xargs.c: implement -p, -P and -0, add TODO for -L

From 2350f520a6dd7e293c5505aaa0983853cdd41ee6 Mon Sep 17 00:00:00 2001
From: Hiltjo Posthuma <hiltjo@codemadness.org>
Date: Thu, 31 Jul 2025 14:40:43 +0200
Subject: [PATCH 3/4] xargs.c: implement -p, -P and -0, add TODO for -L

- Add option to read arguments separated by NUL (-0).
  Useful with find -print0 for example.
- Add very useful parallel option (-P).
  POSIX vaguely mentions parallel operations, but this is commonly supported
  and very useful.  For example OpenBSD xargs supports it since at least 2003.
  GNU xargs since at least 1996.
- Add prompt option (-p), (POSIX).
- Add a TODO for xargs -L (POSIX extension).
- Documentation and man page improvements.

TODO

@@ -85,3 +85,8 @@ tr
 sbase-box
 ---------
 * List of commands does not contain `install` (only `xinstall`).
+
+
+xargs
+-----
+* Add -L.

xargs.1

@@ -1,4 +1,4 @@
-.Dd July 30, 2023
+.Dd July 30, 2025
 .Dt XARGS 1
 .Os sbase
 .Sh NAME
@@ -6,10 +6,11 @@
 .Nd construct argument lists and execute command
 .Sh SYNOPSIS
 .Nm
-.Op Fl rtx
+.Op Fl 0prtx
 .Op Fl E Ar eofstr
 .Op Fl I Ar replstr
 .Op Fl n Ar num
+.Op Fl P Ar maxprocs
 .Op Fl s Ar num
 .Op Ar cmd Op Ar arg ...
 .Sh DESCRIPTION
@@ -26,7 +27,7 @@ stdin.
 The command is repeatedly executed one or more times until stdin is exhausted.
 .Pp
 Spaces, tabs and newlines may be embedded in arguments using single (`'')
-or double (`"') quotes or backslashes ('\\').
+or double (`"') quotes or backslashes ('\e').
 Single quotes escape all non-single quote characters, excluding newlines, up
 to the matching single quote.
 Double quotes escape all non-double quote characters, excluding newlines, up
@@ -34,13 +35,12 @@ to the matching double quote.
 Any single character, including newlines, may be escaped by a backslash.
 .Sh OPTIONS
 .Bl -tag -width Ds
-.It Fl n Ar num
-Use at most
-.Ar num
-arguments per command line.
-.It Fl r
-Do not run the command if there are no arguments.
-Normally the command is executed at least once even if there are no arguments.
+.It Fl 0
+Change
+.Nm
+to expect NUL ('\e0') characters as separators, instead of spaces
+and newlines.
+The quoting mechanisms described above are not performed.
 .It Fl E Ar eofstr
 Use
 .Ar eofstr
@@ -51,11 +51,32 @@ Use
 as the placeholder for the argument.
 Sets the arguments count to 1 per command line.
 It also implies the option x.
+.It Fl n Ar num
+Use at most
+.Ar num
+arguments per command line.
+.It Fl p
+Prompt mode: the user is asked whether to execute
+.Ar cmd
+at each invocation.
+Trace mode (-t) is turned on to write the command instance to be executed,
+followed by a prompt to standard error.
+An affirmative response read from
+.Pa /dev/tty
+executes the command, otherwise it is skipped.
+.It Fl P Ar maxprocs
+Parallel mode: run at most maxprocs invocations of
+.Ar cmd
+at once.
+.It Fl r
+Do not run the command if there are no arguments.
+Normally the command is executed at least once even if there are no arguments.
 .It Fl s Ar num
 Use at most
 .Ar num
 bytes per command line.
 .It Fl t
+Enable trace mode.
 Write the command line to stderr before executing it.
 .It Fl x
 Terminate if the command line exceeds the system limit or the number of bytes
@@ -95,9 +116,7 @@ The
 .Nm
 utility is compliant with the
 .St -p1003.1-2013
-specification except from the
-.Op Fl p
-flag.
+specification.
 .Pp
 The
 .Op Fl r

xargs.c

@@ -19,14 +19,15 @@ static int eatspace(void);
 static int parsequote(int);
 static int parseescape(void);
 static char *poparg(void);
-static void waitchld(void);
+static void waitchld(int);
 static void spawn(void);
 
 static size_t argbsz;
 static size_t argbpos;
-static size_t maxargs = 0;
-static int    nerrors = 0;
-static int    rflag = 0, nflag = 0, tflag = 0, xflag = 0, Iflag = 0;
+static size_t maxargs;
+static size_t curprocs, maxprocs = 1;
+static int    nerrors;
+static int    nulflag, nflag, pflag, rflag, tflag, xflag, Iflag;
 static char  *argb;
 static char  *cmd[NARGS];
 static char  *eofstr;
@@ -59,10 +60,7 @@ eatspace(void)
 	int ch;
 
 	while ((ch = inputc()) != EOF) {
-		switch (ch) {
-		case ' ': case '\t': case '\n':
-			break;
-		default:
+		if (nulflag || !(ch == ' ' || ch == '\t' || ch == '\n')) {
 			ungetc(ch, stdin);
 			return ch;
 		}
@@ -129,6 +127,10 @@ poparg(void)
 			if (parseescape() < 0)
 				eprintf("backslash at EOF\n");
 			break;
+		case '\0':
+			/* NUL separator: no escaping */
+			if (nulflag)
+				goto out;
 		default:
 		fill:
 			fillargbuf(ch);
@@ -143,22 +145,55 @@ out:
 }
 
 static void
-waitchld(void)
+waitchld(int waitall)
 {
+	pid_t pid;
 	int status;
 
-	wait(&status);
-	if (WIFEXITED(status)) {
-		if (WEXITSTATUS(status) == 255)
-			exit(124);
-		if (WEXITSTATUS(status) == 127 ||
-		    WEXITSTATUS(status) == 126)
-			exit(WEXITSTATUS(status));
-		if (status)
-			nerrors++;
+	while ((pid = waitpid(-1, &status, !waitall && curprocs < maxprocs ?
+	       WNOHANG : 0)) > 0) {
+	       curprocs--;
+
+		if (WIFEXITED(status)) {
+			if (WEXITSTATUS(status) == 255)
+				exit(124);
+			if (WEXITSTATUS(status) == 127 ||
+			    WEXITSTATUS(status) == 126)
+				exit(WEXITSTATUS(status));
+			if (WEXITSTATUS(status))
+				nerrors++;
+		}
+		if (WIFSIGNALED(status))
+			exit(125);
 	}
-	if (WIFSIGNALED(status))
-		exit(125);
+	if (pid == -1 && errno != ECHILD)
+		eprintf("waitpid:");
+}
+
+static int
+prompt(void)
+{
+	FILE *fp;
+	int ch, ret;
+
+	if (!(fp = fopen("/dev/tty", "r")))
+		return -1;
+
+	fputs("?...", stderr);
+	fflush(stderr);
+
+	ch = fgetc(fp);
+	ret = (ch == 'y' || ch == 'Y');
+	if (ch != EOF && ch != '\n') {
+		while ((ch = fgetc(fp)) != EOF) {
+			if (ch == '\n')
+				break;
+		}
+	}
+
+	fclose(fp);
+
+	return ret;
 }
 
 static void
@@ -168,16 +203,25 @@ spawn(void)
 	int first = 1;
 	char **p;
 
-	if (tflag) {
+	if (pflag || tflag) {
 		for (p = cmd; *p; p++) {
 			if (!first)
 				fputc(' ', stderr);
 			fputs(*p, stderr);
 			first = 0;
 		}
+		if (pflag) {
+			switch (prompt()) {
+			case -1: break; /* error */
+			case 0: return; /* no */
+			case 1: goto dospawn; /* yes */
+			}
+		}
 		fputc('\n', stderr);
+		fflush(stderr);
 	}
 
+dospawn:
 	switch (fork()) {
 	case -1:
 		eprintf("fork:");
@@ -187,13 +231,14 @@ spawn(void)
 		weprintf("execvp %s:", *cmd);
 		_exit(126 + (savederrno == ENOENT));
 	}
-	waitchld();
+	curprocs++;
+	waitchld(0);
 }
 
 static void
 usage(void)
 {
-	eprintf("usage: %s [-rtx] [-E eofstr] [-n num] [-s num] "
+	eprintf("usage: %s [-0prtx] [-E eofstr] [-n num] [-P maxprocs] [-s num] "
 	        "[cmd [arg ...]]\n", argv0);
 }
 
@@ -212,10 +257,16 @@ main(int argc, char *argv[])
 	argmaxsz -= 4096;
 
 	ARGBEGIN {
+	case '0':
+		nulflag = 1;
+		break;
 	case 'n':
 		nflag = 1;
 		maxargs = estrtonum(EARGF(usage()), 1, MIN(SIZE_MAX, LLONG_MAX));
 		break;
+	case 'p':
+		pflag = 1;
+		break;
 	case 'r':
 		rflag = 1;
 		break;
@@ -238,6 +289,9 @@ main(int argc, char *argv[])
 		maxargs = 1;
 		replstr = EARGF(usage());
 		break;
+	case 'P':
+		maxprocs = estrtonum(EARGF(usage()), 1, MIN(SIZE_MAX, LLONG_MAX));
+		break;
 	default:
 		usage();
 	} ARGEND
@@ -295,6 +349,8 @@ main(int argc, char *argv[])
 
 	free(argb);
 
+	waitchld(1);
+
 	if (nerrors || (fshut(stdin, "<stdin>") | fshut(stdout, "<stdout>")))
 		ret = 123;