X-Git-Url: https://git.distorted.org.uk/~mdw/anag/blobdiff_plain/650bb9da7cf5b677960c03e0a6a5616d48340845..904ac13a4c7d86c18d031ad3f113659440ab86b8:/wildcard.c?ds=sidebyside

diff --git a/wildcard.c b/wildcard.c
index 78b3168..82f33f0 100644
--- a/wildcard.c
+++ b/wildcard.c
@@ -51,81 +51,96 @@ typedef struct node_wild {
 
 static int match(const char *p, const char *s)
 {
-  for (;;) {
-    char pch = *p++, pche, sch;
-    int sense;
+  char pch, pche, sch;
+  int sense;
 
+  for (;;) {
+    pch = *p++;
     switch (pch) {
+
       case '?':
-	if (!*s)
-	  return (0);
-	s++;
+	/* If there's no character left, then we fail; otherwise consume
+	 * anything and move on.
+	 */
+
+	if (!*s++) return (0);
 	break;
+
       case '*':
-	if (!*p)
-	  return (1);
-	while (*s) {
-	  if (match(p, s))
-	    return (1);
+	/* If there's no more pattern then we win. */
+	if (!*p) return (1);
+
+	/* Try skipping any number of characters from the pattern looking for
+	 * a match.
+	 */
+	do {
+	  if (match(p, s)) return (1);
 	  s++;
-	}
+	} while (*s);
 	return (0);
+
       case '[':
-	if (!*s)
-	  return (0);
+	/* Character sets.  This is the hard part. */
+
+	/* If there is no character left, then we fail. */
+	if (!*s) return (0);
+
+	/* Fetch the string character, and start munching through the
+	 * pattern.
+	 */
 	sch = *s++;
-	pch = *p++;
-	sense = 1;
-	if (pch == '^' || pch == '!') {
-	  sense = !sense;
-	  pch = *p++;
-	}
+	pch = *p++; sense = 1;
+
+	/* Maybe we need to negate. */
+	if (pch == '^' || pch == '!') { sense = !sense; pch = *p++; }
+
+	/* A close bracket here is literal.  Watch for ranges. */
 	if (pch == ']') {
 	  if (*p == '-' && p[1] && p[1] != ']') {
-	    pche = p[1];
-	    p += 2;
-	    if (pch <= sch && sch <= pche)
-	      goto class_match;
-	  } else if (pch == sch)
-	    goto class_match;
+	    pche = p[1]; p += 2;
+	    if (pch <= sch && sch <= pche) goto class_match;
+	  } else if (pch == sch) goto class_match;
 	  pch = *p++;
 	}
+
+	/* Work through the other characters and ranges in the set. */
 	for (;; pch = *p++) {
-	  if (!pch || pch == ']')
-	    goto class_nomatch;
+	  if (!pch || pch == ']') goto class_nomatch;
 	  if (*p == '-' && p[1] && p[1] != ']') {
-	    pche = p[1];
-	    p += 2;
-	    if (pch <= sch && sch <= pche)
-	      goto class_match;
-	  } else if (pch == sch)
-	    goto class_match;
+	    pche = p[1]; p += 2;
+	    if (pch <= sch && sch <= pche) goto class_match;
+	  } else if (pch == sch) goto class_match;
 	}
+
       class_match:
-	if (!sense)
-	  return (0);
+	/* Found a match.  Chew through the rest of the pattern. */
+	if (!sense) return (0);
 	for (;;) {
-	  pch = *p++;
-	  if (!pch)
-	    return (0);
-	  if (pch == ']')
-	    break;
-	  if (*p == '-' && p[1] && p[1] != ']')
-	    p += 2;
+	  pch = *p++; if (!pch) return (0);
+	  if (pch == ']') break;
+	  if (*p == '-' && p[1] && p[1] != ']') p += 2;
 	}
 	break;
+
       class_nomatch:
-	if (sense)
-	  return (0);
+	/* Found the end of the set, so it's a mismatch. */
+	if (sense) return (0);
 	break;
+
       case '\\':
+	/* Treat the next thing literally. */
 	pch = *p++;
+	/* fall through... */
+
       default:
-	if (pch != *s)
-	  return (0);
-	if (!pch)
-	  return (1);
-	s++;
+	/* A plain character match.
+	 *
+	 * Trick: If this is the end of the pattern, we expect the end of
+	 * the string.
+	 */
+
+	if (pch != *s++) return (0);
+	if (!pch) return (1);
 	break;
     }
   }