Merge pull request #3564 from ethomson/merge_drivers

Custom merge drivers and proper gitattributes `merge` handling

CHANGELOG.md

@@ -17,6 +17,10 @@ v0.24
 
 ### Changes or improvements
 
+* Custom merge drivers can now be registered, which allows callers to
+  configure callbacks to honor `merge=driver` configuration in
+  `.gitattributes`.
+
 * Custom filters can now be registered with wildcard attributes, for
   example `filter=*`.  Consumers should examine the attributes parameter
   of the `check` function for details.
@@ -83,6 +87,10 @@ v0.24
 
 ### Breaking API changes
 
+* `git_merge_options` now provides a `default_driver` that can be used
+  to provide the name of a merge driver to be used to handle files changed
+  during a merge.
+
 * The `git_merge_tree_flag_t` is now `git_merge_flag_t`.  Subsequently,
   its members are no longer prefixed with `GIT_MERGE_TREE_FLAG` but are
   now prefixed with `GIT_MERGE_FLAG`, and the `tree_flags` field of the

include/git2/merge.h

@@ -273,7 +273,16 @@ typedef struct {
 	 */
 	unsigned int recursion_limit;
 
-	/** Flags for handling conflicting content. */
+	/**
+	 * Default merge driver to be used when both sides of a merge have
+	 * changed.  The default is the `text` driver.
+	 */
+	const char *default_driver;
+
+	/**
+	 * Flags for handling conflicting content, to be used with the standard
+	 * (`text`) merge driver.
+	 */
 	git_merge_file_favor_t file_favor;
 
 	/** see `git_merge_file_flag_t` above */

include/git2/sys/merge.h

+/*
+ * 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_sys_git_merge_h__
+#define INCLUDE_sys_git_merge_h__
+
+/**
+ * @file git2/sys/merge.h
+ * @brief Git merge driver backend and plugin routines
+ * @defgroup git_backend Git custom backend APIs
+ * @ingroup Git
+ * @{
+ */
+GIT_BEGIN_DECL
+
+typedef struct git_merge_driver git_merge_driver;
+
+/**
+ * Look up a merge driver by name
+ *
+ * @param name The name of the merge driver
+ * @return Pointer to the merge driver object or NULL if not found
+ */
+GIT_EXTERN(git_merge_driver *) git_merge_driver_lookup(const char *name);
+
+#define GIT_MERGE_DRIVER_TEXT   "text"
+#define GIT_MERGE_DRIVER_BINARY "binary"
+#define GIT_MERGE_DRIVER_UNION  "union"
+
+/**
+ * A merge driver source represents the file to be merged
+ */
+typedef struct git_merge_driver_source git_merge_driver_source;
+
+/** Get the repository that the source data is coming from. */
+GIT_EXTERN(git_repository *) git_merge_driver_source_repo(
+	const git_merge_driver_source *src);
+
+/** Gets the ancestor of the file to merge. */
+GIT_EXTERN(git_index_entry *) git_merge_driver_source_ancestor(
+	const git_merge_driver_source *src);
+
+/** Gets the ours side of the file to merge. */
+GIT_EXTERN(git_index_entry *) git_merge_driver_source_ours(
+	const git_merge_driver_source *src);
+
+/** Gets the theirs side of the file to merge. */
+GIT_EXTERN(git_index_entry *) git_merge_driver_source_theirs(
+	const git_merge_driver_source *src);
+
+/** Gets the merge file options that the merge was invoked with */
+GIT_EXTERN(git_merge_file_options *) git_merge_driver_source_file_options(
+	const git_merge_driver_source *src);
+
+
+/**
+ * Initialize callback on merge driver
+ *
+ * Specified as `driver.initialize`, this is an optional callback invoked
+ * before a merge driver is first used.  It will be called once at most
+ * per library lifetime.
+ *
+ * If non-NULL, the merge driver's `initialize` callback will be invoked
+ * right before the first use of the driver, so you can defer expensive
+ * initialization operations (in case libgit2 is being used in a way that
+ * doesn't need the merge driver).
+ */
+typedef int (*git_merge_driver_init_fn)(git_merge_driver *self);
+
+/**
+ * Shutdown callback on merge driver
+ *
+ * Specified as `driver.shutdown`, this is an optional callback invoked
+ * when the merge driver is unregistered or when libgit2 is shutting down.
+ * It will be called once at most and should release resources as needed.
+ * This may be called even if the `initialize` callback was not made.
+ *
+ * Typically this function will free the `git_merge_driver` object itself.
+ */
+typedef void (*git_merge_driver_shutdown_fn)(git_merge_driver *self);
+
+/**
+ * Callback to perform the merge.
+ *
+ * Specified as `driver.apply`, this is the callback that actually does the
+ * merge.  If it can successfully perform a merge, it should populate
+ * `path_out` with a pointer to the filename to accept, `mode_out` with
+ * the resultant mode, and `merged_out` with the buffer of the merged file
+ * and then return 0.  If the driver returns `GIT_PASSTHROUGH`, then the
+ * default merge driver should instead be run.  It can also return
+ * `GIT_EMERGECONFLICT` if the driver is not able to produce a merge result,
+ * and the file will remain conflicted.  Any other errors will fail and
+ * return to the caller.
+ *
+ * The `filter_name` contains the name of the filter that was invoked, as
+ * specified by the file's attributes.
+ *
+ * The `src` contains the data about the file to be merged.
+ */
+typedef int (*git_merge_driver_apply_fn)(
+	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 structure used to register custom merge drivers.
+ *
+ * To associate extra data with a driver, allocate extra data and put the
+ * `git_merge_driver` struct at the start of your data buffer, then cast
+ * the `self` pointer to your larger structure when your callback is invoked.
+ */
+struct git_merge_driver {
+	/** The `version` should be set to `GIT_MERGE_DRIVER_VERSION`. */
+	unsigned int                 version;
+
+	/** Called when the merge driver is first used for any file. */
+	git_merge_driver_init_fn     initialize;
+
+	/** Called when the merge driver is unregistered from the system. */
+	git_merge_driver_shutdown_fn shutdown;
+
+	/**
+	 * Called to merge the contents of a conflict.  If this function
+	 * returns `GIT_PASSTHROUGH` then the default (`text`) merge driver
+	 * will instead be invoked.  If this function returns
+	 * `GIT_EMERGECONFLICT` then the file will remain conflicted.
+	 */
+	git_merge_driver_apply_fn    apply;
+};
+
+#define GIT_MERGE_DRIVER_VERSION 1
+
+/**
+ * Register a merge driver under a given name.
+ *
+ * As mentioned elsewhere, the initialize callback will not be invoked
+ * immediately.  It is deferred until the driver is used in some way.
+ *
+ * Currently the merge driver registry is not thread safe, so any
+ * registering or deregistering of merge drivers must be done outside of
+ * any possible usage of the drivers (i.e. during application setup or
+ * shutdown).
+ *
+ * @param name The name of this driver to match an attribute.  Attempting
+ * 			to register with an in-use name will return GIT_EEXISTS.
+ * @param driver The merge driver definition.  This pointer will be stored
+ *			as is by libgit2 so it must be a durable allocation (either
+ *			static or on the heap).
+ * @return 0 on successful registry, error code <0 on failure
+ */
+GIT_EXTERN(int) git_merge_driver_register(
+	const char *name, git_merge_driver *driver);
+
+/**
+ * Remove the merge driver with the given name.
+ *
+ * Attempting to remove the builtin libgit2 merge drivers is not permitted
+ * and will return an error.
+ *
+ * Currently the merge driver registry is not thread safe, so any
+ * registering or deregistering of drivers must be done outside of any
+ * possible usage of the drivers (i.e. during application setup or shutdown).
+ *
+ * @param name The name under which the merge driver was registered
+ * @return 0 on success, error code <0 on failure
+ */
+GIT_EXTERN(int) git_merge_driver_unregister(const char *name);
+
+/** @} */
+GIT_END_DECL
+#endif

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();
 

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

src/merge_driver.c

+/*
+ * 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;
+}
+

src/merge_driver.h

+/*
+ * 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

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)

tests/merge/driver.c

+#include "clar_libgit2.h"
+#include "git2/repository.h"
+#include "git2/merge.h"
+#include "buffer.h"
+#include "merge.h"
+
+#define TEST_REPO_PATH "merge-resolve"
+#define BRANCH_ID "7cb63eed597130ba4abb87b3e544b85021905520"
+
+#define AUTOMERGEABLE_IDSTR "f2e1550a0c9e53d5811175864a29536642ae3821"
+
+static git_repository *repo;
+static git_index *repo_index;
+static git_oid automergeable_id;
+
+static void test_drivers_register(void);
+static void test_drivers_unregister(void);
+
+void test_merge_driver__initialize(void)
+{
+    git_config *cfg;
+
+    repo = cl_git_sandbox_init(TEST_REPO_PATH);
+    git_repository_index(&repo_index, repo);
+
+	git_oid_fromstr(&automergeable_id, AUTOMERGEABLE_IDSTR);
+
+    /* Ensure that the user's merge.conflictstyle doesn't interfere */
+    cl_git_pass(git_repository_config(&cfg, repo));
+
+    cl_git_pass(git_config_set_string(cfg, "merge.conflictstyle", "merge"));
+    cl_git_pass(git_config_set_bool(cfg, "core.autocrlf", false));
+
+	test_drivers_register();
+
+    git_config_free(cfg);
+}
+
+void test_merge_driver__cleanup(void)
+{
+	test_drivers_unregister();
+
+    git_index_free(repo_index);
+	cl_git_sandbox_cleanup();
+}
+
+struct test_merge_driver {
+	git_merge_driver base;
+	int initialized;
+	int shutdown;
+};
+
+static int test_driver_init(git_merge_driver *s)
+{
+	struct test_merge_driver *self = (struct test_merge_driver *)s;
+	self->initialized = 1;
+	return 0;
+}
+
+static void test_driver_shutdown(git_merge_driver *s)
+{
+	struct test_merge_driver *self = (struct test_merge_driver *)s;
+	self->shutdown = 1;
+}
+
+static int test_driver_apply(
+	git_merge_driver *s,
+	const char **path_out,
+	uint32_t *mode_out,
+	git_buf *merged_out,
+	const char *filter_name,
+	const git_merge_driver_source *src)
+{
+	GIT_UNUSED(s);
+	GIT_UNUSED(src);
+
+	*path_out = "applied.txt";
+	*mode_out = GIT_FILEMODE_BLOB;
+
+	return git_buf_printf(merged_out, "This is the `%s` driver.\n",
+		filter_name);
+}
+
+static struct test_merge_driver test_driver_custom = {
+	{
+		GIT_MERGE_DRIVER_VERSION,
+		test_driver_init,
+		test_driver_shutdown,
+		test_driver_apply,
+	},
+	0,
+	0,
+};
+
+static struct test_merge_driver test_driver_wildcard = {
+	{
+		GIT_MERGE_DRIVER_VERSION,
+		test_driver_init,
+		test_driver_shutdown,
+		test_driver_apply,
+	},
+	0,
+	0,
+};
+
+static void test_drivers_register(void)
+{
+	cl_git_pass(git_merge_driver_register("custom", &test_driver_custom.base));
+	cl_git_pass(git_merge_driver_register("*", &test_driver_wildcard.base));
+}
+
+static void test_drivers_unregister(void)
+{
+	cl_git_pass(git_merge_driver_unregister("custom"));
+	cl_git_pass(git_merge_driver_unregister("*"));
+}
+
+static void set_gitattributes_to(const char *driver)
+{
+	git_buf line = GIT_BUF_INIT;
+
+	if (driver && strcmp(driver, ""))
+		git_buf_printf(&line, "automergeable.txt merge=%s\n", driver);
+	else if (driver)
+		git_buf_printf(&line, "automergeable.txt merge\n");
+	else
+		git_buf_printf(&line, "automergeable.txt -merge\n");
+
+	cl_assert(!git_buf_oom(&line));
+
+	cl_git_mkfile(TEST_REPO_PATH "/.gitattributes", line.ptr);
+	git_buf_free(&line);
+}
+
+static void merge_branch(void)
+{
+	git_oid their_id;
+	git_annotated_commit *their_head;
+
+	cl_git_pass(git_oid_fromstr(&their_id, BRANCH_ID));
+	cl_git_pass(git_annotated_commit_lookup(&their_head, repo, &their_id));
+
+	cl_git_pass(git_merge(repo, (const git_annotated_commit **)&their_head,
+		1, NULL, NULL));
+
+	git_annotated_commit_free(their_head);
+}
+
+void test_merge_driver__custom(void)
+{
+	const char *expected = "This is the `custom` driver.\n";
+	set_gitattributes_to("custom");
+	merge_branch();
+
+	cl_assert_equal_file(expected, strlen(expected),
+		TEST_REPO_PATH "/applied.txt");
+}
+
+void test_merge_driver__wildcard(void)
+{
+	const char *expected = "This is the `foobar` driver.\n";
+	set_gitattributes_to("foobar");
+	merge_branch();
+
+	cl_assert_equal_file(expected, strlen(expected),
+		TEST_REPO_PATH "/applied.txt");
+}
+
+void test_merge_driver__shutdown_is_called(void)
+{
+    test_driver_custom.initialized = 0;
+    test_driver_custom.shutdown = 0;
+    test_driver_wildcard.initialized = 0;
+    test_driver_wildcard.shutdown = 0;
+    
+    /* run the merge with the custom driver */
+    set_gitattributes_to("custom");
+    merge_branch();
+    
+	/* unregister the drivers, ensure their shutdown function is called */
+	test_drivers_unregister();
+
+    /* since the `custom` driver was used, it should have been initialized and
+     * shutdown, but the wildcard driver was not used at all and should not
+     * have been initialized or shutdown.
+     */
+	cl_assert(test_driver_custom.initialized);
+	cl_assert(test_driver_custom.shutdown);
+	cl_assert(!test_driver_wildcard.initialized);
+	cl_assert(!test_driver_wildcard.shutdown);
+
+	test_drivers_register();
+}
+
+static int defer_driver_apply(
+	git_merge_driver *s,
+	const char **path_out,
+	uint32_t *mode_out,
+	git_buf *merged_out,
+	const char *filter_name,
+	const git_merge_driver_source *src)
+{
+	GIT_UNUSED(s);
+	GIT_UNUSED(path_out);
+	GIT_UNUSED(mode_out);
+	GIT_UNUSED(merged_out);
+	GIT_UNUSED(filter_name);
+	GIT_UNUSED(src);
+
+	return GIT_PASSTHROUGH;
+}
+
+static struct test_merge_driver test_driver_defer_apply = {
+	{
+		GIT_MERGE_DRIVER_VERSION,
+		test_driver_init,
+		test_driver_shutdown,
+		defer_driver_apply,
+	},
+	0,
+	0,
+};
+
+void test_merge_driver__apply_can_defer(void)
+{
+	const git_index_entry *idx;
+
+	cl_git_pass(git_merge_driver_register("defer",
+		&test_driver_defer_apply.base));
+
+    set_gitattributes_to("defer");
+    merge_branch();
+
+	cl_assert((idx = git_index_get_bypath(repo_index, "automergeable.txt", 0)));
+	cl_assert_equal_oid(&automergeable_id, &idx->id);
+
+	git_merge_driver_unregister("defer");
+}
+
+static int conflict_driver_apply(
+	git_merge_driver *s,
+	const char **path_out,
+	uint32_t *mode_out,
+	git_buf *merged_out,
+	const char *filter_name,
+	const git_merge_driver_source *src)
+{
+	GIT_UNUSED(s);
+	GIT_UNUSED(path_out);
+	GIT_UNUSED(mode_out);
+	GIT_UNUSED(merged_out);
+	GIT_UNUSED(filter_name);
+	GIT_UNUSED(src);
+
+	return GIT_EMERGECONFLICT;
+}
+
+static struct test_merge_driver test_driver_conflict_apply = {
+	{
+		GIT_MERGE_DRIVER_VERSION,
+		test_driver_init,
+		test_driver_shutdown,
+		conflict_driver_apply,
+	},
+	0,
+	0,
+};
+
+void test_merge_driver__apply_can_conflict(void)
+{
+	const git_index_entry *ancestor, *ours, *theirs;
+
+	cl_git_pass(git_merge_driver_register("conflict",
+		&test_driver_conflict_apply.base));
+
+    set_gitattributes_to("conflict");
+    merge_branch();
+
+	cl_git_pass(git_index_conflict_get(&ancestor, &ours, &theirs,
+		repo_index, "automergeable.txt"));
+
+	git_merge_driver_unregister("conflict");
+}
+
+void test_merge_driver__default_can_be_specified(void)
+{
+	git_oid their_id;
+	git_annotated_commit *their_head;
+	git_merge_options merge_opts = GIT_MERGE_OPTIONS_INIT;
+	const char *expected = "This is the `custom` driver.\n";
+
+	merge_opts.default_driver = "custom";
+
+	cl_git_pass(git_oid_fromstr(&their_id, BRANCH_ID));
+	cl_git_pass(git_annotated_commit_lookup(&their_head, repo, &their_id));
+
+	cl_git_pass(git_merge(repo, (const git_annotated_commit **)&their_head,
+		1, &merge_opts, NULL));
+
+	git_annotated_commit_free(their_head);
+
+	cl_assert_equal_file(expected, strlen(expected),
+		TEST_REPO_PATH "/applied.txt");
+}
+
+void test_merge_driver__honors_builtin_mergedefault(void)
+{
+	const git_index_entry *ancestor, *ours, *theirs;
+
+	cl_repo_set_string(repo, "merge.default", "binary");
+	merge_branch();
+
+	cl_git_pass(git_index_conflict_get(&ancestor, &ours, &theirs,
+		repo_index, "automergeable.txt"));
+}
+
+void test_merge_driver__honors_custom_mergedefault(void)
+{
+	const char *expected = "This is the `custom` driver.\n";
+
+	cl_repo_set_string(repo, "merge.default", "custom");
+	merge_branch();
+
+	cl_assert_equal_file(expected, strlen(expected),
+		TEST_REPO_PATH "/applied.txt");
+}
+
+void test_merge_driver__mergedefault_deferring_falls_back_to_text(void)
+{
+	const git_index_entry *idx;
+
+	cl_git_pass(git_merge_driver_register("defer",
+		&test_driver_defer_apply.base));
+
+	cl_repo_set_string(repo, "merge.default", "defer");
+	merge_branch();
+
+	cl_assert((idx = git_index_get_bypath(repo_index, "automergeable.txt", 0)));
+	cl_assert_equal_oid(&automergeable_id, &idx->id);
+
+	git_merge_driver_unregister("defer");
+}
+
+void test_merge_driver__set_forces_text(void)
+{
+	const git_index_entry *idx;
+
+	/* `merge` without specifying a driver indicates `text` */
+	set_gitattributes_to("");
+	cl_repo_set_string(repo, "merge.default", "custom");
+
+	merge_branch();
+
+	cl_assert((idx = git_index_get_bypath(repo_index, "automergeable.txt", 0)));
+	cl_assert_equal_oid(&automergeable_id, &idx->id);
+}
+
+void test_merge_driver__unset_forces_binary(void)
+{
+	const git_index_entry *ancestor, *ours, *theirs;
+
+	/* `-merge` without specifying a driver indicates `binary` */
+	set_gitattributes_to(NULL);
+	cl_repo_set_string(repo, "merge.default", "custom");
+
+	merge_branch();
+
+	cl_git_pass(git_index_conflict_get(&ancestor, &ours, &theirs,
+		repo_index, "automergeable.txt"));
+}
+
+void test_merge_driver__not_configured_driver_falls_back(void)
+{
+	const git_index_entry *idx;
+
+	test_drivers_unregister();
+
+	/* `merge` without specifying a driver indicates `text` */
+	set_gitattributes_to("notfound");
+
+	merge_branch();
+
+	cl_assert((idx = git_index_get_bypath(repo_index, "automergeable.txt", 0)));
+	cl_assert_equal_oid(&automergeable_id, &idx->id);
+
+	test_drivers_register();
+}
+

tests/merge/workdir/simple.c

@@ -330,6 +330,42 @@ void test_merge_workdir_simple__union(void)
 	cl_assert(merge_test_reuc(repo_index, merge_reuc_entries, 4));
 }
 
+void test_merge_workdir_simple__gitattributes_union(void)
+{
+	git_buf conflicting_buf = GIT_BUF_INIT;
+
+	struct merge_index_entry merge_index_entries[] = {
+		ADDED_IN_MASTER_INDEX_ENTRY,
+		AUTOMERGEABLE_INDEX_ENTRY,
+		CHANGED_IN_BRANCH_INDEX_ENTRY,
+		CHANGED_IN_MASTER_INDEX_ENTRY,
+
+		{ 0100644, "72cdb057b340205164478565e91eb71647e66891", 0, "conflicting.txt" },
+
+		UNCHANGED_INDEX_ENTRY,
+	};
+
+	struct merge_reuc_entry merge_reuc_entries[] = {
+		AUTOMERGEABLE_REUC_ENTRY,
+		CONFLICTING_REUC_ENTRY,
+		REMOVED_IN_BRANCH_REUC_ENTRY,
+		REMOVED_IN_MASTER_REUC_ENTRY
+	};
+
+	set_core_autocrlf_to(repo, false);
+	cl_git_mkfile(TEST_REPO_PATH "/.gitattributes", "conflicting.txt merge=union\n");
+
+	merge_simple_branch(GIT_MERGE_FILE_FAVOR_NORMAL, 0);
+
+	cl_git_pass(git_futils_readbuffer(&conflicting_buf,
+		TEST_REPO_PATH "/conflicting.txt"));
+	cl_assert(strcmp(conflicting_buf.ptr, CONFLICTING_UNION_FILE) == 0);
+	git_buf_free(&conflicting_buf);
+
+	cl_assert(merge_test_index(repo_index, merge_index_entries, 6));
+	cl_assert(merge_test_reuc(repo_index, merge_reuc_entries, 4));
+}
+
 void test_merge_workdir_simple__diff3_from_config(void)
 {
 	git_config *config;