Support wrapping with arbitrary I/O channels

Issue: #78

core-tools/src/dgsh-wrap.1

@@ -18,10 +18,17 @@
 dgsh-wrap \- allow any program to participate in an dgsh pipeline
 .SH SYNOPSIS
 \fBdgsh-wrap\fP
-[\fB-S\fP] [\fB-deImO\fP] \fIprogram\fP [\fIprogram-arguments\fP ...]
+[\fB-S\fP]
+[\fB-i\fP \fB0\fP|\fBa\fP]
+[\fB-o\fP \fB0\fP|\fBa\fP]
+[\fB-eIO\fP]
+\fIprogram\fP [\fIprogram-arguments\fP ...]
 
 #!/usr/libexec/dgsh/\fBdgsh-wrap\fP
-\fB-s\fP [\fB-deImO\fP] [\fIprogram-arguments\fP ...]
+\fB-s\fP
+[\fB-i\fP \fB0\fP|\fBa\fP]
+[\fB-o\fP \fB0\fP|\fBa\fP]
+[\fB-eIO\fP] [\fIprogram-arguments\fP ...]
 
 .SH DESCRIPTION
 \fIdgsh-wrap\fP takes as arguments an absolute path or the name
@@ -49,32 +56,42 @@ wrapped with \fIdgsh-wrap\fP in order to communicate their particular
 I/O requirements.
 
 .SH OPTIONS
-.IP "\fB\-d\fP
-The wrapped program is deaf, it will not read any input.
-
 .IP "\fB\-e\fP
 Replace \fI<|\fP and \fI>|\fP strings embedded within arguments,
 with names of input file descriptor paths.
 By default, only standalone arguments are thus replaced.
 
+.IP "\fB\-i\fP \fB0\fP|\fBa\fP
+Specify the wrapped program's number of input channels.
+The \fB0\fP character specifies that the program does not read any input.
+The \fBa\fP character specifies that the program can read an arbitrary
+number of input streams;
+these will be automatically supplied by \fIdgsh-wrap\fP as file descriptor
+command line arguments.
+
 .IP "\fB\-I\fP
 Do not include the standard input to the command line arguments,
 when replacing \fI<|\fP arguments with names of input file descriptor paths.
 This will result in the standard input becoming available to the
 command as its standard input, rather than as a command line argument.
-Without this option, the first path with be \fI/proc/self/fd/0\fP.
+Without this option, the first path with be \fI/dev/fd/0\fP.
 When this option is given, the program will require one input channel
 more than those specified by the \fI<|\fP arguments.
 
-.IP "\fB\-m\fP
-The wrapped program is mute, it will not produce any output.
+.IP "\fB\-o\fP \fB0\fP|\fBa\fP
+Specify the wrapped program's number of output channels.
+The \fB0\fP character specifies that the program does not produce any output.
+The \fBa\fP character specifies that the program can write to an arbitrary
+number of output streams;
+these will be automatically supplied by \fIdgsh-wrap\fP as file descriptor
+command line arguments.
 
 .IP "\fB\-O\fP
 Do not include the standard output to the command line arguments,
 when replacing \fI>|\fP arguments with names of output file descriptor paths.
 This will result in the standard output becoming available to the
 command as its standard output, rather than as a command line argument.
-Without this option the first path with be \fI/proc/self/fd/1\fP.
+Without this option the first path with be \fI/dev/fd/1\fP.
 When this option is given, the program will require one output channel
 more than those specified by the \fI>|\fP arguments.
 
@@ -117,7 +134,7 @@ Wrap the \fIecho\fP command, specifying that it accepts no input.
 .ft C
 .ps -1
 .nf
-dgsh-wrap -d echo hi
+dgsh-wrap -i 0 echo hi
 .fi
 .ps +1
 .ft P
@@ -126,7 +143,7 @@ Wrap the \fIcp\fP command, specifying that it does not perform any I/O.
 .ft C
 .ps -1
 .nf
-dgsh-wrap -dm cp src-file dest-file
+dgsh-wrap -i 0 -o 0 cp src-file dest-file
 .fi
 .ps +1
 .ft P
@@ -135,7 +152,7 @@ Wrap the \fIpaste\fP command, supplying its two arguments.
 .ft C
 .ps -1
 .nf
-dgsh-wrap -I /usr/bin/paste "<|" "<|"
+dgsh-wrap /usr/bin/paste "<|" "<|"
 .fi
 .ps +1
 .ft P
@@ -145,7 +162,17 @@ input and uses an additional input argument.
 .ft C
 .ps -1
 .nf
-dgsh-wrap /usr/bin/paste - "<|"
+dgsh-wrap -I /usr/bin/paste - "<|"
+.fi
+.ps +1
+.ft P
+.PP
+Wrap the \fIpaste\fP command, so that it will process all input channels
+provided to it.
+.ft C
+.ps -1
+.nf
+dgsh-wrap -i a /usr/bin/paste
 .fi
 .ps +1
 .ft P

core-tools/src/dgsh-wrap.c

@@ -19,8 +19,8 @@
 
 /*
  * Examples:
- * dgsh-wrap -d yes | fsck
- * tar cf - / | dgsh-wrap -m dd of=/dev/st0
+ * dgsh-wrap -i 0 yes | fsck
+ * tar cf - / | dgsh-wrap -o 0 dd of=/dev/st0
  * ls | dgsh-wrap /usr/bin/sort -k5n | more
  */
 
@@ -46,11 +46,12 @@
 static void
 usage(void)
 {
-	fputs("Usage:\tdgsh-wrap [-S] [-dImO] program [program-arguments ...]\n"
-		"\tdgsh-wrap -s [-diImoO] [program-arguments ...]\n"
-		"-d\t"		"Requires no input (deaf)\n"
+	fputs("Usage:\tdgsh-wrap [-S] [-i 0|a] [-o 0|a] [-eIO] program [program-arguments ...]\n"
+		"\tdgsh-wrap -s [-i 0|a] [-o 0|a] [-eIO] [program-arguments ...]\n"
+		"-e\t"		"Process <| and >| embedded in arguments\n"
+		"-i 0|a\t"	"Process no (0) or arbitrary (a) input channels\n"
 		"-I\t"		"Do not provide standard input as a <| arg\n"
-		"-m\t"		"Provides no output; (mute)\n"
+		"-o 0|a\t"	"Process no (0) or arbitrary (a) output channels\n"
 		"-O\t"		"Do not provide standard output as a >| arg\n"
 		"-S\t"		"Process flags and program as a #! interpreter\n"
 		"-s\t"		"Process flags as a #! interpreter\n"
@@ -274,17 +275,34 @@ process_embedded_io_arg(char **arg, const char *special, int **fdptr)
  * Replace an instance of the string special (e.g. "<|") matching an arg
  * with /dev/fd/N, where N is the integer pointed
  * by fdptr, and increase fdptr to point to the next integer.
+ * If special is null, then the replacement is always made.
  */
 static void
 process_standalone_io_arg(char **arg, const char *special, int **fdptr)
 {
-	if (strcmp(*arg, special) != 0)
+	if (special != NULL && strcmp(*arg, special) != 0)
 		return;
 	if (asprintf(arg, "/dev/fd/%d", **fdptr) == -1)
 		err(1, "asprintf out of memory");
 	(*fdptr)++;
 }
 
+/*
+ * Increment the channels specified by the given variable.
+ * Ensure that the corresponding variable is not
+ * already set to an arbitrary number of channels.
+ */
+static void
+increment_channels(int *var)
+{
+	if (*var == -1) {
+		fputs("I/O channel arguments cannot be combined with an arbitrary I/O file specification\n",
+				stderr);
+		exit(1);
+	}
+	(*var)++;
+}
+
 int
 main(int argc, char *argv[])
 {
@@ -299,6 +317,7 @@ main(int argc, char *argv[])
 	bool embedded_args = false;
 	/* Pass stdin/stdout as a command-line argument */
 	bool stdin_as_arg = true, stdout_as_arg = true;
+	bool supply_input_args = false, supply_output_args = false;
 
 
 	debug_level = getenv("DGSH_DEBUG_LEVEL");
@@ -324,12 +343,18 @@ main(int argc, char *argv[])
 	 * first non-flag argument.
 	 * Therefore, adjust argc, argv on entry and optind on exit.
 	 */
-        while ((ch = getopt(argc, argv, "+deImOSs")) != -1) {
+        while ((ch = getopt(argc, argv, "+ei:Io:OSs")) != -1) {
 		DPRINTF(4, "getopt switch=%c", ch);
 		switch (ch) {
-		case 'd':
+		case 'i':
 			nflags++;
-			ninputs = 0;
+			if (strcmp(optarg, "0") == 0)
+				ninputs = 0;
+			else if (strcmp(optarg, "a") == 0) {
+				ninputs = -1;
+				supply_input_args = true;
+			} else
+				usage();
 			break;
 		case 'e':
 			embedded_args = true;
@@ -339,9 +364,15 @@ main(int argc, char *argv[])
 			stdin_as_arg = false;
 			nflags++;
 			break;
-		case 'm':
+		case 'o':
 			nflags++;
-			noutputs = 0;
+			if (strcmp(optarg, "0") == 0)
+				noutputs = 0;
+			else if (strcmp(optarg, "a") == 0) {
+				noutputs = -1;
+				supply_output_args = true;
+			} else
+				usage();
 			break;
 		case 'O':
 			stdout_as_arg = false;
@@ -406,14 +437,14 @@ main(int argc, char *argv[])
 	for (i = optind + 1; i < argc; i++) {
 		if (embedded_args) {
 			for (p = argv[i]; p = strstr(p, "<|"); p += 2)
-				ninputs++;
+				increment_channels(&ninputs);
 			for (p = argv[i]; p = strstr(p, ">|"); p += 2)
-				noutputs++;
+				increment_channels(&noutputs);
 		} else {
 			if (strcmp(argv[i], "<|") == 0)
-				ninputs++;
+				increment_channels(&ninputs);
 			if (strcmp(argv[i], ">|") == 0)
-				noutputs++;
+				increment_channels(&noutputs);
 		}
 	}
 	/*
@@ -434,20 +465,57 @@ main(int argc, char *argv[])
 					&input_fds, &output_fds);
 
 	/*
-	 * Substitute special arguments "<|" and ">|" with file descriptor
+	 * Substitute special arguments "<|" and ">|" with or add file descriptor
 	 * paths /dev/fd/N using the fds received from negotiation.
 	 */
 	int *inptr = stdin_as_arg ? input_fds : input_fds + 1;
+	if (supply_input_args) {
+		if (!stdin_as_arg)
+			ninputs--;
+
+		/* Create space for arguments to add */
+		char **nargv = xmalloc((argc + ninputs + 1) * sizeof(char *));
+		memcpy(nargv, argv, argc * sizeof(char *));
+		memset(argv + argc, 0, (ninputs + 1) * sizeof(char *));
+		argv = nargv;
+
+		/* Add arguments */
+		for (i = argc; i < argc + ninputs; i++)
+			process_standalone_io_arg(&argv[i], NULL, &inptr);
+		argc += ninputs;
+		argv[argc] = NULL;
+	} else {
+		for (i = optind + 1; i < argc; i++) {
+			if (embedded_args)
+				while (process_embedded_io_arg(&argv[i], "<|", &inptr))
+					;
+			else
+				process_standalone_io_arg(&argv[i], "<|", &inptr);
+		}
+	}
 	int *outptr = stdout_as_arg ? output_fds : output_fds + 1;
-	for (i = optind + 1; i < argc; i++) {
-		if (embedded_args) {
-			while (process_embedded_io_arg(&argv[i], "<|", &inptr))
-				;
-			while (process_embedded_io_arg(&argv[i], ">|", &outptr))
-				;
-		} else {
-			process_standalone_io_arg(&argv[i], "<|", &inptr);
-			process_standalone_io_arg(&argv[i], ">|", &outptr);
+	if (supply_output_args) {
+		if (!stdout_as_arg)
+			noutputs--;
+
+		/* Create space for arguments to add */
+		char **nargv = xmalloc((argc + noutputs + 1) * sizeof(char *));
+		memcpy(nargv, argv, argc * sizeof(char *));
+		memset(argv + argc, 0, (noutputs + 1) * sizeof(char *));
+		argv = nargv;
+
+		/* Add arguments */
+		for (i = argc; i < argc + noutputs; i++)
+			process_standalone_io_arg(&argv[i], NULL, &outptr);
+		argc += noutputs;
+		argv[argc] = NULL;
+	} else {
+		for (i = optind + 1; i < argc; i++) {
+			if (embedded_args)
+				while (process_embedded_io_arg(&argv[i], ">|", &outptr))
+					;
+			else
+				process_standalone_io_arg(&argv[i], ">|", &outptr);
 		}
 	}
 	DPRINTF(4, "Arguments to execvp after substitung <| and >|");

core-tools/tests-regression/dgsh-wrap/paste3.ok

@@ -0,0 +1 @@
+0	1

core-tools/tests-regression/dgsh-wrap/paste4.ok

@@ -0,0 +1 @@
+0	1

core-tools/tests-regression/dgsh-wrap/tee1.ok

@@ -0,0 +1,2 @@
+ahi
+bhi

core-tools/tests-regression/dgsh-wrap/tee2.ok

@@ -0,0 +1,2 @@
+ahi
+bhi

core-tools/tests-regression/test-wrap.sh

@@ -27,17 +27,17 @@ ensure_same()
 export PATH="$PATH:bin"
 
 # Test that echo is wrapped as deaf
-$DGSH -c 'dgsh-enumerate 1 | {{ dgsh-wrap -d echo hi ; dgsh-wrap dd 2>/dev/null ; }} | cat' >dgsh-wrap/echo-deaf.test
+$DGSH -c 'dgsh-enumerate 1 | {{ dgsh-wrap -i 0 echo hi ; dgsh-wrap dd 2>/dev/null ; }} | cat' >dgsh-wrap/echo-deaf.test
 ensure_same echo-deaf
 
 # Test that echo is wrapped as deaf when wrapped as script with supplied exec
-echo "#!$TOP/build/libexec/dgsh/dgsh-wrap -S  -d `which echo`" >dgsh-wrap/echo-S
+echo "#!$TOP/build/libexec/dgsh/dgsh-wrap -S  -i 0 `which echo`" >dgsh-wrap/echo-S
 chmod +x dgsh-wrap/echo-S
 $DGSH -c 'dgsh-enumerate 1 | {{ dgsh-wrap/echo-S hi ; dgsh-wrap dd 2>/dev/null ; }} | cat' >dgsh-wrap/echo-S.test
 ensure_same echo-S
 
 # Test that echo is wrapped as deaf when wrapped as script with implied exec
-echo "#!$TOP/build/libexec/dgsh/dgsh-wrap -s  -d" >dgsh-wrap/echo
+echo "#!$TOP/build/libexec/dgsh/dgsh-wrap -s  -i 0" >dgsh-wrap/echo
 chmod +x dgsh-wrap/echo
 $DGSH -c 'dgsh-enumerate 1 | {{ dgsh-wrap/echo hi ; dgsh-wrap dd 2>/dev/null ; }} | cat' >dgsh-wrap/echo-s.test
 ensure_same echo-s
@@ -59,18 +59,34 @@ the first three file descriptors under /dev/fd.  In FreeBSD systems
 consider mounting fdescfs(5) on /dev/fd to avoid this problem.
 EOF
   exit 1
-else
-  # Test stand-alone path substitution (with stdin)
-  $DGSH -c 'dgsh-enumerate 2 | dgsh-wrap paste "<|" "<|" ' >dgsh-wrap/paste2.test
-  ensure_same paste2
-
-  # Test stand-alone path substitution (without stdin)
-  $DGSH -c 'dgsh-enumerate 2 | dgsh-wrap -I /usr/bin/paste - "<|" ' >dgsh-wrap/paste1.test
-  ensure_same paste1
-
-  # Test substitution of embedded arguments
-  $DGSH -c 'dgsh-enumerate 1 | {{ dgsh-wrap -e dd "if=<|" "of=>|" 2>/dev/null ; }} | cat' >dgsh-wrap/dd-args.test
-  ensure_same dd-args
 fi
 
+# Test stand-alone input path substitution (with stdin)
+$DGSH -c 'dgsh-enumerate 2 | dgsh-wrap paste "<|" "<|" ' >dgsh-wrap/paste1.test
+ensure_same paste1
+
+# Test stand-alone input path substitution (without stdin)
+$DGSH -c 'dgsh-enumerate 2 | dgsh-wrap -I /usr/bin/paste - "<|" ' >dgsh-wrap/paste2.test
+ensure_same paste2
+
+# Test arbitrary input argument provision (with stdin)
+$DGSH -c 'dgsh-enumerate 2 | dgsh-wrap -i a paste' >dgsh-wrap/paste3.test
+ensure_same paste3
+
+# Test arbitrary input argument provision (without stdin)
+$DGSH -c 'dgsh-enumerate 2 | dgsh-wrap -i a -I /usr/bin/paste -' >dgsh-wrap/paste4.test
+ensure_same paste4
+
+# Test stand-alone output path substitution (without stdout)
+$DGSH -c 'echo hi | dgsh-wrap -O /usr/bin/tee ">|" | {{ sed "s/^/a/" ; sed "s/^/b/" ; }} | cat' >dgsh-wrap/tee1.test
+ensure_same tee1
+
+# Test arbitrary input argument provision (without stdout)
+$DGSH -c 'echo hi | dgsh-wrap -o a -O /usr/bin/tee | {{ sed "s/^/a/" ; sed "s/^/b/" ; }} | cat' >dgsh-wrap/tee2.test
+ensure_same tee2
+
+# Test substitution of embedded arguments
+$DGSH -c 'dgsh-enumerate 1 | {{ dgsh-wrap -e dd "if=<|" "of=>|" 2>/dev/null ; }} | cat' >dgsh-wrap/dd-args.test
+ensure_same dd-args
+
 exit 0