hush: fix backslash and terminator handling in <<[-]["]heredoc["]

function old new delta parse_stream 2339 2395 +56 expand_pseudo_dquoted 104 118 +14 parse_stream_dquoted 296 300 +4 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 3/0 up/down: 74/0) Total: 74 bytes Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
author: Denys Vlasenko <dvlasenk@redhat.com> 2010-09-06 11:27:32 +0200
committer: Denys Vlasenko <dvlasenk@redhat.com> 2010-09-06 11:27:32 +0200
commit: 77b32ccbf2a1a77911b486b673008a4cb82bb8b7 (patch)
tree: c68a1bef8013018843321d6807aa87ca432808b1
parent: c49d2d97939d77be3d1f3bbbbf9db30a55771c15 (diff)
download: busybox-w32-77b32ccbf2a1a77911b486b673008a4cb82bb8b7.tar.gz
busybox-w32-77b32ccbf2a1a77911b486b673008a4cb82bb8b7.tar.bz2
busybox-w32-77b32ccbf2a1a77911b486b673008a4cb82bb8b7.zip
3 files changed, 92 insertions, 5 deletions
diff --git a/shell/hush.c b/shell/hush.c
index ef46372de..e8aef2d7e 100644
--- a/shell/hush.c
+++ b/shell/hush.c
@@ -3162,17 +3162,20 @@ static int redirect_opt_num(o_string *o)
 static char *fetch_till_str(o_string *as_string,
                struct in_str *input,
                const char *word,
-                int skip_tabs)
+                int heredoc_flags)
 {
        o_string heredoc = NULL_O_STRING;
        int past_EOL = 0;
+        int prev = 0; /* not \ */
        int ch;
        goto jump_in;
        while (1) {
                ch = i_getch(input);
                nommu_addchr(as_string, ch);
-                if (ch == '\n') {
+                if (ch == '\n'
+                 && ((heredoc_flags & HEREDOC_QUOTED) || prev != '\\')
+                ) {
                        if (strcmp(heredoc.data + past_EOL, word) == 0) {
                                heredoc.data[past_EOL] = '\0';
                                debug_printf_parse("parsed heredoc '%s'\n", heredoc.data);
@@ -3185,7 +3188,7 @@ static char *fetch_till_str(o_string *as_string,
                                do {
                                        ch = i_getch(input);
                                        nommu_addchr(as_string, ch);
-                                } while (skip_tabs && ch == '\t');
+                                } while ((heredoc_flags & HEREDOC_SKIPTABS) && ch == '\t');
                        } while (ch == '\n');
                }
                if (ch == EOF) {
@@ -3194,6 +3197,7 @@ static char *fetch_till_str(o_string *as_string,
                }
                o_addchr(&heredoc, ch);
                nommu_addchr(as_string, ch);
+                prev = ch;
        }
 }
@@ -3223,7 +3227,7 @@ static int fetch_heredocs(int heredoc_cnt, struct parse_context *ctx, struct in_
                                        redir->rd_type = REDIRECT_HEREDOC2;
                                        /* redir->rd_dup is (ab)used to indicate <<- */
                                        p = fetch_till_str(&ctx->as_string, input,
-                                                redir->rd_filename, redir->rd_dup & HEREDOC_SKIPTABS);
+                                                        redir->rd_filename, redir->rd_dup);
                                        if (!p) {
                                                syntax_error("unexpected EOF in here document");
                                                return 1;
@@ -3778,8 +3782,9 @@ static int parse_stream_dquoted(o_string *as_string,
                 * only when followed by one of the following characters:
                 * $, `, ", \, or <newline>.  A double quote may be quoted
                 * within double quotes by preceding it with a backslash."
+                 * NB: in (unquoted) heredoc, above does not apply to ".
                 */
-                if (strchr("$`\"\\\n", next) != NULL) {
+                if (next == dquote_end || strchr("$`\\\n", next) != NULL) {
                        ch = i_getch(input);
                        if (ch != '\n') {
                                o_addqchr(dest, ch);
@@ -4412,6 +4417,7 @@ static char *expand_pseudo_dquoted(const char *str)
        o_string dest = NULL_O_STRING;
        if (!strchr(str, '$')
+         && !strchr(str, '\\')
 #if ENABLE_HUSH_TICK
         && !strchr(str, '`')
 #endif
diff --git a/shell/hush_test/hush-misc/heredoc_backslash1.right b/shell/hush_test/hush-misc/heredoc_backslash1.right
new file mode 100644
index 000000000..234c0172f
--- /dev/null
+++ b/shell/hush_test/hush-misc/heredoc_backslash1.right
@@ -0,0 +1,27 @@
+Quoted heredoc:
+a\
+        b
+ 123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+        -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+c\
+Unquoted heredoc:
+a       b
+ 123456 -qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
+        -qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
+cEOF2
+Quoted -heredoc:
+a\
+b
+ 123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+-$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+c\
+Unquoted -heredoc:
+a       b
+ 123456 -qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
+-qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
+cEOF4
+Done: 0
diff --git a/shell/hush_test/hush-misc/heredoc_backslash1.tests b/shell/hush_test/hush-misc/heredoc_backslash1.tests
new file mode 100755
index 000000000..b70467df8
--- /dev/null
+++ b/shell/hush_test/hush-misc/heredoc_backslash1.tests
@@ -0,0 +1,54 @@
+# Test for correct handling of backslashes.
+# Note that some lines in each heredoc start with a tab.
+a=qwerty
+echo Quoted heredoc:
+cat <<"EOF1"
+a\
+        b
+ 123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+        -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+c\
+EOF1
+echo
+echo Unquoted heredoc:
+cat <<EOF2
+a\
+        b
+ 123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+        -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+c\
+EOF2
+EOF2
+echo
+echo Quoted -heredoc:
+cat <<-"EOF3"
+a\
+        b
+ 123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+        -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+c\
+        EOF3
+# In -heredoc case the marker is detected even if it is indented.
+echo
+echo Unquoted -heredoc:
+cat <<-EOF4
+a\
+        b
+ 123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+        -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
+c\
+EOF4
+        EOF4
+# The marker is not detected if preceding line ends in backslash.
+# TODO: marker should be detected even if it is split by line continuation:
+# EOF\
+# 4
+# but currently hush doesn't do it. (Tab before "4" is not allowed, though.)
+echo
+echo "Done: $?"
author	Denys Vlasenko <dvlasenk@redhat.com>	2010-09-06 11:27:32 +0200
committer	Denys Vlasenko <dvlasenk@redhat.com>	2010-09-06 11:27:32 +0200
commit	77b32ccbf2a1a77911b486b673008a4cb82bb8b7 (patch)
tree	c68a1bef8013018843321d6807aa87ca432808b1
parent	c49d2d97939d77be3d1f3bbbbf9db30a55771c15 (diff)
download	busybox-w32-77b32ccbf2a1a77911b486b673008a4cb82bb8b7.tar.gz busybox-w32-77b32ccbf2a1a77911b486b673008a4cb82bb8b7.tar.bz2 busybox-w32-77b32ccbf2a1a77911b486b673008a4cb82bb8b7.zip

diff --git a/shell/hush.c b/shell/hush.c index ef46372de..e8aef2d7e 100644 --- a/shell/hush.c +++ b/shell/hush.c
@@ -3162,17 +3162,20 @@ static int redirect_opt_num(o_string *o)
3162	static char fetch_till_str(o_string as_string,	3162	static char fetch_till_str(o_string as_string,
3163	struct in_str *input,	3163	struct in_str *input,
3164	const char *word,	3164	const char *word,
3165	int skip_tabs)	3165	int heredoc_flags)
3166	{	3166	{
3167	o_string heredoc = NULL_O_STRING;	3167	o_string heredoc = NULL_O_STRING;
3168	int past_EOL = 0;	3168	int past_EOL = 0;
		3169	int prev = 0; /* not \ */
3169	int ch;	3170	int ch;
3170		3171
3171	goto jump_in;	3172	goto jump_in;
3172	while (1) {	3173	while (1) {
3173	ch = i_getch(input);	3174	ch = i_getch(input);
3174	nommu_addchr(as_string, ch);	3175	nommu_addchr(as_string, ch);
3175	if (ch == '\n') {	3176	if (ch == '\n'
		3177	&& ((heredoc_flags & HEREDOC_QUOTED) \|\| prev != '\\')
		3178	) {
3176	if (strcmp(heredoc.data + past_EOL, word) == 0) {	3179	if (strcmp(heredoc.data + past_EOL, word) == 0) {
3177	heredoc.data[past_EOL] = '\0';	3180	heredoc.data[past_EOL] = '\0';
3178	debug_printf_parse("parsed heredoc '%s'\n", heredoc.data);	3181	debug_printf_parse("parsed heredoc '%s'\n", heredoc.data);
@@ -3185,7 +3188,7 @@ static char fetch_till_str(o_string as_string,
3185	do {	3188	do {
3186	ch = i_getch(input);	3189	ch = i_getch(input);
3187	nommu_addchr(as_string, ch);	3190	nommu_addchr(as_string, ch);
3188	} while (skip_tabs && ch == '\t');	3191	} while ((heredoc_flags & HEREDOC_SKIPTABS) && ch == '\t');
3189	} while (ch == '\n');	3192	} while (ch == '\n');
3190	}	3193	}
3191	if (ch == EOF) {	3194	if (ch == EOF) {
@@ -3194,6 +3197,7 @@ static char fetch_till_str(o_string as_string,
3194	}	3197	}
3195	o_addchr(&heredoc, ch);	3198	o_addchr(&heredoc, ch);
3196	nommu_addchr(as_string, ch);	3199	nommu_addchr(as_string, ch);
		3200	prev = ch;
3197	}	3201	}
3198	}	3202	}
3199		3203
@@ -3223,7 +3227,7 @@ static int fetch_heredocs(int heredoc_cnt, struct parse_context *ctx, struct in_
3223	redir->rd_type = REDIRECT_HEREDOC2;	3227	redir->rd_type = REDIRECT_HEREDOC2;
3224	/* redir->rd_dup is (ab)used to indicate <<- */	3228	/* redir->rd_dup is (ab)used to indicate <<- */
3225	p = fetch_till_str(&ctx->as_string, input,	3229	p = fetch_till_str(&ctx->as_string, input,
3226	redir->rd_filename, redir->rd_dup & HEREDOC_SKIPTABS);	3230	redir->rd_filename, redir->rd_dup);
3227	if (!p) {	3231	if (!p) {
3228	syntax_error("unexpected EOF in here document");	3232	syntax_error("unexpected EOF in here document");
3229	return 1;	3233	return 1;
@@ -3778,8 +3782,9 @@ static int parse_stream_dquoted(o_string *as_string,
3778	* only when followed by one of the following characters:	3782	* only when followed by one of the following characters:
3779	* $, `, ", \, or <newline>. A double quote may be quoted	3783	* $, `, ", \, or <newline>. A double quote may be quoted
3780	* within double quotes by preceding it with a backslash."	3784	* within double quotes by preceding it with a backslash."
		3785	* NB: in (unquoted) heredoc, above does not apply to ".
3781	*/	3786	*/
3782	if (strchr("$`\"\\\n", next) != NULL) {	3787	if (next == dquote_end \|\| strchr("$`\\\n", next) != NULL) {
3783	ch = i_getch(input);	3788	ch = i_getch(input);
3784	if (ch != '\n') {	3789	if (ch != '\n') {
3785	o_addqchr(dest, ch);	3790	o_addqchr(dest, ch);
@@ -4412,6 +4417,7 @@ static char expand_pseudo_dquoted(const char str)
4412	o_string dest = NULL_O_STRING;	4417	o_string dest = NULL_O_STRING;
4413		4418
4414	if (!strchr(str, '$')	4419	if (!strchr(str, '$')
		4420	&& !strchr(str, '\\')
4415	#if ENABLE_HUSH_TICK	4421	#if ENABLE_HUSH_TICK
4416	&& !strchr(str, '`')	4422	&& !strchr(str, '`')
4417	#endif	4423	#endif


diff --git a/shell/hush_test/hush-misc/heredoc_backslash1.right b/shell/hush_test/hush-misc/heredoc_backslash1.right new file mode 100644 index 000000000..234c0172f --- /dev/null +++ b/shell/hush_test/hush-misc/heredoc_backslash1.right
@@ -0,0 +1,27 @@
		1	Quoted heredoc:
		2	a\
		3	b
		4	123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		5	-$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		6	c\
		7
		8	Unquoted heredoc:
		9	a b
		10	123456 -qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
		11	-qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
		12	cEOF2
		13
		14	Quoted -heredoc:
		15	a\
		16	b
		17	123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		18	-$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		19	c\
		20
		21	Unquoted -heredoc:
		22	a b
		23	123456 -qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
		24	-qwerty-\t-\-\"-\'-`-\--\z-\*-\?-
		25	cEOF4
		26
		27	Done: 0


diff --git a/shell/hush_test/hush-misc/heredoc_backslash1.tests b/shell/hush_test/hush-misc/heredoc_backslash1.tests new file mode 100755 index 000000000..b70467df8 --- /dev/null +++ b/shell/hush_test/hush-misc/heredoc_backslash1.tests
@@ -0,0 +1,54 @@
		1	# Test for correct handling of backslashes.
		2	# Note that some lines in each heredoc start with a tab.
		3
		4	a=qwerty
		5
		6	echo Quoted heredoc:
		7	cat <<"EOF1"
		8	a\
		9	b
		10	123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		11	-$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		12	c\
		13	EOF1
		14	echo
		15
		16	echo Unquoted heredoc:
		17	cat <<EOF2
		18	a\
		19	b
		20	123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		21	-$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		22	c\
		23	EOF2
		24	EOF2
		25	echo
		26
		27	echo Quoted -heredoc:
		28	cat <<-"EOF3"
		29	a\
		30	b
		31	123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		32	-$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		33	c\
		34	EOF3
		35	# In -heredoc case the marker is detected even if it is indented.
		36	echo
		37
		38	echo Unquoted -heredoc:
		39	cat <<-EOF4
		40	a\
		41	b
		42	123456 -$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		43	-$a-\t-\\-\"-\'-\`-\--\z-\*-\?-
		44	c\
		45	EOF4
		46	EOF4
		47	# The marker is not detected if preceding line ends in backslash.
		48	# TODO: marker should be detected even if it is split by line continuation:
		49	# EOF\
		50	# 4
		51	# but currently hush doesn't do it. (Tab before "4" is not allowed, though.)
		52	echo
		53
		54	echo "Done: $?"