PERFORCE change 147064 for review

[Gabor Kovesdan]

http://perforce.freebsd.org/chv.cgi?CH=147064

Change 147064 by gabor at gabor_server on 2008/08/10 09:48:11

	- Cut out fixed string -w in a working way

Affected files ...

.. //depot/projects/soc2008/gabor_textproc/grep/fastgrep.c#4 edit

Differences ...

==== //depot/projects/soc2008/gabor_textproc/grep/fastgrep.c#4 (text+ko) ====

@@ -52,7 +52,6 @@
 	fg->patternLen = strlen(pattern);
 	fg->bol = 0;
 	fg->eol = 0;
-	fg->wmatch = wflag;
 	fg->reversedSearch = 0;
 
 	/*
@@ -101,7 +100,6 @@
 	fg->patternLen = strlen(pattern);
 	fg->bol = 0;
 	fg->eol = 0;
-	fg->wmatch = 0;
 	fg->reversedSearch = 0;
 
 	/* Remove end-of-line character ('$'). */
@@ -118,17 +116,11 @@
 		fg->patternLen--;
 	}
 
-	/* Remove enclosing [[:<:]] and [[:>:]] (word match). */
-	if (wflag) {
-		/* basic re's use \( \), extended re's ( ) */
-		int extra = (grepbehave == GREP_EXTENDED) ? 1 : 2;
-		fg->patternLen -= 14 + 2 * extra;
-		fg->wmatch = 7 + extra;
-	} else if (fg->patternLen >= 14 &&
+	if (fg->patternLen >= 14 &&
 	    strncmp(pattern + fg->bol, "[[:<:]]", 7) == 0 &&
 	    strncmp(pattern + fg->bol + fg->patternLen - 7, "[[:>:]]", 7) == 0) {
 		fg->patternLen -= 14;
-		fg->wmatch = 7;
+		wflag = 1;
 	}
 
 	/*
@@ -137,7 +129,7 @@
 	 * string respectively.
 	 */
 	fg->pattern = grep_malloc(fg->patternLen + 1);
-	memcpy(fg->pattern, pattern + bol + fg->wmatch, fg->patternLen);
+	memcpy(fg->pattern, pattern + bol + wflag, fg->patternLen);
 	fg->pattern[fg->patternLen] = '\0';
 
 	/* Look for ways to cheat...er...avoid the full regex engine. */
@@ -230,18 +222,6 @@
 	return (0);
 }
 
-/*
- * Word boundaries using regular expressions are defined as the point
- * of transition from a non-word char to a word char, or vice versa.
- * This means that grep -w +a and grep -w a+ never match anything,
- * because they lack a starting or ending transition, but grep -w a+b
- * does match a line containing a+b.
- */
-#define isword(x) (isalnum(x) || (x) == '_')
-#define wmatch(d, l, s, e)	\
-	((s == 0 || !isword(d[s-1])) && (e == l || !isword(d[e])) && \
-	  e > s && isword(d[s]) && isword(d[e-1]))
-
 int
 grep_search(fastgrep_t *fg, unsigned char *data, size_t dataLen, regmatch_t *pmatch)
 {
@@ -270,8 +250,6 @@
 				    fg->patternLen) == -1) {
 					pmatch->rm_so = j;
 					pmatch->rm_eo = j + fg->patternLen;
-					if (!fg->wmatch || wmatch(data, dataLen,
-					    pmatch->rm_so, pmatch->rm_eo))
 						rtrnVal = 0;
 				}
 		}
@@ -283,11 +261,8 @@
 			    fg->patternLen) == -1) {
 				pmatch->rm_so = j - fg->patternLen;
 				pmatch->rm_eo = j;
-				if (!fg->wmatch || wmatch(data, dataLen,
-				    pmatch->rm_so, pmatch->rm_eo)) {
-					rtrnVal = 0;
-					break;
-				}
+				rtrnVal = 0;
+				break;
 			}
 			/* Shift if within bounds, otherwise, we are done. */
 			if (j == fg->patternLen)
@@ -301,12 +276,8 @@
 			if (grep_cmp(fg->pattern, data + j, fg->patternLen) == -1) {
 				pmatch->rm_so = j;
 				pmatch->rm_eo = j + fg->patternLen;
-				if (fg->patternLen == 0 || !fg->wmatch ||
-				    wmatch(data, dataLen, pmatch->rm_so,
-				    pmatch->rm_eo)) {
-					rtrnVal = 0;
-					break;
-				}
+				rtrnVal = 0;
+				break;
 			}
 
 			/* Shift if within bounds, otherwise, we are done. */