patch application: apply binary patches

Handle the application of binary patches.  Include tests that
produce a binary patch (an in-memory `git_patch` object),
then enusre that the patch applies correctly.

src/apply.c

@@ -13,6 +13,8 @@
 #include "diff_patch.h"
 #include "fileops.h"
 #include "apply.h"
+#include "delta.h"
+#include "zstream.h"
 
 #define apply_err(...) \
 	( giterr_set(GITERR_PATCH, __VA_ARGS__), -1 )
@@ -239,6 +241,87 @@ done:
 	return error;
 }
 
+static int apply_binary_delta(
+	git_buf *out,
+	const char *source,
+	size_t source_len,
+	git_diff_binary_file *binary_file)
+{
+	git_buf inflated = GIT_BUF_INIT;
+	int error = 0;
+
+	/* no diff means identical contents */
+	if (binary_file->datalen == 0)
+		return git_buf_put(out, source, source_len);
+
+	error = git_zstream_inflatebuf(&inflated,
+		binary_file->data, binary_file->datalen);
+
+	if (!error && inflated.size != binary_file->inflatedlen) {
+		error = apply_err("inflated delta does not match expected length");
+		git_buf_free(out);
+	}
+
+	if (error < 0)
+		goto done;
+
+	if (binary_file->type == GIT_DIFF_BINARY_DELTA) {
+		void *data;
+		size_t data_len;
+
+		error = git_delta_apply(&data, &data_len, (void *)source, source_len,
+			(void *)inflated.ptr, inflated.size);
+
+		out->ptr = data;
+		out->size = data_len;
+		out->asize = data_len;
+	}
+	else if (binary_file->type == GIT_DIFF_BINARY_LITERAL) {
+		git_buf_swap(out, &inflated);
+	}
+	else {
+		error = apply_err("unknown binary delta type");
+		goto done;
+	}
+
+done:
+	git_buf_free(&inflated);
+	return error;
+}
+
+static int apply_binary(
+	git_buf *out,
+	const char *source,
+	size_t source_len,
+	git_patch *patch)
+{
+	git_buf reverse = GIT_BUF_INIT;
+	int error;
+
+	/* first, apply the new_file delta to the given source */
+	if ((error = apply_binary_delta(out, source, source_len,
+			&patch->binary.new_file)) < 0)
+		goto done;
+
+	/* second, apply the old_file delta to sanity check the result */
+	if ((error = apply_binary_delta(&reverse, out->ptr, out->size,
+			&patch->binary.old_file)) < 0)
+		goto done;
+
+	if (source_len != reverse.size ||
+		memcmp(source, reverse.ptr, source_len) != 0) {
+		error = apply_err("binary patch did not apply cleanly");
+		goto done;
+	}
+
+done:
+	if (error < 0)
+		git_buf_free(out);
+
+	git_buf_free(&reverse);
+	return error;
+}
+
 int git_apply__patch(
 	git_buf *contents_out,
 	char **filename_out,
@@ -262,10 +345,14 @@ int git_apply__patch(
 			patch->nfile.file->mode : GIT_FILEMODE_BLOB;
 	}
 
-	/* If the patch is empty, simply keep the source unchanged */
-	if (patch->hunks.size == 0)
-		git_buf_put(contents_out, source, source_len);
-	else if ((error = apply_hunks(contents_out, source, source_len, patch)) < 0)
+	if (patch->delta->flags & GIT_DIFF_FLAG_BINARY)
+		error = apply_binary(contents_out, source, source_len, patch);
+	else if (patch->hunks.size)
+		error = apply_hunks(contents_out, source, source_len, patch);
+	else
+		error = git_buf_put(contents_out, source, source_len);
+
+	if (error)
 		goto done;
 
 	if (patch->delta->status == GIT_DELTA_DELETED &&

src/diff_print.c

@@ -415,6 +415,9 @@ static int diff_print_patch_file_binary(
 		(error = diff_print_load_content(pi, delta)) < 0)
 		return error;
 
+	if (binary->new_file.datalen == 0 && binary->old_file.datalen == 0)
+		return 0;
+
 	pre_binary_size = pi->buf->size;
 	git_buf_printf(pi->buf, "GIT binary patch\n");
 	pi->line.num_lines++;

tests/apply/apply_common.h

@@ -473,3 +473,102 @@
 	"+patch file\n" \
 	"-it's something else\n" \
 	" entirely!"
+
+/* binary contents */
+
+#define FILE_BINARY_LITERAL_ORIGINAL "\x00\x00\x0a"
+#define FILE_BINARY_LITERAL_ORIGINAL_LEN 3
+
+#define FILE_BINARY_LITERAL_MODIFIED "\x00\x00\x01\x02\x0a"
+#define FILE_BINARY_LITERAL_MODIFIED_LEN 5
+
+#define PATCH_BINARY_LITERAL \
+	"diff --git a/binary.bin b/binary.bin\n" \
+	"index bd474b2519cc15eab801ff851cc7d50f0dee49a1..9ac35ff15cd8864aeafd889e4826a3150f0b06c4 100644\n" \
+	"GIT binary patch\n" \
+	"literal 5\n" \
+	"Mc${NkU}WL~000&M4gdfE\n" \
+	"\n" \
+	"literal 3\n" \
+	"Kc${Nk-~s>u4FC%O\n\n"
+
+#define FILE_BINARY_DELTA_ORIGINAL \
+	"\x00\x00\x01\x02\x00\x00\x01\x02\x00\x00\x01\x02\x0a\x54\x68\x69" \
+	"\x73\x20\x69\x73\x20\x61\x20\x62\x69\x6e\x61\x72\x79\x20\x66\x69" \
+	"\x6c\x65\x2c\x20\x62\x79\x20\x76\x69\x72\x74\x75\x65\x20\x6f\x66" \
+	"\x20\x68\x61\x76\x69\x6e\x67\x20\x73\x6f\x6d\x65\x20\x6e\x75\x6c" \
+	"\x6c\x73\x2e\x0a\x00\x00\x01\x02\x00\x00\x01\x02\x00\x00\x01\x02" \
+	"\x0a\x57\x65\x27\x72\x65\x20\x67\x6f\x69\x6e\x67\x20\x74\x6f\x20" \
+	"\x63\x68\x61\x6e\x67\x65\x20\x70\x6f\x72\x74\x69\x6f\x6e\x73\x20" \
+	"\x6f\x66\x20\x69\x74\x2e\x0a\x00\x00\x01\x02\x00\x00\x01\x02\x00" \
+	"\x00\x01\x02\x0a\x53\x6f\x20\x74\x68\x61\x74\x20\x77\x65\x20\x67" \
+	"\x69\x74\x20\x61\x20\x62\x69\x6e\x61\x72\x79\x20\x64\x65\x6c\x74" \
+	"\x61\x20\x69\x6e\x73\x74\x65\x61\x64\x20\x6f\x66\x20\x74\x68\x65" \
+	"\x20\x64\x65\x66\x6c\x61\x74\x65\x64\x20\x63\x6f\x6e\x74\x65\x6e" \
+	"\x74\x73\x2e\x0a\x00\x00\x01\x02\x00\x00\x01\x02\x00\x00\x01\x02" \
+	"\x0a"
+#define FILE_BINARY_DELTA_ORIGINAL_LEN 209
+
+#define FILE_BINARY_DELTA_MODIFIED \
+	"\x00\x00\x01\x02\x00\x00\x01\x02\x00\x00\x01\x02\x0a\x5a\x5a\x5a" \
+	"\x5a\x20\x69\x73\x20\x61\x20\x62\x69\x6e\x61\x72\x79\x20\x66\x69" \
+	"\x6c\x65\x2c\x20\x62\x79\x20\x76\x69\x72\x74\x75\x65\x20\x6f\x66" \
+	"\x20\x68\x61\x76\x69\x6e\x67\x20\x73\x6f\x6d\x65\x20\x6e\x75\x6c" \
+	"\x6c\x73\x2e\x0a\x00\x00\x01\x02\x00\x00\x01\x02\x00\x00\x01\x02" \
+	"\x0a\x57\x65\x27\x72\x65\x20\x67\x6f\x69\x6e\x67\x20\x74\x6f\x20" \
+	"\x63\x68\x61\x6e\x67\x65\x20\x70\x6f\x72\x74\x69\x6f\x6e\x73\x20" \
+	"\x6f\x66\x20\x49\x54\x2e\x0a\x00\x00\x01\x02\x00\x00\x01\x02\x00" \
+	"\x00\x01\x02\x0a\x53\x4f\x20\x74\x68\x61\x74\x20\x77\x65\x20\x67" \
+	"\x69\x74\x20\x61\x20\x62\x69\x6e\x61\x72\x79\x20\x64\x65\x6c\x74" \
+	"\x61\x20\x69\x6e\x73\x74\x65\x61\x64\x20\x6f\x66\x20\x74\x68\x65" \
+	"\x20\x64\x65\x66\x6c\x61\x74\x65\x64\x20\x63\x6f\x6e\x74\x65\x6e" \
+	"\x74\x73\x2e\x0a\x00\x00\x01\x02\x00\x00\x01\x02\x00\x00\x01\x02" \
+	"\x0a"
+#define FILE_BINARY_DELTA_MODIFIED_LEN 209
+
+#define PATCH_BINARY_DELTA \
+	"diff --git a/binary.bin b/binary.bin\n" \
+	"index 27184d9883b12c4c9c54b4a31137603586169f51..7c94f9e60bf366033d98e0d551ae37d30faef74a 100644\n" \
+	"GIT binary patch\n" \
+	"delta 48\n" \
+	"kc$~Y)c#%<%fq{_;hPk4EV4`4>uxE%K7m7r%|HL+L0In7XGynhq\n" \
+	"\n" \
+	"delta 48\n" \
+	"mc$~Y)c#%<%fq{_;hPgsAGK(h)CJASj=y9P)1m{m|^9BI99|yz$\n\n"
+
+#define PATCH_BINARY_ADD \
+	"diff --git a/binary.bin b/binary.bin\n" \
+	"new file mode 100644\n" \
+	"index 0000000000000000000000000000000000000000..7c94f9e60bf366033d98e0d551ae37d30faef74a\n" \
+	"GIT binary patch\n" \
+	"literal 209\n" \
+	"zc${60u?oUK5JXSQe8qG&;(u6KC<u0&+$Ohh?#kUJlD{_rLCL^0!@QXgcKh&k^H>C_\n" \
+	"zAhe=XX7rNzh<3&##YcwqNHmEKsP<&&m~%Zf;eX@Khr$?aExDmfqyyt+#l^I)3+LMg\n" \
+	"kxnAIj9Pfn_|Gh`fP7tlm6j#y{FJYg_IifRlR^R@A08f862mk;8\n" \
+	"\n" \
+	"literal 0\n" \
+	"Hc$@<O00001\n\n"
+
+#define PATCH_BINARY_DELETE \
+	"diff --git a/binary.bin b/binary.bin\n" \
+	"deleted file mode 100644\n" \
+	"index 7c94f9e60bf366033d98e0d551ae37d30faef74a..0000000000000000000000000000000000000000\n" \
+	"GIT binary patch\n" \
+	"literal 0\n" \
+	"Hc$@<O00001\n" \
+	"\n" \
+	"literal 209\n" \
+	"zc${60u?oUK5JXSQe8qG&;(u6KC<u0&+$Ohh?#kUJlD{_rLCL^0!@QXgcKh&k^H>C_\n" \
+	"zAhe=XX7rNzh<3&##YcwqNHmEKsP<&&m~%Zf;eX@Khr$?aExDmfqyyt+#l^I)3+LMg\n" \
+	"kxnAIj9Pfn_|Gh`fP7tlm6j#y{FJYg_IifRlR^R@A08f862mk;8\n\n"
+
+/* contains an old side that does not match the expected source */
+#define PATCH_BINARY_NOT_REVERSIBLE \
+	"diff --git a/binary.bin b/binary.bin\n" \
+	"index 27184d9883b12c4c9c54b4a31137603586169f51..7c94f9e60bf366033d98e0d551ae37d30faef74a 100644\n" \
+	"GIT binary patch\n" \
+	"literal 5\n" \
+	"Mc${NkU}WL~000&M4gdfE\n" \
+	"\n" \
+	"delta 48\n" \
+	"mc$~Y)c#%<%fq{_;hPgsAGK(h)CJASj=y9P)1m{m|^9BI99|yz$\n\n"

tests/apply/fromdiff.c

@@ -8,10 +8,13 @@
 #include "apply_common.h"
 
 static git_repository *repo = NULL;
+static git_diff_options binary_opts = GIT_DIFF_OPTIONS_INIT;
 
 void test_apply_fromdiff__initialize(void)
 {
 	repo = cl_git_sandbox_init("renames");
+
+	binary_opts.flags |= GIT_DIFF_SHOW_BINARY;
 }
 
 void test_apply_fromdiff__cleanup(void)
@@ -19,10 +22,10 @@ void test_apply_fromdiff__cleanup(void)
 	cl_git_sandbox_cleanup();
 }
 
-static int apply_buf(
-	const char *old,
+static int apply_gitbuf(
+	const git_buf *old,
 	const char *oldname,
-	const char *new,
+	const git_buf *new,
 	const char *newname,
 	const char *patch_expected,
 	const git_diff_options *diff_opts)
@@ -35,22 +38,27 @@ static int apply_buf(
 	int error;
 
 	cl_git_pass(git_patch_from_buffers(&patch,
-		old, old ? strlen(old) : 0, oldname,
-		new, new ? strlen(new) : 0, newname,
+		old ? old->ptr : NULL, old ? old->size : 0,
+		oldname,
+		new ? new->ptr : NULL, new ? new->size : 0,
+		newname,
 		diff_opts));
-	cl_git_pass(git_patch_to_buf(&patchbuf, patch));
 
+	if (patch_expected) {
+		cl_git_pass(git_patch_to_buf(&patchbuf, patch));
 		cl_assert_equal_s(patch_expected, patchbuf.ptr);
+	}
 
-	error = git_apply__patch(&result, &filename, &mode, old, old ? strlen(old) : 0, patch);
+	error = git_apply__patch(&result, &filename, &mode, old ? old->ptr : NULL, old ? old->size : 0, patch);
 
 	if (error == 0 && new == NULL) {
 		cl_assert_equal_i(0, result.size);
 		cl_assert_equal_p(NULL, filename);
 		cl_assert_equal_i(0, mode);
-	} else {
-		cl_assert_equal_s(new, result.ptr);
-		cl_assert_equal_s("file.txt", filename);
+	}
+	else if (error == 0) {
+		cl_assert_equal_s(new->ptr, result.ptr);
+		cl_assert_equal_s(newname ? newname : oldname, filename);
 		cl_assert_equal_i(0100644, mode);
 	}
 
@@ -62,6 +70,32 @@ static int apply_buf(
 	return error;
 }
 
+static int apply_buf(
+	const char *old,
+	const char *oldname,
+	const char *new,
+	const char *newname,
+	const char *patch_expected,
+	const git_diff_options *diff_opts)
+{
+	git_buf o = GIT_BUF_INIT, n = GIT_BUF_INIT,
+		*optr = NULL, *nptr = NULL;
+
+	if (old) {
+		o.ptr = (char *)old;
+		o.size = strlen(old);
+		optr = &o;
+	}
+
+	if (new) {
+		n.ptr = (char *)new;
+		n.size = strlen(new);
+		nptr = &n;
+	}
+
+	return apply_gitbuf(optr, oldname, nptr, newname, patch_expected, diff_opts);
+}
+
 void test_apply_fromdiff__change_middle(void)
 {
 	cl_git_pass(apply_buf(
@@ -182,3 +216,74 @@ void test_apply_fromdiff__no_change(void)
 		FILE_ORIGINAL, "file.txt",
 		"", NULL));
 }
+
+void test_apply_fromdiff__binary_add(void)
+{
+	git_buf newfile = GIT_BUF_INIT;
+
+	newfile.ptr = FILE_BINARY_DELTA_MODIFIED;
+	newfile.size = FILE_BINARY_DELTA_MODIFIED_LEN;
+
+	cl_git_pass(apply_gitbuf(
+		NULL, NULL,
+		&newfile, "binary.bin",
+		NULL, &binary_opts));
+}
+
+void test_apply_fromdiff__binary_no_change(void)
+{
+	git_buf original = GIT_BUF_INIT;
+
+	original.ptr = FILE_BINARY_DELTA_ORIGINAL;
+	original.size = FILE_BINARY_DELTA_ORIGINAL_LEN;
+
+	cl_git_pass(apply_gitbuf(
+		&original, "binary.bin",
+		&original, "binary.bin",
+		"", &binary_opts));
+}
+
+void test_apply_fromdiff__binary_change_delta(void)
+{
+	git_buf original = GIT_BUF_INIT, modified = GIT_BUF_INIT;
+
+	original.ptr = FILE_BINARY_DELTA_ORIGINAL;
+	original.size = FILE_BINARY_DELTA_ORIGINAL_LEN;
+
+	modified.ptr = FILE_BINARY_DELTA_MODIFIED;
+	modified.size = FILE_BINARY_DELTA_MODIFIED_LEN;
+
+	cl_git_pass(apply_gitbuf(
+		&original, "binary.bin",
+		&modified, "binary.bin",
+		NULL, &binary_opts));
+}
+
+void test_apply_fromdiff__binary_change_literal(void)
+{
+	git_buf original = GIT_BUF_INIT, modified = GIT_BUF_INIT;
+
+	original.ptr = FILE_BINARY_LITERAL_ORIGINAL;
+	original.size = FILE_BINARY_LITERAL_ORIGINAL_LEN;
+
+	modified.ptr = FILE_BINARY_LITERAL_MODIFIED;
+	modified.size = FILE_BINARY_LITERAL_MODIFIED_LEN;
+
+	cl_git_pass(apply_gitbuf(
+		&original, "binary.bin",
+		&modified, "binary.bin",
+		NULL, &binary_opts));
+}
+
+void test_apply_fromdiff__binary_delete(void)
+{
+	git_buf original = GIT_BUF_INIT;
+
+	original.ptr = FILE_BINARY_DELTA_MODIFIED;
+	original.size = FILE_BINARY_DELTA_MODIFIED_LEN;
+
+	cl_git_pass(apply_gitbuf(
+		&original, "binary.bin",
+		NULL, NULL,
+		NULL, &binary_opts));
+}

tests/apply/fromfile.c

@@ -299,3 +299,30 @@ void test_apply_fromfile__fail_not_a_patch(void)
 	cl_git_fail(git_patch_from_patchfile(&patch, PATCH_NOT_A_PATCH,
 		strlen(PATCH_NOT_A_PATCH)));
 }
+
+/*
+void test_apply_fromdiff__binary_change_must_be_reversible(void)
+{
+	git_buf original = GIT_BUF_INIT, modified = GIT_BUF_INIT,
+		result = GIT_BUF_INIT;
+	char *filename;
+	unsigned int mode;
+
+	original.ptr = FILE_BINARY_DELTA_ORIGINAL;
+	original.size = FILE_BINARY_DELTA_ORIGINAL_LEN;
+
+	modified.ptr = FILE_BINARY_DELTA_MODIFIED;
+	modified.size = FILE_BINARY_DELTA_MODIFIED_LEN;
+
+	cl_git_fail(git_apply__patch(&result, &filename, &mode, old ? old->ptr : NULL, old ? old->size : 0, patch);
+
+	cl_git_fail(apply_gitbuf(
+		&original, "binary.bin",
+		&modified, "binary.bin",
+		PATCH_BINARY_NOT_REVERSIBLE, &binary_opts));
+	cl_assert_equal_s("binary patch did not apply cleanly", giterr_last()->message);
+
+	git_buf_free(&result);
+	git__free(filename);
+}
+*/