Merge pull request #3564 from ethomson/merge_drivers

Custom merge drivers and proper gitattributes `merge` handling

6 files changed, 690 insertions, 114 deletions

diff --git a/src/global.c b/src/global.c
index c725b5184..cbd12ddda 100644
--- a/src/global.c
+++ b/src/global.c
@@ -9,6 +9,7 @@
 #include "hash.h"
 #include "sysdir.h"
 #include "filter.h"
+#include "merge_driver.h"
 #include "openssl_stream.h"
 #include "thread-utils.h"
 #include "git2/global.h"
@@ -59,6 +60,7 @@ static int init_common(void)
 	if ((ret = git_hash_global_init()) == 0 &&
 		(ret = git_sysdir_global_init()) == 0 &&
 		(ret = git_filter_global_init()) == 0 &&
+		(ret = git_merge_driver_global_init()) == 0 &&
 		(ret = git_transport_ssh_global_init()) == 0)
 		ret = git_openssl_stream_global_init();
 
diff --git a/src/merge.c b/src/merge.c
index d2f92ccce..a0f2405ff 100644
--- a/src/merge.c
+++ b/src/merge.c
@@ -29,6 +29,7 @@
 #include "annotated_commit.h"
 #include "commit.h"
 #include "oidarray.h"
+#include "merge_driver.h"
 
 #include "git2/types.h"
 #include "git2/repository.h"
@@ -50,18 +51,6 @@
 #define GIT_MERGE_INDEX_ENTRY_ISFILE(X) S_ISREG((X).mode)
 
 
-/** Internal merge flags. */
-enum {
-	/** The merge is for a virtual base in a recursive merge. */
-	GIT_MERGE__VIRTUAL_BASE = (1 << 31),
-};
-
-enum {
-	/** Accept the conflict file, staging it as the merge result. */
-	GIT_MERGE_FILE_FAVOR__CONFLICTED = 4,
-};
-
-
 typedef enum {
 	TREE_IDX_ANCESTOR = 0,
 	TREE_IDX_OURS = 1,
@@ -273,7 +262,7 @@ int git_merge_base(git_oid *out, git_repository *repo, const git_oid *one, const
 int git_merge_bases(git_oidarray *out, git_repository *repo, const git_oid *one, const git_oid *two)
 {
 	int error;
-        git_revwalk *walk;
+	git_revwalk *walk;
 	git_commit_list *result, *list;
 	git_array_oid_t array;
 
@@ -810,76 +799,158 @@ static int merge_conflict_resolve_one_renamed(
 	return error;
 }
 
-static int merge_conflict_resolve_automerge(
-	int *resolved,
-	git_merge_diff_list *diff_list,
-	const git_merge_diff *conflict,
-	const git_merge_file_options *file_opts)
+static bool merge_conflict_can_resolve_contents(
+	const git_merge_diff *conflict)
 {
-	const git_index_entry *ancestor = NULL, *ours = NULL, *theirs = NULL;
-	git_merge_file_result result = {0};
-	git_index_entry *index_entry;
-	git_odb *odb = NULL;
-	git_oid automerge_oid;
-	int error = 0;
-
-	assert(resolved && diff_list && conflict);
-
-	*resolved = 0;
-
 	if (!GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->our_entry) ||
 		!GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->their_entry))
-		return 0;
+		return false;
 
 	/* Reject D/F conflicts */
 	if (conflict->type == GIT_MERGE_DIFF_DIRECTORY_FILE)
-		return 0;
+		return false;
 
 	/* Reject submodules. */
 	if (S_ISGITLINK(conflict->ancestor_entry.mode) ||
 		S_ISGITLINK(conflict->our_entry.mode) ||
 		S_ISGITLINK(conflict->their_entry.mode))
-		return 0;
+		return false;
 
 	/* Reject link/file conflicts. */
-	if ((S_ISLNK(conflict->ancestor_entry.mode) ^ S_ISLNK(conflict->our_entry.mode)) ||
-		(S_ISLNK(conflict->ancestor_entry.mode) ^ S_ISLNK(conflict->their_entry.mode)))
-		return 0;
+	if ((S_ISLNK(conflict->ancestor_entry.mode) ^
+			S_ISLNK(conflict->our_entry.mode)) ||
+		(S_ISLNK(conflict->ancestor_entry.mode) ^
+			S_ISLNK(conflict->their_entry.mode)))
+		return false;
 
 	/* Reject name conflicts */
 	if (conflict->type == GIT_MERGE_DIFF_BOTH_RENAMED_2_TO_1 ||
 		conflict->type == GIT_MERGE_DIFF_RENAMED_ADDED)
-		return 0;
+		return false;
 
 	if ((conflict->our_status & GIT_DELTA_RENAMED) == GIT_DELTA_RENAMED &&
 		(conflict->their_status & GIT_DELTA_RENAMED) == GIT_DELTA_RENAMED &&
 		strcmp(conflict->ancestor_entry.path, conflict->their_entry.path) != 0)
+		return false;
+
+	return true;
+}
+
+static int merge_conflict_invoke_driver(
+	git_index_entry **out,
+	const char *name,
+	git_merge_driver *driver,
+	git_merge_diff_list *diff_list,
+	git_merge_driver_source *src)
+{
+	git_index_entry *result;
+	git_buf buf = GIT_BUF_INIT;
+	const char *path;
+	uint32_t mode;
+	git_odb *odb = NULL;
+	git_oid oid;
+	int error;
+
+	*out = NULL;
+
+	if ((error = driver->apply(driver, &path, &mode, &buf, name, src)) < 0 ||
+		(error = git_repository_odb(&odb, src->repo)) < 0 ||
+		(error = git_odb_write(&oid, odb, buf.ptr, buf.size, GIT_OBJ_BLOB)) < 0)
+		goto done;
+
+	result = git_pool_mallocz(&diff_list->pool, sizeof(git_index_entry));
+	GITERR_CHECK_ALLOC(result);
+
+	git_oid_cpy(&result->id, &oid);
+	result->mode = mode;
+	result->file_size = buf.size;
+
+	result->path = git_pool_strdup(&diff_list->pool, path);
+	GITERR_CHECK_ALLOC(result->path);
+
+	*out = result;
+
+done:
+	git_buf_free(&buf);
+	git_odb_free(odb);
+
+	return error;
+}
+
+static int merge_conflict_resolve_contents(
+	int *resolved,
+	git_merge_diff_list *diff_list,
+	const git_merge_diff *conflict,
+	const git_merge_options *merge_opts,
+	const git_merge_file_options *file_opts)
+{
+	git_merge_driver_source source = {0};
+	git_merge_file_result result = {0};
+	git_merge_driver *driver;
+	git_merge_driver__builtin builtin = {{0}};
+	git_index_entry *merge_result;
+	git_odb *odb = NULL;
+	const char *name;
+	bool fallback = false;
+	int error;
+
+	assert(resolved && diff_list && conflict);
+
+	*resolved = 0;
+
+	if (!merge_conflict_can_resolve_contents(conflict))
 		return 0;
 
-	ancestor = GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->ancestor_entry) ?
+	source.repo = diff_list->repo;
+	source.default_driver = merge_opts->default_driver;
+	source.file_opts = file_opts;
+	source.ancestor = GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->ancestor_entry) ?
 		&conflict->ancestor_entry : NULL;
-	ours = GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->our_entry) ?
+	source.ours = GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->our_entry) ?
 		&conflict->our_entry : NULL;
-	theirs = GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->their_entry) ?
+	source.theirs = GIT_MERGE_INDEX_ENTRY_EXISTS(conflict->their_entry) ?
 		&conflict->their_entry : NULL;
 
-	if ((error = git_repository_odb(&odb, diff_list->repo)) < 0 ||
-		(error = git_merge_file_from_index(&result, diff_list->repo, ancestor, ours, theirs, file_opts)) < 0 ||
-		(!result.automergeable && !(file_opts->flags & GIT_MERGE_FILE_FAVOR__CONFLICTED)) ||
-		(error = git_odb_write(&automerge_oid, odb, result.ptr, result.len, GIT_OBJ_BLOB)) < 0)
-		goto done;
+	if (file_opts->favor != GIT_MERGE_FILE_FAVOR_NORMAL) {
+		/* if the user requested a particular type of resolution (via the
+		 * favor flag) then let that override the gitattributes and use
+		 * the builtin driver.
+		 */
+		name = "text";
+		builtin.base.apply = git_merge_driver__builtin_apply;
+		builtin.favor = file_opts->favor;
+
+		driver = &builtin.base;
+	} else {
+		/* find the merge driver for this file */
+		if ((error = git_merge_driver_for_source(&name, &driver, &source)) < 0)
+			goto done;
+
+		if (driver == NULL)
+			fallback = true;
+	}
 
-	if ((index_entry = git_pool_mallocz(&diff_list->pool, sizeof(git_index_entry))) == NULL)
-	GITERR_CHECK_ALLOC(index_entry);
+	if (driver) {
+		error = merge_conflict_invoke_driver(&merge_result, name, driver,
+			diff_list, &source);
 
-	index_entry->path = git_pool_strdup(&diff_list->pool, result.path);
-	GITERR_CHECK_ALLOC(index_entry->path);
+		if (error == GIT_PASSTHROUGH)
+			fallback = true;
+	}
 
-	index_entry->file_size = result.len;
-	index_entry->mode = result.mode;
-	git_oid_cpy(&index_entry->id, &automerge_oid);
+	if (fallback) {
+		error = merge_conflict_invoke_driver(&merge_result, "text",
+			&git_merge_driver__text.base, diff_list, &source);
+	}
 
-	git_vector_insert(&diff_list->staged, index_entry);
+	if (error < 0) {
+		if (error == GIT_EMERGECONFLICT)
+			error = 0;
+
+		goto done;
+	}
+
+	git_vector_insert(&diff_list->staged, merge_result);
 	git_vector_insert(&diff_list->resolved, (git_merge_diff *)conflict);
 
 	*resolved = 1;
@@ -895,6 +966,7 @@ static int merge_conflict_resolve(
 	int *out,
 	git_merge_diff_list *diff_list,
 	const git_merge_diff *conflict,
+	const git_merge_options *merge_opts,
 	const git_merge_file_options *file_opts)
 {
 	int resolved = 0;
@@ -902,16 +974,20 @@ static int merge_conflict_resolve(
 
 	*out = 0;
 
-	if ((error = merge_conflict_resolve_trivial(&resolved, diff_list, conflict)) < 0)
+	if ((error = merge_conflict_resolve_trivial(
+			&resolved, diff_list, conflict)) < 0)
 		goto done;
 
-	if (!resolved && (error = merge_conflict_resolve_one_removed(&resolved, diff_list, conflict)) < 0)
+	if (!resolved && (error = merge_conflict_resolve_one_removed(
+			&resolved, diff_list, conflict)) < 0)
 		goto done;
 
-	if (!resolved && (error = merge_conflict_resolve_one_renamed(&resolved, diff_list, conflict)) < 0)
+	if (!resolved && (error = merge_conflict_resolve_one_renamed(
+			&resolved, diff_list, conflict)) < 0)
 		goto done;
 
-	if (!resolved && (error = merge_conflict_resolve_automerge(&resolved, diff_list, conflict, file_opts)) < 0)
+	if (!resolved && (error = merge_conflict_resolve_contents(
+			&resolved, diff_list, conflict, merge_opts, file_opts)) < 0)
 		goto done;
 
 	*out = resolved;
@@ -1627,6 +1703,7 @@ static int merge_normalize_opts(
 	const git_merge_options *given)
 {
 	git_config *cfg = NULL;
+	git_config_entry *entry = NULL;
 	int error = 0;
 
 	assert(repo && opts);
@@ -1644,6 +1721,22 @@ static int merge_normalize_opts(
 		opts->rename_threshold = GIT_MERGE_DEFAULT_RENAME_THRESHOLD;
 	}
 
+	if (given && given->default_driver) {
+		opts->default_driver = git__strdup(given->default_driver);
+		GITERR_CHECK_ALLOC(opts->default_driver);
+	} else {
+		error = git_config_get_entry(&entry, cfg, "merge.default");
+
+		if (error == 0) {
+			opts->default_driver = git__strdup(entry->value);
+			GITERR_CHECK_ALLOC(opts->default_driver);
+		} else if (error == GIT_ENOTFOUND) {
+			error = 0;
+		} else {
+			goto done;
+		}
+	}
+
 	if (!opts->target_limit) {
 		int limit = git_config__get_int_force(cfg, "merge.renamelimit", 0);
 
@@ -1666,7 +1759,9 @@ static int merge_normalize_opts(
 		opts->metric->payload = (void *)GIT_HASHSIG_SMART_WHITESPACE;
 	}
 
-	return 0;
+done:
+	git_config_entry_free(entry);
+	return error;
 }
 
 
@@ -1878,7 +1973,7 @@ int git_merge__iterators(
 		int resolved = 0;
 
 		if ((error = merge_conflict_resolve(
-			&resolved, diff_list, conflict, &file_opts)) < 0)
+			&resolved, diff_list, conflict, &opts, &file_opts)) < 0)
 			goto done;
 
 		if (!resolved) {
@@ -1899,6 +1994,8 @@ done:
 	if (!given_opts || !given_opts->metric)
 		git__free(opts.metric);
 
+	git__free((char *)opts.default_driver);
+
 	git_merge_diff_list__free(diff_list);
 	git_iterator_free(empty_ancestor);
 	git_iterator_free(empty_ours);
@@ -2111,14 +2208,14 @@ static int merge_annotated_commits(
 	git_iterator *base_iter = NULL, *our_iter = NULL, *their_iter = NULL;
 	int error;
 
-    if ((error = compute_base(&base, repo, ours, theirs, opts,
+	if ((error = compute_base(&base, repo, ours, theirs, opts,
 		recursion_level)) < 0) {
 
-        if (error != GIT_ENOTFOUND)
-            goto done;
+		if (error != GIT_ENOTFOUND)
+			goto done;
 
-        giterr_clear();
-    }
+		giterr_clear();
+	}
 
 	if ((error = iterator_for_annotated_commit(&base_iter, base)) < 0 ||
 		(error = iterator_for_annotated_commit(&our_iter, ours)) < 0 ||
diff --git a/src/merge.h b/src/merge.h
index bd839be49..f8cac161f 100644
--- a/src/merge.h
+++ b/src/merge.h
@@ -12,8 +12,9 @@
 #include "pool.h"
 #include "iterator.h"
 
-#include "git2/merge.h"
 #include "git2/types.h"
+#include "git2/merge.h"
+#include "git2/sys/merge.h"
 
 #define GIT_MERGE_MSG_FILE		"MERGE_MSG"
 #define GIT_MERGE_MODE_FILE		"MERGE_MODE"
@@ -22,6 +23,19 @@
 #define GIT_MERGE_DEFAULT_RENAME_THRESHOLD	50
 #define GIT_MERGE_DEFAULT_TARGET_LIMIT		1000
 
+
+/** Internal merge flags. */
+enum {
+	/** The merge is for a virtual base in a recursive merge. */
+	GIT_MERGE__VIRTUAL_BASE = (1 << 31),
+};
+
+enum {
+	/** Accept the conflict file, staging it as the merge result. */
+	GIT_MERGE_FILE_FAVOR__CONFLICTED = 4,
+};
+
+
 /** Types of changes when files are merged from branch to branch. */
 typedef enum {
 	/* No conflict - a change only occurs in one branch. */
@@ -70,7 +84,6 @@ typedef enum {
 	GIT_MERGE_DIFF_DF_CHILD = (1 << 11),
 } git_merge_diff_type_t;
 
-
 typedef struct {
 	git_repository *repo;
 	git_pool pool;
@@ -152,4 +165,50 @@ int git_merge__check_result(git_repository *repo, git_index *index_new);
 
 int git_merge__append_conflicts_to_merge_msg(git_repository *repo, git_index *index);
 
+/* Merge files */
+
+GIT_INLINE(const char *) git_merge_file__best_path(
+	const char *ancestor,
+	const char *ours,
+	const char *theirs)
+{
+	if (!ancestor) {
+		if (ours && theirs && strcmp(ours, theirs) == 0)
+			return ours;
+
+		return NULL;
+	}
+
+	if (ours && strcmp(ancestor, ours) == 0)
+		return theirs;
+	else if(theirs && strcmp(ancestor, theirs) == 0)
+		return ours;
+
+	return NULL;
+}
+
+GIT_INLINE(uint32_t) git_merge_file__best_mode(
+	uint32_t ancestor, uint32_t ours, uint32_t theirs)
+{
+	/*
+	 * If ancestor didn't exist and either ours or theirs is executable,
+	 * assume executable.  Otherwise, if any mode changed from the ancestor,
+	 * use that one.
+	 */
+	if (!ancestor) {
+		if (ours == GIT_FILEMODE_BLOB_EXECUTABLE ||
+			theirs == GIT_FILEMODE_BLOB_EXECUTABLE)
+			return GIT_FILEMODE_BLOB_EXECUTABLE;
+
+		return GIT_FILEMODE_BLOB;
+	} else if (ours && theirs) {
+		if (ancestor == ours)
+			return theirs;
+
+		return ours;
+	}
+
+	return 0;
+}
+
 #endif
diff --git a/src/merge_driver.c b/src/merge_driver.c
new file mode 100644
index 000000000..cc039dbb5
--- /dev/null
+++ b/src/merge_driver.c
@@ -0,0 +1,396 @@
+/*
+ * Copyright (C) the libgit2 contributors. All rights reserved.
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+
+#include "common.h"
+#include "vector.h"
+#include "global.h"
+#include "merge.h"
+#include "merge_driver.h"
+#include "git2/merge.h"
+#include "git2/sys/merge.h"
+
+static const char *merge_driver_name__text = "text";
+static const char *merge_driver_name__union = "union";
+static const char *merge_driver_name__binary = "binary";
+
+struct merge_driver_registry {
+	git_rwlock lock;
+	git_vector drivers;
+};
+
+typedef struct {
+	git_merge_driver *driver;
+	int initialized;
+	char name[GIT_FLEX_ARRAY];
+} git_merge_driver_entry;
+
+static struct merge_driver_registry merge_driver_registry;
+
+static void git_merge_driver_global_shutdown(void);
+
+
+int git_merge_driver__builtin_apply(
+	git_merge_driver *self,
+	const char **path_out,
+	uint32_t *mode_out,
+	git_buf *merged_out,
+	const char *filter_name,
+	const git_merge_driver_source *src)
+{
+	git_merge_driver__builtin *driver = (git_merge_driver__builtin *)self;
+	git_merge_file_options file_opts = GIT_MERGE_FILE_OPTIONS_INIT;
+	git_merge_file_result result = {0};
+	int error;
+
+	GIT_UNUSED(filter_name);
+
+	if (src->file_opts)
+		memcpy(&file_opts, src->file_opts, sizeof(git_merge_file_options));
+
+	if (driver->favor)
+		file_opts.favor = driver->favor;
+
+	if ((error = git_merge_file_from_index(&result, src->repo,
+		src->ancestor, src->ours, src->theirs, &file_opts)) < 0)
+		goto done;
+
+	if (!result.automergeable &&
+		!(file_opts.flags & GIT_MERGE_FILE_FAVOR__CONFLICTED)) {
+		error = GIT_EMERGECONFLICT;
+		goto done;
+	}
+
+	*path_out = git_merge_file__best_path(
+		src->ancestor ? src->ancestor->path : NULL,
+		src->ours ? src->ours->path : NULL,
+		src->theirs ? src->theirs->path : NULL);
+
+	*mode_out = git_merge_file__best_mode(
+		src->ancestor ? src->ancestor->mode : 0,
+		src->ours ? src->ours->mode : 0,
+		src->theirs ? src->theirs->mode : 0);
+
+	merged_out->ptr = (char *)result.ptr;
+	merged_out->size = result.len;
+	merged_out->asize = result.len;
+	result.ptr = NULL;
+
+done:
+	git_merge_file_result_free(&result);
+	return error;
+}
+
+static int merge_driver_binary_apply(
+	git_merge_driver *self,
+	const char **path_out,
+	uint32_t *mode_out,
+	git_buf *merged_out,
+	const char *filter_name,
+	const git_merge_driver_source *src)
+{
+	GIT_UNUSED(self);
+	GIT_UNUSED(path_out);
+	GIT_UNUSED(mode_out);
+	GIT_UNUSED(merged_out);
+	GIT_UNUSED(filter_name);
+	GIT_UNUSED(src);
+
+	return GIT_EMERGECONFLICT;
+}
+
+static int merge_driver_entry_cmp(const void *a, const void *b)
+{
+	const git_merge_driver_entry *entry_a = a;
+	const git_merge_driver_entry *entry_b = b;
+
+	return strcmp(entry_a->name, entry_b->name);
+}
+
+static int merge_driver_entry_search(const void *a, const void *b)
+{
+	const char *name_a = a;
+	const git_merge_driver_entry *entry_b = b;
+
+	return strcmp(name_a, entry_b->name);
+}
+
+git_merge_driver__builtin git_merge_driver__text = {
+	{
+		GIT_MERGE_DRIVER_VERSION,
+		NULL,
+		NULL,
+		git_merge_driver__builtin_apply,
+	},
+	GIT_MERGE_FILE_FAVOR_NORMAL
+};
+
+git_merge_driver__builtin git_merge_driver__union = {
+	{
+		GIT_MERGE_DRIVER_VERSION,
+		NULL,
+		NULL,
+		git_merge_driver__builtin_apply,
+	},
+	GIT_MERGE_FILE_FAVOR_UNION
+};
+
+git_merge_driver git_merge_driver__binary = {
+	GIT_MERGE_DRIVER_VERSION,
+	NULL,
+	NULL,
+	merge_driver_binary_apply
+};
+
+/* Note: callers must lock the registry before calling this function */
+static int merge_driver_registry_insert(
+	const char *name, git_merge_driver *driver)
+{
+	git_merge_driver_entry *entry;
+
+	entry = git__calloc(1, sizeof(git_merge_driver_entry) + strlen(name) + 1);
+	GITERR_CHECK_ALLOC(entry);
+
+	strcpy(entry->name, name);
+	entry->driver = driver;
+
+	return git_vector_insert_sorted(
+		&merge_driver_registry.drivers, entry, NULL);
+}
+
+int git_merge_driver_global_init(void)
+{
+	int error;
+
+	if (git_rwlock_init(&merge_driver_registry.lock) < 0)
+		return -1;
+
+	if ((error = git_vector_init(&merge_driver_registry.drivers, 3,
+		merge_driver_entry_cmp)) < 0)
+		goto done;
+
+	if ((error = merge_driver_registry_insert(
+			merge_driver_name__text, &git_merge_driver__text.base)) < 0 ||
+		(error = merge_driver_registry_insert(
+			merge_driver_name__union, &git_merge_driver__union.base)) < 0 ||
+		(error = merge_driver_registry_insert(
+			merge_driver_name__binary, &git_merge_driver__binary)) < 0)
+
+	git__on_shutdown(git_merge_driver_global_shutdown);
+
+done:
+	if (error < 0)
+		git_vector_free_deep(&merge_driver_registry.drivers);
+
+	return error;
+}
+
+static void git_merge_driver_global_shutdown(void)
+{
+	git_merge_driver_entry *entry;
+	size_t i;
+
+	if (git_rwlock_wrlock(&merge_driver_registry.lock) < 0)
+		return;
+
+	git_vector_foreach(&merge_driver_registry.drivers, i, entry) {
+		if (entry->driver->shutdown)
+			entry->driver->shutdown(entry->driver);
+
+		git__free(entry);
+	}
+
+	git_vector_free(&merge_driver_registry.drivers);
+
+	git_rwlock_wrunlock(&merge_driver_registry.lock);
+	git_rwlock_free(&merge_driver_registry.lock);
+}
+
+/* Note: callers must lock the registry before calling this function */
+static int merge_driver_registry_find(size_t *pos, const char *name)
+{
+	return git_vector_search2(pos, &merge_driver_registry.drivers,
+		merge_driver_entry_search, name);
+}
+
+/* Note: callers must lock the registry before calling this function */
+static git_merge_driver_entry *merge_driver_registry_lookup(
+	size_t *pos, const char *name)
+{
+	git_merge_driver_entry *entry = NULL;
+
+	if (!merge_driver_registry_find(pos, name))
+		entry = git_vector_get(&merge_driver_registry.drivers, *pos);
+
+	return entry;
+}
+
+int git_merge_driver_register(const char *name, git_merge_driver *driver)
+{
+	int error;
+
+	assert(name && driver);
+
+	if (git_rwlock_wrlock(&merge_driver_registry.lock) < 0) {
+		giterr_set(GITERR_OS, "failed to lock merge driver registry");
+		return -1;
+	}
+
+	if (!merge_driver_registry_find(NULL, name)) {
+		giterr_set(GITERR_MERGE, "attempt to reregister existing driver '%s'",
+			name);
+		error = GIT_EEXISTS;
+		goto done;
+	}
+
+	error = merge_driver_registry_insert(name, driver);
+
+done:
+	git_rwlock_wrunlock(&merge_driver_registry.lock);
+	return error;
+}
+
+int git_merge_driver_unregister(const char *name)
+{
+	git_merge_driver_entry *entry;
+	size_t pos;
+	int error = 0;
+
+	if (git_rwlock_wrlock(&merge_driver_registry.lock) < 0) {
+		giterr_set(GITERR_OS, "failed to lock merge driver registry");
+		return -1;
+	}
+
+	if ((entry = merge_driver_registry_lookup(&pos, name)) == NULL) {
+		giterr_set(GITERR_MERGE, "cannot find merge driver '%s' to unregister",
+			name);
+		error = GIT_ENOTFOUND;
+		goto done;
+	}
+
+	git_vector_remove(&merge_driver_registry.drivers, pos);
+
+	if (entry->initialized && entry->driver->shutdown) {
+		entry->driver->shutdown(entry->driver);
+		entry->initialized = false;
+	}
+
+	git__free(entry);
+
+done:
+	git_rwlock_wrunlock(&merge_driver_registry.lock);
+	return error;
+}
+
+git_merge_driver *git_merge_driver_lookup(const char *name)
+{
+	git_merge_driver_entry *entry;
+	size_t pos;
+	int error;
+
+	/* If we've decided the merge driver to use internally - and not
+	 * based on user configuration (in merge_driver_name_for_path)
+	 * then we can use a hardcoded name to compare instead of bothering
+	 * to take a lock and look it up in the vector.
+	 */
+	if (name == merge_driver_name__text)
+		return &git_merge_driver__text.base;
+	else if (name == merge_driver_name__binary)
+		return &git_merge_driver__binary;
+
+	if (git_rwlock_rdlock(&merge_driver_registry.lock) < 0) {
+		giterr_set(GITERR_OS, "failed to lock merge driver registry");
+		return NULL;
+	}
+
+	entry = merge_driver_registry_lookup(&pos, name);
+
+	git_rwlock_rdunlock(&merge_driver_registry.lock);
+
+	if (entry == NULL) {
+		giterr_set(GITERR_MERGE, "cannot use an unregistered filter");
+		return NULL;
+	}
+
+	if (!entry->initialized) {
+		if (entry->driver->initialize &&
+			(error = entry->driver->initialize(entry->driver)) < 0)
+			return NULL;
+
+		entry->initialized = 1;
+	}
+
+	return entry->driver;
+}
+
+static int merge_driver_name_for_path(
+	const char **out,
+	git_repository *repo,
+	const char *path,
+	const char *default_driver)
+{
+	const char *value;
+	int error;
+
+	*out = NULL;
+
+	if ((error = git_attr_get(&value, repo, 0, path, "merge")) < 0)
+		return error;
+
+	/* set: use the built-in 3-way merge driver ("text") */
+	if (GIT_ATTR_TRUE(value))
+		*out = merge_driver_name__text;
+
+	/* unset: do not merge ("binary") */
+	else if (GIT_ATTR_FALSE(value))
+		*out = merge_driver_name__binary;
+
+	else if (GIT_ATTR_UNSPECIFIED(value) && default_driver)
+		*out = default_driver;
+
+	else if (GIT_ATTR_UNSPECIFIED(value))
+		*out = merge_driver_name__text;
+
+	else
+		*out = value;
+
+	return 0;
+}
+
+
+GIT_INLINE(git_merge_driver *) merge_driver_lookup_with_wildcard(
+	const char *name)
+{
+	git_merge_driver *driver = git_merge_driver_lookup(name);
+
+	if (driver == NULL)
+		driver = git_merge_driver_lookup("*");
+
+	return driver;
+}
+
+int git_merge_driver_for_source(
+	const char **name_out,
+	git_merge_driver **driver_out,
+	const git_merge_driver_source *src)
+{
+	const char *path, *driver_name;
+	int error = 0;
+
+	path = git_merge_file__best_path(
+		src->ancestor ? src->ancestor->path : NULL,
+		src->ours ? src->ours->path : NULL,
+		src->theirs ? src->theirs->path : NULL);
+
+	if ((error = merge_driver_name_for_path(
+			&driver_name, src->repo, path, src->default_driver)) < 0)
+		return error;
+
+	*name_out = driver_name;
+	*driver_out = merge_driver_lookup_with_wildcard(driver_name);
+	return error;
+}
+
diff --git a/src/merge_driver.h b/src/merge_driver.h
new file mode 100644
index 000000000..bde27502c
--- /dev/null
+++ b/src/merge_driver.h
@@ -0,0 +1,60 @@
+/*
+ * Copyright (C) the libgit2 contributors. All rights reserved.
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+#ifndef INCLUDE_merge_driver_h__
+#define INCLUDE_merge_driver_h__
+
+#include "git2/merge.h"
+#include "git2/index.h"
+#include "git2/sys/merge.h"
+
+struct git_merge_driver_source {
+	git_repository *repo;
+	const char *default_driver;
+	const git_merge_file_options *file_opts;
+
+	const git_index_entry *ancestor;
+	const git_index_entry *ours;
+	const git_index_entry *theirs;
+};
+
+typedef struct git_merge_driver__builtin {
+	git_merge_driver base;
+	git_merge_file_favor_t favor;
+} git_merge_driver__builtin;
+
+extern int git_merge_driver_global_init(void);
+
+extern int git_merge_driver_for_path(
+	char **name_out,
+	git_merge_driver **driver_out,
+	git_repository *repo,
+	const char *path);
+
+/* Merge driver configuration */
+extern int git_merge_driver_for_source(
+	const char **name_out,
+	git_merge_driver **driver_out,
+	const git_merge_driver_source *src);
+
+extern int git_merge_driver__builtin_apply(
+	git_merge_driver *self,
+	const char **path_out,
+	uint32_t *mode_out,
+	git_buf *merged_out,
+	const char *filter_name,
+	const git_merge_driver_source *src);
+
+/* Merge driver for text files, performs a standard three-way merge */
+extern git_merge_driver__builtin git_merge_driver__text;
+
+/* Merge driver for union-style merging */
+extern git_merge_driver__builtin git_merge_driver__union;
+
+/* Merge driver for unmergeable (binary) files: always produces conflicts */
+extern git_merge_driver git_merge_driver__binary;
+
+#endif
diff --git a/src/merge_file.c b/src/merge_file.c
index 6d4738065..731f4b724 100644
--- a/src/merge_file.c
+++ b/src/merge_file.c
@@ -11,6 +11,7 @@
 #include "fileops.h"
 #include "index.h"
 #include "diff_xdiff.h"
+#include "merge.h"
 
 #include "git2/repository.h"
 #include "git2/object.h"
@@ -26,52 +27,6 @@
 
 #define GIT_MERGE_FILE_SIDE_EXISTS(X)	((X)->mode != 0)
 
-GIT_INLINE(const char *) merge_file_best_path(
-	const git_merge_file_input *ancestor,
-	const git_merge_file_input *ours,
-	const git_merge_file_input *theirs)
-{
-	if (!ancestor) {
-		if (ours && theirs && strcmp(ours->path, theirs->path) == 0)
-			return ours->path;
-
-		return NULL;
-	}
-
-	if (ours && strcmp(ancestor->path, ours->path) == 0)
-		return theirs ? theirs->path : NULL;
-	else if(theirs && strcmp(ancestor->path, theirs->path) == 0)
-		return ours ? ours->path : NULL;
-
-	return NULL;
-}
-
-GIT_INLINE(int) merge_file_best_mode(
-	const git_merge_file_input *ancestor,
-	const git_merge_file_input *ours,
-	const git_merge_file_input *theirs)
-{
-	/*
-	 * If ancestor didn't exist and either ours or theirs is executable,
-	 * assume executable.  Otherwise, if any mode changed from the ancestor,
-	 * use that one.
-	 */
-	if (!ancestor) {
-		if ((ours && ours->mode == GIT_FILEMODE_BLOB_EXECUTABLE) ||
-			(theirs && theirs->mode == GIT_FILEMODE_BLOB_EXECUTABLE))
-			return GIT_FILEMODE_BLOB_EXECUTABLE;
-
-		return GIT_FILEMODE_BLOB;
-	} else if (ours && theirs) {
-		if (ancestor->mode == ours->mode)
-			return theirs->mode;
-
-		return ours->mode;
-	}
-
-	return 0;
-}
-
 int git_merge_file__input_from_index(
 	git_merge_file_input *input_out,
 	git_odb_object **odb_object_out,
@@ -177,8 +132,12 @@ static int merge_file__xdiff(
 		goto done;
 	}
 
-	if ((path = merge_file_best_path(ancestor, ours, theirs)) != NULL &&
-		(out->path = strdup(path)) == NULL) {
+	path = git_merge_file__best_path(
+		ancestor ? ancestor->path : NULL,
+		ours ? ours->path : NULL,
+		theirs ? theirs->path : NULL);
+
+	if (path != NULL && (out->path = git__strdup(path)) == NULL) {
 		error = -1;
 		goto done;
 	}
@@ -186,7 +145,10 @@ static int merge_file__xdiff(
 	out->automergeable = (xdl_result == 0);
 	out->ptr = (const char *)mmbuffer.ptr;
 	out->len = mmbuffer.size;
-	out->mode = merge_file_best_mode(ancestor, ours, theirs);
+	out->mode = git_merge_file__best_mode(
+		ancestor ? ancestor->mode : 0,
+		ours ? ours->mode : 0,
+		theirs ? theirs->mode : 0);
 
 done:
 	if (error < 0)