Add lax parsing for commit and tag objectsrb/object-parse-flexibility

This changes the behavior of object parsing for commits and tags
so that even when bad data is found inside the object, we will
continue to try to parse as much of the object as we can.  The
existing functions (`git_object_lookup` for example) will still
delete the partially parsed object before returning an error, but
this also adds a new function `git_object_lookup_lax` that will
still return the error, but will also return the object with the
partial data (if we got far enough along in the parsing process to
even create the base object).

3 files changed, 57 insertions, 43 deletions

diff --git a/tests/commit/parse.c b/tests/commit/parse.c
index 41e162440..00e763d9c 100644
--- a/tests/commit/parse.c
+++ b/tests/commit/parse.c
@@ -22,50 +22,62 @@ typedef struct {
 } parse_test_case;
 
 static parse_test_case passing_header_cases[] = {
-	{ "parent 05452d6349abcd67aa396dfb28660d765d8b2a36\n", "parent " },
-	{ "tree 05452d6349abcd67aa396dfb28660d765d8b2a36\n", "tree " },
-	{ "random_heading 05452d6349abcd67aa396dfb28660d765d8b2a36\n", "random_heading " },
-	{ "stuck_heading05452d6349abcd67aa396dfb28660d765d8b2a36\n", "stuck_heading" },
-	{ "tree 5F4BEFFC0759261D015AA63A3A85613FF2F235DE\n", "tree " },
-	{ "tree 1A669B8AB81B5EB7D9DB69562D34952A38A9B504\n", "tree " },
-	{ "tree 5B20DCC6110FCC75D31C6CEDEBD7F43ECA65B503\n", "tree " },
-	{ "tree 173E7BF00EA5C33447E99E6C1255954A13026BE4\n", "tree " },
+	{ "parent 05452d6349abcd67aa396dfb28660d765d8b2a36\n", "parent" },
+	{ "tree 05452d6349abcd67aa396dfb28660d765d8b2a36\n", "tree" },
+	{ "random_heading 05452d6349abcd67aa396dfb28660d765d8b2a36\n", "random_heading" },
+	{ "tree 5F4BEFFC0759261D015AA63A3A85613FF2F235DE\n", "tree" },
+	{ "tree 1A669B8AB81B5EB7D9DB69562D34952A38A9B504\n", "tree" },
+	{ "tree 5B20DCC6110FCC75D31C6CEDEBD7F43ECA65B503\n", "tree" },
+	{ "tree 173E7BF00EA5C33447E99E6C1255954A13026BE4\n", "tree" },
 	{ NULL, NULL }
 };
 
 static parse_test_case failing_header_cases[] = {
-	{ "parent 05452d6349abcd67aa396dfb28660d765d8b2a36", "parent " },
-	{ "05452d6349abcd67aa396dfb28660d765d8b2a36\n", "tree " },
-	{ "parent05452d6349abcd67aa396dfb28660d765d8b2a6a\n", "parent " },
-	{ "parent 05452d6349abcd67aa396dfb280d765d8b2a6\n", "parent " },
-	{ "tree  05452d6349abcd67aa396dfb28660d765d8b2a36\n", "tree " },
-	{ "parent 0545xd6349abcd67aa396dfb28660d765d8b2a36\n", "parent " },
-	{ "parent 0545xd6349abcd67aa396dfb28660d765d8b2a36FF\n", "parent " },
-	{ "", "tree " },
+	{ "parent 05452d6349abcd67aa396dfb28660d765d8b2a36", "parent" },
+	{ "05452d6349abcd67aa396dfb28660d765d8b2a36\n", "tree" },
+	{ "parent05452d6349abcd67aa396dfb28660d765d8b2a6a\n", "parent" },
+	{ "parent 05452d6349abcd67aa396dfb280d765d8b2a6\n", "parent" },
+	{ "tree  05452d6349abcd67aa396dfb28660d765d8b2a36\n", "tree" },
+	{ "parent 0545xd6349abcd67aa396dfb28660d765d8b2a36\n", "parent" },
+	{ "parent 0545xd6349abcd67aa396dfb28660d765d8b2a36FF\n", "parent" },
+	{ "", "tree" },
 	{ "", "" },
+	{ "stuck_heading05452d6349abcd67aa396dfb28660d765d8b2a36\n", "stuck_heading" },
 	{ NULL, NULL }
 };
 
 void test_commit_parse__header(void)
 {
-	git_oid oid;
+	git_oid oid, exp;
+	git_object_parse_t template[2] = {
+		{ NULL, 0, GIT_PARSE_OID, { .id = &oid } },
+		{ NULL, 0, GIT_PARSE_BODY_OPTIONAL },
+	};
+	parse_test_case *test;
 
-	parse_test_case *testcase;
-	for (testcase = passing_header_cases; testcase->line != NULL; testcase++)
-	{
-		const char *line = testcase->line;
+	for (test = passing_header_cases; test->line != NULL; test++) {
+		const char *line = test->line;
 		const char *line_end = line + strlen(line);
 
-		cl_git_pass(git_oid__parse(&oid, &line, line_end, testcase->header));
-		cl_assert(line == line_end);
+		template[0].tag = test->header;
+		template[0].taglen = strlen(test->header);
+
+		cl_git_pass(git_object__parse_lines(
+			GIT_OBJ_COMMIT, template, line, line_end));
+
+		cl_git_pass(git_oid_fromstr(&exp, line + strlen(test->header) + 1));
+		cl_assert(git_oid_equal(&exp, &oid));
 	}
 
-	for (testcase = failing_header_cases; testcase->line != NULL; testcase++)
-	{
-		const char *line = testcase->line;
+	for (test = failing_header_cases; test->line != NULL; test++) {
+		const char *line = test->line;
 		const char *line_end = line + strlen(line);
 
-		cl_git_fail(git_oid__parse(&oid, &line, line_end, testcase->header));
+		template[0].tag = test->header;
+		template[0].taglen = strlen(test->header);
+
+		cl_git_fail(git_object__parse_lines(
+			GIT_OBJ_COMMIT, template, line, line_end));
 	}
 }
 
@@ -152,12 +164,13 @@ void test_commit_parse__signature(void)
 		size_t len = strlen(passcase->string);
 		struct git_signature person = {0};
 
-		cl_git_pass(git_signature__parse(&person, &str, str + len, passcase->header, '\n'));
+		cl_git_pass(git_signature__parse(
+			&person, &str, str + len, passcase->header, '\n'));
 		cl_assert_equal_s(passcase->name, person.name);
 		cl_assert_equal_s(passcase->email, person.email);
 		cl_assert_equal_i((int)passcase->time, (int)person.when.time);
 		cl_assert_equal_i(passcase->offset, person.when.offset);
-		git__free(person.name); git__free(person.email);
+		git_signature__clear(&person);
 	}
 
 	for (failcase = failing_signature_cases; failcase->string != NULL; failcase++)
@@ -165,8 +178,9 @@ void test_commit_parse__signature(void)
 		const char *str = failcase->string;
 		size_t len = strlen(failcase->string);
 		git_signature person = {0};
-		cl_git_fail(git_signature__parse(&person, &str, str + len, failcase->header, '\n'));
-		git__free(person.name); git__free(person.email);
+		cl_git_fail(git_signature__parse(
+			&person, &str, str + len, failcase->header, '\n'));
+		git_signature__clear(&person);
 	}
 }
 
diff --git a/tests/object/raw/type2string.c b/tests/object/raw/type2string.c
index a3585487f..45045a062 100644
--- a/tests/object/raw/type2string.c
+++ b/tests/object/raw/type2string.c
@@ -23,17 +23,17 @@ void test_object_raw_type2string__convert_type_to_string(void)
 
 void test_object_raw_type2string__convert_string_to_type(void)
 {
-	cl_assert(git_object_string2type(NULL) == GIT_OBJ_BAD);
-	cl_assert(git_object_string2type("") == GIT_OBJ_BAD);
-	cl_assert(git_object_string2type("commit") == GIT_OBJ_COMMIT);
-	cl_assert(git_object_string2type("tree") == GIT_OBJ_TREE);
-	cl_assert(git_object_string2type("blob") == GIT_OBJ_BLOB);
-	cl_assert(git_object_string2type("tag") == GIT_OBJ_TAG);
-	cl_assert(git_object_string2type("OFS_DELTA") == GIT_OBJ_OFS_DELTA);
-	cl_assert(git_object_string2type("REF_DELTA") == GIT_OBJ_REF_DELTA);
-
-	cl_assert(git_object_string2type("CoMmIt") == GIT_OBJ_BAD);
-	cl_assert(git_object_string2type("hohoho") == GIT_OBJ_BAD);
+	cl_assert(git_object_string2type(NULL, 0) == GIT_OBJ_BAD);
+	cl_assert(git_object_string2type("", 0) == GIT_OBJ_BAD);
+	cl_assert(git_object_string2type("commit", 0) == GIT_OBJ_COMMIT);
+	cl_assert(git_object_string2type("tree", 0) == GIT_OBJ_TREE);
+	cl_assert(git_object_string2type("blob", 0) == GIT_OBJ_BLOB);
+	cl_assert(git_object_string2type("tag", 0) == GIT_OBJ_TAG);
+	cl_assert(git_object_string2type("OFS_DELTA", 0) == GIT_OBJ_OFS_DELTA);
+	cl_assert(git_object_string2type("REF_DELTA", 0) == GIT_OBJ_REF_DELTA);
+
+	cl_assert(git_object_string2type("CoMmIt", 0) == GIT_OBJ_BAD);
+	cl_assert(git_object_string2type("hohoho", 0) == GIT_OBJ_BAD);
 }
 
 void test_object_raw_type2string__check_type_is_loose(void)
diff --git a/tests/odb/loose.c b/tests/odb/loose.c
index c91927c4a..da5518990 100644
--- a/tests/odb/loose.c
+++ b/tests/odb/loose.c
@@ -24,7 +24,7 @@ static void write_object_files(object_data *d)
 
 static void cmp_objects(git_rawobj *o, object_data *d)
 {
-	cl_assert(o->type == git_object_string2type(d->type));
+	cl_assert(o->type == git_object_string2type(d->type, 0));
 	cl_assert(o->len == d->dlen);
 
 	if (o->len > 0)