list-objects-filter: implement composite filters

Allow combining filters such that only objects accepted by all filters
are shown. The motivation for this is to allow getting directory
listings without also fetching blobs. This can be done by combining
blob:none with tree:<depth>. There are massive repositories that have
larger-than-expected trees - even if you include only a single commit.

A combined filter supports any number of subfilters, and is written in
the following form:

	combine:<filter 1>+<filter 2>+<filter 3>

Certain non-alphanumeric characters in each filter must be
URL-encoded.

For now, combined filters must be specified in this form. In a
subsequent commit, rev-list will support multiple --filter arguments
which will have the same effect as specifying one filter argument
starting with "combine:". The documentation will be updated in that
commit, as the URL-encoding scheme is in general not meant to be used
directly by the user, and it is better to describe the URL-encoding
feature in terms of the repeated flag.

Helped-by: Emily Shaffer <emilyshaffer@google.com>
Helped-by: Jeff Hostetler <git@jeffhostetler.com>
Helped-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Helped-by: Jonathan Tan <jonathantanmy@google.com>
Helped-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Matthew DeVore <matvore@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

1 files changed, 150 insertions, 1 deletions

diff --git a/t/t6112-rev-list-filters-objects.sh b/t/t6112-rev-list-filters-objects.sh
index acd7f5ab80..05d4f2e9c2 100755
--- a/t/t6112-rev-list-filters-objects.sh
+++ b/t/t6112-rev-list-filters-objects.sh
@@ -278,7 +278,19 @@ test_expect_success 'verify skipping tree iteration when not collecting omits' '
 	test_line_count = 2 actual &&
 
 	# Make sure no other trees were considered besides the root.
-	! grep "Skipping contents of tree [^.]" filter_trace
+	! grep "Skipping contents of tree [^.]" filter_trace &&
+
+	# Try this again with "combine:". If both sub-filters are skipping
+	# trees, the composite filter should also skip trees. This is not
+	# important unless the user does combine:tree:X+tree:Y or another filter
+	# besides "tree:" is implemented in the future which can skip trees.
+	GIT_TRACE=1 git -C r3 rev-list \
+		--objects --filter=combine:tree:1+tree:3 HEAD 2>filter_trace &&
+
+	# Only skip the dir1/ tree, which is shared between the two commits.
+	grep "Skipping contents of tree " filter_trace >actual &&
+	test_write_lines "Skipping contents of tree dir1/..." >expected &&
+	test_cmp expected actual
 '
 
 # Test tree:# filters.
@@ -330,6 +342,112 @@ test_expect_success 'verify tree:3 includes everything expected' '
 	test_line_count = 10 actual
 '
 
+test_expect_success 'combine:... for a simple combination' '
+	git -C r3 rev-list --objects --filter=combine:tree:2+blob:none HEAD \
+		>actual &&
+
+	expect_has HEAD "" &&
+	expect_has HEAD~1 "" &&
+	expect_has HEAD dir1 &&
+
+	# There are also 2 commit objects
+	test_line_count = 5 actual
+'
+
+test_expect_success 'combine:... with URL encoding' '
+	git -C r3 rev-list --objects \
+		--filter=combine:tree%3a2+blob:%6Eon%65 HEAD >actual &&
+
+	expect_has HEAD "" &&
+	expect_has HEAD~1 "" &&
+	expect_has HEAD dir1 &&
+
+	# There are also 2 commit objects
+	test_line_count = 5 actual
+'
+
+expect_invalid_filter_spec () {
+	spec="$1" &&
+	err="$2" &&
+
+	test_must_fail git -C r3 rev-list --objects --filter="$spec" HEAD \
+		>actual 2>actual_stderr &&
+	test_must_be_empty actual &&
+	test_i18ngrep "$err" actual_stderr
+}
+
+test_expect_success 'combine:... while URL-encoding things that should not be' '
+	expect_invalid_filter_spec combine%3Atree:2+blob:none \
+		"invalid filter-spec"
+'
+
+test_expect_success 'combine: with nothing after the :' '
+	expect_invalid_filter_spec combine: "expected something after combine:"
+'
+
+test_expect_success 'parse error in first sub-filter in combine:' '
+	expect_invalid_filter_spec combine:tree:asdf+blob:none \
+		"expected .tree:<depth>."
+'
+
+test_expect_success 'combine:... with non-encoded reserved chars' '
+	expect_invalid_filter_spec combine:tree:2+sparse:@xyz \
+		"must escape char in sub-filter-spec: .@." &&
+	expect_invalid_filter_spec combine:tree:2+sparse:\` \
+		"must escape char in sub-filter-spec: .\`." &&
+	expect_invalid_filter_spec combine:tree:2+sparse:~abc \
+		"must escape char in sub-filter-spec: .\~."
+'
+
+test_expect_success 'validate err msg for "combine:<valid-filter>+"' '
+	expect_invalid_filter_spec combine:tree:2+ "expected .tree:<depth>."
+'
+
+test_expect_success 'combine:... with edge-case hex digits: Ff Aa 0 9' '
+	git -C r3 rev-list --objects --filter="combine:tree:2+bl%6Fb:n%6fne" \
+		HEAD >actual &&
+	test_line_count = 5 actual &&
+	git -C r3 rev-list --objects --filter="combine:tree%3A2+blob%3anone" \
+		HEAD >actual &&
+	test_line_count = 5 actual &&
+	git -C r3 rev-list --objects --filter="combine:tree:%30" HEAD >actual &&
+	test_line_count = 2 actual &&
+	git -C r3 rev-list --objects --filter="combine:tree:%39+blob:none" \
+		HEAD >actual &&
+	test_line_count = 5 actual
+'
+
+test_expect_success 'add a sparse pattern blob whose path has reserved chars' '
+	cp r3/pattern r3/pattern1+renamed% &&
+	git -C r3 add pattern1+renamed% &&
+	git -C r3 commit -m "add sparse pattern file with reserved chars"
+'
+
+test_expect_success 'combine:... with more than two sub-filters' '
+	git -C r3 rev-list --objects \
+		--filter=combine:tree:3+blob:limit=40+sparse:oid=master:pattern \
+		HEAD >actual &&
+
+	expect_has HEAD "" &&
+	expect_has HEAD~1 "" &&
+	expect_has HEAD~2 "" &&
+	expect_has HEAD dir1 &&
+	expect_has HEAD dir1/sparse1 &&
+	expect_has HEAD dir1/sparse2 &&
+
+	# Should also have 3 commits
+	test_line_count = 9 actual &&
+
+	# Try again, this time making sure the last sub-filter is only
+	# URL-decoded once.
+	cp actual expect &&
+
+	git -C r3 rev-list --objects \
+		--filter=combine:tree:3+blob:limit=40+sparse:oid=master:pattern1%2brenamed%25 \
+		HEAD >actual &&
+	test_cmp expect actual
+'
+
 # Test provisional omit collection logic with a repo that has objects appearing
 # at multiple depths - first deeper than the filter's threshold, then shallow.
 
@@ -373,6 +491,37 @@ test_expect_success 'verify skipping tree iteration when collecting omits' '
 	test_cmp expect actual
 '
 
+test_expect_success 'setup r5' '
+	git init r5 &&
+	mkdir -p r5/subdir &&
+
+	echo 1     >r5/short-root          &&
+	echo 12345 >r5/long-root           &&
+	echo a     >r5/subdir/short-subdir &&
+	echo abcde >r5/subdir/long-subdir  &&
+
+	git -C r5 add short-root long-root subdir &&
+	git -C r5 commit -m "commit msg"
+'
+
+test_expect_success 'verify collecting omits in combined: filter' '
+	# Note that this test guards against the naive implementation of simply
+	# giving both filters the same "omits" set and expecting it to
+	# automatically merge them.
+	git -C r5 rev-list --objects --quiet --filter-print-omitted \
+		--filter=combine:tree:2+blob:limit=3 HEAD >actual &&
+
+	# Expect 0 trees/commits, 3 blobs omitted (all blobs except short-root)
+	omitted_1=$(echo 12345 | git hash-object --stdin) &&
+	omitted_2=$(echo a     | git hash-object --stdin) &&
+	omitted_3=$(echo abcde | git hash-object --stdin) &&
+
+	grep ~$omitted_1 actual &&
+	grep ~$omitted_2 actual &&
+	grep ~$omitted_3 actual &&
+	test_line_count = 3 actual
+'
+
 # Test tree:<depth> where a tree is iterated to twice - once where a subentry is
 # too deep to be included, and again where the blob inside it is shallow enough
 # to be included. This makes sure we don't use LOFR_MARK_SEEN incorrectly (we