xargs: add support for -I and -i. Closes 493

function                                             old     new   delta
process_stdin_with_replace                             -     195    +195
xmalloc_substitute_string                              -     145    +145
xargs_main                                           808     884     +76
count_strstr                                           -      45     +45
packed_usage                                       29580   29571      -9
parse_params                                        1445    1416     -29
func_exec                                            285     138    -147
------------------------------------------------------------------------------
(add/remove: 4/0 grow/shrink: 1/3 up/down: 461/-185)          Total: 276 bytes
   text	   data	    bss	    dec	    hex	filename
 922156	    932	  17692	 940780	  e5aec	busybox_old
 922440	    932	  17692	 941064	  e5c08	busybox_unstripped

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

4 files changed, 176 insertions, 56 deletions

diff --git a/findutils/find.c b/findutils/find.c
index 044f010..6d34f4d 100644
--- a/findutils/find.c
+++ b/findutils/find.c
@@ -402,34 +402,6 @@ struct globals {
 	G.recurse_flags = ACTION_RECURSE; \
 } while (0)
 
-#if ENABLE_FEATURE_FIND_EXEC
-static unsigned count_subst(const char *str)
-{
-	unsigned count = 0;
-	while ((str = strstr(str, "{}")) != NULL) {
-		count++;
-		str++;
-	}
-	return count;
-}
-
-
-static char* subst(const char *src, unsigned count, const char* filename)
-{
-	char *buf, *dst, *end;
-	size_t flen = strlen(filename);
-	/* we replace each '{}' with filename: growth by strlen-2 */
-	buf = dst = xmalloc(strlen(src) + count*(flen-2) + 1);
-	while ((end = strstr(src, "{}")) != NULL) {
-		dst = mempcpy(dst, src, end - src);
-		dst = mempcpy(dst, filename, flen);
-		src = end + 2;
-	}
-	strcpy(dst, src);
-	return buf;
-}
-#endif
-
 /* Return values of ACTFs ('action functions') are a bit mask:
  * bit 1=1: prune (use SKIP constant for setting it)
  * bit 0=1: matched successfully (TRUE)
@@ -613,7 +585,7 @@ ACTF(exec)
 	char *argv[ap->exec_argc + 1];
 #endif
 	for (i = 0; i < ap->exec_argc; i++)
-		argv[i] = subst(ap->exec_argv[i], ap->subst_count[i], fileName);
+		argv[i] = xmalloc_substitute_string(ap->exec_argv[i], ap->subst_count[i], "{}", fileName);
 	argv[i] = NULL; /* terminate the list */
 
 	rc = spawn_and_wait(argv);
@@ -1091,7 +1063,7 @@ static action*** parse_params(char **argv)
 			ap->subst_count = xmalloc(ap->exec_argc * sizeof(int));
 			i = ap->exec_argc;
 			while (i--)
-				ap->subst_count[i] = count_subst(ap->exec_argv[i]);
+				ap->subst_count[i] = count_strstr(ap->exec_argv[i], "{}");
 		}
 #endif
 #if ENABLE_FEATURE_FIND_PAREN
diff --git a/findutils/xargs.c b/findutils/xargs.c
index 0d1bb43..ed6dbd3 100644
--- a/findutils/xargs.c
+++ b/findutils/xargs.c
@@ -53,6 +53,13 @@
 //config:	  Support -0: input items are terminated by a NUL character
 //config:	  instead of whitespace, and the quotes and backslash
 //config:	  are not special.
+//config:
+//config:config FEATURE_XARGS_SUPPORT_REPL_STR
+//config:	bool "Enable -I STR: string to replace"
+//config:	default y
+//config:	depends on XARGS
+//config:	help
+//config:	  Support -I STR and -i[STR] options.
 
 //applet:IF_XARGS(APPLET_NOEXEC(xargs, xargs, BB_DIR_USR_BIN, BB_SUID_DROP, xargs))
 
@@ -85,19 +92,22 @@
 
 struct globals {
 	char **args;
+#if ENABLE_FEATURE_XARGS_SUPPORT_REPL_STR
+	char **argv;
+	const char *repl_str;
+	char eol_ch;
+#endif
 	const char *eof_str;
 	int idx;
 } FIX_ALIASING;
 #define G (*(struct globals*)&bb_common_bufsiz1)
 #define INIT_G() do { \
 	G.eof_str = NULL; /* need to clear by hand because we are NOEXEC applet */ \
+	IF_FEATURE_XARGS_SUPPORT_REPL_STR(G.repl_str = "{}";) \
+	IF_FEATURE_XARGS_SUPPORT_REPL_STR(G.eol_ch = '\n';) \
 } while (0)
 
 
-/*
- * This function has special algorithm.
- * Don't use fork and include to main!
- */
 static int xargs_exec(void)
 {
 	int status;
@@ -301,7 +311,7 @@ static char* FAST_FUNC process0_stdin(int n_max_chars, int n_max_arg, char *buf)
 			c = '\0';
 		}
 		*p++ = c;
-		if (c == '\0') {   /* word's delimiter or EOF detected */
+		if (c == '\0') {   /* NUL or EOF detected */
 			/* A full word is loaded */
 			store_param(s);
 			dbg_msg("args[]:'%s'", s);
@@ -323,10 +333,71 @@ static char* FAST_FUNC process0_stdin(int n_max_chars, int n_max_arg, char *buf)
 }
 #endif /* FEATURE_XARGS_SUPPORT_ZERO_TERM */
 
+#if ENABLE_FEATURE_XARGS_SUPPORT_REPL_STR
+/*
+ * Used if -I<repl> was specified.
+ * In this mode, words aren't appended to PROG ARGS.
+ * Instead, entire input line is read, then <repl> string
+ * in every PROG and ARG is replaced with the line:
+ *  echo -e "ho ho\nhi" | xargs -I_ cmd __ _
+ * results in "cmd 'ho hoho ho' 'ho ho'"; "cmd 'hihi' 'hi'".
+ * -n MAX_ARGS seems to be ignored.
+ * Tested with GNU findutils 4.5.10.
+ */
+//FIXME: n_max_chars is not handled the same way as in GNU findutils.
+//FIXME: quoting is not implemented.
+static char* FAST_FUNC process_stdin_with_replace(int n_max_chars, int n_max_arg UNUSED_PARAM, char *buf)
+{
+	int i;
+	char *end, *p;
+
+	/* Free strings from last invocation, if any */
+	for (i = 0; G.args && G.args[i]; i++)
+		if (G.args[i] != G.argv[i])
+			free(G.args[i]);
+
+	end = buf + n_max_chars;
+	p = buf;
+
+	while (1) {
+		int c = getchar();
+		if (c == EOF || c == G.eol_ch) {
+			if (p == buf)
+				goto ret; /* empty line */
+			c = '\0';
+		}
+		*p++ = c;
+		if (c == '\0') {   /* EOL or EOF detected */
+			i = 0;
+			while (G.argv[i]) {
+				char *arg = G.argv[i];
+				int count = count_strstr(arg, G.repl_str);
+				if (count != 0)
+					arg = xmalloc_substitute_string(arg, count, G.repl_str, s);
+				store_param(arg);
+				dbg_msg("args[]:'%s'", arg);
+				i++;
+			}
+			p = buf;
+			goto ret;
+		}
+		if (p == end) {
+			goto ret;
+		}
+	}
+ ret:
+	*p = '\0';
+	/* store_param(NULL) - caller will do it */
+	dbg_msg("return:'%s'", buf);
+	return buf;
+}
+#endif
+
 #if ENABLE_FEATURE_XARGS_SUPPORT_CONFIRMATION
 /* Prompt the user for a response, and
-   if the user responds affirmatively, return true;
-   otherwise, return false. Uses "/dev/tty", not stdin. */
+ * if user responds affirmatively, return true;
+ * otherwise, return false. Uses "/dev/tty", not stdin.
+ */
 static int xargs_ask_confirmation(void)
 {
 	FILE *tty_stream;
@@ -360,6 +431,9 @@ static int xargs_ask_confirmation(void)
 //usage:     "\n	-e[STR]	STR stops input processing"
 //usage:     "\n	-n N	Pass no more than N args to PROG"
 //usage:     "\n	-s N	Pass command line of no more than N bytes"
+//usage:	IF_FEATURE_XARGS_SUPPORT_REPL_STR(
+//usage:     "\n	-I STR	Replace STR within PROG ARGS with input line"
+//usage:	)
 //usage:	IF_FEATURE_XARGS_SUPPORT_TERMOPT(
 //usage:     "\n	-x	Exit if size is exceeded"
 //usage:	)
@@ -378,6 +452,8 @@ enum {
 	IF_FEATURE_XARGS_SUPPORT_CONFIRMATION(OPTBIT_INTERACTIVE,)
 	IF_FEATURE_XARGS_SUPPORT_TERMOPT(     OPTBIT_TERMINATE  ,)
 	IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(   OPTBIT_ZEROTERM   ,)
+	IF_FEATURE_XARGS_SUPPORT_REPL_STR(    OPTBIT_REPLSTR    ,)
+	IF_FEATURE_XARGS_SUPPORT_REPL_STR(    OPTBIT_REPLSTR1   ,)
 
 	OPT_VERBOSE     = 1 << OPTBIT_VERBOSE    ,
 	OPT_NO_EMPTY    = 1 << OPTBIT_NO_EMPTY   ,
@@ -388,11 +464,14 @@ enum {
 	OPT_INTERACTIVE = IF_FEATURE_XARGS_SUPPORT_CONFIRMATION((1 << OPTBIT_INTERACTIVE)) + 0,
 	OPT_TERMINATE   = IF_FEATURE_XARGS_SUPPORT_TERMOPT(     (1 << OPTBIT_TERMINATE  )) + 0,
 	OPT_ZEROTERM    = IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(   (1 << OPTBIT_ZEROTERM   )) + 0,
+	OPT_REPLSTR     = IF_FEATURE_XARGS_SUPPORT_REPL_STR(    (1 << OPTBIT_REPLSTR    )) + 0,
+	OPT_REPLSTR1    = IF_FEATURE_XARGS_SUPPORT_REPL_STR(    (1 << OPTBIT_REPLSTR1   )) + 0,
 };
 #define OPTION_STR "+trn:s:e::E:" \
 	IF_FEATURE_XARGS_SUPPORT_CONFIRMATION("p") \
 	IF_FEATURE_XARGS_SUPPORT_TERMOPT(     "x") \
-	IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(   "0")
+	IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(   "0") \
+	IF_FEATURE_XARGS_SUPPORT_REPL_STR(    "I:i::")
 
 int xargs_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 int xargs_main(int argc, char **argv)
@@ -405,7 +484,8 @@ int xargs_main(int argc, char **argv)
 	unsigned opt;
 	int n_max_chars;
 	int n_max_arg;
-#if ENABLE_FEATURE_XARGS_SUPPORT_ZERO_TERM
+#if ENABLE_FEATURE_XARGS_SUPPORT_ZERO_TERM \
+ || ENABLE_FEATURE_XARGS_SUPPORT_REPL_STR
 	char* FAST_FUNC (*read_args)(int, int, char*) = process_stdin;
 #else
 #define read_args process_stdin
@@ -419,7 +499,10 @@ int xargs_main(int argc, char **argv)
 		"no-run-if-empty\0" No_argument "r"
 		;
 #endif
-	opt = getopt32(argv, OPTION_STR, &max_args, &max_chars, &G.eof_str, &G.eof_str);
+	opt = getopt32(argv, OPTION_STR,
+		&max_args, &max_chars, &G.eof_str, &G.eof_str
+		IF_FEATURE_XARGS_SUPPORT_REPL_STR(, &G.repl_str, &G.repl_str)
+	);
 
 	/* -E ""? You may wonder why not just omit -E?
 	 * This is used for portability:
@@ -427,8 +510,10 @@ int xargs_main(int argc, char **argv)
 	if ((opt & OPT_EOF_STRING1) && G.eof_str[0] == '\0')
 		G.eof_str = NULL;
 
-	if (opt & OPT_ZEROTERM)
-		IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(read_args = process0_stdin);
+	if (opt & OPT_ZEROTERM) {
+		IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(read_args = process0_stdin;)
+		IF_FEATURE_XARGS_SUPPORT_REPL_STR(G.eol_ch = '\0';)
+	}
 
 	argv += optind;
 	argc -= optind;
@@ -486,20 +571,36 @@ int xargs_main(int argc, char **argv)
 		/* if (n_max_arg > n_max_chars) n_max_arg = n_max_chars */
 	}
 
-	/* Allocate pointers for execvp */
-	/* We can statically allocate (argc + n_max_arg + 1) elements
-	 * and do not bother with resizing args[], but on 64-bit machines
-	 * this results in args[] vector which is ~8 times bigger
-	 * than n_max_chars! That is, with n_max_chars == 20k,
-	 * args[] will take 160k (!), which will most likely be
-	 * almost entirely unused.
-	 */
-	/* See store_param() for matching 256-step growth logic */
-	G.args = xmalloc(sizeof(G.args[0]) * ((argc + 0xff) & ~0xff));
-
-	/* Store the command to be executed, part 1 */
-	for (i = 0; argv[i]; i++)
-		G.args[i] = argv[i];
+#if ENABLE_FEATURE_XARGS_SUPPORT_REPL_STR
+	if (opt & (OPT_REPLSTR | OPT_REPLSTR1)) {
+		/*
+		 * -I<str>:
+		 * Unmodified args are kept in G.argv[i],
+		 * G.args[i] receives malloced G.argv[i] with <str> replaced
+		 * with input line. Setting this up:
+		 */
+		G.args = NULL;
+		G.argv = argv;
+		argc = 0;
+		read_args = process_stdin_with_replace;
+	} else
+#endif
+	{
+		/* Allocate pointers for execvp.
+		 * We can statically allocate (argc + n_max_arg + 1) elements
+		 * and do not bother with resizing args[], but on 64-bit machines
+		 * this results in args[] vector which is ~8 times bigger
+		 * than n_max_chars! That is, with n_max_chars == 20k,
+		 * args[] will take 160k (!), which will most likely be
+		 * almost entirely unused.
+		 *
+		 * See store_param() for matching 256-step growth logic
+		 */
+		G.args = xmalloc(sizeof(G.args[0]) * ((argc + 0xff) & ~0xff));
+		/* Store the command to be executed, part 1 */
+		for (i = 0; argv[i]; i++)
+			G.args[i] = argv[i];
+	}
 
 	while (1) {
 		char *rem;
diff --git a/include/libbb.h b/include/libbb.h
index 96f3334..1cbe2c8 100644
--- a/include/libbb.h
+++ b/include/libbb.h
@@ -650,6 +650,8 @@ char *xstrndup(const char *s, int n) FAST_FUNC RETURNS_MALLOC;
 void overlapping_strcpy(char *dst, const char *src) FAST_FUNC;
 char *safe_strncpy(char *dst, const char *src, size_t size) FAST_FUNC;
 char *strncpy_IFNAMSIZ(char *dst, const char *src) FAST_FUNC;
+unsigned count_strstr(const char *str, const char *sub) FAST_FUNC;
+char *xmalloc_substitute_string(const char *src, int count, const char *sub, const char *repl) FAST_FUNC;
 /* Guaranteed to NOT be a macro (smallest code). Saves nearly 2k on uclibc.
  * But potentially slow, don't use in one-billion-times loops */
 int bb_putchar(int ch) FAST_FUNC;
diff --git a/libbb/replace.c b/libbb/replace.c
new file mode 100644
index 0000000..8711f95
--- a/dev/null
+++ b/libbb/replace.c
@@ -0,0 +1,45 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * Utility routines.
+ *
+ * Copyright (C) 1999-2004 by Erik Andersen <andersen@codepoet.org>
+ *
+ * Licensed under GPLv2 or later, see file LICENSE in this source tree.
+ */
+
+//kbuild:lib-y += replace.o
+
+#include "libbb.h"
+
+unsigned FAST_FUNC count_strstr(const char *str, const char *sub)
+{
+	size_t sub_len = strlen(sub);
+	unsigned count = 0;
+
+	while ((str = strstr(str, sub)) != NULL) {
+		count++;
+		str += sub_len;
+	}
+	return count;
+}
+
+char* FAST_FUNC xmalloc_substitute_string(const char *src, int count, const char *sub, const char *repl)
+{
+	char *buf, *dst, *end;
+	size_t sub_len = strlen(sub);
+	size_t repl_len = strlen(repl);
+
+	//dbg_msg("subst(s:'%s',count:%d,sub:'%s',repl:'%s'", src, count, sub, repl);
+
+	buf = dst = xmalloc(strlen(src) + count * ((int)repl_len - (int)sub_len) + 1);
+	/* we replace each sub with repl */
+	while ((end = strstr(src, sub)) != NULL) {
+		dst = mempcpy(dst, src, end - src);
+		dst = mempcpy(dst, repl, repl_len);
+		/*src = end + 1; - GNU findutils 4.5.10 doesn't do this... */
+		src = end + sub_len; /* but this. Try "xargs -Iaa echo aaa" */
+	}
+	strcpy(dst, src);
+	//dbg_msg("subst9:'%s'", buf);
+	return buf;
+}