[git commit] grep: fix two bugs with -w

Denys Vlasenko vda.linux at googlemail.com
Tue Jan 7 13:59:30 UTC 2014


commit: http://git.busybox.net/busybox/commit/?id=cd55f2d9332489432ba2b3093903949c6c2e3e33
branch: http://git.busybox.net/busybox/commit/?id=refs/heads/master

Unfortunately, with !EXTRA_COMPAT, "grep -w ^str" still erroneously matches "strstr".

function                                             old     new   delta
grep_file                                           1499    1510     +11

Signed-off-by: Denys Vlasenko <vda.linux at googlemail.com>
---
 findutils/grep.c     |   28 +++++++++++++++++++++++++---
 testsuite/grep.tests |   12 ++++++++++++
 2 files changed, 37 insertions(+), 3 deletions(-)

diff --git a/findutils/grep.c b/findutils/grep.c
index b8ad1bf..a7bc4ca 100644
--- a/findutils/grep.c
+++ b/findutils/grep.c
@@ -373,6 +373,9 @@ static int grep_file(FILE *file)
  opt_f_not_found: ;
 				}
 			} else {
+#if ENABLE_EXTRA_COMPAT
+				unsigned start_pos;
+#endif
 				char *match_at;
 
 				if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
@@ -389,15 +392,18 @@ static int grep_file(FILE *file)
 #if !ENABLE_EXTRA_COMPAT
 				gl->matched_range.rm_so = 0;
 				gl->matched_range.rm_eo = 0;
+#else
+				start_pos = 0;
 #endif
 				match_at = line;
  opt_w_again:
+//bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len);
 				if (
 #if !ENABLE_EXTRA_COMPAT
 					regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, 0) == 0
 #else
 					re_search(&gl->compiled_regex, match_at, line_len,
-							/*start:*/ 0, /*range:*/ line_len,
+							start_pos, /*range:*/ line_len,
 							&gl->matched_range) >= 0
 #endif
 				) {
@@ -416,8 +422,24 @@ static int grep_file(FILE *file)
 							if (!c || (!isalnum(c) && c != '_')) {
 								found = 1;
 							} else {
-								match_at += gl->matched_range.rm_eo;
-								goto opt_w_again;
+			/*
+			 * Why check gl->matched_range.rm_eo?
+			 * Zero-length match makes -w skip the line:
+			 * "echo foo | grep ^" prints "foo",
+			 * "echo foo | grep -w ^" prints nothing.
+			 * Without such check, we can loop forever.
+			 */
+#if !ENABLE_EXTRA_COMPAT
+								if (gl->matched_range.rm_eo != 0) {
+									match_at += gl->matched_range.rm_eo;
+									goto opt_w_again;
+								}
+#else
+								if (gl->matched_range.rm_eo > start_pos) {
+									start_pos = gl->matched_range.rm_eo;
+									goto opt_w_again;
+								}
+#endif
 							}
 						}
 					}
diff --git a/testsuite/grep.tests b/testsuite/grep.tests
index 64d99a9..412efff 100755
--- a/testsuite/grep.tests
+++ b/testsuite/grep.tests
@@ -147,6 +147,18 @@ testing "grep -w doesn't stop on 1st mismatch" \
 	"foop foo\n" \
 	""
 
+testing "grep -w ^str doesn't match str not at the beginning" \
+	"grep -w ^str input" \
+	"" \
+	"strstr\n" \
+	""
+
+testing "grep -w ^ doesn't hang" \
+	"grep -w ^ input" \
+	"" \
+	"anything\n" \
+	""
+
 # testing "test name" "commands" "expected result" "file input" "stdin"
 #   file input will be file called "input"
 #   test can create a file "actual" instead of writing to stdout


More information about the busybox-cvs mailing list