From fbf29150887290e0944913771dee7af1ed98fdab Mon Sep 17 00:00:00 2001
From: "Eric S. Raymond" <esr@thyrsus.com>
Date: Wed, 17 Sep 2003 21:01:02 +0000
Subject: [PATCH] Now seems to work.

---
 linebyline.c | 105 +++++++++++++++++++++++++++++----------------------
 1 file changed, 60 insertions(+), 45 deletions(-)

diff --git a/linebyline.c b/linebyline.c
index d1b86fc..f15611d 100644
--- a/linebyline.c
+++ b/linebyline.c
@@ -13,31 +13,29 @@ static char *c_patterns[] = {
     "/\\* *NOTREACHED *\\*/", 
     "/\\* *FALL *THRO?UG?H? *\\*/",
     /* Bare C keywords */
-
-    "[:<:]break[:>:]",  "[:<:]case[:>:]","[:<:]continue[:>:]", 
-    "[:<:]default[:>:]", "[:<:]do[:>:]", "[:<:]else[:>:]", 
-    "[:<:]enum[:>:]", "[:<:]if[:>:]", "[:<:]goto[:>:]",
-    "[:<:]return[:>:]", "[:<:]switch[:>:]", "[:<:]while[:>:]",
-    "[:<:]enum[:>:]", "[:<:]int[:>:]", "[:<:]long[:>:]",
-    "[:<:]short[:>:]", "[:<:]static[:>:]", "[:<:]struct[:>:]",
-    "[:<:]typedef[:>:]", "[:<:]union[:>:]", "[:<:]void[:>:]",
+    " break ",  " case "," continue ", 
+    " default ", " do ", " else ", 
+    " enum ", " if ", " goto ",
+    " return ", " switch ", " while ",
+    " enum ", " int ", " long ",
+    " short ", " static ", " struct ",
+    " typedef ", " union ", " void ",
     /* Preprocessor constructs */
-    "/* *define","# *endif","# *else","# *if[:>:]",
-    "# *ifdef[:>:]","# *ifndef[:>:]",
+    "# *define","# *endif","# *else","# *if ", "# *ifdef ","# *ifndef ",
     /* Comment delimiters with no content, and line continuation */
     "/\\*+", "\\*+/", "\\*+", "^ *\\s*\\n", "\\\n",
     /* Common preprocessor macros, not significant by themselves. */
-    "[:<:]ASSERT[:>:]", 
-    "[:<:]FALSE[:>:]", 
-    "[:<:]NULL[:>:]",
-    "[:<:]STATIC[:>:]", 
-    "[:<:]TRUE[:>:]",
+    " ASSERT ", 
+    " FALSE ", 
+    " NULL ",
+    " STATIC ", 
+    " TRUE ",
     /* Macro include lines are noise, too. */
     "\\s*#include.*","#\\s*line.*",
     /* Common error macros. */
-    "[:<:]EFAULT[:>:]",
-    "[:<:]EINVAL[:>:]",
-    "[:<:]ENOSYS[:>:]",
+    " EFAULT ",
+    " EINVAL ",
+    " ENOSYS ",
 };
 static regex_t c_regexps[sizeof(c_patterns)/sizeof(*c_patterns)];
 
@@ -53,10 +51,10 @@ static char *shell_patterns[] = {
     /* Idioms that don't convey any meaning in isolation */
     "exit *[01];?",
     /* Bare shell keywords */
-    "[:<:]break[:>:]", "[:<:]case[:>:]", "[:<:]done[:>:]", 
-    "[:<:]do[:>:]", "[:<:]else[:>:]", "[:<:]esac[:>:]",
-    "[:<:]fi[:>:]", "[:<:]if[:>:]", "[:<:]return[:>:]",
-    "[:<:]shift[:>:]", "[:<:]true[:>:]", "[:<:]while[:>:]", 
+    " break ", " case ", " done ", 
+    " do ", " else ", " esac ",
+    " fi ", " if ", " return ",
+    " shift ", " true ", " while ", 
     /* Blank comment */
     "^#\n",
 };
@@ -64,7 +62,7 @@ static regex_t shell_regexps[sizeof(shell_patterns)/sizeof(*shell_patterns)];
 
 static char shell_punct[] = {
     /* Punctuation and whitespace */
-    ':', ';', '\n','\t', '\v',
+    ' ', ':', ';', '\n','\t', '\v',
 };
 
 void filter_init(void)
@@ -89,7 +87,9 @@ void filter_init(void)
 }
 
 static filtertype	active = none;
-static char *punct, **patterns;
+static char *punct;
+static regex_t *regexps;
+static int nregexps;
 
 void filter_set(filtertype newfilter)
 {
@@ -97,13 +97,15 @@ void filter_set(filtertype newfilter)
 
     if (newfilter == c)
     {
-	patterns = c_patterns;
+	regexps = c_regexps;
 	punct = c_punct;
+	nregexps = sizeof(c_regexps)/sizeof(*c_regexps);
     }
     else
     {
-	patterns = shell_patterns;
+	regexps = shell_regexps;
 	punct = shell_punct;
+	nregexps = sizeof(shell_regexps)/sizeof(*shell_regexps);
     }
 }
 
@@ -111,38 +113,50 @@ int filter_pass(const char *line)
 {
     if (active == none)
 	return(1);
+    else if (strspn(line, punct) == strlen(line))
+	return(0);
     else
     {
-	char	buf[2][BUFSIZ];
-	int	changed , i, flip = 0;
-
-	buf[0][0] = ' ';
-	strncpy(buf[0]+1, line, BUFSIZ-2);
+	char	*sp, *tp;
+	char	buf[BUFSIZ];
+	int	changed;
+
+	/* change all punctuation to spaces */
+	buf[0] = ' ';
+	for (sp = (char *)line, tp = buf+1; *sp; sp++)
+	    if (strchr(punct, *sp) && *sp != ' ')
+		*tp++ = ' ';
+	    else
+		*tp++ = *sp;
+	*tp = '\0';
 
 	do {
-	    char *s, *t, *sp, *tp;
+	    regex_t	*re;
 
-	    s = buf[flip]; t = buf[!flip]; flip ^= 1; 
 	    changed = 0;
 
-	    /* remove all punctuation */
-	    for (sp = s, tp = t; *sp; sp++)
-		if (strchr(punct, *sp))
-		    changed++;
-	        else
-		    *tp++ = *sp;
-	    *tp = '\0';
-
 	    /* replace all regexps with the empty string */
+	    for (re = regexps; re < regexps + nregexps; re++)
+	    {
+		regmatch_t f;
 
-	    if (t[0] == '\0')
-		return(0);
-	} while 
+		if (!regexec(re, buf, 1, &f, 0))
+		{
+		    char	*start, *end;
+
+		    for (start = buf+f.rm_so, end = buf+f.rm_eo; *end; end++)
+			*start++ = *end;
+		    *start = '\0'; 
+		    changed++;
+		}
+	    }
+	} while
 	    (changed);
-	return(1);
+	return !(buf[0] == '\0' || strspn(buf, " ") == strlen(buf));
     }
 }
 
+#ifdef TEST
 int main(int argc, char *argv[])
 {
     char	buf[BUFSIZ];
@@ -158,5 +172,6 @@ int main(int argc, char *argv[])
 	fputs(buf, stdout);
     }
 }
+#endif /* TEST */
 
 /* filter.c ends here */
-- 
2.43.0