Merge pull request #778 from ben/clone

Clone

14 files changed, 767 insertions, 21 deletions

diff --git a/src/checkout.c b/src/checkout.c
new file mode 100644
index 000000000..252d9c4ae
--- /dev/null
+++ b/src/checkout.c
@@ -0,0 +1,230 @@
+/*
+ * Copyright (C) 2009-2012 the libgit2 contributors
+ *
+ * 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 <assert.h>
+
+#include "git2/checkout.h"
+#include "git2/repository.h"
+#include "git2/refs.h"
+#include "git2/tree.h"
+#include "git2/commit.h"
+#include "git2/blob.h"
+#include "git2/config.h"
+
+#include "common.h"
+#include "refs.h"
+#include "buffer.h"
+#include "repository.h"
+#include "filter.h"
+#include "blob.h"
+
+GIT_BEGIN_DECL
+
+
+typedef struct tree_walk_data
+{
+	git_indexer_stats *stats;
+	git_checkout_opts *opts;
+	git_repository *repo;
+	git_odb *odb;
+	bool do_symlinks;
+} tree_walk_data;
+
+
+static int blob_contents_to_link(tree_walk_data *data, git_buf *fnbuf,
+											const git_oid *id)
+{
+	int retcode = GIT_ERROR;
+	git_blob *blob;
+
+	/* Get the link target */
+	if (!(retcode = git_blob_lookup(&blob, data->repo, id))) {
+		git_buf linktarget = GIT_BUF_INIT;
+		if (!(retcode = git_blob__getbuf(&linktarget, blob))) {
+			/* Create the link */
+			const char *new = git_buf_cstr(&linktarget),
+						  *old = git_buf_cstr(fnbuf);
+			retcode = data->do_symlinks
+				? p_symlink(new, old)
+				: git_futils_fake_symlink(new, old);
+		}
+		git_buf_free(&linktarget);
+		git_blob_free(blob);
+	}
+
+	return retcode;
+}
+
+
+static int blob_contents_to_file(git_repository *repo, git_buf *fnbuf,
+											const git_tree_entry *entry, tree_walk_data *data)
+{
+	int retcode = GIT_ERROR;
+	int fd = -1;
+	git_buf contents = GIT_BUF_INIT;
+	const git_oid *id = git_tree_entry_id(entry);
+	int file_mode = data->opts->file_mode;
+
+	/* Deal with pre-existing files */
+	if (git_path_exists(git_buf_cstr(fnbuf)) &&
+		 data->opts->existing_file_action == GIT_CHECKOUT_SKIP_EXISTING)
+		return 0;
+
+	/* Allow disabling of filters */
+	if (data->opts->disable_filters) {
+		git_blob *blob;
+		if (!(retcode = git_blob_lookup(&blob, repo, id))) {
+			retcode = git_blob__getbuf(&contents, blob);
+			git_blob_free(blob);
+		}
+	} else {
+		retcode = git_filter_blob_contents(&contents, repo, id, git_buf_cstr(fnbuf));
+	}
+	if (retcode < 0) goto bctf_cleanup;
+
+	/* Allow overriding of file mode */
+	if (!file_mode)
+		file_mode = git_tree_entry_attributes(entry);
+
+	if ((retcode = git_futils_mkpath2file(git_buf_cstr(fnbuf), data->opts->dir_mode)) < 0)
+		goto bctf_cleanup;
+
+	fd = p_open(git_buf_cstr(fnbuf), data->opts->file_open_flags, file_mode);
+	if (fd < 0) goto bctf_cleanup;
+
+	if (!p_write(fd, git_buf_cstr(&contents), git_buf_len(&contents)))
+		retcode = 0;
+	else
+		retcode = GIT_ERROR;
+	p_close(fd);
+
+bctf_cleanup:
+	git_buf_free(&contents);
+	return retcode;
+}
+
+static int checkout_walker(const char *path, const git_tree_entry *entry, void *payload)
+{
+	int retcode = 0;
+	tree_walk_data *data = (tree_walk_data*)payload;
+	int attr = git_tree_entry_attributes(entry);
+	git_buf fnbuf = GIT_BUF_INIT;
+	git_buf_join_n(&fnbuf, '/', 3,
+						git_repository_workdir(data->repo),
+						path,
+						git_tree_entry_name(entry));
+
+	switch(git_tree_entry_type(entry))
+	{
+	case GIT_OBJ_TREE:
+		/* Nothing to do; the blob handling creates necessary directories. */
+		break;
+
+	case GIT_OBJ_COMMIT:
+		/* Submodule */
+		git_futils_mkpath2file(git_buf_cstr(&fnbuf), data->opts->dir_mode);
+		retcode = p_mkdir(git_buf_cstr(&fnbuf), data->opts->dir_mode);
+		break;
+
+	case GIT_OBJ_BLOB:
+		if (S_ISLNK(attr)) {
+			retcode = blob_contents_to_link(data, &fnbuf,
+													  git_tree_entry_id(entry));
+		} else {
+			retcode = blob_contents_to_file(data->repo, &fnbuf, entry, data);
+		}
+		break;
+
+	default:
+		retcode = -1;
+		break;
+	}
+
+	git_buf_free(&fnbuf);
+	data->stats->processed++;
+	return retcode;
+}
+
+
+int git_checkout_head(git_repository *repo, git_checkout_opts *opts, git_indexer_stats *stats)
+{
+	int retcode = GIT_ERROR;
+	git_indexer_stats dummy_stats;
+	git_checkout_opts default_opts = {0};
+	git_tree *tree;
+	tree_walk_data payload;
+	git_config *cfg;
+
+	assert(repo);
+	if (!opts) opts = &default_opts;
+	if (!stats) stats = &dummy_stats;
+
+	/* Default options */
+	if (!opts->existing_file_action)
+		opts->existing_file_action = GIT_CHECKOUT_OVERWRITE_EXISTING;
+	/* opts->disable_filters is false by default */
+	if (!opts->dir_mode) opts->dir_mode = GIT_DIR_MODE;
+	if (!opts->file_open_flags)
+		opts->file_open_flags = O_CREAT | O_TRUNC | O_WRONLY;
+
+	if (git_repository_is_bare(repo)) {
+		giterr_set(GITERR_INVALID, "Checkout is not allowed for bare repositories");
+		return GIT_ERROR;
+	}
+
+	/* Determine if symlinks should be handled */
+	if (!git_repository_config(&cfg, repo)) {
+		int temp = true;
+		if (!git_config_get_bool(&temp, cfg, "core.symlinks")) {
+			payload.do_symlinks = !!temp;
+		}
+		git_config_free(cfg);
+	}
+
+	stats->total = stats->processed = 0;
+	payload.stats = stats;
+	payload.opts = opts;
+	payload.repo = repo;
+	if (git_repository_odb(&payload.odb, repo) < 0) return GIT_ERROR;
+
+	if (!git_repository_head_tree(&tree, repo)) {
+		git_index *idx;
+		if (!(retcode = git_repository_index(&idx, repo))) {
+			if (!(retcode = git_index_read_tree(idx, tree, stats))) {
+				git_index_write(idx);
+				retcode = git_tree_walk(tree, checkout_walker, GIT_TREEWALK_POST, &payload);
+			}
+			git_index_free(idx);
+		}
+		git_tree_free(tree);
+	}
+
+	git_odb_free(payload.odb);
+	return retcode;
+}
+
+
+int git_checkout_reference(git_reference *ref,
+									git_checkout_opts *opts,
+									git_indexer_stats *stats)
+{
+	git_repository *repo= git_reference_owner(ref);
+	git_reference *head = NULL;
+	int retcode = GIT_ERROR;
+
+	if ((retcode = git_reference_create_symbolic(&head, repo, GIT_HEAD_FILE,
+																git_reference_name(ref), true)) < 0)
+		return retcode;
+
+	retcode = git_checkout_head(git_reference_owner(ref), opts, stats);
+
+	git_reference_free(head);
+	return retcode;
+}
+
+
+GIT_END_DECL
diff --git a/src/clone.c b/src/clone.c
new file mode 100644
index 000000000..33953d7a0
--- /dev/null
+++ b/src/clone.c
@@ -0,0 +1,254 @@
+/*
+ * Copyright (C) 2009-2012 the libgit2 contributors
+ *
+ * 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 <assert.h>
+
+#ifndef GIT_WIN32
+#include <dirent.h>
+#endif
+
+#include "git2/clone.h"
+#include "git2/remote.h"
+#include "git2/revparse.h"
+#include "git2/branch.h"
+#include "git2/config.h"
+#include "git2/checkout.h"
+#include "git2/commit.h"
+#include "git2/tree.h"
+
+#include "common.h"
+#include "remote.h"
+#include "fileops.h"
+#include "refs.h"
+#include "path.h"
+
+GIT_BEGIN_DECL
+
+struct HeadInfo {
+	git_repository *repo;
+	git_oid remote_head_oid;
+	git_buf branchname;
+};
+
+static int create_tracking_branch(git_repository *repo, const git_oid *target, const char *name)
+{
+	git_object *head_obj = NULL;
+	git_reference *branch_ref;
+	int retcode = GIT_ERROR;
+
+	/* Find the target commit */
+	if (git_object_lookup(&head_obj, repo, target, GIT_OBJ_ANY) < 0)
+		return GIT_ERROR;
+
+	/* Create the new branch */
+	if (!git_branch_create(&branch_ref, repo, name, head_obj, 0)) {
+		git_config *cfg;
+
+		git_reference_free(branch_ref);
+		/* Set up tracking */
+		if (!git_repository_config(&cfg, repo)) {
+			git_buf remote = GIT_BUF_INIT;
+			git_buf merge = GIT_BUF_INIT;
+			git_buf merge_target = GIT_BUF_INIT;
+			if (!git_buf_printf(&remote, "branch.%s.remote", name) &&
+				 !git_buf_printf(&merge, "branch.%s.merge", name) &&
+				 !git_buf_printf(&merge_target, "refs/heads/%s", name) &&
+				 !git_config_set_string(cfg, git_buf_cstr(&remote), "origin") &&
+				 !git_config_set_string(cfg, git_buf_cstr(&merge), git_buf_cstr(&merge_target))) {
+				retcode = 0;
+			}
+			git_buf_free(&remote);
+			git_buf_free(&merge);
+			git_buf_free(&merge_target);
+			git_config_free(cfg);
+		}
+	}
+
+	git_object_free(head_obj);
+	return retcode;
+}
+
+static int reference_matches_remote_head(const char *head_name, void *payload)
+{
+	struct HeadInfo *head_info = (struct HeadInfo *)payload;
+	git_oid oid;
+
+	/* Stop looking if we've already found a match */
+	if (git_buf_len(&head_info->branchname) > 0) return 0;
+
+	if (!git_reference_name_to_oid(&oid, head_info->repo, head_name) &&
+		 !git_oid_cmp(&head_info->remote_head_oid, &oid)) {
+		git_buf_puts(&head_info->branchname,
+						 head_name+strlen("refs/remotes/origin/"));
+	}
+	return 0;
+}
+
+static int update_head_to_new_branch(git_repository *repo, const git_oid *target, const char *name)
+{
+	int retcode = GIT_ERROR;
+
+	if (!create_tracking_branch(repo, target, name)) {
+		git_reference *head;
+		if (!git_reference_lookup(&head, repo, GIT_HEAD_FILE)) {
+			git_buf targetbuf = GIT_BUF_INIT;
+			if (!git_buf_printf(&targetbuf, "refs/heads/%s", name)) {
+				retcode = git_reference_set_target(head, git_buf_cstr(&targetbuf));
+			}
+			git_buf_free(&targetbuf);
+			git_reference_free(head);
+		}
+	}
+
+	return retcode;
+}
+
+static int update_head_to_remote(git_repository *repo, git_remote *remote)
+{
+	int retcode = GIT_ERROR;
+	git_remote_head *remote_head;
+	git_oid oid;
+	struct HeadInfo head_info;
+
+	/* Get the remote's HEAD. This is always the first ref in remote->refs. */
+	remote_head = remote->refs.contents[0];
+	git_oid_cpy(&head_info.remote_head_oid, &remote_head->oid);
+	git_buf_init(&head_info.branchname, 16);
+	head_info.repo = repo;
+
+	/* Check to see if "master" matches the remote head */
+	if (!git_reference_name_to_oid(&oid, repo, "refs/remotes/origin/master") &&
+		 !git_oid_cmp(&remote_head->oid, &oid)) {
+		retcode = update_head_to_new_branch(repo, &oid, "master");
+	}
+	/* Not master. Check all the other refs. */
+	else if (!git_reference_foreach(repo, GIT_REF_LISTALL,
+												 reference_matches_remote_head,
+												 &head_info) &&
+				git_buf_len(&head_info.branchname) > 0) {
+		retcode = update_head_to_new_branch(repo, &head_info.remote_head_oid,
+														git_buf_cstr(&head_info.branchname));
+	}
+
+	git_buf_free(&head_info.branchname);
+	return retcode;
+}
+
+/*
+ * submodules?
+ */
+
+
+
+static int setup_remotes_and_fetch(git_repository *repo,
+											  const char *origin_url,
+											  git_indexer_stats *fetch_stats)
+{
+	int retcode = GIT_ERROR;
+	git_remote *origin = NULL;
+	git_off_t bytes = 0;
+	git_indexer_stats dummy_stats;
+
+	if (!fetch_stats) fetch_stats = &dummy_stats;
+
+	/* Create the "origin" remote */
+	if (!git_remote_add(&origin, repo, "origin", origin_url)) {
+		/* Connect and download everything */
+		if (!git_remote_connect(origin, GIT_DIR_FETCH)) {
+			if (!git_remote_download(origin, &bytes, fetch_stats)) {
+				/* Create "origin/foo" branches for all remote branches */
+				if (!git_remote_update_tips(origin)) {
+					/* Point HEAD to the same ref as the remote's head */
+					if (!update_head_to_remote(repo, origin)) {
+						retcode = 0;
+					}
+				}
+			}
+			git_remote_disconnect(origin);
+		}
+		git_remote_free(origin);
+	}
+
+	return retcode;
+}
+
+
+static bool path_is_okay(const char *path)
+{
+	/* The path must either not exist, or be an empty directory */
+	if (!git_path_exists(path)) return true;
+	if (!git_path_is_empty_dir(path)) {
+		giterr_set(GITERR_INVALID,
+					  "'%s' exists and is not an empty directory", path);
+		return false;
+	}
+	return true;
+}
+
+
+static int clone_internal(git_repository **out,
+								  const char *origin_url,
+								  const char *path,
+								  git_indexer_stats *fetch_stats,
+								  int is_bare)
+{
+	int retcode = GIT_ERROR;
+	git_repository *repo = NULL;
+	git_indexer_stats dummy_stats;
+
+	if (!fetch_stats) fetch_stats = &dummy_stats;
+
+	if (!path_is_okay(path)) {
+		return GIT_ERROR;
+	}
+
+	if (!(retcode = git_repository_init(&repo, path, is_bare))) {
+		if ((retcode = setup_remotes_and_fetch(repo, origin_url, fetch_stats)) < 0) {
+			/* Failed to fetch; clean up */
+			git_repository_free(repo);
+			git_futils_rmdir_r(path, GIT_DIRREMOVAL_FILES_AND_DIRS);
+		} else {
+			*out = repo;
+			retcode = 0;
+		}
+	}
+
+	return retcode;
+}
+
+int git_clone_bare(git_repository **out,
+						 const char *origin_url,
+						 const char *dest_path,
+						 git_indexer_stats *fetch_stats)
+{
+	assert(out && origin_url && dest_path);
+	return clone_internal(out, origin_url, dest_path, fetch_stats, 1);
+}
+
+
+int git_clone(git_repository **out,
+				  const char *origin_url,
+				  const char *workdir_path,
+				  git_indexer_stats *fetch_stats,
+				  git_indexer_stats *checkout_stats,
+				  git_checkout_opts *checkout_opts)
+{
+	int retcode = GIT_ERROR;
+
+	assert(out && origin_url && workdir_path);
+
+	if (!(retcode = clone_internal(out, origin_url, workdir_path, fetch_stats, 0))) {
+		retcode = git_checkout_head(*out, checkout_opts, checkout_stats);
+	}
+
+	return retcode;
+}
+
+
+
+
+GIT_END_DECL
diff --git a/src/crlf.c b/src/crlf.c
index 303a46d3b..509e55897 100644
--- a/src/crlf.c
+++ b/src/crlf.c
@@ -184,7 +184,87 @@ static int crlf_apply_to_odb(git_filter *self, git_buf *dest, const git_buf *sou
 	return drop_crlf(dest, source);
 }
 
-int git_filter_add__crlf_to_odb(git_vector *filters, git_repository *repo, const char *path)
+static int convert_line_endings(git_buf *dest, const git_buf *source, const char *ending)
+{
+	const char *scan = git_buf_cstr(source),
+				  *next,
+				  *scan_end = git_buf_cstr(source) + git_buf_len(source);
+
+	while ((next = memchr(scan, '\n', scan_end - scan)) != NULL) {
+		if (next > scan)
+			git_buf_put(dest, scan, next-scan);
+		git_buf_puts(dest, ending);
+		scan = next + 1;
+	}
+
+	git_buf_put(dest, scan, scan_end - scan);
+	return 0;
+}
+
+static const char *line_ending(struct crlf_filter *filter)
+{
+	switch (filter->attrs.crlf_action) {
+	case GIT_CRLF_BINARY:
+	case GIT_CRLF_INPUT:
+		return "\n";
+
+	case GIT_CRLF_CRLF:
+		return "\r\n";
+
+	case GIT_CRLF_AUTO:
+	case GIT_CRLF_TEXT:
+	case GIT_CRLF_GUESS:
+		break;
+
+	default:
+		goto line_ending_error;
+	}
+
+	switch (filter->attrs.eol) {
+	case GIT_EOL_UNSET:
+		return GIT_EOL_NATIVE == GIT_EOL_CRLF
+			? "\r\n"
+			: "\n";
+
+	case GIT_EOL_CRLF:
+		return "\r\n";
+
+	case GIT_EOL_LF:
+		return "\n";
+
+	default:
+		goto line_ending_error;
+	}
+
+line_ending_error:
+	giterr_set(GITERR_INVALID, "Invalid input to line ending filter");
+	return NULL;
+}
+
+static int crlf_apply_to_workdir(git_filter *self, git_buf *dest, const git_buf *source)
+{
+	struct crlf_filter *filter = (struct crlf_filter *)self;
+	const char *workdir_ending = NULL;
+
+	assert (self && dest && source);
+
+	/* Empty file? Nothing to do. */
+	if (git_buf_len(source) == 0)
+		return 0;
+
+	/* Determine proper line ending */
+	workdir_ending = line_ending(filter);
+	if (!workdir_ending) return -1;
+
+	/* If the line ending is '\n', just copy the input */
+	if (!strcmp(workdir_ending, "\n"))
+		return git_buf_puts(dest, git_buf_cstr(source));
+
+	return convert_line_endings(dest, source, workdir_ending);
+}
+
+static int find_and_add_filter(git_vector *filters, git_repository *repo, const char *path,
+										 int (*apply)(struct git_filter *self, git_buf *dest, const git_buf *source))
 {
 	struct crlf_attrs ca;
 	struct crlf_filter *filter;
@@ -206,8 +286,7 @@ int git_filter_add__crlf_to_odb(git_vector *filters, git_repository *repo, const
 	if (ca.crlf_action == GIT_CRLF_GUESS) {
 		int auto_crlf;
 
-		if ((error = git_repository__cvar(
-			&auto_crlf, repo, GIT_CVAR_AUTO_CRLF)) < 0)
+		if ((error = git_repository__cvar(&auto_crlf, repo, GIT_CVAR_AUTO_CRLF)) < 0)
 			return error;
 
 		if (auto_crlf == GIT_AUTO_CRLF_FALSE)
@@ -219,10 +298,19 @@ int git_filter_add__crlf_to_odb(git_vector *filters, git_repository *repo, const
 	filter = git__malloc(sizeof(struct crlf_filter));
 	GITERR_CHECK_ALLOC(filter);
 
-	filter->f.apply = &crlf_apply_to_odb;
+	filter->f.apply = apply;
 	filter->f.do_free = NULL;
 	memcpy(&filter->attrs, &ca, sizeof(struct crlf_attrs));
 
 	return git_vector_insert(filters, filter);
 }
 
+int git_filter_add__crlf_to_odb(git_vector *filters, git_repository *repo, const char *path)
+{
+	return find_and_add_filter(filters, repo, path, &crlf_apply_to_odb);
+}
+
+int git_filter_add__crlf_to_workdir(git_vector *filters, git_repository *repo, const char *path)
+{
+	return find_and_add_filter(filters, repo, path, &crlf_apply_to_workdir);
+}
diff --git a/src/fileops.c b/src/fileops.c
index e936c3e2b..4de58b0cc 100644
--- a/src/fileops.c
+++ b/src/fileops.c
@@ -484,3 +484,14 @@ int git_futils_find_global_file(git_buf *path, const char *filename)
 	return 0;
 #endif
 }
+
+int git_futils_fake_symlink(const char *old, const char *new)
+{
+	int retcode = GIT_ERROR;
+	int fd = git_futils_creat_withpath(new, 0755, 0644);
+	if (fd >= 0) {
+		retcode = p_write(fd, old, strlen(old));
+		p_close(fd);
+	}
+	return retcode;
+}
diff --git a/src/fileops.h b/src/fileops.h
index b0c5779e5..594eacbd0 100644
--- a/src/fileops.h
+++ b/src/fileops.h
@@ -179,4 +179,14 @@ extern int git_futils_find_global_file(git_buf *path, const char *filename);
  */
 extern int git_futils_find_system_file(git_buf *path, const char *filename);
 
+
+/**
+ * Create a "fake" symlink (text file containing the target path).
+ *
+ * @param new symlink file to be created
+ * @param old original symlink target
+ * @return 0 on success, -1 on error
+ */
+extern int git_futils_fake_symlink(const char *new, const char *old);
+
 #endif /* INCLUDE_fileops_h__ */
diff --git a/src/filter.c b/src/filter.c
index 8fa3eb684..e9517a259 100644
--- a/src/filter.c
+++ b/src/filter.c
@@ -11,6 +11,7 @@
 #include "filter.h"
 #include "repository.h"
 #include "git2/config.h"
+#include "blob.h"
 
 /* Tweaked from Core Git. I wonder what we could use this for... */
 void git_text_gather_stats(git_text_stats *stats, const git_buf *text)
@@ -95,8 +96,9 @@ int git_filters_load(git_vector *filters, git_repository *repo, const char *path
 		if (error < 0)
 			return error;
 	} else {
-		giterr_set(GITERR_INVALID, "Worktree filters are not implemented yet");
-		return -1;
+		error = git_filter_add__crlf_to_workdir(filters, repo, path);
+		if (error < 0)
+			return error;
 	}
 
 	return (int)filters->length;
@@ -163,3 +165,37 @@ int git_filters_apply(git_buf *dest, git_buf *source, git_vector *filters)
 	return 0;
 }
 
+static int unfiltered_blob_contents(git_buf *out, git_repository *repo, const git_oid *blob_id)
+{
+	int retcode = GIT_ERROR;
+	git_blob *blob;
+
+	if (!(retcode = git_blob_lookup(&blob, repo, blob_id)))
+	{
+		retcode = git_blob__getbuf(out, blob);
+		git_blob_free(blob);
+	}
+
+	return retcode;
+}
+
+int git_filter_blob_contents(git_buf *out, git_repository *repo, const git_oid *oid, const char *path)
+{
+	int retcode = GIT_ERROR;
+
+	git_buf unfiltered = GIT_BUF_INIT;
+	if (!unfiltered_blob_contents(&unfiltered, repo, oid)) {
+		git_vector filters = GIT_VECTOR_INIT;
+		if (git_filters_load(&filters,
+									repo, path, GIT_FILTER_TO_WORKTREE) >= 0) {
+			git_buf_clear(out);
+			retcode = git_filters_apply(out, &unfiltered, &filters);
+		}
+
+		git_filters_free(&filters);
+	}
+
+	git_buf_free(&unfiltered);
+	return retcode;
+}
+
diff --git a/src/filter.h b/src/filter.h
index 66e370aef..5b7a25b04 100644
--- a/src/filter.h
+++ b/src/filter.h
@@ -96,6 +96,9 @@ extern void git_filters_free(git_vector *filters);
 /* Strip CRLF, from Worktree to ODB */
 extern int git_filter_add__crlf_to_odb(git_vector *filters, git_repository *repo, const char *path);
 
+/* Add CRLF, from ODB to worktree */
+extern int git_filter_add__crlf_to_workdir(git_vector *filters, git_repository *repo, const char *path);
+
 
 /*
  * PLAINTEXT API
@@ -116,4 +119,16 @@ extern void git_text_gather_stats(git_text_stats *stats, const git_buf *text);
  */
 extern int git_text_is_binary(git_text_stats *stats);
 
+
+/**
+ * Get the content of a blob after all filters have been run.
+ *
+ * @param out buffer to receive the contents
+ * @param repo repository containing the blob
+ * @param oid object id for the blob
+ * @param path path to the blob's output file, relative to the workdir root
+ * @return 0 on success, an error code otherwise
+ */
+extern int git_filter_blob_contents(git_buf *out, git_repository *repo, const git_oid *oid, const char *path);
+
 #endif
diff --git a/src/index.c b/src/index.c
index b6b1b779e..a1042b723 100644
--- a/src/index.c
+++ b/src/index.c
@@ -986,12 +986,19 @@ int git_index_entry_stage(const git_index_entry *entry)
 	return (entry->flags & GIT_IDXENTRY_STAGEMASK) >> GIT_IDXENTRY_STAGESHIFT;
 }
 
+typedef struct read_tree_data {
+	git_index *index;
+	git_indexer_stats *stats;
+} read_tree_data;
+
 static int read_tree_cb(const char *root, const git_tree_entry *tentry, void *data)
 {
-	git_index *index = data;
+	read_tree_data *rtd = data;
 	git_index_entry *entry = NULL;
 	git_buf path = GIT_BUF_INIT;
 
+	rtd->stats->total++;
+
 	if (git_tree_entry__is_tree(tentry))
 		return 0;
 
@@ -1006,7 +1013,7 @@ static int read_tree_cb(const char *root, const git_tree_entry *tentry, void *da
 	entry->path = git_buf_detach(&path);
 	git_buf_free(&path);
 
-	if (index_insert(index, entry, 0) < 0) {
+	if (index_insert(rtd->index, entry, 0) < 0) {
 		index_entry_free(entry);
 		return -1;
 	}
@@ -1014,9 +1021,16 @@ static int read_tree_cb(const char *root, const git_tree_entry *tentry, void *da
 	return 0;
 }
 
-int git_index_read_tree(git_index *index, git_tree *tree)
+int git_index_read_tree(git_index *index, git_tree *tree, git_indexer_stats *stats)
 {
+	git_indexer_stats dummy_stats;
+	read_tree_data rtd = {index, NULL};
+
+	if (!stats) stats = &dummy_stats;
+	stats->total = 0;
+	rtd.stats = stats;
+
 	git_index_clear(index);
 
-	return git_tree_walk(tree, read_tree_cb, GIT_TREEWALK_POST, index);
+	return git_tree_walk(tree, read_tree_cb, GIT_TREEWALK_POST, &rtd);
 }
diff --git a/src/path.c b/src/path.c
index e9bc4871c..22391c52b 100644
--- a/src/path.c
+++ b/src/path.c
@@ -387,6 +387,69 @@ bool git_path_isfile(const char *path)
 	return S_ISREG(st.st_mode) != 0;
 }
 
+#ifdef GIT_WIN32
+
+bool git_path_is_empty_dir(const char *path)
+{
+	git_buf pathbuf = GIT_BUF_INIT;
+	HANDLE hFind = INVALID_HANDLE_VALUE;
+	wchar_t *wbuf;
+	WIN32_FIND_DATAW ffd;
+	bool retval = true;
+
+	if (!git_path_isdir(path)) return false;
+
+	git_buf_printf(&pathbuf, "%s\\*", path);
+	wbuf = gitwin_to_utf16(git_buf_cstr(&pathbuf));
+
+	hFind = FindFirstFileW(wbuf, &ffd);
+	if (INVALID_HANDLE_VALUE == hFind) {
+		giterr_set(GITERR_OS, "Couldn't open '%s'", path);
+		return false;
+	}
+
+	do {
+		if (!git_path_is_dot_or_dotdotW(ffd.cFileName)) {
+			retval = false;
+		}
+	} while (FindNextFileW(hFind, &ffd) != 0);
+
+	FindClose(hFind);
+	git_buf_free(&pathbuf);
+	git__free(wbuf);
+	return retval;
+}
+
+#else
+
+bool git_path_is_empty_dir(const char *path)
+{
+	DIR *dir = NULL;
+	struct dirent *e;
+	bool retval = true;
+
+	if (!git_path_isdir(path)) return false;
+
+	dir = opendir(path);
+	if (!dir) {
+		giterr_set(GITERR_OS, "Couldn't open '%s'", path);
+		return false;
+	}
+
+	while ((e = readdir(dir)) != NULL) {
+		if (!git_path_is_dot_or_dotdot(e->d_name)) {
+			giterr_set(GITERR_INVALID,
+						  "'%s' exists and is not an empty directory", path);
+			retval = false;
+			break;
+		}
+	}
+	closedir(dir);
+
+	return retval;
+}
+#endif
+
 int git_path_lstat(const char *path, struct stat *st)
 {
 	int err = 0;
@@ -551,14 +614,6 @@ int git_path_cmp(
 	return (c1 < c2) ? -1 : (c1 > c2) ? 1 : 0;
 }
 
-/* Taken from git.git */
-GIT_INLINE(int) is_dot_or_dotdot(const char *name)
-{
-	return (name[0] == '.' &&
-		(name[1] == '\0' ||
-		 (name[1] == '.' && name[2] == '\0')));
-}
-
 int git_path_direach(
 	git_buf *path,
 	int (*fn)(void *, git_buf *),
@@ -587,7 +642,7 @@ int git_path_direach(
 	while (p_readdir_r(dir, de_buf, &de) == 0 && de != NULL) {
 		int result;
 
-		if (is_dot_or_dotdot(de->d_name))
+		if (git_path_is_dot_or_dotdot(de->d_name))
 			continue;
 
 		if (git_buf_puts(path, de->d_name) < 0) {
@@ -646,7 +701,7 @@ int git_path_dirload(
 		char *entry_path;
 		size_t entry_len;
 
-		if (is_dot_or_dotdot(de->d_name))
+		if (git_path_is_dot_or_dotdot(de->d_name))
 			continue;
 
 		entry_len = strlen(de->d_name);
diff --git a/src/path.h b/src/path.h
index d611428c1..14618b2fc 100644
--- a/src/path.h
+++ b/src/path.h
@@ -80,7 +80,24 @@ extern int git_path_to_dir(git_buf *path);
  */
 extern void git_path_string_to_dir(char* path, size_t size);
 
+/**
+ * Taken from git.git; returns nonzero if the given path is "." or "..".
+ */
+GIT_INLINE(int) git_path_is_dot_or_dotdot(const char *name)
+{
+	return (name[0] == '.' &&
+			  (name[1] == '\0' ||
+				(name[1] == '.' && name[2] == '\0')));
+}
+
 #ifdef GIT_WIN32
+GIT_INLINE(int) git_path_is_dot_or_dotdotW(const wchar_t *name)
+{
+	return (name[0] == L'.' &&
+			  (name[1] == L'\0' ||
+				(name[1] == L'.' && name[2] == L'\0')));
+}
+
 /**
  * Convert backslashes in path to forward slashes.
  */
@@ -130,6 +147,11 @@ extern bool git_path_isdir(const char *path);
 extern bool git_path_isfile(const char *path);
 
 /**
+ * Check if the given path is a directory, and is empty.
+ */
+extern bool git_path_is_empty_dir(const char *path);
+
+/**
  * Stat a file and/or link and set error if needed.
  */
 extern int git_path_lstat(const char *path, struct stat *st);
diff --git a/src/reset.c b/src/reset.c
index 14f7a236a..1379f6442 100644
--- a/src/reset.c
+++ b/src/reset.c
@@ -80,7 +80,7 @@ int git_reset(
 		goto cleanup;
 	}
 
-	if (git_index_read_tree(index, tree) < 0) {
+	if (git_index_read_tree(index, tree, NULL) < 0) {
 		giterr_set(GITERR_INDEX, "%s - Failed to update the index.", ERROR_MSG);
 		goto cleanup;
 	}
diff --git a/src/unix/posix.h b/src/unix/posix.h
index 83fd8a189..7a3a388ec 100644
--- a/src/unix/posix.h
+++ b/src/unix/posix.h
@@ -19,6 +19,7 @@
 #define p_lstat(p,b) lstat(p,b)
 #define p_readlink(a, b, c) readlink(a, b, c)
 #define p_link(o,n) link(o, n)
+#define p_symlink(o,n) symlink(o,n)
 #define p_unlink(p) unlink(p)
 #define p_mkdir(p,m) mkdir(p, m)
 #define p_fsync(fd) fsync(fd)
diff --git a/src/win32/posix.h b/src/win32/posix.h
index baa4a3b4e..14caae418 100644
--- a/src/win32/posix.h
+++ b/src/win32/posix.h
@@ -33,6 +33,7 @@ GIT_INLINE(int) p_mkdir(const char *path, mode_t mode)
 extern int p_unlink(const char *path);
 extern int p_lstat(const char *file_name, struct stat *buf);
 extern int p_readlink(const char *link, char *target, size_t target_len);
+extern int p_symlink(const char *old, const char *new);
 extern int p_hide_directory__w32(const char *path);
 extern char *p_realpath(const char *orig_path, char *buffer);
 extern int p_vsnprintf(char *buffer, size_t count, const char *format, va_list argptr);
diff --git a/src/win32/posix_w32.c b/src/win32/posix_w32.c
index e1471cab4..aa34ad3ac 100644
--- a/src/win32/posix_w32.c
+++ b/src/win32/posix_w32.c
@@ -7,6 +7,7 @@
 #include "../posix.h"
 #include "path.h"
 #include "utf-conv.h"
+#include "repository.h"
 #include <errno.h>
 #include <io.h>
 #include <fcntl.h>
@@ -224,6 +225,14 @@ int p_readlink(const char *link, char *target, size_t target_len)
 	return dwRet;
 }
 
+int p_symlink(const char *old, const char *new)
+{
+	/* Real symlinks on NTFS require admin privileges. Until this changes,
+	 * libgit2 just creates a text file with the link target in the contents.
+	 */
+	return git_futils_fake_symlink(old, new);
+}
+
 int p_open(const char *path, int flags, ...)
 {
 	int fd;