Merge pull request #3087 from ethomson/pr/3054

Performance Improvements to Status on Windows

src/attr.c

@@ -55,7 +55,7 @@ int git_attr_get(
 
 	*value = NULL;
 
-	if (git_attr_path__init(&path, pathname, git_repository_workdir(repo)) < 0)
+	if (git_attr_path__init(&path, pathname, git_repository_workdir(repo), GIT_DIR_FLAG_UNKNOWN) < 0)
 		return -1;
 
 	if ((error = collect_attr_files(repo, NULL, flags, pathname, &files)) < 0)
@@ -114,7 +114,7 @@ int git_attr_get_many_with_session(
 
 	assert(values && repo && names);
 
-	if (git_attr_path__init(&path, pathname, git_repository_workdir(repo)) < 0)
+	if (git_attr_path__init(&path, pathname, git_repository_workdir(repo), GIT_DIR_FLAG_UNKNOWN) < 0)
 		return -1;
 
 	if ((error = collect_attr_files(repo, attr_session, flags, pathname, &files)) < 0)
@@ -193,7 +193,7 @@ int git_attr_foreach(
 
 	assert(repo && callback);
 
-	if (git_attr_path__init(&path, pathname, git_repository_workdir(repo)) < 0)
+	if (git_attr_path__init(&path, pathname, git_repository_workdir(repo), GIT_DIR_FLAG_UNKNOWN) < 0)
 		return -1;
 
 	if ((error = collect_attr_files(repo, NULL, flags, pathname, &files)) < 0 ||

src/attr_file.c

@@ -457,7 +457,7 @@ git_attr_assignment *git_attr_rule__lookup_assignment(
 }
 
 int git_attr_path__init(
-	git_attr_path *info, const char *path, const char *base)
+	git_attr_path *info, const char *path, const char *base, git_dir_flag dir_flag)
 {
 	ssize_t root;
 
@@ -488,7 +488,21 @@ int git_attr_path__init(
 	if (!info->basename || !*info->basename)
 		info->basename = info->path;
 
-	info->is_dir = (int)git_path_isdir(info->full.ptr);
+	switch (dir_flag)
+	{
+	case GIT_DIR_FLAG_FALSE:
+		info->is_dir = 0;
+		break;
+
+	case GIT_DIR_FLAG_TRUE:
+		info->is_dir = 1;
+		break;
+
+	case GIT_DIR_FLAG_UNKNOWN:
+	default:
+		info->is_dir = (int)git_path_isdir(info->full.ptr);
+		break;
+	}
 
 	return 0;
 }

src/attr_file.h

@@ -202,8 +202,10 @@ extern bool git_attr_rule__match(
 extern git_attr_assignment *git_attr_rule__lookup_assignment(
 	git_attr_rule *rule, const char *name);
 
+typedef enum { GIT_DIR_FLAG_TRUE = 1, GIT_DIR_FLAG_FALSE = 0, GIT_DIR_FLAG_UNKNOWN = -1 } git_dir_flag;
+
 extern int git_attr_path__init(
-	git_attr_path *info, const char *path, const char *base);
+	git_attr_path *info, const char *path, const char *base, git_dir_flag is_dir);
 
 extern void git_attr_path__free(git_attr_path *info);
 

src/ignore.c

@@ -388,7 +388,7 @@ static bool ignore_lookup_in_rules(
 }
 
 int git_ignore__lookup(
-	int *out, git_ignores *ignores, const char *pathname)
+	int *out, git_ignores *ignores, const char *pathname, git_dir_flag dir_flag)
 {
 	unsigned int i;
 	git_attr_file *file;
@@ -397,7 +397,7 @@ int git_ignore__lookup(
 	*out = GIT_IGNORE_NOTFOUND;
 
 	if (git_attr_path__init(
-		&path, pathname, git_repository_workdir(ignores->repo)) < 0)
+		&path, pathname, git_repository_workdir(ignores->repo), dir_flag) < 0)
 		return -1;
 
 	/* first process builtins - success means path was found */
@@ -470,7 +470,7 @@ int git_ignore_path_is_ignored(
 	memset(&path, 0, sizeof(path));
 	memset(&ignores, 0, sizeof(ignores));
 
-	if ((error = git_attr_path__init(&path, pathname, workdir)) < 0 ||
+	if ((error = git_attr_path__init(&path, pathname, workdir, GIT_DIR_FLAG_UNKNOWN)) < 0 ||
 		(error = git_ignore__for_path(repo, path.path, &ignores)) < 0)
 		goto cleanup;
 

src/ignore.h

@@ -49,7 +49,7 @@ enum {
 	GIT_IGNORE_TRUE = 1,
 };
 
-extern int git_ignore__lookup(int *out, git_ignores *ign, const char *path);
+extern int git_ignore__lookup(int *out, git_ignores *ign, const char *path, git_dir_flag dir_flag);
 
 /* command line Git sometimes generates an error message if given a
  * pathspec that contains an exact match to an ignored file (provided

src/iterator.c

@@ -920,12 +920,31 @@ struct fs_iterator {
 
 #define FS_MAX_DEPTH 100
 
+typedef struct {
+	struct stat st;
+	size_t      path_len;
+	char        path[GIT_FLEX_ARRAY];
+} fs_iterator_path_with_stat;
+
+static int fs_iterator_path_with_stat_cmp(const void *a, const void *b)
+{
+	const fs_iterator_path_with_stat *psa = a, *psb = b;
+	return strcmp(psa->path, psb->path);
+}
+
+static int fs_iterator_path_with_stat_cmp_icase(const void *a, const void *b)
+{
+	const fs_iterator_path_with_stat *psa = a, *psb = b;
+	return strcasecmp(psa->path, psb->path);
+}
+
 static fs_iterator_frame *fs_iterator__alloc_frame(fs_iterator *fi)
 {
 	fs_iterator_frame *ff = git__calloc(1, sizeof(fs_iterator_frame));
 	git_vector_cmp entry_compare = CASESELECT(
 		iterator__ignore_case(fi),
-		git_path_with_stat_cmp_icase, git_path_with_stat_cmp);
+		fs_iterator_path_with_stat_cmp_icase,
+		fs_iterator_path_with_stat_cmp);
 
 	if (ff && git_vector_init(&ff->entries, 0, entry_compare) < 0) {
 		git__free(ff);
@@ -967,7 +986,7 @@ static int fs_iterator__advance_over(
 static int fs_iterator__entry_cmp(const void *i, const void *item)
 {
 	const fs_iterator *fi = (const fs_iterator *)i;
-	const git_path_with_stat *ps = item;
+	const fs_iterator_path_with_stat *ps = item;
 	return fi->base.prefixcomp(fi->base.start, ps->path);
 }
 
@@ -984,6 +1003,96 @@ static void fs_iterator__seek_frame_start(
 		ff->index = 0;
 }
 
+static int dirload_with_stat(
+	const char *dirpath,
+	size_t prefix_len,
+	unsigned int flags,
+	const char *start_stat,
+	const char *end_stat,
+	git_vector *contents)
+{
+	git_path_diriter diriter = GIT_PATH_DIRITER_INIT;
+	const char *path;
+	int (*strncomp)(const char *a, const char *b, size_t sz);
+	size_t start_len = start_stat ? strlen(start_stat) : 0;
+	size_t end_len = end_stat ? strlen(end_stat) : 0;
+	fs_iterator_path_with_stat *ps;
+	size_t path_len, cmp_len, ps_size;
+	int error;
+
+	strncomp = (flags & GIT_PATH_DIR_IGNORE_CASE) != 0 ?
+		git__strncasecmp : git__strncmp;
+
+	if ((error = git_path_diriter_init(&diriter, dirpath, flags)) < 0)
+		goto done;
+
+	while ((error = git_path_diriter_next(&diriter)) == 0) {
+		if ((error = git_path_diriter_fullpath(&path, &path_len, &diriter)) < 0)
+			goto done;
+
+		assert(path_len > prefix_len);
+
+		/* remove the prefix if requested */
+		path += prefix_len;
+		path_len -= prefix_len;
+
+		/* skip if before start_stat or after end_stat */
+		cmp_len = min(start_len, path_len);
+		if (cmp_len && strncomp(path, start_stat, cmp_len) < 0)
+			continue;
+		cmp_len = min(end_len, path_len);
+		if (cmp_len && strncomp(path, end_stat, cmp_len) > 0)
+			continue;
+
+		/* Make sure to append two bytes, one for the path's null
+		 * termination, one for a possible trailing '/' for folders.
+		 */
+		GITERR_CHECK_ALLOC_ADD(&ps_size, sizeof(fs_iterator_path_with_stat), path_len);
+		GITERR_CHECK_ALLOC_ADD(&ps_size, ps_size, 2);
+
+		ps = git__calloc(1, ps_size);
+		ps->path_len = path_len;
+
+		memcpy(ps->path, path, path_len);
+
+		if ((error = git_path_diriter_stat(&ps->st, &diriter)) < 0) {
+			if (error == GIT_ENOTFOUND) {
+				/* file was removed between readdir and lstat */
+				git__free(ps);
+				continue;
+			}
+
+			/* Treat the file as unreadable if we get any other error */
+			memset(&ps->st, 0, sizeof(ps->st));
+			ps->st.st_mode = GIT_FILEMODE_UNREADABLE;
+
+			giterr_clear();
+			error = 0;
+		} else if (S_ISDIR(ps->st.st_mode)) {
+			/* Suffix directory paths with a '/' */
+			ps->path[ps->path_len++] = '/';
+			ps->path[ps->path_len] = '\0';
+		} else if(!S_ISREG(ps->st.st_mode) && !S_ISLNK(ps->st.st_mode)) {
+			/* Ignore wacky things in the filesystem */
+			git__free(ps);
+			continue;
+		}
+
+		git_vector_insert(contents, ps);
+	}
+
+	if (error == GIT_ITEROVER)
+		error = 0;
+
+	/* sort now that directory suffix is added */
+	git_vector_sort(contents);
+
+done:
+	git_path_diriter_free(&diriter);
+	return error;
+}
+
+
 static int fs_iterator__expand_dir(fs_iterator *fi)
 {
 	int error;
@@ -998,7 +1107,7 @@ static int fs_iterator__expand_dir(fs_iterator *fi)
 	ff = fs_iterator__alloc_frame(fi);
 	GITERR_CHECK_ALLOC(ff);
 
-	error = git_path_dirload_with_stat(
+	error = dirload_with_stat(
 		fi->path.ptr, fi->root_len, fi->dirload_flags,
 		fi->base.start, fi->base.end, &ff->entries);
 
@@ -1086,7 +1195,7 @@ static int fs_iterator__advance_over(
 	int error = 0;
 	fs_iterator *fi = (fs_iterator *)self;
 	fs_iterator_frame *ff;
-	git_path_with_stat *next;
+	fs_iterator_path_with_stat *next;
 
 	if (entry != NULL)
 		*entry = NULL;
@@ -1176,7 +1285,7 @@ static void fs_iterator__free(git_iterator *self)
 
 static int fs_iterator__update_entry(fs_iterator *fi)
 {
-	git_path_with_stat *ps;
+	fs_iterator_path_with_stat *ps;
 
 	memset(&fi->entry, 0, sizeof(fi->entry));
 
@@ -1307,7 +1416,7 @@ GIT_INLINE(bool) workdir_path_is_dotgit(const git_buf *path)
  * We consider it a submodule if the path is listed as a submodule in
  * either the tree or the index.
  */
-static int is_submodule(workdir_iterator *wi, git_path_with_stat *ie)
+static int is_submodule(workdir_iterator *wi, fs_iterator_path_with_stat *ie)
 {
 	int error, is_submodule = 0;
 
@@ -1344,17 +1453,29 @@ static int is_submodule(workdir_iterator *wi, git_path_with_stat *ie)
 	return is_submodule;
 }
 
+GIT_INLINE(git_dir_flag) git_entry__dir_flag(git_index_entry *entry) {
+#if defined(GIT_WIN32) && !defined(__MINGW32__)
+	return (entry && entry->mode)
+		? S_ISDIR(entry->mode) ? GIT_DIR_FLAG_TRUE : GIT_DIR_FLAG_FALSE
+		: GIT_DIR_FLAG_UNKNOWN;
+#else
+	GIT_UNUSED(entry);
+	return GIT_DIR_FLAG_UNKNOWN;
+#endif
+}
+
 static int workdir_iterator__enter_dir(fs_iterator *fi)
 {
 	workdir_iterator *wi = (workdir_iterator *)fi;
 	fs_iterator_frame *ff = fi->stack;
 	size_t pos;
-	git_path_with_stat *entry;
+	fs_iterator_path_with_stat *entry;
 	bool found_submodules = false;
 
+	git_dir_flag dir_flag = git_entry__dir_flag(&fi->entry);
+
 	/* check if this directory is ignored */
-	if (git_ignore__lookup(
-			&ff->is_ignored, &wi->ignores, fi->path.ptr + fi->root_len) < 0) {
+	if (git_ignore__lookup(&ff->is_ignored, &wi->ignores, fi->path.ptr + fi->root_len, dir_flag) < 0) {
 		giterr_clear();
 		ff->is_ignored = GIT_IGNORE_NOTFOUND;
 	}
@@ -1483,7 +1604,6 @@ int git_iterator_for_workdir_ext(
 	return fs_iterator__initialize(out, &wi->fi, repo_workdir);
 }
 
-
 void git_iterator_free(git_iterator *iter)
 {
 	if (iter == NULL)
@@ -1574,8 +1694,9 @@ int git_iterator_current_parent_tree(
 
 static void workdir_iterator_update_is_ignored(workdir_iterator *wi)
 {
-	if (git_ignore__lookup(
-			&wi->is_ignored, &wi->ignores, wi->fi.entry.path) < 0) {
+	git_dir_flag dir_flag = git_entry__dir_flag(&wi->fi.entry);
+
+	if (git_ignore__lookup(&wi->is_ignored, &wi->ignores, wi->fi.entry.path, dir_flag) < 0) {
 		giterr_clear();
 		wi->is_ignored = GIT_IGNORE_NOTFOUND;
 	}

src/path.h

@@ -273,6 +273,7 @@ extern int git_path_apply_relative(git_buf *target, const char *relpath);
 enum {
 	GIT_PATH_DIR_IGNORE_CASE = (1u << 0),
 	GIT_PATH_DIR_PRECOMPOSE_UNICODE = (1u << 1),
+	GIT_PATH_DIR_INCLUDE_DOT_AND_DOTDOT = (1u << 2),
 };
 
 /**
@@ -326,66 +327,6 @@ extern int git_path_walk_up(
 	int (*callback)(void *payload, const char *path),
 	void *payload);
 
-/**
- * Load all directory entries (except '.' and '..') into a vector.
- *
- * For cases where `git_path_direach()` is not appropriate, this
- * allows you to load the filenames in a directory into a vector
- * of strings. That vector can then be sorted, iterated, or whatever.
- * Remember to free alloc of the allocated strings when you are done.
- *
- * @param path The directory to read from.
- * @param prefix_len When inserting entries, the trailing part of path
- * 		will be prefixed after this length.  I.e. given path "/a/b" and
- * 		prefix_len 3, the entries will look like "b/e1", "b/e2", etc.
- * @param alloc_extra Extra bytes to add to each string allocation in
- * 		case you want to append anything funny.
- * @param flags Combination of GIT_PATH_DIR flags.
- * @param contents Vector to fill with directory entry names.
- */
-extern int git_path_dirload(
-	const char *path,
-	size_t prefix_len,
-	size_t alloc_extra,
-	uint32_t flags,
-	git_vector *contents);
-
-
-typedef struct {
-	struct stat st;
-	size_t      path_len;
-	char        path[GIT_FLEX_ARRAY];
-} git_path_with_stat;
-
-extern int git_path_with_stat_cmp(const void *a, const void *b);
-extern int git_path_with_stat_cmp_icase(const void *a, const void *b);
-
-/**
- * Load all directory entries along with stat info into a vector.
- *
- * This adds four things on top of plain `git_path_dirload`:
- *
- * 1. Each entry in the vector is a `git_path_with_stat` struct that
- *    contains both the path and the stat info
- * 2. The entries will be sorted alphabetically
- * 3. Entries that are directories will be suffixed with a '/'
- * 4. Optionally, you can be a start and end prefix and only elements
- *    after the start and before the end (inclusively) will be stat'ed.
- *
- * @param path The directory to read from
- * @param prefix_len The trailing part of path to prefix to entry paths
- * @param flags GIT_PATH_DIR flags from above
- * @param start_stat As optimization, only stat values after this prefix
- * @param end_stat As optimization, only stat values before this prefix
- * @param contents Vector to fill with git_path_with_stat structures
- */
-extern int git_path_dirload_with_stat(
-	const char *path,
-	size_t prefix_len,
-	uint32_t flags,
-	const char *start_stat,
-	const char *end_stat,
-	git_vector *contents);
 
 enum { GIT_PATH_NOTEQUAL = 0, GIT_PATH_EQUAL = 1, GIT_PATH_PREFIX = 2 };
 
@@ -472,6 +413,137 @@ extern int git_path_iconv(git_path_iconv_t *ic, char **in, size_t *inlen);
 
 extern bool git_path_does_fs_decompose_unicode(const char *root);
 
+
+typedef struct git_path_diriter git_path_diriter;
+
+#if defined(GIT_WIN32) && !defined(__MINGW32__)
+
+struct git_path_diriter
+{
+	git_win32_path path;
+	size_t parent_len;
+
+	git_buf path_utf8;
+	size_t parent_utf8_len;
+
+	HANDLE handle;
+
+	unsigned int flags;
+
+	WIN32_FIND_DATAW current;
+	unsigned int needs_next;
+};
+
+#define GIT_PATH_DIRITER_INIT { {0}, 0, GIT_BUF_INIT, 0, INVALID_HANDLE_VALUE }
+
+#else
+
+struct git_path_diriter
+{
+	git_buf path;
+	size_t parent_len;
+
+	unsigned int flags;
+
+	DIR *dir;
+
+#ifdef GIT_USE_ICONV
+	git_path_iconv_t ic;
+#endif
+};
+
+#define GIT_PATH_DIRITER_INIT { GIT_BUF_INIT }
+
+#endif
+
+/**
+ * Initialize a directory iterator.
+ *
+ * @param diriter Pointer to a diriter structure that will be setup.
+ * @param path The path that will be iterated over
+ * @param flags Directory reader flags
+ * @return 0 or an error code
+ */
+extern int git_path_diriter_init(
+	git_path_diriter *diriter,
+	const char *path,
+	unsigned int flags);
+
+/**
+ * Advance the directory iterator.  Will return GIT_ITEROVER when
+ * the iteration has completed successfully.
+ *
+ * @param diriter The directory iterator
+ * @return 0, GIT_ITEROVER, or an error code
+ */
+extern int git_path_diriter_next(git_path_diriter *diriter);
+
+/**
+ * Returns the file name of the current item in the iterator.
+ *
+ * @param out Pointer to store the path in
+ * @param out_len Pointer to store the length of the path in
+ * @param diriter The directory iterator
+ * @return 0 or an error code
+ */
+extern int git_path_diriter_filename(
+	const char **out,
+	size_t *out_len,
+	git_path_diriter *diriter);
+
+/**
+ * Returns the full path of the current item in the iterator; that
+ * is the current filename plus the path of the directory that the
+ * iterator was constructed with.
+ *
+ * @param out Pointer to store the path in
+ * @param out_len Pointer to store the length of the path in
+ * @param diriter The directory iterator
+ * @return 0 or an error code
+ */
+extern int git_path_diriter_fullpath(
+	const char **out,
+	size_t *out_len,
+	git_path_diriter *diriter);
+
+/**
+ * Performs an `lstat` on the current item in the iterator.
+ *
+ * @param out Pointer to store the stat data in
+ * @param diriter The directory iterator
+ * @return 0 or an error code
+ */
+extern int git_path_diriter_stat(struct stat *out, git_path_diriter *diriter);
+
+/**
+ * Closes the directory iterator.
+ *
+ * @param diriter The directory iterator
+ */
+extern void git_path_diriter_free(git_path_diriter *diriter);
+
+/**
+ * Load all directory entries (except '.' and '..') into a vector.
+ *
+ * For cases where `git_path_direach()` is not appropriate, this
+ * allows you to load the filenames in a directory into a vector
+ * of strings. That vector can then be sorted, iterated, or whatever.
+ * Remember to free alloc of the allocated strings when you are done.
+ *
+ * @param contents Vector to fill with directory entry names.
+ * @param path The directory to read from.
+ * @param prefix_len When inserting entries, the trailing part of path
+ * 		will be prefixed after this length.  I.e. given path "/a/b" and
+ * 		prefix_len 3, the entries will look like "b/e1", "b/e2", etc.
+ * @param flags Combination of GIT_PATH_DIR flags.
+ */
+extern int git_path_dirload(
+	git_vector *contents,
+	const char *path,
+	size_t prefix_len,
+	uint32_t flags);
+
+
 /* Used for paths to repositories on the filesystem */
 extern bool git_path_is_local_file_url(const char *file_url);
 extern int git_path_from_url_or_path(git_buf *local_path_out, const char *url_or_path);

src/posix.h

@@ -122,7 +122,6 @@ extern int git__page_size(size_t *page_size);
 #include "strnlen.h"
 
 #ifdef NO_READDIR_R
-#	include <dirent.h>
 GIT_INLINE(int) p_readdir_r(DIR *dirp, struct dirent *entry, struct dirent **result)
 {
 	GIT_UNUSED(entry);

src/unix/posix.h

@@ -8,6 +8,7 @@
 #define INCLUDE_posix__unix_h__
 
 #include <stdio.h>
+#include <dirent.h>
 #include <sys/param.h>
 
 typedef int GIT_SOCKET;

src/win32/buffer.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 "buffer.h"
+#include "../buffer.h"
+#include "utf-conv.h"
+
+GIT_INLINE(int) handle_wc_error(void)
+{
+	if (GetLastError() == ERROR_INSUFFICIENT_BUFFER)
+		errno = ENAMETOOLONG;
+	else
+		errno = EINVAL;
+
+	return -1;
+}
+
+int git_buf_put_w(git_buf *buf, const wchar_t *string_w, size_t len_w)
+{
+	int utf8_len, utf8_write_len;
+	size_t new_size;
+
+	if (!len_w)
+		return 0;
+
+	assert(string_w);
+
+	/* Measure the string necessary for conversion */
+	if ((utf8_len = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, string_w, len_w, NULL, 0, NULL, NULL)) == 0)
+		return 0;
+
+	assert(utf8_len > 0);
+
+	GITERR_CHECK_ALLOC_ADD(&new_size, buf->size, (size_t)utf8_len);
+	GITERR_CHECK_ALLOC_ADD(&new_size, new_size, 1);
+
+	if (git_buf_grow(buf, new_size) < 0)
+		return -1;
+
+	if ((utf8_write_len = WideCharToMultiByte(
+			CP_UTF8, WC_ERR_INVALID_CHARS, string_w, len_w, &buf->ptr[buf->size], utf8_len, NULL, NULL)) == 0)
+		return handle_wc_error();
+
+	assert(utf8_write_len == utf8_len);
+
+	buf->size += utf8_write_len;
+	buf->ptr[buf->size] = '\0';
+	return 0;
+}
+

src/win32/buffer.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_git_win32_buffer_h__
+#define INCLUDE_git_win32_buffer_h__
+
+#include "../buffer.h"
+
+/**
+ * Convert a wide character string to UTF-8 and append the results to the
+ * buffer.
+ */
+int git_buf_put_w(git_buf *buf, const wchar_t *string_w, size_t len_w);
+
+#endif

src/win32/path_w32.c

@@ -9,6 +9,9 @@
 #include "path.h"
 #include "path_w32.h"
 #include "utf-conv.h"
+#include "posix.h"
+#include "reparse.h"
+#include "dir.h"
 
 #define PATH__NT_NAMESPACE     L"\\\\?\\"
 #define PATH__NT_NAMESPACE_LEN 4
@@ -303,3 +306,75 @@ char *git_win32_path_8dot3_name(const char *path)
 
 	return shortname;
 }
+
+static bool path_is_volume(wchar_t *target, size_t target_len)
+{
+	return (target_len && wcsncmp(target, L"\\??\\Volume{", 11) == 0);
+}
+
+/* On success, returns the length, in characters, of the path stored in dest.
+* On failure, returns a negative value. */
+int git_win32_path_readlink_w(git_win32_path dest, const git_win32_path path)
+{
+	BYTE buf[MAXIMUM_REPARSE_DATA_BUFFER_SIZE];
+	GIT_REPARSE_DATA_BUFFER *reparse_buf = (GIT_REPARSE_DATA_BUFFER *)buf;
+	HANDLE handle = NULL;
+	DWORD ioctl_ret;
+	wchar_t *target;
+	size_t target_len;
+
+	int error = -1;
+
+	handle = CreateFileW(path, GENERIC_READ,
+		FILE_SHARE_READ | FILE_SHARE_DELETE, NULL, OPEN_EXISTING,
+		FILE_FLAG_OPEN_REPARSE_POINT | FILE_FLAG_BACKUP_SEMANTICS, NULL);
+
+	if (handle == INVALID_HANDLE_VALUE) {
+		errno = ENOENT;
+		return -1;
+	}
+
+	if (!DeviceIoControl(handle, FSCTL_GET_REPARSE_POINT, NULL, 0,
+		reparse_buf, sizeof(buf), &ioctl_ret, NULL)) {
+		errno = EINVAL;
+		goto on_error;
+	}
+
+	switch (reparse_buf->ReparseTag) {
+	case IO_REPARSE_TAG_SYMLINK:
+		target = reparse_buf->SymbolicLinkReparseBuffer.PathBuffer +
+			(reparse_buf->SymbolicLinkReparseBuffer.SubstituteNameOffset / sizeof(WCHAR));
+		target_len = reparse_buf->SymbolicLinkReparseBuffer.SubstituteNameLength / sizeof(WCHAR);
+	break;
+	case IO_REPARSE_TAG_MOUNT_POINT:
+		target = reparse_buf->MountPointReparseBuffer.PathBuffer +
+			(reparse_buf->MountPointReparseBuffer.SubstituteNameOffset / sizeof(WCHAR));
+		target_len = reparse_buf->MountPointReparseBuffer.SubstituteNameLength / sizeof(WCHAR);
+	break;
+	default:
+		errno = EINVAL;
+		goto on_error;
+	}
+
+	if (path_is_volume(target, target_len)) {
+		/* This path is a reparse point that represents another volume mounted
+		* at this location, it is not a symbolic link our input was canonical.
+		*/
+		errno = EINVAL;
+		error = -1;
+	} else if (target_len) {
+		/* The path may need to have a prefix removed. */
+		target_len = git_win32__canonicalize_path(target, target_len);
+
+		/* Need one additional character in the target buffer
+		* for the terminating NULL. */
+		if (GIT_WIN_PATH_UTF16 > target_len) {
+			wcscpy(dest, target);
+			error = (int)target_len;
+		}
+	}
+
+on_error:
+	CloseHandle(handle);
+	return error;
+}

src/win32/path_w32.h

@@ -8,6 +8,7 @@
 #define INCLUDE_git_path_w32_h__
 
 #include "common.h"
+#include "vector.h"
 
 /*
  * Provides a large enough buffer to support Windows paths:  MAX_PATH is
@@ -79,4 +80,6 @@ extern int git_win32_path_to_utf8(git_win32_utf8_path dest, const wchar_t *src);
  */
 extern char *git_win32_path_8dot3_name(const char *path);
 
+extern int git_win32_path_readlink_w(git_win32_path dest, const git_win32_path path);
+
 #endif

src/win32/posix.h

@@ -49,7 +49,7 @@ extern int p_ftruncate(int fd, git_off_t size);
  */
 extern int p_lstat_posixly(const char *filename, struct stat *buf);
 
-extern struct tm * p_localtime_r (const time_t *timer, struct tm *result);
-extern struct tm * p_gmtime_r (const time_t *timer, struct tm *result);
+extern struct tm * p_localtime_r(const time_t *timer, struct tm *result);
+extern struct tm * p_gmtime_r(const time_t *timer, struct tm *result);
 
 #endif

src/win32/posix_w32.c

@@ -130,88 +130,6 @@ int p_fsync(int fd)
 	return 0;
 }
 
-GIT_INLINE(time_t) filetime_to_time_t(const FILETIME *ft)
-{
-	long long winTime = ((long long)ft->dwHighDateTime << 32) + ft->dwLowDateTime;
-	winTime -= 116444736000000000LL; /* Windows to Unix Epoch conversion */
-	winTime /= 10000000;		 /* Nano to seconds resolution */
-	return (time_t)winTime;
-}
-
-static bool path_is_volume(wchar_t *target, size_t target_len)
-{
-	return (target_len && wcsncmp(target, L"\\??\\Volume{", 11) == 0);
-}
-
-/* On success, returns the length, in characters, of the path stored in dest.
- * On failure, returns a negative value. */
-static int readlink_w(
-	git_win32_path dest,
-	const git_win32_path path)
-{
-	BYTE buf[MAXIMUM_REPARSE_DATA_BUFFER_SIZE];
-	GIT_REPARSE_DATA_BUFFER *reparse_buf = (GIT_REPARSE_DATA_BUFFER *)buf;
-	HANDLE handle = NULL;
-	DWORD ioctl_ret;
-	wchar_t *target;
-	size_t target_len;
-
-	int error = -1;
-
-	handle = CreateFileW(path, GENERIC_READ,
-		FILE_SHARE_READ | FILE_SHARE_DELETE, NULL, OPEN_EXISTING,
-		FILE_FLAG_OPEN_REPARSE_POINT | FILE_FLAG_BACKUP_SEMANTICS, NULL);
-
-	if (handle == INVALID_HANDLE_VALUE) {
-		errno = ENOENT;
-		return -1;
-	}
-
-	if (!DeviceIoControl(handle, FSCTL_GET_REPARSE_POINT, NULL, 0,
-		reparse_buf, sizeof(buf), &ioctl_ret, NULL)) {
-		errno = EINVAL;
-		goto on_error;
-	}
-
-	switch (reparse_buf->ReparseTag) {
-	case IO_REPARSE_TAG_SYMLINK:
-		target = reparse_buf->SymbolicLinkReparseBuffer.PathBuffer +
-			(reparse_buf->SymbolicLinkReparseBuffer.SubstituteNameOffset / sizeof(WCHAR));
-		target_len = reparse_buf->SymbolicLinkReparseBuffer.SubstituteNameLength / sizeof(WCHAR);
-		break;
-	case IO_REPARSE_TAG_MOUNT_POINT:
-		target = reparse_buf->MountPointReparseBuffer.PathBuffer +
-			(reparse_buf->MountPointReparseBuffer.SubstituteNameOffset / sizeof(WCHAR));
-		target_len = reparse_buf->MountPointReparseBuffer.SubstituteNameLength / sizeof(WCHAR);
-		break;
-	default:
-		errno = EINVAL;
-		goto on_error;
-	}
-
-	if (path_is_volume(target, target_len)) {
-		/* This path is a reparse point that represents another volume mounted
-		 * at this location, it is not a symbolic link our input was canonical.
-		 */
-		errno = EINVAL;
-		error = -1;
-	} else if (target_len) {
-		/* The path may need to have a prefix removed. */
-		target_len = git_win32__canonicalize_path(target, target_len);
-
-		/* Need one additional character in the target buffer
-		 * for the terminating NULL. */
-		if (GIT_WIN_PATH_UTF16 > target_len) {
-			wcscpy(dest, target);
-			error = (int)target_len;
-		}
-	}
-
-on_error:
-	CloseHandle(handle);
-	return error;
-}
-
 #define WIN32_IS_WSEP(CH) ((CH) == L'/' || (CH) == L'\\')
 
 static int lstat_w(
@@ -222,44 +140,10 @@ static int lstat_w(
 	WIN32_FILE_ATTRIBUTE_DATA fdata;
 
 	if (GetFileAttributesExW(path, GetFileExInfoStandard, &fdata)) {
-		int fMode = S_IREAD;
-
 		if (!buf)
 			return 0;
 
-		if (fdata.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY)
-			fMode |= S_IFDIR;
-		else
-			fMode |= S_IFREG;
-
-		if (!(fdata.dwFileAttributes & FILE_ATTRIBUTE_READONLY))
-			fMode |= S_IWRITE;
-
-		buf->st_ino = 0;
-		buf->st_gid = 0;
-		buf->st_uid = 0;
-		buf->st_nlink = 1;
-		buf->st_mode = (mode_t)fMode;
-		buf->st_size = ((git_off_t)fdata.nFileSizeHigh << 32) + fdata.nFileSizeLow;
-		buf->st_dev = buf->st_rdev = (_getdrive() - 1);
-		buf->st_atime = filetime_to_time_t(&(fdata.ftLastAccessTime));
-		buf->st_mtime = filetime_to_time_t(&(fdata.ftLastWriteTime));
-		buf->st_ctime = filetime_to_time_t(&(fdata.ftCreationTime));
-
-		if (fdata.dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) {
-			git_win32_path target;
-
-			if (readlink_w(target, path) >= 0) {
-				buf->st_mode = (buf->st_mode & ~S_IFMT) | S_IFLNK;
-
-				/* st_size gets the UTF-8 length of the target name, in bytes,
-				 * not counting the NULL terminator */
-				if ((buf->st_size = git__utf16_to_8(NULL, 0, target)) < 0)
-					return -1;
-			}
-		}
-
-		return 0;
+		return git_win32__file_attribute_to_stat(buf, &fdata, path);
 	}
 
 	errno = ENOENT;
@@ -331,7 +215,7 @@ int p_readlink(const char *path, char *buf, size_t bufsiz)
 	 * we need to buffer the result on the stack. */
 
 	if (git_win32_path_from_utf8(path_w, path) < 0 ||
-		readlink_w(target_w, path_w) < 0 ||
+		git_win32_path_readlink_w(target_w, path_w) < 0 ||
 		(len = git_win32_path_to_utf8(target, target_w)) < 0)
 		return -1;
 

src/win32/utf-conv.c

@@ -8,10 +8,6 @@
 #include "common.h"
 #include "utf-conv.h"
 
-#ifndef WC_ERR_INVALID_CHARS
-# define WC_ERR_INVALID_CHARS	0x80
-#endif
-
 GIT_INLINE(DWORD) get_wc_flags(void)
 {
 	static char inited = 0;

src/win32/utf-conv.h

@@ -10,6 +10,10 @@
 #include <wchar.h>
 #include "common.h"
 
+#ifndef WC_ERR_INVALID_CHARS
+# define WC_ERR_INVALID_CHARS	0x80
+#endif
+
 /**
  * Converts a UTF-8 string to wide characters.
  *

src/win32/w32_util.h

@@ -9,8 +9,21 @@
 #define INCLUDE_w32_util_h__
 
 #include "utf-conv.h"
+#include "posix.h"
 #include "path_w32.h"
 
+/*
+
+#include "common.h"
+#include "path.h"
+#include "path_w32.h"
+#include "utf-conv.h"
+#include "posix.h"
+#include "reparse.h"
+#include "dir.h"
+*/
+
+
 GIT_INLINE(bool) git_win32__isalpha(wchar_t c)
 {
 	return ((c >= L'A' && c <= L'Z') || (c >= L'a' && c <= L'z'));
@@ -52,4 +65,63 @@ size_t git_win32__path_trim_end(wchar_t *str, size_t len);
  */
 size_t git_win32__canonicalize_path(wchar_t *str, size_t len);
 
+/**
+ * Converts a FILETIME structure to a time_t.
+ *
+ * @param FILETIME A pointer to a FILETIME
+ * @return A time_t containing the same time
+ */
+GIT_INLINE(time_t) git_win32__filetime_to_time_t(const FILETIME *ft)
+{
+	long long winTime = ((long long)ft->dwHighDateTime << 32) + ft->dwLowDateTime;
+	winTime -= 116444736000000000LL; /* Windows to Unix Epoch conversion */
+	winTime /= 10000000;             /* Nano to seconds resolution */
+	return (time_t)winTime;
+}
+
+GIT_INLINE(int) git_win32__file_attribute_to_stat(
+	struct stat *st,
+	const WIN32_FILE_ATTRIBUTE_DATA *attrdata,
+	const wchar_t *path)
+{
+	mode_t mode = S_IREAD;
+
+	if (attrdata->dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY)
+		mode |= S_IFDIR;
+	else
+		mode |= S_IFREG;
+
+	if ((attrdata->dwFileAttributes & FILE_ATTRIBUTE_READONLY) == 0)
+		mode |= S_IWRITE;
+
+	st->st_ino = 0;
+	st->st_gid = 0;
+	st->st_uid = 0;
+	st->st_nlink = 1;
+	st->st_mode = mode;
+	st->st_size = ((git_off_t)attrdata->nFileSizeHigh << 32) + attrdata->nFileSizeLow;
+	st->st_dev = _getdrive() - 1;
+	st->st_rdev = st->st_dev;
+	st->st_atime = git_win32__filetime_to_time_t(&(attrdata->ftLastAccessTime));
+	st->st_mtime = git_win32__filetime_to_time_t(&(attrdata->ftLastWriteTime));
+	st->st_ctime = git_win32__filetime_to_time_t(&(attrdata->ftCreationTime));
+
+	if (attrdata->dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT && path) {
+		git_win32_path target;
+
+		if (git_win32_path_readlink_w(target, path) >= 0) {
+			st->st_mode = (st->st_mode & ~S_IFMT) | S_IFLNK;
+
+			/* st_size gets the UTF-8 length of the target name, in bytes,
+				* not counting the NULL terminator */
+			if ((st->st_size = git__utf16_to_8(NULL, 0, target)) < 0) {
+				giterr_set(GITERR_OS, "Could not convert reparse point name for '%s'", path);
+				return -1;
+			}
+		}
+	}
+
+	return 0;
+}
+
 #endif

tests/attr/lookup.c

@@ -13,7 +13,7 @@ void test_attr_lookup__simple(void)
 	cl_assert_equal_s(cl_fixture("attr/attr0"), file->entry->path);
 	cl_assert(file->rules.length == 1);
 
-	cl_git_pass(git_attr_path__init(&path, "test", NULL));
+	cl_git_pass(git_attr_path__init(&path, "test", NULL, GIT_DIR_FLAG_UNKNOWN));
 	cl_assert_equal_s("test", path.path);
 	cl_assert_equal_s("test", path.basename);
 	cl_assert(!path.is_dir);
@@ -36,7 +36,7 @@ static void run_test_cases(git_attr_file *file, struct attr_expected *cases, int
 	int error;
 
 	for (c = cases; c->path != NULL; c++) {
-		cl_git_pass(git_attr_path__init(&path, c->path, NULL));
+		cl_git_pass(git_attr_path__init(&path, c->path, NULL, GIT_DIR_FLAG_UNKNOWN));
 
 		if (force_dir)
 			path.is_dir = 1;
@@ -133,7 +133,7 @@ void test_attr_lookup__match_variants(void)
 	cl_assert_equal_s(cl_fixture("attr/attr1"), file->entry->path);
 	cl_assert(file->rules.length == 10);
 
-	cl_git_pass(git_attr_path__init(&path, "/testing/for/pat0", NULL));
+	cl_git_pass(git_attr_path__init(&path, "/testing/for/pat0", NULL, GIT_DIR_FLAG_UNKNOWN));
 	cl_assert_equal_s("pat0", path.basename);
 
 	run_test_cases(file, cases, 0);

tests/core/dirent.c

@@ -67,10 +67,23 @@ static void check_counts(walk_data *d)
 	}
 }
 
+static int update_count(name_data *data, const char *name)
+{
+	name_data *n;
+
+	for (n = data; n->name; n++) {
+		if (!strcmp(n->name, name)) {
+			n->count++;
+			return 0;
+		}
+	}
+
+	return GIT_ERROR;
+}
+
 static int one_entry(void *state, git_buf *path)
 {
 	walk_data *d = (walk_data *) state;
-	name_data *n;
 
 	if (state != state_loc)
 		return GIT_ERROR;
@@ -78,14 +91,7 @@ static int one_entry(void *state, git_buf *path)
 	if (path != &d->path)
 		return GIT_ERROR;
 
-	for (n = d->names; n->name; n++) {
-		if (!strcmp(n->name, path->ptr)) {
-			n->count++;
-			return 0;
-		}
-	}
-
-	return GIT_ERROR;
+	return update_count(d->names, path->ptr);
 }
 
 
@@ -234,3 +240,38 @@ void test_core_dirent__empty_dir(void)
 
 	cl_must_pass(p_rmdir("empty_dir"));
 }
+
+static void handle_next(git_path_diriter *diriter, walk_data *walk)
+{
+	const char *fullpath, *filename;
+	size_t fullpath_len, filename_len;
+
+	cl_git_pass(git_path_diriter_fullpath(&fullpath, &fullpath_len, diriter));
+	cl_git_pass(git_path_diriter_filename(&filename, &filename_len, diriter));
+
+	cl_assert_equal_strn(fullpath, "sub/", 4);
+	cl_assert_equal_s(fullpath+4, filename);
+
+	update_count(walk->names, fullpath);
+}
+
+/* test directory iterator */
+void test_core_dirent__diriter_with_fullname(void)
+{
+	git_path_diriter diriter = GIT_PATH_DIRITER_INIT;
+	int error;
+
+	cl_set_cleanup(&dirent_cleanup__cb, &sub);
+	setup(&sub);
+
+	cl_git_pass(git_path_diriter_init(&diriter, sub.path.ptr, 0));
+
+	while ((error = git_path_diriter_next(&diriter)) == 0)
+		handle_next(&diriter, &sub);
+
+	cl_assert_equal_i(error, GIT_ITEROVER);
+
+	git_path_diriter_free(&diriter);
+
+	check_counts(&sub);
+}

tests/diff/drivers.c

@@ -186,7 +186,7 @@ void test_diff_drivers__builtins(void)
 
 	g_repo = cl_git_sandbox_init("userdiff");
 
-	cl_git_pass(git_path_dirload("userdiff/files", 9, 0, 0, &files));
+	cl_git_pass(git_path_dirload(&files, "userdiff/files", 9, 0));
 
 	opts.interhunk_lines = 1;
 	opts.context_lines = 1;