[git commit] grep: add proper support for pattern_list

Denys Vlasenko vda.linux at googlemail.com
Wed Apr 29 22:20:18 UTC 2020


commit: https://git.busybox.net/busybox/commit/?id=42a8984abc5eed709addac263aab43d64d47257d
branch: https://git.busybox.net/busybox/commit/?id=refs/heads/master

>From POSIX.1-2008:

	The pattern_list's value shall consist of one or more patterns
	separated by <newline> characters;

As such, given patterns need to be split at newline characters. Without
doing so, busybox grep will interpret the newline as part of the pattern
which is not in accordance with POSIX.

See also: https://bugs.busybox.net/show_bug.cgi?id=12721

Signed-off-by: Denys Vlasenko <vda.linux at googlemail.com>
---
 findutils/grep.c     | 19 +++++++++++++------
 testsuite/grep.tests | 12 ++++++++++++
 2 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/findutils/grep.c b/findutils/grep.c
index 84a6f7b1c..55e9c0a8f 100644
--- a/findutils/grep.c
+++ b/findutils/grep.c
@@ -650,6 +650,13 @@ static void load_regexes_from_file(llist_t *fopt)
 	}
 }
 
+static void load_pattern_list(llist_t **lst, char *pattern)
+{
+	char *p;
+	while ((p = strsep(&pattern, "\n")) != NULL)
+		llist_add_to(lst, new_grep_list_data(p, 0));
+}
+
 static int FAST_FUNC file_action_grep(const char *filename,
 			struct stat *statbuf,
 			void* matched,
@@ -754,10 +761,12 @@ int grep_main(int argc UNUSED_PARAM, char **argv)
 #endif
 	invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 
-	{	/* convert char **argv to grep_list_data_t */
-		llist_t *cur;
+	{	/* convert char **argv to pattern_list */
+		llist_t *cur, *new = NULL;
 		for (cur = pattern_head; cur; cur = cur->link)
-			cur->data = new_grep_list_data(cur->data, 0);
+			load_pattern_list(&new, cur->data);
+		llist_free(pattern_head, NULL);
+		pattern_head = new;
 	}
 	if (option_mask32 & OPT_f) {
 		load_regexes_from_file(fopt);
@@ -806,11 +815,9 @@ int grep_main(int argc UNUSED_PARAM, char **argv)
 	/* if we didn't get a pattern from -e and no command file was specified,
 	 * first parameter should be the pattern. no pattern, no worky */
 	if (pattern_head == NULL) {
-		char *pattern;
 		if (*argv == NULL)
 			bb_show_usage();
-		pattern = new_grep_list_data(*argv++, 0);
-		llist_add_to(&pattern_head, pattern);
+		load_pattern_list(&pattern_head, *argv++);
 	}
 
 	/* argv[0..(argc-1)] should be names of file to grep through. If
diff --git a/testsuite/grep.tests b/testsuite/grep.tests
index 9c1f35499..e38278810 100755
--- a/testsuite/grep.tests
+++ b/testsuite/grep.tests
@@ -190,6 +190,18 @@ testing "grep -x -v -e EXP1 -e EXP2 finds nothing if either EXP matches" \
 	"" \
 	"  aa bb cc\n"
 
+testing "grep PATTERN can be a newline-delimited list" \
+	'grep -Fv "$(printf "foo\nbar\n")"' \
+	"baz\n" \
+	"" \
+	"foo\nbar\nbaz\n"
+
+testing "grep -e PATTERN can be a newline-delimited list" \
+	'grep -Fv -e "$(printf "foo\nbar\n")"' \
+	"baz\n" \
+	"" \
+	"foo\nbar\nbaz\n"
+
 # -r on symlink to dir should recurse into dir
 mkdir -p grep.testdir/foo
 echo bar > grep.testdir/foo/file


More information about the busybox-cvs mailing list