -139,7 +153,8 @@ evalcond(Estate state, char *fromtest)
 		    return !cd->handler(strs, cd->condid);
 		} else {
 		    zwarnnam(fromtest,
-			     "unrecognized condition: `%s'", name);
+			     "unrecognized condition: `%s'",
+			     name ? name : "<null>");
 		}
 	    }
 	    /* module not found, error */
Index: Src/parse.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/parse.c,v
retrieving revision 1.64
diff -p -u -r1.64 parse.c
--- Src/parse.c	23 Apr 2007 17:24:23 -0000	1.64
+++ Src/parse.c	26 Apr 2007 04:05:19 -0000
@@ -2124,6 +2124,12 @@ par_cond_triple(char *a, char *b, char *
 	ecstr(a);
 	ecstr(c);
 	ecadd(ecnpats++);
+    } else if ((b[0] == Equals || b[0] == '=') &&
+               (b[1] == '~' || b[1] == Tilde) && ~b[2]) {
+	ecadd(WCB_COND(COND_REGEX, 0));
+	ecstr(a);
+	ecstr(c);
+	ecadd(ecnpats++);
     } else if (b[0] == '-') {
 	if ((t0 = get_cond_num(b + 1)) > -1) {
 	    ecadd(WCB_COND(t0 + COND_NT, 0));
Index: Src/text.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/text.c,v
retrieving revision 1.19
diff -p -u -r1.19 text.c
--- Src/text.c	23 Apr 2007 15:24:00 -0000	1.19
+++ Src/text.c	26 Apr 2007 04:05:19 -0000
@@ -640,7 +640,7 @@ gettext2(Estate state)
 	    {
 		static char *c1[] = {
 		    "=", "!=", "<", ">", "-nt", "-ot", "-ef", "-eq",
-		    "-ne", "-lt", "-gt", "-le", "-ge"
+		    "-ne", "-lt", "-gt", "-le", "-ge", "=~"
 		};
 
 		int ctype;
@@ -724,7 +724,7 @@ gettext2(Estate state)
 			}
 			break;
 		    default:
-			if (ctype <= COND_GE) {
+			if (ctype < COND_MOD) {
 			    /* Binary test: `a = b' etc. */
 			    taddstr(ecgetstr(state, EC_NODUP, NULL));
 			    taddstr(" ");
Index: Src/zsh.h
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/zsh.h,v
retrieving revision 1.112
diff -p -u -r1.112 zsh.h
--- Src/zsh.h	29 Mar 2007 21:35:39 -0000	1.112
+++ Src/zsh.h	26 Apr 2007 04:05:22 -0000
@@ -519,8 +519,9 @@ struct timedfn {
 #define COND_GT    13
 #define COND_LE    14
 #define COND_GE    15
-#define COND_MOD   16
-#define COND_MODI  17
+#define COND_REGEX 16
+#define COND_MOD   17
+#define COND_MODI  18
 
 typedef int (*CondHandler) _((char **, int));
 
Index: Src/Modules/pcre.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Modules/pcre.c,v
retrieving revision 1.11
diff -p -u -r1.11 pcre.c
--- Src/Modules/pcre.c	5 Apr 2007 16:20:15 -0000	1.11
+++ Src/Modules/pcre.c	26 Apr 2007 04:05:22 -0000
@@ -173,6 +173,10 @@ cond_pcre_match(char **a, int id)
     switch(id) {
 	 case CPCRE_PLAIN:
 		 pcre_pat = pcre_compile(rhre, pcre_opts, &pcre_err, &pcre_errptr, NULL);
+		 if (pcre_pat == NULL) {
+		     zwarn("failed to compile regexp: %s", rhre);
+		     return 0;
+		 }
                  pcre_fullinfo(pcre_pat, NULL, PCRE_INFO_CAPTURECOUNT, &capcnt);
     		 ovsize = (capcnt+1)*3;
 		 ov = zalloc(ovsize*sizeof(int));
@@ -191,6 +195,7 @@ cond_pcre_match(char **a, int id)
 
 static struct conddef cotab[] = {
     CONDDEF("pcre-match", CONDF_INFIX, cond_pcre_match, 0, 0, CPCRE_PLAIN)
+    /* CONDDEF can register =~ but it won't be found */
 };
 
 /**/