Skip to contentSkip to conversation

ls-tree: fix expansion of repeated %(path)

PATCH In review René Scharfe 2 participants 4 messages Jan 14, 2023

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ls-tree: fix expansion of repeated %(path)

expand_show_tree() borrows the base strbuf given to us by read_tree() to build the full path of the current entry when handling %(path). Only its indirect caller, show_tree_fmt(), removes the added entry name. That works fine as long as %(path) is only included once in the format string, but accumulates duplicates if it's repeated:

   $ git ls-tree --format='%(path) %(path) %(path)' HEAD M*
   Makefile MakefileMakefile MakefileMakefileMakefile

Reset the length after each use to get the same expansion every time; here's the behavior with this patch:

   $ ./git ls-tree --format='%(path) %(path) %(path)' HEAD M*
   Makefile Makefile Makefile
---
 builtin/ls-tree.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
builtin/ls-tree.c +2 -3
index c3ea09281a..120fff9fa0 100644
@@ -98,9 +98,11 @@ static size_t expand_show_tree(struct strbuf *sb, const char *start,
 		const char *prefix = chomp_prefix ? ls_tree_prefix : NULL;
 		struct strbuf quoted = STRBUF_INIT;
 		struct strbuf sbuf = STRBUF_INIT;
+		size_t baselen = data->base->len;
 		strbuf_addstr(data->base, data->pathname);
 		name = relative_path(data->base->buf, prefix, &sbuf);
 		quote_c_style(name, &quoted, NULL, 0);
+		strbuf_setlen(data->base, baselen);
 		strbuf_addbuf(sb, &quoted);
 		strbuf_release(&sbuf);
 		strbuf_release(&quoted);
@@ -144,7 +146,6 @@ static int show_recursive(const char *base, size_t baselen, const char *pathname
 static int show_tree_fmt(const struct object_id *oid, struct strbuf *base,
 			 const char *pathname, unsigned mode, void *context UNUSED)
 {
-	size_t baselen;
 	int recurse = 0;
 	struct strbuf sb = STRBUF_INIT;
 	enum object_type type = object_type(mode);
@@ -164,12 +165,10 @@ static int show_tree_fmt(const struct object_id *oid, struct strbuf *base,
 	if (type == OBJ_BLOB && (ls_options & LS_TREE_ONLY))
 		return 0;
 
-	baselen = base->len;
 	strbuf_expand(&sb, format, expand_show_tree, &data);
 	strbuf_addch(&sb, line_termination);
 	fwrite(sb.buf, sb.len, 1, stdout);
 	strbuf_release(&sb);
-	strbuf_setlen(base, baselen);
 	return recurse;
 }
 

--
2.39.0

#1 lore raw

PATCH 1/1 ls-tree: remove dead store and strbuf for quote_c_style()

Stop initializing "name" because it is set again before use.

Let quote_c_style() write directly to "sb" instead of taking a detour through "quoted". This avoids an allocation and a string copy. The result is the same because the function only appends.

---
 builtin/ls-tree.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
builtin/ls-tree.c +2 -5
index 120fff9fa0..9b804cd13b 100644
@@ -94,18 +94,15 @@ static size_t expand_show_tree(struct strbuf *sb, const char *start,
 	} else if (skip_prefix(start, "(objectname)", &p)) {
 		strbuf_add_unique_abbrev(sb, data->oid, abbrev);
 	} else if (skip_prefix(start, "(path)", &p)) {
-		const char *name = data->base->buf;
+		const char *name;
 		const char *prefix = chomp_prefix ? ls_tree_prefix : NULL;
-		struct strbuf quoted = STRBUF_INIT;
 		struct strbuf sbuf = STRBUF_INIT;
 		size_t baselen = data->base->len;
 		strbuf_addstr(data->base, data->pathname);
 		name = relative_path(data->base->buf, prefix, &sbuf);
-		quote_c_style(name, &quoted, NULL, 0);
+		quote_c_style(name, sb, NULL, 0);
 		strbuf_setlen(data->base, baselen);
-		strbuf_addbuf(sb, &quoted);
 		strbuf_release(&sbuf);
-		strbuf_release(&quoted);
 	} else {
 		errlen = (unsigned long)len;
 		die(_("bad ls-tree format: %%%.*s"), errlen, start);

--
2.39.0

#2 lore raw
In reply to René Scharfe
Quoted 12 lines · René Scharfe

expand_show_tree() borrows the base strbuf given to us by read_tree() to build the full path of the current entry when handling %(path). Only its indirect caller, show_tree_fmt(), removes the added entry name. That works fine as long as %(path) is only included once in the format string, but accumulates duplicates if it's repeated:

   $ git ls-tree --format='%(path) %(path) %(path)' HEAD M*
   Makefile MakefileMakefile MakefileMakefileMakefile

Reset the length after each use to get the same expansion every time; here's the behavior with this patch:

Quoted 7 lines
   $ ./git ls-tree --format='%(path) %(path) %(path)' HEAD M*
   Makefile Makefile Makefile
---
 builtin/ls-tree.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

I wonder if this was broken from its introduction at 455923e0 (ls-tree: introduce "--format" option, 2022-03-23)?

It seems to be the case. With the following applied on top of 455923e0, the new test fails as expected, and your patch fixes the breakage, so I am tempted to squash the tests in ;-)

Thanks.

 t/t3104-ls-tree-format.sh | 6 ++++++
 1 file changed, 6 insertions(+)
t/t3104-ls-tree-format.sh +6 -0
index 7f1eb699d3..7e6c4dc5da 100755
@@ -37,6 +37,12 @@ test_ls_tree_format () {
 	'
 }
 
+test_expect_success "ls-tree --format='%(path) %(path) %(path)' HEAD top-file" '
+	git ls-tree --format="%(path) %(path) %(path)" HEAD top-file.t >actual &&
+	echo top-file.t top-file.t top-file.t >expect &&
+	test_cmp expect actual
+'
+
 test_ls_tree_format \
 	"%(objectmode) %(objecttype) %(objectname)%x09%(path)" \
 	""
#3 lore raw

Am 14.01.2023 um 17:46 schrieb Junio C Hamano:

Quoted 17 lines
Quoted 15 lines · René Scharfe

expand_show_tree() borrows the base strbuf given to us by read_tree() to build the full path of the current entry when handling %(path). Only its indirect caller, show_tree_fmt(), removes the added entry name. That works fine as long as %(path) is only included once in the format string, but accumulates duplicates if it's repeated:

   $ git ls-tree --format='%(path) %(path) %(path)' HEAD M*
   Makefile MakefileMakefile MakefileMakefileMakefile

Reset the length after each use to get the same expansion every time; here's the behavior with this patch:

   $ ./git ls-tree --format='%(path) %(path) %(path)' HEAD M*
   Makefile Makefile Makefile
Quoted 7 lines
Quoted 4 lines
---
 builtin/ls-tree.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

I wonder if this was broken from its introduction at 455923e0 (ls-tree: introduce "--format" option, 2022-03-23)?

Yes.

Quoted 3 lines

It seems to be the case. With the following applied on top of 455923e0, the new test fails as expected, and your patch fixes the breakage, so I am tempted to squash the tests in ;-)

I didn't include a test because it's an odd bug which I didn't expect to ever return. But its current existence proves that it can happen. So I don't mind this addition.

Quoted 23 lines

Thanks.

 t/t3104-ls-tree-format.sh | 6 ++++++
 1 file changed, 6 insertions(+)
t/t3104-ls-tree-format.sh +6 -0
index 7f1eb699d3..7e6c4dc5da 100755
@@ -37,6 +37,12 @@ test_ls_tree_format () {
 	'
 }
 
+test_expect_success "ls-tree --format='%(path) %(path) %(path)' HEAD top-file" '
+	git ls-tree --format="%(path) %(path) %(path)" HEAD top-file.t >actual &&
+	echo top-file.t top-file.t top-file.t >expect &&
+	test_cmp expect actual
+'
+
 test_ls_tree_format \
 	"%(objectmode) %(objecttype) %(objectname)%x09%(path)" \
 	""