New option parsing infrastructure (doesn't use getopt). Hook it up to

existing applets.  Still a bit buggy, but bits of it work.

10 files changed, 339 insertions, 42 deletions

diff --git a/Config.in b/Config.in
index b26f94ba..51b3cdaf 100644
--- a/Config.in
+++ b/Config.in
@@ -12,6 +12,12 @@ config TOYS_FREE
 	  without a real OS (ala newlib+libgloss), enable this to make toybox
 	  clean up after itself.
 
+config DEBUG
+	bool "Debugging tests"
+	default n
+	help
+	  Enable extra checks for debugging purposes.
+
 endmenu
 
 source toys/Config.in
diff --git a/lib/args.c b/lib/args.c
new file mode 100644
index 00000000..09b04d83
--- /dev/null
+++ b/lib/args.c
@@ -0,0 +1,284 @@
+/* vi: set sw=4 ts=4 :
+ * args.c - Command line argument parsing.
+ *
+ * Copyright 2006 Rob Landley <rob@landley.net>
+ */
+
+#include "toys.h"
+
+// Design goals:
+//   Don't use getopt()
+//   Don't permute original arguments.
+//   handle --long gracefully "(noshort)a(along)b(blong1)(blong2)"
+//   After each argument:
+//       Note that pointer and long are always the same size, even on 64 bit.
+//     : plus a string argument, keep most recent if more than one
+//     * plus a string argument, appended to a list
+//     ? plus a signed long argument (TODO: Bounds checking?)
+//     @ plus an occurrence counter (which is a long)
+//     | this is required.  If more than one marked, only one required.
+//     (longopt)
+//     +X enabling this enables X (switch on)
+//     ~X enabling this disables X (switch off)
+//       x~x means toggle x, I.E. specifying it again switches it off.
+//     !X die with error if X already set (x!x die if x supplied twice)
+//     [yz] needs at least one of y or z.
+//   at the beginning:
+//     + stop at first nonoption argument
+//     ? return array of remaining arguments in first vararg
+//     <0 at least # leftover arguments needed (default 0)
+//     >9 at most # leftover arguments needed (default MAX_INT)
+//     # don't show_usage() on unknown argument.
+//     & first argument has imaginary dash (ala tar/ps)
+//       If given twice, all arguments have imaginary dash
+
+// Notes from getopt man page
+//   - and -- cannot be arguments.
+//     -- force end of arguments
+//     - is a synonym for stdin in file arguments
+//   -abc means -a -b -c
+
+/* This uses a getopt-like option string, but not getopt() itself.
+ * 
+ *   Each option in options corresponds to a bit position in the return
+ * value (last argument is (1<<0), the next to last is (1<<1) and so on.
+ * If the option isn't seen in argv its bit is 0.  Options which have an
+ * argument use the next vararg.  (So varargs used by options go from left to
+ * right, but bits set by arguments go from right to left.)
+ *
+ * Example:
+ *   get_optflags("ab:c:d", NULL, &bstring, &cstring);
+ *   argv = ["command", "-b", "fruit", "-d"]
+ *   flags = 5, bstring="fruit", cstring=NULL;
+ */
+
+struct opts {
+	struct opts *next;
+	char c;
+	int type;
+	int shift;
+	void *arg;
+};
+
+struct getoptflagstate
+{
+	int argc;
+	char *arg;
+	struct opts *opts, *this;
+	int noerror, nodash_now;
+};
+
+static struct getoptflagstate gof;
+
+// Returns zero if it didn't consume the rest of the current -abcdef
+static int gotflag(void)
+{
+	char *arg = NULL;
+	int type;
+	int ret = 0;
+
+	// Did we recognize this option?
+	if (!gof.this && !gof.noerror) error_exit("Unknown option %s\n", gof.arg);
+	else toys.optflags |= 1 << gof.this->shift;
+
+	// Does this option take an argument?
+	gof.arg++;
+	if (gof.this->type & 255) {
+		// Make "tar xCjfv blah1 blah2 thingy" work like
+		// "tar -x -C blah1 -j -f blah2 -v thingy"
+		if (!gof.nodash_now && !*gof.arg) {
+			gof.arg = toys.argv[++gof.argc];
+			if (!gof.arg) error_exit("Missing argument");
+		} else {
+			arg = gof.arg;
+			ret++;
+		}
+	} else gof.this = NULL;
+
+	// If the last option had an argument, grab it.
+	if (!gof.this)  return 0;
+	type = gof.this->type & 255;
+	if (!gof.arg && !(gof.arg = toys.argv[++gof.argc]))
+		error_exit("Missing argument");
+	if (type == ':') gof.this->arg = arg;
+	else if (type == '*') {
+		struct arg_list *temp, **list;
+		list = (struct arg_list **)gof.this->arg;
+		temp = xmalloc(sizeof(struct arg_list));
+		temp->arg = arg;
+		temp->next = *list;
+		*list = temp;
+	} else if (type == '?') {
+	} else if (type == '@') {
+	}
+
+	return ret;
+}
+
+// Fill out toys.optflags and toys.optargs.  This isn't reentrant because
+// we don't bzero(&gof, sizeof(gof));
+
+void get_optflags(void)
+{
+	int stopearly = 0, optarg = 0, nodash = 0, minargs = 0, maxargs = INT_MAX;
+	struct longopts {
+		struct longopts *next;
+		struct opts *opt;
+		char *str;
+		int len;
+	} *longopts = NULL;
+	long *nextarg = (long *)&toy;
+	char *options = toys.which->options;
+
+	// Parse leading special behavior indicators
+	for (;;) {
+		if (*options == '+') stopearly++;
+		else if (*options == '<') minargs=*(++options)-'0';
+		else if (*options == '>') maxargs=*(++options)-'0';
+		else if (*options == '#') gof.noerror++;
+		else if (*options == '&') nodash++;
+		else break;
+		options++;
+	}
+
+	// Parse rest of opts into array
+	while (*options) {
+
+		// Allocate a new option entry when necessary
+		if (!gof.this) {
+			gof.this = xzalloc(sizeof(struct opts));
+			gof.this->next = gof.opts;
+			gof.opts = gof.this;
+		}
+		// Each option must start with (or an option character.  (Bare
+		// longopts only come at the start of the string.)
+		if (*options == '(') {
+			char *end;
+			struct longopts *lo = xmalloc(sizeof(struct longopts));
+
+			// Find the end of the longopt
+			for (end = ++options; *end && *end != ')'; end++);
+			if (CFG_DEBUG && !*end) error_exit("Unterminated optstring");
+
+			// Allocate and init a new struct longopts
+			lo = xmalloc(sizeof(struct longopts));
+			lo->next = longopts;
+			lo->opt = gof.this;
+			lo->str = options;
+			lo->len = end-options;
+			longopts = lo;
+			options = end;
+
+			// For leading longopts (with no corresponding short opt), note
+			// that this option struct has been used.
+			gof.this->shift++;
+
+		// If this is the start of a new option that wasn't a longopt,
+
+		} else if (index(":*?@", *options)) {
+			gof.this->type |= *options;
+			// Pointer and long guaranteed to be the same size by LP64.
+			*(++nextarg) = 0;
+			gof.this->arg = (void *)nextarg;
+		} else if (*options == '|') {
+		} else if (*options == '+') {
+		} else if (*options == '~') {
+		} else if (*options == '!') {
+		} else if (*options == '[') {
+
+		// At this point, we've hit the end of the previous option.  The
+		// current character is the start of a new option.  If we've already
+		// assigned an option to this struct, loop to allocate a new one.
+		// (It'll get back here afterwards.)
+		} else if(gof.this->shift || gof.this->c) {
+			gof.this = NULL;
+			continue;
+
+		// Claim this option, loop to see what's after it.
+		} else gof.this->c = *options;
+
+		options++;
+	}
+
+	// Initialize shift bits (have to calculate this ahead of time because
+	// longopts jump into the middle of the list), and allocate space to
+	// store optargs.
+	gof.argc = 0;
+	for (gof.this = gof.opts; gof.this; gof.this = gof.this->next)
+		gof.this->shift = gof.argc++;
+	toys.optargs = xzalloc(sizeof(char *)*(++gof.argc));
+
+	// Iterate through command line arguments, skipping argv[0]
+	for (gof.argc=1; toys.argv[gof.argc]; gof.argc++) {
+		char *arg = toys.argv[gof.argc];
+
+		// Parse this argument
+		if (stopearly>1) goto notflag;
+
+		gof.nodash_now = 0;
+
+		// Various things with dashes
+		if (*arg == '-') {
+
+			// Handle -
+			if (!arg[1]) goto notflag;
+			arg++;
+			if (*arg=='-') {
+				struct longopts *lo;
+
+				arg++;
+				// Handle --
+				if (!*arg) {
+					stopearly += 2;
+					goto notflag;
+				}
+				// Handle --longopt
+
+				for (lo = longopts; lo; lo = lo->next) {
+					if (!strncmp(arg, lo->str, lo->len)) {
+						if (arg[lo->len]) {
+							if (arg[lo->len]=='='
+								&& (lo->opt->type & 255))
+							{
+								arg += lo->len;
+							} else continue;
+
+						// *options should be nul, this makes sure
+						// that the while (*arg) loop terminates;
+						} arg = options-1;
+						gof.this = lo->opt;
+						break;
+					}
+				}
+				// Long option parsed, jump to option handling.
+				gotflag();
+				continue;
+			}
+
+		// Handle things that don't start with a dash.
+		} else {
+			if (nodash && (nodash>1 || gof.argc == 1)) gof.nodash_now = 1;
+			else goto notflag;
+		}
+
+		// At this point, we have the args part of -args.  Loop through
+		// each entry (could be -abc meaning -a -b -c)
+		while (*arg) {
+			// Identify next option char.
+			for (gof.this = gof.opts; gof.this && *arg != gof.this->c;
+					gof.this = gof.this->next);
+			if (gotflag()) break;
+			arg++;
+		}
+		continue;
+
+		// Not a flag, save value in toys.optargs[]
+notflag:
+		if (stopearly) stopearly++;
+		toys.optargs[optarg++] = toys.argv[gof.argc];
+	}
+
+	// Sanity check
+	if (optarg<minargs) error_exit("Need %d arguments", minargs);
+	if (optarg>maxargs) error_exit("Max %d arguments", maxargs);
+}
diff --git a/lib/lib.h b/lib/lib.h
index d013af9d..013e4919 100644
--- a/lib/lib.h
+++ b/lib/lib.h
@@ -13,6 +13,14 @@ struct string_list {
 	char str[0];
 };
 
+struct arg_list {
+	struct arg_list *next;
+	char *arg;
+};
+
+// args.c
+void get_optflags(void);
+
 // functions.c
 void verror_msg(char *msg, int err, va_list va);
 void error_msg(char *msg, ...);
diff --git a/main.c b/main.c
index 89977f69..20f4cf4b 100644
--- a/main.c
+++ b/main.c
@@ -51,6 +51,7 @@ void toy_init(struct toy_list *which, char *argv[])
 	toys.which = which;
 	toys.argv = argv;
 	toys.exitval = 1;
+	if (which->options) get_optflags();
 }
 
 // Run a toy.
diff --git a/toys.h b/toys.h
index 003a827c..67cc2757 100644
--- a/toys.h
+++ b/toys.h
@@ -38,7 +38,11 @@ void toy_exec(char *argv[]);
 extern struct toy_context {
 	struct toy_list *which;  // Which entry in toy_list is this one?
 	int exitval;             // Value error_exit feeds to exit()
-	int optflags;            // Command line option flags
 	char **argv;             // Command line arguments
-	char buf[4096];
+	unsigned optflags;       // Command line option flags from get_optflags()
+	char **optargs;          // Arguments left over from get_optflags()
 } toys;
+
+// One big temporary buffer, for use by applets (not library functions).
+
+char buf[4096];
diff --git a/toys/Config.in b/toys/Config.in
index 2120636c..77e59e4d 100644
--- a/toys/Config.in
+++ b/toys/Config.in
@@ -161,7 +161,7 @@ config TOYSH_BUILTINS
 	  unset, read, alias.
 
 config WHICH
-	bool "Which"
+	bool "which"
 	default n
 	help
 	  usage: which [-a] filename ...
diff --git a/toys/df.c b/toys/df.c
index cbebd157..d4528f96 100644
--- a/toys/df.c
+++ b/toys/df.c
@@ -34,14 +34,13 @@ static void show_mt(struct mtab_list *mt)
 	// Figure out how much total/used/free space this filesystem has,
 	// forcing 64-bit math because filesystems are big now.
 	block = mt->statvfs.f_bsize ? : 1;
-
 	size = (long)((block * mt->statvfs.f_blocks) / toy.df.units);
 	used = (long)((block * (mt->statvfs.f_blocks-mt->statvfs.f_bfree))
 			/ toy.df.units);
 	avail = (long)((block
 				* (getuid() ? mt->statvfs.f_bavail : mt->statvfs.f_bfree))
 			/ toy.df.units);
-	percent = 100-(long)((100*(uint64_t)avail)/size);
+	percent = size ? 100-(long)((100*(uint64_t)avail)/size) : 0;
 
 	// Figure out appropriate spacing
 	len = 25 - strlen(mt->device);
@@ -58,10 +57,6 @@ static void show_mt(struct mtab_list *mt)
 int df_main(void)
 {
 	struct mtab_list *mt, *mt2, *mtlist;
-	char **argv;
-
-	// get_optflags("Pkt:a",&(toy.df.fstype));
-	argv = NULL;
 
 	// Handle -P and -k
 	toy.df.units = 1024;
@@ -75,14 +70,14 @@ int df_main(void)
 	mtlist = getmountlist(1);
 
 	// If we have a list of filesystems on the command line, loop through them.
-	if (argv) {
-		char *next;
+	if (*toys.optargs) {
+		char **next;
 
-		for(next = *argv; *next; next++) {
+		for(next = toys.optargs; *next; next++) {
 			struct stat st;
 
 			// Stat it (complain if we can't).
-			if(!stat(next, &st)) {
+			if(!stat(*next, &st)) {
 				perror_msg("`%s'", next);
 				toys.exitval = 1;
 				continue;
@@ -117,9 +112,7 @@ int df_main(void)
 		}
 	}
 
-	if (CFG_TOYS_FREE) {
-		llist_free(mtlist, NULL);
-		free(argv);
-	}
+	if (CFG_TOYS_FREE) llist_free(mtlist, NULL);
+
 	return 0;
 }
diff --git a/toys/toylist.h b/toys/toylist.h
index 87716b0f..a0eef1c1 100644
--- a/toys/toylist.h
+++ b/toys/toylist.h
@@ -10,15 +10,15 @@
 #ifdef FROM_MAIN
 #undef NEWTOY
 #undef OLDTOY
-#define NEWTOY(name, flags) {#name, name##_main, flags},
-#define OLDTOY(name, oldname, flags) {#name, oldname##_main, flags},
+#define NEWTOY(name, opts, flags) {#name, name##_main, opts, flags},
+#define OLDTOY(name, oldname, opts, flags) {#name, oldname##_main, opts, flags},
 
 // When #included from toys.h, provide function declarations and structs.
 // The #else is because main.c #includes this file twice.
 
 #else
-#define NEWTOY(name, flags) int name##_main(void);
-#define OLDTOY(name, oldname, flags)
+#define NEWTOY(name, opts, flags) int name##_main(void);
+#define OLDTOY(name, oldname, opts, flags)
 
 struct df_data {
 	struct string_list *fstype;
@@ -39,6 +39,7 @@ union toy_union {
 extern struct toy_list {
 	char *name;
 	int (*toy_main)(void);
+	char *options;
 	int flags;
 } toy_list[];
 
@@ -48,15 +49,15 @@ extern struct toy_list {
 
 // This one is out of order on purpose.
 
-NEWTOY(toybox, 0)
+NEWTOY(toybox, NULL, 0)
 
 // The rest of these are alphabetical, for binary search.
 
-USE_TOYSH(NEWTOY(cd, TOYFLAG_NOFORK))
-USE_DF(NEWTOY(df, TOYFLAG_USR|TOYFLAG_SBIN))
-USE_TOYSH(NEWTOY(exit, TOYFLAG_NOFORK))
-USE_HELLO(NEWTOY(hello, TOYFLAG_NOFORK|TOYFLAG_USR))
-USE_PWD(NEWTOY(pwd, TOYFLAG_BIN))
-USE_TOYSH(OLDTOY(sh, toysh, TOYFLAG_BIN))
-USE_TOYSH(NEWTOY(toysh, TOYFLAG_BIN))
-USE_WHICH(NEWTOY(which, TOYFLAG_USR|TOYFLAG_BIN))
+USE_TOYSH(NEWTOY(cd, NULL, TOYFLAG_NOFORK))
+USE_DF(NEWTOY(df, "Pkt:a", TOYFLAG_USR|TOYFLAG_SBIN))
+USE_TOYSH(NEWTOY(exit, NULL, TOYFLAG_NOFORK))
+USE_HELLO(NEWTOY(hello, NULL, TOYFLAG_NOFORK|TOYFLAG_USR))
+USE_PWD(NEWTOY(pwd, NULL, TOYFLAG_BIN))
+USE_TOYSH(OLDTOY(sh, toysh, "c:i", TOYFLAG_BIN))
+USE_TOYSH(NEWTOY(toysh, "c:i", TOYFLAG_BIN))
+USE_WHICH(NEWTOY(which, "a", TOYFLAG_USR|TOYFLAG_BIN))
diff --git a/toys/toysh.c b/toys/toysh.c
index 7ac430c9..074907ef 100644
--- a/toys/toysh.c
+++ b/toys/toysh.c
@@ -125,12 +125,15 @@ static void run_pipeline(struct pipeline *line)
 	tl = toy_find(cmd->argv[0]);
 	// Is this command a builtin that should run in this process?
 	if (tl && (tl->flags & TOYFLAG_NOFORK)) {
-		struct toy_list *which = toys.which;
-		char **argv = toys.argv;
+		struct toy_context temp;
 
+		// This fakes lots of what toybox_main() does.
+		memcpy(&temp, &toys, sizeof(struct toy_context));
+		bzero(&toys, sizeof(struct toy_context));
 		toy_init(tl, cmd->argv);
 		cmd->pid = tl->toy_main();
-		toy_init(which, argv);
+		free(toys.optargs);
+		memcpy(&toys, &temp, sizeof(struct toy_context));
 	} else {
 		int status;
 
@@ -196,8 +199,10 @@ int toysh_main(void)
 	char *command=NULL;
 	FILE *f;
 
-	// TODO get_optflags(argv, "c:", &command);
-
+	// Set up signal handlers and grab control of this tty.
+	if (CFG_TOYSH_TTY) {
+		if (isatty(0)) toys.optflags |= 1;
+	}
 	f = toys.argv[1] ? xfopen(toys.argv[1], "r") : NULL;
 	if (command) handle(command);
 	else {
diff --git a/toys/which.c b/toys/which.c
index 6d00bc82..16e19d00 100644
--- a/toys/which.c
+++ b/toys/which.c
@@ -7,7 +7,6 @@
 
 #include "toys.h"
 
-#define OPTIONS "a"
 #define OPT_a   1
 
 // Find an exectuable file either at a path with a slash in it (absolute or
@@ -55,16 +54,12 @@ static int which_in_path(char *filename)
 
 int which_main(void)
 {
-	char **argv;
 	int rc = 0;
 
-	// get_optflags(OPTIONS);
-	argv = toys.argv+1;
-
-	if (!*argv) rc++;
+	if (!*toys.optargs) rc++;
 	else {
 		int i;
-		for (i=0; argv[i]; i++)	rc |= which_in_path(argv[i]);
+		for (i=0; toys.optargs[i]; i++) rc |= which_in_path(toys.optargs[i]);
 	}
 	// if (CFG_TOYS_FREE) free(argv);