ash: fix globbing bugs when using glibc glob()

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
author: Denys Vlasenko <vda.linux@googlemail.com> 2016-10-02 15:17:15 +0200
committer: Denys Vlasenko <vda.linux@googlemail.com> 2016-10-02 15:17:15 +0200
commit: 8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b (patch)
tree: cc2a99f8b7840b57adcf0998d7e3916a337644dc /shell
parent: 37dc08b874d6a24bd5e4b6aff62577ecc1364275 (diff)
download: busybox-w32-8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b.tar.gz
busybox-w32-8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b.tar.bz2
busybox-w32-8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b.zip
3 files changed, 66 insertions, 7 deletions
diff --git a/shell/ash.c b/shell/ash.c
index f8c3e0c19..40b3ef3e3 100644
--- a/shell/ash.c
+++ b/shell/ash.c
@@ -7038,27 +7038,42 @@ expandmeta(struct strlist *str /*, int flag*/)
                        goto nometa;
                INT_OFF;
                p = preglob(str->text, RMESCAPE_ALLOC | RMESCAPE_HEAP);
-                /*
+// GLOB_NOMAGIC (GNU): if no *?[ chars in pattern, return it even if no match
-                 * GLOB_NOMAGIC (GNU): if no *?[ chars in pattern, return it even if no match
+// GLOB_NOCHECK: if no match, return unchanged pattern (sans \* escapes?)
-                 * TODO?: GLOB_NOCHECK: if no match, return unchanged pattern (sans \* escapes?)
+//
-                 */
+// glibc 2.24.90 glob(GLOB_NOMAGIC) does not remove backslashes used for escaping:
-                i = glob(p, GLOB_NOMAGIC, NULL, &pglob);
+// if you pass it "file\?", it returns "file\?", not "file?", if no match.
+// Which means you need to unescape the string, right? Not so fast:
+// if there _is_ a file named "file\?" (with backslash), it is returned
+// as "file\?" too (whichever pattern you used to find it, say, "file*").
+// You DONT KNOW by looking at the result whether you need to unescape it.
+//
+// Worse, globbing of "file\?" in a directory with two files, "file?" and "file\?",
+// returns "file\?" - which is WRONG: "file\?" pattern matches "file?" file.
+// Without GLOB_NOMAGIC, this works correctly ("file?" is returned as a match).
+// With GLOB_NOMAGIC | GLOB_NOCHECK, this also works correctly.
+//              i = glob(p, GLOB_NOMAGIC | GLOB_NOCHECK, NULL, &pglob);
+//              i = glob(p, GLOB_NOMAGIC, NULL, &pglob);
+                i = glob(p, 0, NULL, &pglob);
+                //bb_error_msg("glob('%s'):%d '%s'...", p, i, pglob.gl_pathv ? pglob.gl_pathv[0] : "-");
                if (p != str->text)
                        free(p);
                switch (i) {
                case 0:
+#if 0 // glibc 2.24.90 bug? Patterns like "*/file", when match, don't set GLOB_MAGCHAR
                        /* GLOB_MAGCHAR is set if *?[ chars were seen (GNU) */
                        if (!(pglob.gl_flags & GLOB_MAGCHAR))
                                goto nometa2;
+#endif
                        addglob(&pglob);
                        globfree(&pglob);
                        INT_ON;
                        break;
                case GLOB_NOMATCH:
-nometa2:
+ //nometa2:
                        globfree(&pglob);
                        INT_ON;
-nometa:
+ nometa:
                        *exparg.lastp = str;
                        rmescapes(str->text, 0);
                        exparg.lastp = &str->next;
diff --git a/shell/ash_test/ash-glob/glob_dir.right b/shell/ash_test/ash-glob/glob_dir.right
new file mode 100644
index 000000000..aa90514d5
--- /dev/null
+++ b/shell/ash_test/ash-glob/glob_dir.right
@@ -0,0 +1,19 @@
+dirtest/z.tmp
+dirtest/z.tmp
+dirtest/z.tmp
+dirtest/z.tmp
+dirtest/z.tmp
+dirtest/z.tmp
+dirtest/z.tmp
+dirtest/z.tmp
+dirtest/z.tmp
+*/z.tmp
+*/z.*
+*/?.*
+*/z*p
+d*r*e*t/z*p
+*\/z.tmp
+*/z.*
+*/z*p
+d*r*e*t/z*p
diff --git a/shell/ash_test/ash-glob/glob_dir.tests b/shell/ash_test/ash-glob/glob_dir.tests
new file mode 100755
index 000000000..dc4c4fdb5
--- /dev/null
+++ b/shell/ash_test/ash-glob/glob_dir.tests
@@ -0,0 +1,25 @@
+mkdir dirtest
+     >dirtest/z.tmp
+echo */z.tmp
+echo */z.*
+echo */?.*
+echo */z*p
+echo d*r*e*t/z*p
+echo *"/z.t"mp
+echo */z"."*
+echo *"/z"*"p"
+echo "d"*r*e*t"/"z*p
+echo
+echo \*/z.tmp
+echo "*"/z.*
+echo */"?".*
+echo */z"*p"
+echo d*r*e\*t/z*p
+echo *"\\/z.t"mp
+echo */z".*"
+echo *"/z"\*"p"
+echo "d*"r*e*t"/"z*p
+rm dirtest/z.tmp
+rmdir dirtest
author	Denys Vlasenko <vda.linux@googlemail.com>	2016-10-02 15:17:15 +0200
committer	Denys Vlasenko <vda.linux@googlemail.com>	2016-10-02 15:17:15 +0200
commit	8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b (patch)
tree	cc2a99f8b7840b57adcf0998d7e3916a337644dc /shell
parent	37dc08b874d6a24bd5e4b6aff62577ecc1364275 (diff)
download	busybox-w32-8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b.tar.gz busybox-w32-8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b.tar.bz2 busybox-w32-8e2c9cc2fcfe23625e1c5845f99dfc93e022fc9b.zip

diff --git a/shell/ash.c b/shell/ash.c index f8c3e0c19..40b3ef3e3 100644 --- a/shell/ash.c +++ b/shell/ash.c
@@ -7038,27 +7038,42 @@ expandmeta(struct strlist str /, int flag*/)
7038	goto nometa;	7038	goto nometa;
7039	INT_OFF;	7039	INT_OFF;
7040	p = preglob(str->text, RMESCAPE_ALLOC \| RMESCAPE_HEAP);	7040	p = preglob(str->text, RMESCAPE_ALLOC \| RMESCAPE_HEAP);
7041	/*	7041	// GLOB_NOMAGIC (GNU): if no *?[ chars in pattern, return it even if no match
7042	* GLOB_NOMAGIC (GNU): if no *?[ chars in pattern, return it even if no match	7042	// GLOB_NOCHECK: if no match, return unchanged pattern (sans \* escapes?)
7043	* TODO?: GLOB_NOCHECK: if no match, return unchanged pattern (sans \* escapes?)	7043	//
7044	*/	7044	// glibc 2.24.90 glob(GLOB_NOMAGIC) does not remove backslashes used for escaping:
7045	i = glob(p, GLOB_NOMAGIC, NULL, &pglob);	7045	// if you pass it "file\?", it returns "file\?", not "file?", if no match.
		7046	// Which means you need to unescape the string, right? Not so fast:
		7047	// if there _is_ a file named "file\?" (with backslash), it is returned
		7048	// as "file\?" too (whichever pattern you used to find it, say, "file*").
		7049	// You DONT KNOW by looking at the result whether you need to unescape it.
		7050	//
		7051	// Worse, globbing of "file\?" in a directory with two files, "file?" and "file\?",
		7052	// returns "file\?" - which is WRONG: "file\?" pattern matches "file?" file.
		7053	// Without GLOB_NOMAGIC, this works correctly ("file?" is returned as a match).
		7054	// With GLOB_NOMAGIC \| GLOB_NOCHECK, this also works correctly.
		7055	// i = glob(p, GLOB_NOMAGIC \| GLOB_NOCHECK, NULL, &pglob);
		7056	// i = glob(p, GLOB_NOMAGIC, NULL, &pglob);
		7057	i = glob(p, 0, NULL, &pglob);
		7058	//bb_error_msg("glob('%s'):%d '%s'...", p, i, pglob.gl_pathv ? pglob.gl_pathv[0] : "-");
7046	if (p != str->text)	7059	if (p != str->text)
7047	free(p);	7060	free(p);
7048	switch (i) {	7061	switch (i) {
7049	case 0:	7062	case 0:
		7063	#if 0 // glibc 2.24.90 bug? Patterns like "*/file", when match, don't set GLOB_MAGCHAR
7050	/* GLOB_MAGCHAR is set if ?[ chars were seen (GNU) /	7064	/* GLOB_MAGCHAR is set if ?[ chars were seen (GNU) /
7051	if (!(pglob.gl_flags & GLOB_MAGCHAR))	7065	if (!(pglob.gl_flags & GLOB_MAGCHAR))
7052	goto nometa2;	7066	goto nometa2;
		7067	#endif
7053	addglob(&pglob);	7068	addglob(&pglob);
7054	globfree(&pglob);	7069	globfree(&pglob);
7055	INT_ON;	7070	INT_ON;
7056	break;	7071	break;
7057	case GLOB_NOMATCH:	7072	case GLOB_NOMATCH:
7058	nometa2:	7073	//nometa2:
7059	globfree(&pglob);	7074	globfree(&pglob);
7060	INT_ON;	7075	INT_ON;
7061	nometa:	7076	nometa:
7062	*exparg.lastp = str;	7077	*exparg.lastp = str;
7063	rmescapes(str->text, 0);	7078	rmescapes(str->text, 0);
7064	exparg.lastp = &str->next;	7079	exparg.lastp = &str->next;


diff --git a/shell/ash_test/ash-glob/glob_dir.right b/shell/ash_test/ash-glob/glob_dir.right new file mode 100644 index 000000000..aa90514d5 --- /dev/null +++ b/shell/ash_test/ash-glob/glob_dir.right
@@ -0,0 +1,19 @@
		1	dirtest/z.tmp
		2	dirtest/z.tmp
		3	dirtest/z.tmp
		4	dirtest/z.tmp
		5	dirtest/z.tmp
		6	dirtest/z.tmp
		7	dirtest/z.tmp
		8	dirtest/z.tmp
		9	dirtest/z.tmp
		10
		11	*/z.tmp
		12	/z.
		13	/?.
		14	/zp
		15	dret/zp
		16	*\/z.tmp
		17	/z.
		18	/zp
		19	dret/zp


diff --git a/shell/ash_test/ash-glob/glob_dir.tests b/shell/ash_test/ash-glob/glob_dir.tests new file mode 100755 index 000000000..dc4c4fdb5 --- /dev/null +++ b/shell/ash_test/ash-glob/glob_dir.tests
@@ -0,0 +1,25 @@
		1	mkdir dirtest
		2	>dirtest/z.tmp
		3
		4	echo */z.tmp
		5	echo /z.
		6	echo /?.
		7	echo /zp
		8	echo dret/zp
		9	echo *"/z.t"mp
		10	echo /z"."
		11	echo "/z""p"
		12	echo "d"ret"/"zp
		13	echo
		14	echo \*/z.tmp
		15	echo ""/z.
		16	echo /"?".
		17	echo /z"p"
		18	echo dre\t/zp
		19	echo *"\\/z.t"mp
		20	echo /z"."
		21	echo "/z"\"p"
		22	echo "d"ret"/"zp
		23
		24	rm dirtest/z.tmp
		25	rmdir dirtest