Merge pull request #4864 from pks-t/pks/object-parse-fixes

Object parse fixes

7 files changed, 546 insertions, 6 deletions

diff --git a/src/commit.c b/src/commit.c
index 97ac2a189..bda5a8b42 100644
--- a/src/commit.c
+++ b/src/commit.c
@@ -444,7 +444,7 @@ int git_commit__parse_raw(void *_commit, const char *data, size_t size)
 		while (eoln < buffer_end && *eoln != '\n')
 			++eoln;
 
-		if (git__prefixcmp(buffer, "encoding ") == 0) {
+		if (git__prefixncmp(buffer, buffer_end - buffer, "encoding ") == 0) {
 			buffer += strlen("encoding ");
 
 			commit->message_encoding = git__strndup(buffer, eoln - buffer);
diff --git a/src/tag.c b/src/tag.c
index 663c7dabd..c45335151 100644
--- a/src/tag.c
+++ b/src/tag.c
@@ -70,10 +70,9 @@ static int tag_parse(git_tag *tag, const char *buffer, const char *buffer_end)
 	static const char *tag_types[] = {
 		NULL, "commit\n", "tree\n", "blob\n", "tag\n"
 	};
-
-	unsigned int i;
 	size_t text_len, alloc_len;
-	char *search;
+	const char *search;
+	unsigned int i;
 
 	if (git_oid__parse(&tag->target, &buffer, buffer_end, "object ") < 0)
 		return tag_error("object field invalid");
@@ -138,8 +137,9 @@ static int tag_parse(git_tag *tag, const char *buffer, const char *buffer_end)
 	tag->message = NULL;
 	if (buffer < buffer_end) {
 		/* If we're not at the end of the header, search for it */
-		if( *buffer != '\n' ) {
-			search = strstr(buffer, "\n\n");
+		if(*buffer != '\n') {
+			search = git__memmem(buffer, buffer_end - buffer,
+					     "\n\n", 2);
 			if (search)
 				buffer = search + 1;
 			else
diff --git a/src/util.c b/src/util.c
index 20c88a1d3..52495f752 100644
--- a/src/util.c
+++ b/src/util.c
@@ -357,6 +357,47 @@ size_t git__linenlen(const char *buffer, size_t buffer_len)
 	return nl ? (size_t)(nl - buffer) + 1 : buffer_len;
 }
 
+/*
+ * Adapted Not So Naive algorithm from http://www-igm.univ-mlv.fr/~lecroq/string/
+ */
+const void * git__memmem(const void *haystack, size_t haystacklen,
+			 const void *needle, size_t needlelen)
+{
+	const char *h, *n;
+	size_t j, k, l;
+
+	if (needlelen > haystacklen || !haystacklen || !needlelen)
+		return NULL;
+
+	h = (const char *) haystack,
+	n = (const char *) needle;
+
+	if (needlelen == 1)
+		return memchr(haystack, *n, haystacklen);
+
+	if (n[0] == n[1]) {
+		k = 2;
+		l = 1;
+	} else {
+		k = 1;
+		l = 2;
+	}
+
+	j = 0;
+	while (j <= haystacklen - needlelen) {
+		if (n[1] != h[j + 1]) {
+			j += k;
+		} else {
+			if (memcmp(n + 2, h + j + 2, needlelen - 2) == 0 &&
+			    n[0] == h[j])
+				return h + j;
+			j += l;
+		}
+	}
+
+	return NULL;
+}
+
 void git__hexdump(const char *buffer, size_t len)
 {
 	static const size_t LINE_WIDTH = 16;
diff --git a/src/util.h b/src/util.h
index 400024373..4314295f1 100644
--- a/src/util.h
+++ b/src/util.h
@@ -111,6 +111,9 @@ GIT_INLINE(const void *) git__memrchr(const void *s, int c, size_t n)
 	return NULL;
 }
 
+extern const void * git__memmem(const void *haystack, size_t haystacklen,
+				const void *needle, size_t needlelen);
+
 typedef int (*git__tsort_cmp)(const void *a, const void *b);
 
 extern void git__tsort(void **dst, size_t size, git__tsort_cmp cmp);
diff --git a/tests/core/memmem.c b/tests/core/memmem.c
new file mode 100644
index 000000000..fd9986d01
--- /dev/null
+++ b/tests/core/memmem.c
@@ -0,0 +1,46 @@
+#include "clar_libgit2.h"
+
+static void assert_found(const char *haystack, const char *needle, size_t expected_pos)
+{
+	cl_assert_equal_p(git__memmem(haystack, haystack ? strlen(haystack) : 0,
+				      needle, needle ? strlen(needle) : 0),
+			  haystack + expected_pos);
+}
+
+static void assert_absent(const char *haystack, const char *needle)
+{
+	cl_assert_equal_p(git__memmem(haystack, haystack ? strlen(haystack) : 0,
+				      needle, needle ? strlen(needle) : 0),
+			  NULL);
+}
+
+void test_core_memmem__found(void)
+{
+	assert_found("a", "a", 0);
+	assert_found("ab", "a", 0);
+	assert_found("ba", "a", 1);
+	assert_found("aa", "a", 0);
+	assert_found("aab", "aa", 0);
+	assert_found("baa", "aa", 1);
+	assert_found("dabc", "abc", 1);
+	assert_found("abababc", "abc", 4);
+}
+
+void test_core_memmem__absent(void)
+{
+	assert_absent("a", "b");
+	assert_absent("a", "aa");
+	assert_absent("ba", "ab");
+	assert_absent("ba", "ab");
+	assert_absent("abc", "abcd");
+	assert_absent("abcabcabc", "bcac");
+}
+
+void test_core_memmem__edgecases(void)
+{
+	assert_absent(NULL, NULL);
+	assert_absent("a", NULL);
+	assert_absent(NULL, "a");
+	assert_absent("", "a");
+	assert_absent("a", "");
+}
diff --git a/tests/object/commit/parse.c b/tests/object/commit/parse.c
new file mode 100644
index 000000000..a99110f18
--- /dev/null
+++ b/tests/object/commit/parse.c
@@ -0,0 +1,232 @@
+#include "clar_libgit2.h"
+#include "commit.h"
+#include "object.h"
+#include "signature.h"
+
+static void assert_commit_parses(const char *data, size_t datalen,
+	const char *expected_treeid,
+	const char *expected_author,
+	const char *expected_committer,
+	const char *expected_encoding,
+	const char *expected_message,
+	size_t expected_parents)
+{
+	git_commit *commit;
+	if (!datalen)
+		datalen = strlen(data);
+	cl_git_pass(git_object__from_raw((git_object **) &commit, data, datalen, GIT_OBJ_COMMIT));
+
+	if (expected_author) {
+		git_signature *author;
+		cl_git_pass(git_signature_from_buffer(&author, expected_author));
+		cl_assert(git_signature__equal(author, commit->author));
+		cl_assert_equal_s(author->name, commit->author->name);
+		cl_assert_equal_s(author->email, commit->author->email);
+		cl_assert_equal_i(author->when.time, commit->author->when.time);
+		cl_assert_equal_i(author->when.offset, commit->author->when.offset);
+		cl_assert_equal_i(author->when.sign, commit->author->when.sign);
+		git_signature_free(author);
+	}
+
+	if (expected_committer) {
+		git_signature *committer;
+		cl_git_pass(git_signature_from_buffer(&committer, expected_committer));
+		cl_assert_equal_s(committer->name, commit->committer->name);
+		cl_assert_equal_s(committer->email, commit->committer->email);
+		cl_assert_equal_i(committer->when.time, commit->committer->when.time);
+		cl_assert_equal_i(committer->when.offset, commit->committer->when.offset);
+		cl_assert_equal_i(committer->when.sign, commit->committer->when.sign);
+		git_signature_free(committer);
+	}
+
+	if (expected_encoding)
+		cl_assert_equal_s(commit->message_encoding, expected_encoding);
+	else
+		cl_assert_equal_p(commit->message_encoding, NULL);
+
+	if (expected_message)
+		cl_assert_equal_s(commit->raw_message, expected_message);
+	else
+		cl_assert_equal_p(commit->message_encoding, NULL);
+
+	if (expected_treeid) {
+		git_oid tree_oid;
+		cl_git_pass(git_oid_fromstr(&tree_oid, expected_treeid));
+		cl_assert_equal_oid(&tree_oid, &commit->tree_id);
+	}
+
+	cl_assert_equal_i(commit->parent_ids.size, expected_parents);
+
+	git_object__free(&commit->object);
+}
+
+static void assert_commit_fails(const char *data, size_t datalen)
+{
+	git_object *object;
+	if (!datalen)
+		datalen = strlen(data);
+	cl_git_fail(git_object__from_raw(&object, data, datalen, GIT_OBJ_COMMIT));
+}
+
+void test_object_commit_parse__parsing_commit_succeeds(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author Author <author@example.com>\n"
+		"committer Committer <committer@example.com>\n"
+		"encoding Encoding\n"
+		"\n"
+		"Message";
+	assert_commit_parses(commit, 0,
+		"3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8",
+		"Author <author@example.com>",
+		"Committer <committer@example.com>",
+		"Encoding",
+		"Message", 0);
+}
+
+void test_object_commit_parse__parsing_commit_without_encoding_succeeds(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author Author <author@example.com>\n"
+		"committer Committer <committer@example.com>\n"
+		"\n"
+		"Message";
+	assert_commit_parses(commit, 0,
+		"3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8",
+		"Author <author@example.com>",
+		"Committer <committer@example.com>",
+		NULL,
+		"Message", 0);
+}
+
+void test_object_commit_parse__parsing_commit_with_multiple_authors_succeeds(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author Author1 <author@example.com>\n"
+		"author Author2 <author@example.com>\n"
+		"author Author3 <author@example.com>\n"
+		"author Author4 <author@example.com>\n"
+		"committer Committer <committer@example.com>\n"
+		"\n"
+		"Message";
+	assert_commit_parses(commit, 0,
+		"3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8",
+		"Author1 <author@example.com>",
+		"Committer <committer@example.com>",
+		NULL,
+		"Message", 0);
+}
+
+void test_object_commit_parse__parsing_commit_with_multiple_committers_succeeds(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author Author <author@example.com>\n"
+		"committer Committer1 <committer@example.com>\n"
+		"committer Committer2 <committer@example.com>\n"
+		"committer Committer3 <committer@example.com>\n"
+		"committer Committer4 <committer@example.com>\n"
+		"\n"
+		"Message";
+	assert_commit_parses(commit, 0,
+		"3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8",
+		"Author <author@example.com>",
+		"Committer1 <committer@example.com>",
+		NULL,
+		"Message", 0);
+}
+
+void test_object_commit_parse__parsing_commit_without_message_succeeds(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author Author <author@example.com>\n"
+		"committer Committer <committer@example.com>\n";
+	assert_commit_parses(commit, 0,
+		"3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8",
+		"Author <author@example.com>",
+		"Committer <committer@example.com>",
+		NULL,
+		"", 0);
+}
+
+void test_object_commit_parse__parsing_commit_with_unknown_fields_succeeds(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author Author <author@example.com>\n"
+		"committer Committer <committer@example.com>\n"
+		"foo bar\n"
+		"more garbage\n"
+		"\n"
+		"Message";
+	assert_commit_parses(commit, 0,
+		"3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8",
+		"Author <author@example.com>",
+		"Committer <committer@example.com>",
+		NULL,
+		"Message", 0);
+}
+
+void test_object_commit_parse__parsing_commit_with_invalid_tree_fails(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1xxx5f3445895b71d9cb0f8\n"
+		"author Author <author@example.com>\n"
+		"committer Committer <committer@example.com>\n"
+		"\n"
+		"Message";
+	assert_commit_fails(commit, 0);
+}
+
+void test_object_commit_parse__parsing_commit_without_tree_fails(void)
+{
+	const char *commit =
+		"author Author <author@example.com>\n"
+		"committer Committer <committer@example.com>\n"
+		"\n"
+		"Message";
+	assert_commit_fails(commit, 0);
+}
+
+void test_object_commit_parse__parsing_commit_without_author_fails(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"committer Committer <committer@example.com>\n"
+		"\n"
+		"Message";
+	assert_commit_fails(commit, 0);
+}
+
+void test_object_commit_parse__parsing_commit_without_committer_fails(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author Author <author@example.com>\n"
+		"\n"
+		"Message";
+	assert_commit_fails(commit, 0);
+}
+
+void test_object_commit_parse__parsing_encoding_will_not_cause_oob_read(void)
+{
+	const char *commit =
+		"tree 3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8\n"
+		"author <>\n"
+		"committer <>\n"
+		"encoding foo\n";
+	/*
+	 * As we ignore unknown fields, the cut-off encoding field will be
+	 * parsed just fine.
+	 */
+	assert_commit_parses(commit, strlen(commit) - strlen("ncoding foo\n"),
+		"3e7ac388cadacccdf1c6c5f3445895b71d9cb0f8",
+		"<>",
+		"<>",
+		NULL,
+		"", 0);
+}
diff --git a/tests/object/tag/parse.c b/tests/object/tag/parse.c
new file mode 100644
index 000000000..f701f6b89
--- /dev/null
+++ b/tests/object/tag/parse.c
@@ -0,0 +1,218 @@
+#include "clar_libgit2.h"
+#include "object.h"
+#include "signature.h"
+#include "tag.h"
+
+static void assert_tag_parses(const char *data, size_t datalen,
+	const char *expected_oid,
+	const char *expected_name,
+	const char *expected_tagger,
+	const char *expected_message)
+{
+	git_tag *tag;
+
+	if (!datalen)
+		datalen = strlen(data);
+
+	cl_git_pass(git_object__from_raw((git_object **) &tag, data, datalen, GIT_OBJ_TAG));
+	cl_assert_equal_i(tag->type, GIT_OBJ_TAG);
+
+	if (expected_oid) {
+		git_oid oid;
+		cl_git_pass(git_oid_fromstr(&oid, expected_oid));
+		cl_assert_equal_oid(&oid, &tag->target);
+	}
+
+	if (expected_name)
+		cl_assert_equal_s(expected_name, tag->tag_name);
+	else
+		cl_assert_equal_s(tag->message, NULL);
+
+	if (expected_tagger) {
+		git_signature *tagger;
+		cl_git_pass(git_signature_from_buffer(&tagger, expected_tagger));
+		cl_assert_equal_s(tagger->name, tag->tagger->name);
+		cl_assert_equal_s(tagger->email, tag->tagger->email);
+		cl_assert_equal_i(tagger->when.time, tag->tagger->when.time);
+		cl_assert_equal_i(tagger->when.offset, tag->tagger->when.offset);
+		cl_assert_equal_i(tagger->when.sign, tag->tagger->when.sign);
+		git_signature_free(tagger);
+	} else {
+		cl_assert_equal_s(tag->tagger, NULL);
+	}
+
+	if (expected_message)
+		cl_assert_equal_s(expected_message, tag->message);
+	else
+		cl_assert_equal_s(tag->message, NULL);
+
+	git_object__free(&tag->object);
+}
+
+static void assert_tag_fails(const char *data, size_t datalen)
+{
+	git_object *object;
+	if (!datalen)
+		datalen = strlen(data);
+	cl_git_fail(git_object__from_raw(&object, data, datalen, GIT_OBJ_TAG));
+}
+
+void test_object_tag_parse__valid_tag_parses(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tag tagname\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n"
+		"\n"
+		"Message";
+	assert_tag_parses(tag, 0,
+		"a8d447f68076d1520f69649bb52629941be7031f",
+		"tagname",
+		"Taggy Mr. Taggart <taggy@taggart.com>",
+		"Message");
+}
+
+void test_object_tag_parse__missing_tagger_parses(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tag tagname\n"
+		"\n"
+		"Message";
+	assert_tag_parses(tag, 0,
+		"a8d447f68076d1520f69649bb52629941be7031f",
+		"tagname",
+		NULL,
+		"Message");
+}
+
+void test_object_tag_parse__missing_message_parses(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tag tagname\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n";
+	assert_tag_parses(tag, 0,
+		"a8d447f68076d1520f69649bb52629941be7031f",
+		"tagname",
+		"Taggy Mr. Taggart <taggy@taggart.com>",
+		NULL);
+}
+
+void test_object_tag_parse__unknown_field_parses(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tag tagname\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n"
+		"foo bar\n"
+		"frubble frabble\n"
+		"\n"
+		"Message";
+	assert_tag_parses(tag, 0,
+		"a8d447f68076d1520f69649bb52629941be7031f",
+		"tagname",
+		"Taggy Mr. Taggart <taggy@taggart.com>",
+		"Message");
+}
+
+void test_object_tag_parse__missing_object_fails(void)
+{
+	const char *tag =
+		"type tag\n"
+		"tag tagname\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n"
+		"\n"
+		"Message";
+	assert_tag_fails(tag, 0);
+}
+
+void test_object_tag_parse__malformatted_object_fails(void)
+{
+	const char *tag =
+		"object a8d447f68076d15xxxxxxxxxxxxxxxx41be7031f\n"
+		"type tag\n"
+		"tag tagname\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n"
+		"\n"
+		"Message";
+	assert_tag_fails(tag, 0);
+}
+
+void test_object_tag_parse__missing_type_fails(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"tag tagname\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n"
+		"\n"
+		"Message";
+	assert_tag_fails(tag, 0);
+}
+
+void test_object_tag_parse__invalid_type_fails(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type garbage\n"
+		"tag tagname\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n"
+		"\n"
+		"Message";
+	assert_tag_fails(tag, 0);
+}
+
+void test_object_tag_parse__missing_tagname_fails(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tagger Taggy Mr. Taggart <taggy@taggart.com>\n"
+		"\n"
+		"Message";
+	assert_tag_fails(tag, 0);
+}
+
+void test_object_tag_parse__misformatted_tagger_fails(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tag Tag\n"
+		"tagger taggy@taggart.com>\n"
+		"\n"
+		"Message";
+	assert_tag_fails(tag, 0);
+}
+
+void test_object_tag_parse__missing_message_fails(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tag Tag\n"
+		"tagger taggy@taggart.com>\n";
+	assert_tag_fails(tag, 0);
+}
+
+void test_object_tag_parse__no_oob_read_when_searching_message(void)
+{
+	const char *tag =
+		"object a8d447f68076d1520f69649bb52629941be7031f\n"
+		"type tag\n"
+		"tag \n"
+		"tagger <>\n"
+		" \n\n"
+		"Message";
+	/*
+	 * The OOB read previously resulted in an OOM error. We
+	 * thus want to make sure that the resulting error is the
+	 * expected one.
+	 */
+	assert_tag_fails(tag, strlen(tag) - strlen("\n\nMessage"));
+	cl_assert(strstr(giterr_last()->message, "tag contains no message"));
+}