Audit col(1), add UTF-8-support

Nothing special here, only renaming of variables and adding the
Rune-utility-functions.
Also, I refactored the manpage.

README

@@ -19,7 +19,7 @@ The following tools are implemented ('*' == finished, '#' == UTF-8 support,
 =*| cksum           yes                          none
 =*| cmp             yes                          none
 #*| cols            non-posix                    none
-    col             yes                          none
+#*| col             non-posix                    none
 =*| comm            yes                          none
 =*| cp              yes                          none (-i)
 =*| cron            non-posix                    none

col.1

@@ -1,4 +1,4 @@
-.Dd March 6, 2014
+.Dd March 22, 2014
 .Dt COL 1
 .Os sbase
 .Sh NAME
@@ -9,20 +9,15 @@
 .Op Fl bfpx
 .Op Fl l Ar num
 .Sh DESCRIPTION
-The
 .Nm
-utility filters all the reverse (and half reverse) line feeds,
+filters all reverse (and half reverse) line feeds,
-as they are produced by
+as produced by
 .Xr nroff 1
-with .2C of
+with .2C,
 .Xr ms 6
-or by
+or
 .Xr tbl 1 .
-.Nm
+The recognized control sequences are:
-also replaces spaces by tabs when it is possible.
-The control sequences managed by
-.Nm
-are:
 .Bl -tag -width Ds
 .It ESC-7
 Reverse line-feed
@@ -42,25 +37,24 @@ Carriage return
 New line
 .El
 .Pp
-All the other control codes and escape sequences are removed.
+All other control codes and escape sequences are removed.
 .Nm
-transforms all the spaces into tabulators.
+converts all spaces to tabs.
 .Sh OPTIONS
 .Bl -tag -width Ds
 .It Fl p
-Print unknown escape sequences to the output.
+Print unknown escape sequences.
 .It Fl b
-Do not print backspaces in output,
+Do not print backspaces and instead only print the last
-and print only the last overstriked character in the output.
+character written to each column position.
 .It Fl f
 Allow forward half line feeds in the output.
 .It Fl x
-Do not convert spaces in tabulators.
+Do not convert spaces to tabs.
 .It Fl l Ar num
-Increment to
+Buffer
 .Ar num
-the number of lines buffered for
+lines in memory.
-.Nm
 .El
 .Sh SEE ALSO
 .Xr nroff 1 ,
@@ -68,13 +62,9 @@ the number of lines buffered for
 .Xr ms 6
 .Sh BUGS
 .Nm
-only process text with a maximum of 256 lines with 800 bytes per line,
+only buffers up to 256 lines with up to 800 bytes per line
-although the number of lines can be modified with the
+if the line-number hasn't been set differently with the
-.Fl l
+.Op Fl l
-option.
+flag.
-When the number of lines is bigger,
+When the number of lines is bigger, the buffer is flushed and
-the buffer is flushed to the output,
+reverse line feeds can not operate on the flushed lines.
-so new reverse line feeds can not operate in the flushed  lines.
-This implementation ignores SI and SO selection character sets,
-because it is supposed to work only with UTF-8 strings,
-although the UTF-8 support is missed.

col.c

@@ -1,47 +1,47 @@
 /* See LICENSE file for copyright and license details. */
+#include <limits.h>
 #include <stdio.h>
 #include <stdint.h>
 #include <stdlib.h>
 #include <string.h>
-#include <ctype.h>
 
+#include "utf.h"
 #include "util.h"
 
 #define NLINES 256
 #define NCOLS 800
 
-static char **buff;
+static Rune **buf;
 
-static int obackspace, onotabs, ohalfline, oescape;
+static int    backspace, notabs, halfline, escape;
-static unsigned nline, ncol, nchar, nspaces, maxline, bs;
+static size_t nline, ncol, nchar, nspaces, maxline, bs, pagesize = NLINES;
-static size_t pagsize = NLINES;
 
 static void
 flush(void)
 {
-	int c;
+	Rune c;
-	unsigned i, j;
+	size_t i, j;
 
 	for (i = 0; i < maxline; ++i) {
-		for (j = 0; j < NCOLS && (c = buff[i][j]) != '\0'; ++j)
+		for (j = 0; j < NCOLS && (c = buf[i][j]); ++j)
-			putchar(c);
+			efputrune(&c, stdout, "<stdout>");
 		putchar('\n');
 	}
 	bs = nchar = nline = ncol = 0;
 }
 
 static void
-forward(unsigned n)
+forward(size_t n)
 {
-	unsigned lim;
+	size_t lim;
 
-	for (lim = ncol + n; ncol != lim && nchar < NCOLS-1; ++nchar) {
+	for (lim = ncol + n; ncol != lim && nchar < NCOLS - 1; ++nchar) {
-		switch (buff[nline][nchar]) {
+		switch (buf[nline][nchar]) {
 		case '\b':
 			--ncol;
 			break;
 		case '\0':
-			buff[nline][nchar] = ' ';
+			buf[nline][nchar] = ' ';
 			/* FALLTHROUGH */
 		default:
 			++ncol;
@@ -53,18 +53,17 @@ forward(unsigned n)
 static void
 linefeed(int up, int rcarriage)
 {
-	unsigned oncol = ncol;
+	size_t oncol = ncol;
 
 	nspaces = 0;
 	if (up > 0) {
-		if (nline == pagsize-1) {
+		if (nline == pagesize - 1) {
 			flush();
 		}  else {
 			if (++nline > maxline)
 				maxline = nline;
 		}
-	} else {
+	} else if (nline > 0) {
-		if (nline > 0)
 		--nline;
 	}
 	bs = 0;
@@ -75,9 +74,9 @@ linefeed(int up, int rcarriage)
 }
 
 static void
-newchar(int c)
+newchar(Rune c)
 {
-	char *cp;
+	Rune *cp;
 
 	forward(nspaces);
 	nspaces = 0;
@@ -90,7 +89,7 @@ newchar(int c)
 		nchar = ncol = 0;
 		break;
 	case '\t':
-		forward(8 - ncol%8);
+		forward(8 - ncol % 8);
 		break;
 	case '\b':
 		if (ncol > 0)
@@ -100,20 +99,18 @@ newchar(int c)
 		bs = 1;
 		break;
 	default:
-		cp = &buff[nline][nchar];
+		cp = &buf[nline][nchar];
-		if (*cp != '\0' && *cp != ' ' && bs && !obackspace) {
+		if (*cp && *cp != ' ' && bs && !backspace && nchar != NCOLS - 3) {
-			if (nchar != NCOLS-3) {
+			memmove(cp + 3, cp + 1, (NCOLS - nchar - 2) * sizeof(*cp));
-				memmove(cp + 3, cp + 1, NCOLS - nchar - 2);
 			cp[1] = '\b';
 			nchar += 2;
 		}
-		}
+		if (nchar != NCOLS - 1) {
-		if (nchar != NCOLS-1) {
+			for (cp = buf[nline]; cp < &buf[nline][nchar]; ++cp) {
-			for (cp = buff[nline]; cp < &buff[nline][nchar]; ++cp) {
 				if (*cp == '\0')
 					*cp = ' ';
 			}
-			buff[nline][nchar++] = c;
+			buf[nline][nchar++] = c;
 			++ncol;
 		}
 		bs = 0;
@@ -123,50 +120,52 @@ newchar(int c)
 static void
 col(void)
 {
-	int c;
+	Rune r;
+	int ret;
 
-	while ((c = getchar()) != EOF) {
+	while (efgetrune(&r, stdin, "<stdin>")) {
-		switch (c) {
+		switch (r) {
 		case '\x1b':
-			switch (c = getchar()) {
+			ret = efgetrune(&r, stdin, "<stdin>");
+			switch (r) {
 			case '8': /* reverse half-line-feed */
 			case '7': /* reverse line-feed */
 				linefeed(-1, 0);
 				continue;
 			case '9':  /* forward half-line-feed */
-				if (ohalfline)
+				if (halfline)
 					break;
 				linefeed(1, 0);
 				continue;
 			}
-			if (!oescape)
+			if (!escape)
 				continue;
 			newchar('\x1b');
-			if (c != EOF)
+			if (ret)
-				newchar(c);
+				newchar(r);
 			break;
 		case '\v':
 			linefeed(-1, 0);
 			break;
 		case ' ':
-			if (!onotabs) {
+			if (!notabs) {
 				if (++nspaces != 8)
 					continue;
-				c = '\t';
+				r = '\t';
 				nspaces = 0;
 			}
 			/* FALLTHROUGH */
 		case '\r':
 		case '\b':
 		case '\t':
-			newchar(c);
+			newchar(r);
 			break;
 		case '\n':
 			linefeed(1, 1);
 			break;
 		default:
-			if (!iscntrl(c))
+			if (!iscntrlrune(r))
-				newchar(c);
+				newchar(r);
 			break;
 		}
 	}
@@ -175,17 +174,17 @@ col(void)
 static void
 allocbuf(void)
 {
-	char **bp;
+	Rune **bp;
 
-	buff = ereallocarray(NULL, pagsize, sizeof(*buff));
+	buf = ereallocarray(NULL, pagesize, sizeof(*buf));
-	for (bp = buff; bp < &buff[pagsize]; ++bp)
+	for (bp = buf; bp < buf + pagesize; ++bp)
-		*bp = emalloc(NCOLS);
+		*bp = ereallocarray(NULL, NCOLS, sizeof(**buf));
 }
 
 static void
 usage(void)
 {
-	enprintf(2, "usage: %s [-p][-l num][-b][-f][-x]\n", argv0);
+	enprintf(2, "usage: %s [-pbfx] [-l num]\n", argv0);
 }
 
 int
@@ -193,35 +192,30 @@ main(int argc, char *argv[])
 {
 	ARGBEGIN {
 	case 'b':
-		obackspace = 1;
+		backspace = 1;
 		break;
 	case 'f':
-		ohalfline = 1;
+		halfline = 1;
 		break;
 	case 'l':
-		pagsize = estrtonum(EARGF(usage()), 0, SIZE_MAX);
+		pagesize = estrtonum(EARGF(usage()), 0, MIN(SIZE_MAX, LLONG_MAX));
 		break;
 	case 'p':
-		oescape = 1;
+		escape = 1;
 		break;
 	case 'x':
-		onotabs = 1;
+		notabs = 1;
 		break;
 	default:
 		usage();
 	} ARGEND;
 
-	if (argc > 0)
+	if (argc)
 		usage();
 
 	allocbuf();
 	col();
 	flush();
 
-	if (ferror(stdin))
-		enprintf(1, "error reading input");
-	if (ferror(stdout))
-		enprintf(2, "error writing output");
-
 	return 0;
 }