Merge branch 'arrbee/examples-log' into development

11 files changed, 783 insertions, 130 deletions

diff --git a/src/array.h b/src/array.h
index 2d77c71a0..248010425 100644
--- a/src/array.h
+++ b/src/array.h
@@ -30,6 +30,9 @@
 #define git_array_init(a) \
 	do { (a).size = (a).asize = 0; (a).ptr = NULL; } while (0)
 
+#define git_array_init_to_size(a, desired) \
+	do { (a).size = 0; (a).asize = desired; (a).ptr = git__calloc(desired, sizeof(*(a).ptr)); } while (0)
+
 #define git_array_clear(a) \
 	do { git__free((a).ptr); git_array_init(a); } while (0)
 
@@ -63,4 +66,6 @@ GIT_INLINE(void *) git_array_grow(git_array_generic_t *a, size_t item_size)
 
 #define git_array_size(a) (a).size
 
+#define git_array_valid_index(a, i) ((i) < (a).size)
+
 #endif
diff --git a/src/bitvec.h b/src/bitvec.h
new file mode 100644
index 000000000..fd6f0ccf8
--- /dev/null
+++ b/src/bitvec.h
@@ -0,0 +1,75 @@
+/*
+ * 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_bitvec_h__
+#define INCLUDE_bitvec_h__
+
+#include "util.h"
+
+/*
+ * This is a silly little fixed length bit vector type that will store
+ * vectors of 64 bits or less directly in the structure and allocate
+ * memory for vectors longer than 64 bits.  You can use the two versions
+ * transparently through the API and avoid heap allocation completely when
+ * using a short bit vector as a result.
+ */
+typedef struct {
+	size_t length;
+	union {
+		uint64_t *words;
+		uint64_t bits;
+	} u;
+} git_bitvec;
+
+GIT_INLINE(int) git_bitvec_init(git_bitvec *bv, size_t capacity)
+{
+	memset(bv, 0x0, sizeof(*bv));
+
+	if (capacity >= 64) {
+		bv->length = (capacity / 64) + 1;
+		bv->u.words = git__calloc(bv->length, sizeof(uint64_t));
+		if (!bv->u.words)
+			return -1;
+	}
+
+	return 0;
+}
+
+#define GIT_BITVEC_MASK(BIT) ((uint64_t)1 << (BIT % 64))
+#define GIT_BITVEC_WORD(BV, BIT) (BV->length ? &BV->u.words[BIT / 64] : &BV->u.bits)
+
+GIT_INLINE(void) git_bitvec_set(git_bitvec *bv, size_t bit, bool on)
+{
+	uint64_t *word = GIT_BITVEC_WORD(bv, bit);
+	uint64_t mask = GIT_BITVEC_MASK(bit);
+
+	if (on)
+		*word |= mask;
+	else
+		*word &= ~mask;
+}
+
+GIT_INLINE(bool) git_bitvec_get(git_bitvec *bv, size_t bit)
+{
+	uint64_t *word = GIT_BITVEC_WORD(bv, bit);
+	return (*word & GIT_BITVEC_MASK(bit)) != 0;
+}
+
+GIT_INLINE(void) git_bitvec_clear(git_bitvec *bv)
+{
+	if (!bv->length)
+		bv->u.bits = 0;
+	else
+		memset(bv->u.words, 0x0, bv->length * sizeof(uint64_t));
+}
+
+GIT_INLINE(void) git_bitvec_free(git_bitvec *bv)
+{
+	if (bv->length)
+		git__free(bv->u.words);
+}
+
+#endif
diff --git a/src/checkout.c b/src/checkout.c
index 8f9ec64e4..ec9da7e2e 100644
--- a/src/checkout.c
+++ b/src/checkout.c
@@ -246,10 +246,10 @@ static int checkout_action_wd_only(
 	bool remove = false;
 	git_checkout_notify_t notify = GIT_CHECKOUT_NOTIFY_NONE;
 
-	if (!git_pathspec_match_path(
+	if (!git_pathspec__match(
 			pathspec, wd->path,
 			(data->strategy & GIT_CHECKOUT_DISABLE_PATHSPEC_MATCH) != 0,
-			git_iterator_ignore_case(workdir), NULL))
+			git_iterator_ignore_case(workdir), NULL, NULL))
 		return 0;
 
 	/* check if item is tracked in the index but not in the checkout diff */
@@ -607,7 +607,7 @@ static int checkout_get_actions(
 	uint32_t *actions = NULL;
 
 	if (data->opts.paths.count > 0 &&
-		git_pathspec_init(&pathspec, &data->opts.paths, &pathpool) < 0)
+		git_pathspec__vinit(&pathspec, &data->opts.paths, &pathpool) < 0)
 		return -1;
 
 	if ((error = git_iterator_current(&wditem, workdir)) < 0 &&
@@ -659,7 +659,7 @@ static int checkout_get_actions(
 		goto fail;
 	}
 
-	git_pathspec_free(&pathspec);
+	git_pathspec__vfree(&pathspec);
 	git_pool_clear(&pathpool);
 
 	return 0;
@@ -670,7 +670,7 @@ fail:
 	*actions_ptr = NULL;
 	git__free(actions);
 
-	git_pathspec_free(&pathspec);
+	git_pathspec__vfree(&pathspec);
 	git_pool_clear(&pathpool);
 
 	return error;
diff --git a/src/commit.c b/src/commit.c
index 1ab9b34f7..cc912a7be 100644
--- a/src/commit.c
+++ b/src/commit.c
@@ -19,30 +19,19 @@
 
 #include <stdarg.h>
 
-static void clear_parents(git_commit *commit)
-{
-	size_t i;
-
-	for (i = 0; i < commit->parent_ids.length; ++i) {
-		git_oid *parent = git_vector_get(&commit->parent_ids, i);
-		git__free(parent);
-	}
-
-	git_vector_clear(&commit->parent_ids);
-}
-
 void git_commit__free(void *_commit)
 {
 	git_commit *commit = _commit;
 
-	clear_parents(commit);
-	git_vector_free(&commit->parent_ids);
+	git_array_clear(commit->parent_ids);
 
 	git_signature_free(commit->author);
 	git_signature_free(commit->committer);
 
+	git__free(commit->raw_header);
 	git__free(commit->message);
 	git__free(commit->message_encoding);
+
 	git__free(commit);
 }
 
@@ -171,12 +160,34 @@ int git_commit_create(
 int git_commit__parse(void *_commit, git_odb_object *odb_obj)
 {
 	git_commit *commit = _commit;
-	const char *buffer = git_odb_object_data(odb_obj);
-	const char *buffer_end = buffer + git_odb_object_size(odb_obj);
+	const char *buffer_start = git_odb_object_data(odb_obj), *buffer;
+	const char *buffer_end = buffer_start + git_odb_object_size(odb_obj);
 	git_oid parent_id;
+	size_t parent_count = 0, header_len;
 
-	if (git_vector_init(&commit->parent_ids, 4, NULL) < 0)
-		return -1;
+	/* find end-of-header (counting parents as we go) */
+	for (buffer = buffer_start; buffer < buffer_end; ++buffer) {
+		if (!strncmp("\n\n", buffer, 2)) {
+			++buffer;
+			break;
+		}
+		if (!strncmp("\nparent ", buffer, strlen("\nparent ")))
+			++parent_count;
+	}
+
+	header_len = buffer - buffer_start;
+	commit->raw_header = git__strndup(buffer_start, header_len);
+	GITERR_CHECK_ALLOC(commit->raw_header);
+
+	/* point "buffer" to header data */
+	buffer = commit->raw_header;
+	buffer_end = commit->raw_header + header_len;
+
+	if (parent_count < 1)
+		parent_count = 1;
+
+	git_array_init_to_size(commit->parent_ids, parent_count);
+	GITERR_CHECK_ARRAY(commit->parent_ids);
 
 	if (git_oid__parse(&commit->tree_id, &buffer, buffer_end, "tree ") < 0)
 		goto bad_buffer;
@@ -186,13 +197,10 @@ int git_commit__parse(void *_commit, git_odb_object *odb_obj)
 	 */
 
 	while (git_oid__parse(&parent_id, &buffer, buffer_end, "parent ") == 0) {
-		git_oid *new_id = git__malloc(sizeof(git_oid));
+		git_oid *new_id = git_array_alloc(commit->parent_ids);
 		GITERR_CHECK_ALLOC(new_id);
 
 		git_oid_cpy(new_id, &parent_id);
-
-		if (git_vector_insert(&commit->parent_ids, new_id) < 0)
-			return -1;
 	}
 
 	commit->author = git__malloc(sizeof(git_signature));
@@ -208,8 +216,8 @@ int git_commit__parse(void *_commit, git_odb_object *odb_obj)
 	if (git_signature__parse(commit->committer, &buffer, buffer_end, "committer ", '\n') < 0)
 		return -1;
 
-	/* Parse add'l header entries until blank line found */
-	while (buffer < buffer_end && *buffer != '\n') {
+	/* Parse add'l header entries */
+	while (buffer < buffer_end) {
 		const char *eoln = buffer;
 		while (eoln < buffer_end && *eoln != '\n')
 			++eoln;
@@ -223,15 +231,18 @@ int git_commit__parse(void *_commit, git_odb_object *odb_obj)
 
 		if (eoln < buffer_end && *eoln == '\n')
 			++eoln;
-
 		buffer = eoln;
 	}
 
-	/* buffer is now at the end of the header, double-check and move forward into the message */
-	if (buffer < buffer_end && *buffer == '\n')
-		buffer++;
+	/* point "buffer" to data after header */
+	buffer = git_odb_object_data(odb_obj);
+	buffer_end = buffer + git_odb_object_size(odb_obj);
+
+	buffer += header_len;
+	if (*buffer == '\n')
+		++buffer;
 
-	/* parse commit message */
+	/* extract commit message */
 	if (buffer <= buffer_end) {
 		commit->message = git__strndup(buffer, buffer_end - buffer);
 		GITERR_CHECK_ALLOC(commit->message);
@@ -255,9 +266,10 @@ GIT_COMMIT_GETTER(const git_signature *, author, commit->author)
 GIT_COMMIT_GETTER(const git_signature *, committer, commit->committer)
 GIT_COMMIT_GETTER(const char *, message, commit->message)
 GIT_COMMIT_GETTER(const char *, message_encoding, commit->message_encoding)
+GIT_COMMIT_GETTER(const char *, raw_header, commit->raw_header)
 GIT_COMMIT_GETTER(git_time_t, time, commit->committer->when.time)
 GIT_COMMIT_GETTER(int, time_offset, commit->committer->when.offset)
-GIT_COMMIT_GETTER(unsigned int, parentcount, (unsigned int)commit->parent_ids.length)
+GIT_COMMIT_GETTER(unsigned int, parentcount, (unsigned int)git_array_size(commit->parent_ids))
 GIT_COMMIT_GETTER(const git_oid *, tree_id, &commit->tree_id);
 
 int git_commit_tree(git_tree **tree_out, const git_commit *commit)
@@ -271,7 +283,7 @@ const git_oid *git_commit_parent_id(
 {
 	assert(commit);
 
-	return git_vector_get(&commit->parent_ids, n);
+	return git_array_get(commit->parent_ids, n);
 }
 
 int git_commit_parent(
diff --git a/src/commit.h b/src/commit.h
index d0981b125..22fc898a1 100644
--- a/src/commit.h
+++ b/src/commit.h
@@ -10,14 +10,14 @@
 #include "git2/commit.h"
 #include "tree.h"
 #include "repository.h"
-#include "vector.h"
+#include "array.h"
 
 #include <time.h>
 
 struct git_commit {
 	git_object object;
 
-	git_vector parent_ids;
+	git_array_t(git_oid) parent_ids;
 	git_oid tree_id;
 
 	git_signature *author;
@@ -25,6 +25,7 @@ struct git_commit {
 
 	char *message_encoding;
 	char *message;
+	char *raw_header;
 };
 
 void git_commit__free(void *commit);
diff --git a/src/diff.c b/src/diff.c
index 0980f412a..cc7be451f 100644
--- a/src/diff.c
+++ b/src/diff.c
@@ -81,11 +81,11 @@ static int diff_delta__from_one(
 		DIFF_FLAG_IS_SET(diff, GIT_DIFF_IGNORE_SUBMODULES))
 		return 0;
 
-	if (!git_pathspec_match_path(
+	if (!git_pathspec__match(
 			&diff->pathspec, entry->path,
 			DIFF_FLAG_IS_SET(diff, GIT_DIFF_DISABLE_PATHSPEC_MATCH),
 			DIFF_FLAG_IS_SET(diff, GIT_DIFF_DELTAS_ARE_ICASE),
-			&matched_pathspec))
+			&matched_pathspec, NULL))
 		return 0;
 
 	delta = diff_delta__alloc(diff, status, entry->path);
@@ -247,6 +247,11 @@ GIT_INLINE(const char *) diff_delta__path(const git_diff_delta *delta)
 	return str;
 }
 
+const char *git_diff_delta__path(const git_diff_delta *delta)
+{
+	return diff_delta__path(delta);
+}
+
 int git_diff_delta__cmp(const void *a, const void *b)
 {
 	const git_diff_delta *da = a, *db = b;
@@ -387,7 +392,7 @@ static int diff_list_apply_options(
 		DIFF_FLAG_SET(diff, GIT_DIFF_DELTAS_ARE_ICASE, icase);
 
 		/* initialize pathspec from options */
-		if (git_pathspec_init(&diff->pathspec, &opts->pathspec, pool) < 0)
+		if (git_pathspec__vinit(&diff->pathspec, &opts->pathspec, pool) < 0)
 			return -1;
 	}
 
@@ -473,7 +478,7 @@ static void diff_list_free(git_diff_list *diff)
 	}
 	git_vector_free(&diff->deltas);
 
-	git_pathspec_free(&diff->pathspec);
+	git_pathspec__vfree(&diff->pathspec);
 	git_pool_clear(&diff->pool);
 
 	git__memzero(diff, sizeof(*diff));
@@ -634,11 +639,11 @@ static int maybe_modified(
 	bool new_is_workdir = (info->new_iter->type == GIT_ITERATOR_TYPE_WORKDIR);
 	const char *matched_pathspec;
 
-	if (!git_pathspec_match_path(
+	if (!git_pathspec__match(
 			&diff->pathspec, oitem->path,
 			DIFF_FLAG_IS_SET(diff, GIT_DIFF_DISABLE_PATHSPEC_MATCH),
 			DIFF_FLAG_IS_SET(diff, GIT_DIFF_DELTAS_ARE_ICASE),
-			&matched_pathspec))
+			&matched_pathspec, NULL))
 		return 0;
 
 	memset(&noid, 0, sizeof(noid));
@@ -1235,6 +1240,11 @@ size_t git_diff_num_deltas_of_type(git_diff_list *diff, git_delta_t type)
 	return count;
 }
 
+int git_diff_is_sorted_icase(const git_diff_list *diff)
+{
+	return (diff->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0;
+}
+
 int git_diff__paired_foreach(
 	git_diff_list *head2idx,
 	git_diff_list *idx2wd,
diff --git a/src/diff.h b/src/diff.h
index 6ef03ee7c..d09a130bc 100644
--- a/src/diff.h
+++ b/src/diff.h
@@ -76,6 +76,8 @@ extern void git_diff_list_addref(git_diff_list *diff);
 extern int git_diff_delta__cmp(const void *a, const void *b);
 extern int git_diff_delta__casecmp(const void *a, const void *b);
 
+extern const char *git_diff_delta__path(const git_diff_delta *delta);
+
 extern bool git_diff_delta__should_skip(
 	const git_diff_options *opts, const git_diff_delta *delta);
 
diff --git a/src/index.c b/src/index.c
index ffa84e53f..21efd2c63 100644
--- a/src/index.c
+++ b/src/index.c
@@ -101,8 +101,6 @@ static int parse_index(git_index *index, const char *buffer, size_t buffer_size)
 static bool is_index_extended(git_index *index);
 static int write_index(git_index *index, git_filebuf *file);
 
-static int index_find(size_t *at_pos, git_index *index, const char *path, int stage);
-
 static void index_entry_free(git_index_entry *entry);
 static void index_entry_reuc_free(git_index_reuc_entry *reuc);
 
@@ -114,7 +112,7 @@ static int index_srch(const void *key, const void *array_member)
 
 	ret = strcmp(srch_key->path, entry->path);
 
-	if (ret == 0)
+	if (ret == 0 && srch_key->stage != GIT_INDEX_STAGE_ANY)
 		ret = srch_key->stage - GIT_IDXENTRY_STAGE(entry);
 
 	return ret;
@@ -128,7 +126,7 @@ static int index_isrch(const void *key, const void *array_member)
 
 	ret = strcasecmp(srch_key->path, entry->path);
 
-	if (ret == 0)
+	if (ret == 0 && srch_key->stage != GIT_INDEX_STAGE_ANY)
 		ret = srch_key->stage - GIT_IDXENTRY_STAGE(entry);
 
 	return ret;
@@ -562,7 +560,7 @@ const git_index_entry *git_index_get_bypath(
 
 	git_vector_sort(&index->entries);
 
-	if (index_find(&pos, index, path, stage) < 0) {
+	if (git_index__find(&pos, index, path, stage) < 0) {
 		giterr_set(GITERR_INDEX, "Index does not contain %s", path);
 		return NULL;
 	}
@@ -719,7 +717,8 @@ static int index_insert(git_index *index, git_index_entry *entry, int replace)
 		entry->flags |= GIT_IDXENTRY_NAMEMASK;
 
 	/* look if an entry with this path already exists */
-	if (!index_find(&position, index, entry->path, GIT_IDXENTRY_STAGE(entry))) {
+	if (!git_index__find(
+			&position, index, entry->path, GIT_IDXENTRY_STAGE(entry))) {
 		existing = (git_index_entry **)&index->entries.contents[position];
 
 		/* update filemode to existing values if stat is not trusted */
@@ -831,7 +830,7 @@ int git_index_remove(git_index *index, const char *path, int stage)
 
 	git_vector_sort(&index->entries);
 
-	if (index_find(&position, index, path, stage) < 0) {
+	if (git_index__find(&position, index, path, stage) < 0) {
 		giterr_set(GITERR_INDEX, "Index does not contain %s at stage %d",
 			path, stage);
 		return GIT_ENOTFOUND;
@@ -887,7 +886,8 @@ int git_index_remove_directory(git_index *index, const char *dir, int stage)
 	return error;
 }
 
-static int index_find(size_t *at_pos, git_index *index, const char *path, int stage)
+int git_index__find(
+	size_t *at_pos, git_index *index, const char *path, int stage)
 {
 	struct entry_srch_key srch_key;
 
@@ -896,7 +896,8 @@ static int index_find(size_t *at_pos, git_index *index, const char *path, int st
 	srch_key.path = path;
 	srch_key.stage = stage;
 
-	return git_vector_bsearch2(at_pos, &index->entries, index->entries_search, &srch_key);
+	return git_vector_bsearch2(
+		at_pos, &index->entries, index->entries_search, &srch_key);
 }
 
 int git_index_find(size_t *at_pos, git_index *index, const char *path)
@@ -2053,7 +2054,7 @@ int git_index_add_all(
 	git_iterator *wditer = NULL;
 	const git_index_entry *wd = NULL;
 	git_index_entry *entry;
-	git_pathspec_context ps;
+	git_pathspec ps;
 	const char *match;
 	size_t existing;
 	bool no_fnmatch = (flags & GIT_INDEX_ADD_DISABLE_PATHSPEC_MATCH) != 0;
@@ -2074,7 +2075,7 @@ int git_index_add_all(
 	if (git_repository__cvar(&ignorecase, repo, GIT_CVAR_IGNORECASE) < 0)
 		return -1;
 
-	if ((error = git_pathspec_context_init(&ps, paths)) < 0)
+	if ((error = git_pathspec__init(&ps, paths)) < 0)
 		return error;
 
 	/* optionally check that pathspec doesn't mention any ignored files */
@@ -2091,14 +2092,14 @@ int git_index_add_all(
 	while (!(error = git_iterator_advance(&wd, wditer))) {
 
 		/* check if path actually matches */
-		if (!git_pathspec_match_path(
-				&ps.pathspec, wd->path, no_fnmatch, ignorecase, &match))
+		if (!git_pathspec__match(
+				&ps.pathspec, wd->path, no_fnmatch, ignorecase, &match, NULL))
 			continue;
 
 		/* skip ignored items that are not already in the index */
 		if ((flags & GIT_INDEX_ADD_FORCE) == 0 &&
 			git_iterator_current_is_ignored(wditer) &&
-			index_find(&existing, index, wd->path, 0) < 0)
+			git_index__find(&existing, index, wd->path, 0) < 0)
 			continue;
 
 		/* issue notification callback if requested */
@@ -2148,7 +2149,7 @@ int git_index_add_all(
 
 cleanup:
 	git_iterator_free(wditer);
-	git_pathspec_context_free(&ps);
+	git_pathspec__clear(&ps);
 
 	return error;
 }
@@ -2168,13 +2169,13 @@ static int index_apply_to_all(
 {
 	int error = 0;
 	size_t i;
-	git_pathspec_context ps;
+	git_pathspec ps;
 	const char *match;
 	git_buf path = GIT_BUF_INIT;
 
 	assert(index);
 
-	if ((error = git_pathspec_context_init(&ps, paths)) < 0)
+	if ((error = git_pathspec__init(&ps, paths)) < 0)
 		return error;
 
 	git_vector_sort(&index->entries);
@@ -2183,8 +2184,9 @@ static int index_apply_to_all(
 		git_index_entry *entry = git_vector_get(&index->entries, i);
 
 		/* check if path actually matches */
-		if (!git_pathspec_match_path(
-				&ps.pathspec, entry->path, false, index->ignore_case, &match))
+		if (!git_pathspec__match(
+				&ps.pathspec, entry->path, false, index->ignore_case,
+				&match, NULL))
 			continue;
 
 		/* issue notification callback if requested */
@@ -2231,7 +2233,7 @@ static int index_apply_to_all(
 	}
 
 	git_buf_free(&path);
-	git_pathspec_context_free(&ps);
+	git_pathspec__clear(&ps);
 
 	return error;
 }
diff --git a/src/index.h b/src/index.h
index a59107a7b..40577e105 100644
--- a/src/index.h
+++ b/src/index.h
@@ -47,13 +47,17 @@ struct git_index_conflict_iterator {
 	size_t cur;
 };
 
-extern void git_index_entry__init_from_stat(git_index_entry *entry, struct stat *st);
+extern void git_index_entry__init_from_stat(
+	git_index_entry *entry, struct stat *st);
 
 extern size_t git_index__prefix_position(git_index *index, const char *path);
 
 extern int git_index_entry__cmp(const void *a, const void *b);
 extern int git_index_entry__cmp_icase(const void *a, const void *b);
 
+extern int git_index__find(
+	size_t *at_pos, git_index *index, const char *path, int stage);
+
 extern void git_index__set_ignore_case(git_index *index, bool ignore_case);
 
 #endif
diff --git a/src/pathspec.c b/src/pathspec.c
index f029836d0..625726e0b 100644
--- a/src/pathspec.c
+++ b/src/pathspec.c
@@ -5,9 +5,16 @@
  * a Linking Exception. For full terms see the included COPYING file.
  */
 
+#include "git2/pathspec.h"
+#include "git2/diff.h"
 #include "pathspec.h"
 #include "buf_text.h"
 #include "attr_file.h"
+#include "iterator.h"
+#include "repository.h"
+#include "index.h"
+#include "bitvec.h"
+#include "diff.h"
 
 /* what is the common non-wildcard prefix for all items in the pathspec */
 char *git_pathspec_prefix(const git_strarray *pathspec)
@@ -56,7 +63,7 @@ bool git_pathspec_is_empty(const git_strarray *pathspec)
 }
 
 /* build a vector of fnmatch patterns to evaluate efficiently */
-int git_pathspec_init(
+int git_pathspec__vinit(
 	git_vector *vspec, const git_strarray *strspec, git_pool *strpool)
 {
 	size_t i;
@@ -93,7 +100,7 @@ int git_pathspec_init(
 }
 
 /* free data from the pathspec vector */
-void git_pathspec_free(git_vector *vspec)
+void git_pathspec__vfree(git_vector *vspec)
 {
 	git_attr_fnmatch *match;
 	unsigned int i;
@@ -106,88 +113,602 @@ void git_pathspec_free(git_vector *vspec)
 	git_vector_free(vspec);
 }
 
+struct pathspec_match_context {
+	int fnmatch_flags;
+	int (*strcomp)(const char *, const char *);
+	int (*strncomp)(const char *, const char *, size_t);
+};
+
+static void pathspec_match_context_init(
+	struct pathspec_match_context *ctxt,
+	bool disable_fnmatch,
+	bool casefold)
+{
+	if (disable_fnmatch)
+		ctxt->fnmatch_flags = -1;
+	else if (casefold)
+		ctxt->fnmatch_flags = FNM_CASEFOLD;
+	else
+		ctxt->fnmatch_flags = 0;
+
+	if (casefold) {
+		ctxt->strcomp  = git__strcasecmp;
+		ctxt->strncomp = git__strncasecmp;
+	} else {
+		ctxt->strcomp  = git__strcmp;
+		ctxt->strncomp = git__strncmp;
+	}
+}
+
+static int pathspec_match_one(
+	const git_attr_fnmatch *match,
+	struct pathspec_match_context *ctxt,
+	const char *path)
+{
+	int result = (match->flags & GIT_ATTR_FNMATCH_MATCH_ALL) ? 0 : FNM_NOMATCH;
+
+	if (result == FNM_NOMATCH)
+		result = ctxt->strcomp(match->pattern, path) ? FNM_NOMATCH : 0;
+
+	if (ctxt->fnmatch_flags >= 0 && result == FNM_NOMATCH)
+		result = p_fnmatch(match->pattern, path, ctxt->fnmatch_flags);
+
+	/* if we didn't match, look for exact dirname prefix match */
+	if (result == FNM_NOMATCH &&
+		(match->flags & GIT_ATTR_FNMATCH_HASWILD) == 0 &&
+		ctxt->strncomp(path, match->pattern, match->length) == 0 &&
+		path[match->length] == '/')
+		result = 0;
+
+	if (result == 0)
+		return (match->flags & GIT_ATTR_FNMATCH_NEGATIVE) ? 0 : 1;
+	return -1;
+}
+
+static int git_pathspec__match_at(
+	size_t *matched_at,
+	const git_vector *vspec,
+	struct pathspec_match_context *ctxt,
+	const char *path0,
+	const char *path1)
+{
+	int result = GIT_ENOTFOUND;
+	size_t i = 0;
+	const git_attr_fnmatch *match;
+
+	git_vector_foreach(vspec, i, match) {
+		if (path0 && (result = pathspec_match_one(match, ctxt, path0)) >= 0)
+			break;
+		if (path1 && (result = pathspec_match_one(match, ctxt, path1)) >= 0)
+			break;
+	}
+
+	*matched_at = i;
+	return result;
+}
+
 /* match a path against the vectorized pathspec */
-bool git_pathspec_match_path(
-	git_vector *vspec,
+bool git_pathspec__match(
+	const git_vector *vspec,
 	const char *path,
 	bool disable_fnmatch,
 	bool casefold,
-	const char **matched_pathspec)
+	const char **matched_pathspec,
+	size_t *matched_at)
 {
-	size_t i;
-	git_attr_fnmatch *match;
-	int fnmatch_flags = 0;
-	int (*use_strcmp)(const char *, const char *);
-	int (*use_strncmp)(const char *, const char *, size_t);
+	int result;
+	size_t pos;
+	struct pathspec_match_context ctxt;
 
 	if (matched_pathspec)
 		*matched_pathspec = NULL;
+	if (matched_at)
+		*matched_at = GIT_PATHSPEC_NOMATCH;
 
 	if (!vspec || !vspec->length)
 		return true;
 
-	if (disable_fnmatch)
-		fnmatch_flags = -1;
-	else if (casefold)
-		fnmatch_flags = FNM_CASEFOLD;
+	pathspec_match_context_init(&ctxt, disable_fnmatch, casefold);
 
-	if (casefold) {
-		use_strcmp  = git__strcasecmp;
-		use_strncmp = git__strncasecmp;
-	} else {
-		use_strcmp  = git__strcmp;
-		use_strncmp = git__strncmp;
+	result = git_pathspec__match_at(&pos, vspec, &ctxt, path, NULL);
+	if (result >= 0) {
+		if (matched_pathspec) {
+			const git_attr_fnmatch *match = git_vector_get(vspec, pos);
+			*matched_pathspec = match->pattern;
+		}
+
+		if (matched_at)
+			*matched_at = pos;
 	}
 
-	git_vector_foreach(vspec, i, match) {
-		int result = (match->flags & GIT_ATTR_FNMATCH_MATCH_ALL) ? 0 : FNM_NOMATCH;
+	return (result > 0);
+}
+
+
+int git_pathspec__init(git_pathspec *ps, const git_strarray *paths)
+{
+	int error = 0;
+
+	memset(ps, 0, sizeof(*ps));
+
+	ps->prefix = git_pathspec_prefix(paths);
+
+	if ((error = git_pool_init(&ps->pool, 1, 0)) < 0 ||
+		(error = git_pathspec__vinit(&ps->pathspec, paths, &ps->pool)) < 0)
+		git_pathspec__clear(ps);
+
+	return error;
+}
+
+void git_pathspec__clear(git_pathspec *ps)
+{
+	git__free(ps->prefix);
+	git_pathspec__vfree(&ps->pathspec);
+	git_pool_clear(&ps->pool);
+	memset(ps, 0, sizeof(*ps));
+}
+
+int git_pathspec_new(git_pathspec **out, const git_strarray *pathspec)
+{
+	int error = 0;
+	git_pathspec *ps = git__malloc(sizeof(git_pathspec));
+	GITERR_CHECK_ALLOC(ps);
+
+	if ((error = git_pathspec__init(ps, pathspec)) < 0) {
+		git__free(ps);
+		return error;
+	}
+
+	GIT_REFCOUNT_INC(ps);
+	*out = ps;
+	return 0;
+}
+
+static void pathspec_free(git_pathspec *ps)
+{
+	git_pathspec__clear(ps);
+	git__free(ps);
+}
+
+void git_pathspec_free(git_pathspec *ps)
+{
+	if (!ps)
+		return;
+	GIT_REFCOUNT_DEC(ps, pathspec_free);
+}
+
+int git_pathspec_matches_path(
+	const git_pathspec *ps, uint32_t flags, const char *path)
+{
+	bool no_fnmatch = (flags & GIT_PATHSPEC_NO_GLOB) != 0;
+	bool casefold =  (flags & GIT_PATHSPEC_IGNORE_CASE) != 0;
+
+	assert(ps && path);
+
+	return (0 != git_pathspec__match(
+		&ps->pathspec, path, no_fnmatch, casefold, NULL, NULL));
+}
+
+static void pathspec_match_free(git_pathspec_match_list *m)
+{
+	git_pathspec_free(m->pathspec);
+	m->pathspec = NULL;
+
+	git_array_clear(m->matches);
+	git_array_clear(m->failures);
+	git_pool_clear(&m->pool);
+	git__free(m);
+}
+
+static git_pathspec_match_list *pathspec_match_alloc(
+	git_pathspec *ps, int datatype)
+{
+	git_pathspec_match_list *m = git__calloc(1, sizeof(git_pathspec_match_list));
+
+	if (m != NULL && git_pool_init(&m->pool, 1, 0) < 0) {
+		pathspec_match_free(m);
+		m = NULL;
+	}
+
+	/* need to keep reference to pathspec and increment refcount because
+	 * failures array stores pointers to the pattern strings of the
+	 * pathspec that had no matches
+	 */
+	GIT_REFCOUNT_INC(ps);
+	m->pathspec = ps;
+	m->datatype = datatype;
+
+	return m;
+}
+
+GIT_INLINE(size_t) pathspec_mark_pattern(git_bitvec *used, size_t pos)
+{
+	if (!git_bitvec_get(used, pos)) {
+		git_bitvec_set(used, pos, true);
+		return 1;
+	}
+
+	return 0;
+}
+
+static size_t pathspec_mark_remaining(
+	git_bitvec *used,
+	git_vector *patterns,
+	struct pathspec_match_context *ctxt,
+	size_t start,
+	const char *path0,
+	const char *path1)
+{
+	size_t count = 0;
+
+	if (path1 == path0)
+		path1 = NULL;
+
+	for (; start < patterns->length; ++start) {
+		const git_attr_fnmatch *pat = git_vector_get(patterns, start);
+
+		if (git_bitvec_get(used, start))
+			continue;
+
+		if (path0 && pathspec_match_one(pat, ctxt, path0) > 0)
+			count += pathspec_mark_pattern(used, start);
+		else if (path1 && pathspec_match_one(pat, ctxt, path1) > 0)
+			count += pathspec_mark_pattern(used, start);
+	}
+
+	return count;
+}
+
+static int pathspec_build_failure_array(
+	git_pathspec_string_array_t *failures,
+	git_vector *patterns,
+	git_bitvec *used,
+	git_pool *pool)
+{
+	size_t pos;
+	char **failed;
+	const git_attr_fnmatch *pat;
+
+	for (pos = 0; pos < patterns->length; ++pos) {
+		if (git_bitvec_get(used, pos))
+			continue;
+
+		if ((failed = git_array_alloc(*failures)) == NULL)
+			return -1;
+
+		pat = git_vector_get(patterns, pos);
+
+		if ((*failed = git_pool_strdup(pool, pat->pattern)) == NULL)
+			return -1;
+	}
+
+	return 0;
+}
+
+static int pathspec_match_from_iterator(
+	git_pathspec_match_list **out,
+	git_iterator *iter,
+	uint32_t flags,
+	git_pathspec *ps)
+{
+	int error = 0;
+	git_pathspec_match_list *m = NULL;
+	const git_index_entry *entry = NULL;
+	struct pathspec_match_context ctxt;
+	git_vector *patterns = &ps->pathspec;
+	bool find_failures = out && (flags & GIT_PATHSPEC_FIND_FAILURES) != 0;
+	bool failures_only = !out || (flags & GIT_PATHSPEC_FAILURES_ONLY) != 0;
+	size_t pos, used_ct = 0, found_files = 0;
+	git_index *index = NULL;
+	git_bitvec used_patterns;
+	char **file;
+
+	if (git_bitvec_init(&used_patterns, patterns->length) < 0)
+		return -1;
+
+	if (out) {
+		*out = m = pathspec_match_alloc(ps, PATHSPEC_DATATYPE_STRINGS);
+		GITERR_CHECK_ALLOC(m);
+	}
+
+	if ((error = git_iterator_reset(iter, ps->prefix, ps->prefix)) < 0)
+		goto done;
 
-		if (result == FNM_NOMATCH)
-			result = use_strcmp(match->pattern, path) ? FNM_NOMATCH : 0;
+	if (git_iterator_type(iter) == GIT_ITERATOR_TYPE_WORKDIR &&
+		(error = git_repository_index__weakptr(
+			&index, git_iterator_owner(iter))) < 0)
+		goto done;
 
-		if (fnmatch_flags >= 0 && result == FNM_NOMATCH)
-			result = p_fnmatch(match->pattern, path, fnmatch_flags);
+	pathspec_match_context_init(
+		&ctxt, (flags & GIT_PATHSPEC_NO_GLOB) != 0,
+		git_iterator_ignore_case(iter));
 
-		/* if we didn't match, look for exact dirname prefix match */
-		if (result == FNM_NOMATCH &&
-			(match->flags & GIT_ATTR_FNMATCH_HASWILD) == 0 &&
-			use_strncmp(path, match->pattern, match->length) == 0 &&
-			path[match->length] == '/')
-			result = 0;
+	while (!(error = git_iterator_advance(&entry, iter))) {
+		/* search for match with entry->path */
+		int result = git_pathspec__match_at(
+			&pos, patterns, &ctxt, entry->path, NULL);
 
-		if (result == 0) {
-			if (matched_pathspec)
-				*matched_pathspec = match->pattern;
+		/* no matches for this path */
+		if (result < 0)
+			continue;
 
-			return (match->flags & GIT_ATTR_FNMATCH_NEGATIVE) ? false : true;
+		/* if result was a negative pattern match, then don't list file */
+		if (!result) {
+			used_ct += pathspec_mark_pattern(&used_patterns, pos);
+			continue;
 		}
+
+		/* check if path is ignored and untracked */
+		if (index != NULL &&
+			git_iterator_current_is_ignored(iter) &&
+			git_index__find(NULL, index, entry->path, GIT_INDEX_STAGE_ANY) < 0)
+			continue;
+
+		/* mark the matched pattern as used */
+		used_ct += pathspec_mark_pattern(&used_patterns, pos);
+		++found_files;
+
+		/* if find_failures is on, check if any later patterns also match */
+		if (find_failures && used_ct < patterns->length)
+			used_ct += pathspec_mark_remaining(
+				&used_patterns, patterns, &ctxt, pos + 1, entry->path, NULL);
+
+		/* if only looking at failures, exit early or just continue */
+		if (failures_only || !out) {
+			if (used_ct == patterns->length)
+				break;
+			continue;
+		}
+
+		/* insert matched path into matches array */
+		if ((file = (char **)git_array_alloc(m->matches)) == NULL ||
+			(*file = git_pool_strdup(&m->pool, entry->path)) == NULL) {
+			error = -1;
+			goto done;
+		}
+	}
+
+	if (error < 0 && error != GIT_ITEROVER)
+		goto done;
+	error = 0;
+
+	/* insert patterns that had no matches into failures array */
+	if (find_failures && used_ct < patterns->length &&
+		(error = pathspec_build_failure_array(
+			&m->failures, patterns, &used_patterns, &m->pool)) < 0)
+		goto done;
+
+	/* if every pattern failed to match, then we have failed */
+	if ((flags & GIT_PATHSPEC_NO_MATCH_ERROR) != 0 && !found_files) {
+		giterr_set(GITERR_INVALID, "No matching files were found");
+		error = GIT_ENOTFOUND;
+	}
+
+done:
+	git_bitvec_free(&used_patterns);
+
+	if (error < 0) {
+		pathspec_match_free(m);
+		if (out) *out = NULL;
 	}
 
-	return false;
+	return error;
 }
 
+static git_iterator_flag_t pathspec_match_iter_flags(uint32_t flags)
+{
+	git_iterator_flag_t f = 0;
+
+	if ((flags & GIT_PATHSPEC_IGNORE_CASE) != 0)
+		f |= GIT_ITERATOR_IGNORE_CASE;
+	else if ((flags & GIT_PATHSPEC_USE_CASE) != 0)
+		f |= GIT_ITERATOR_DONT_IGNORE_CASE;
 
-int git_pathspec_context_init(
-	git_pathspec_context *ctxt, const git_strarray *paths)
+	return f;
+}
+
+int git_pathspec_match_workdir(
+	git_pathspec_match_list **out,
+	git_repository *repo,
+	uint32_t flags,
+	git_pathspec *ps)
 {
 	int error = 0;
+	git_iterator *iter;
+
+	assert(repo);
 
-	memset(ctxt, 0, sizeof(*ctxt));
+	if (!(error = git_iterator_for_workdir(
+			&iter, repo, pathspec_match_iter_flags(flags), NULL, NULL))) {
 
-	ctxt->prefix = git_pathspec_prefix(paths);
+		error = pathspec_match_from_iterator(out, iter, flags, ps);
 
-	if ((error = git_pool_init(&ctxt->pool, 1, 0)) < 0 ||
-		(error = git_pathspec_init(&ctxt->pathspec, paths, &ctxt->pool)) < 0)
-		git_pathspec_context_free(ctxt);
+		git_iterator_free(iter);
+	}
 
 	return error;
 }
 
-void git_pathspec_context_free(
-	git_pathspec_context *ctxt)
+int git_pathspec_match_index(
+	git_pathspec_match_list **out,
+	git_index *index,
+	uint32_t flags,
+	git_pathspec *ps)
 {
-	git__free(ctxt->prefix);
-	git_pathspec_free(&ctxt->pathspec);
-	git_pool_clear(&ctxt->pool);
-	memset(ctxt, 0, sizeof(*ctxt));
+	int error = 0;
+	git_iterator *iter;
+
+	assert(index);
+
+	if (!(error = git_iterator_for_index(
+			&iter, index, pathspec_match_iter_flags(flags), NULL, NULL))) {
+
+		error = pathspec_match_from_iterator(out, iter, flags, ps);
+
+		git_iterator_free(iter);
+	}
+
+	return error;
 }
+
+int git_pathspec_match_tree(
+	git_pathspec_match_list **out,
+	git_tree *tree,
+	uint32_t flags,
+	git_pathspec *ps)
+{
+	int error = 0;
+	git_iterator *iter;
+
+	assert(tree);
+
+	if (!(error = git_iterator_for_tree(
+			&iter, tree, pathspec_match_iter_flags(flags), NULL, NULL))) {
+
+		error = pathspec_match_from_iterator(out, iter, flags, ps);
+
+		git_iterator_free(iter);
+	}
+
+	return error;
+}
+
+int git_pathspec_match_diff(
+	git_pathspec_match_list **out,
+	git_diff_list *diff,
+	uint32_t flags,
+	git_pathspec *ps)
+{
+	int error = 0;
+	git_pathspec_match_list *m = NULL;
+	struct pathspec_match_context ctxt;
+	git_vector *patterns = &ps->pathspec;
+	bool find_failures = out && (flags & GIT_PATHSPEC_FIND_FAILURES) != 0;
+	bool failures_only = !out || (flags & GIT_PATHSPEC_FAILURES_ONLY) != 0;
+	size_t i, pos, used_ct = 0, found_deltas = 0;
+	const git_diff_delta *delta, **match;
+	git_bitvec used_patterns;
+
+	assert(diff);
+
+	if (git_bitvec_init(&used_patterns, patterns->length) < 0)
+		return -1;
+
+	if (out) {
+		*out = m = pathspec_match_alloc(ps, PATHSPEC_DATATYPE_DIFF);
+		GITERR_CHECK_ALLOC(m);
+	}
+
+	pathspec_match_context_init(
+		&ctxt, (flags & GIT_PATHSPEC_NO_GLOB) != 0,
+		git_diff_is_sorted_icase(diff));
+
+	git_vector_foreach(&diff->deltas, i, delta) {
+		/* search for match with delta */
+		int result = git_pathspec__match_at(
+			&pos, patterns, &ctxt, delta->old_file.path, delta->new_file.path);
+
+		/* no matches for this path */
+		if (result < 0)
+			continue;
+
+		/* mark the matched pattern as used */
+		used_ct += pathspec_mark_pattern(&used_patterns, pos);
+
+		/* if result was a negative pattern match, then don't list file */
+		if (!result)
+			continue;
+
+		++found_deltas;
+
+		/* if find_failures is on, check if any later patterns also match */
+		if (find_failures && used_ct < patterns->length)
+			used_ct += pathspec_mark_remaining(
+				&used_patterns, patterns, &ctxt, pos + 1,
+				delta->old_file.path, delta->new_file.path);
+
+		/* if only looking at failures, exit early or just continue */
+		if (failures_only || !out) {
+			if (used_ct == patterns->length)
+				break;
+			continue;
+		}
+
+		/* insert matched delta into matches array */
+		if (!(match = (const git_diff_delta **)git_array_alloc(m->matches))) {
+			error = -1;
+			goto done;
+		} else {
+			*match = delta;
+		}
+	}
+
+	/* insert patterns that had no matches into failures array */
+	if (find_failures && used_ct < patterns->length &&
+		(error = pathspec_build_failure_array(
+			&m->failures, patterns, &used_patterns, &m->pool)) < 0)
+		goto done;
+
+	/* if every pattern failed to match, then we have failed */
+	if ((flags & GIT_PATHSPEC_NO_MATCH_ERROR) != 0 && !found_deltas) {
+		giterr_set(GITERR_INVALID, "No matching deltas were found");
+		error = GIT_ENOTFOUND;
+	}
+
+done:
+	git_bitvec_free(&used_patterns);
+
+	if (error < 0) {
+		pathspec_match_free(m);
+		if (out) *out = NULL;
+	}
+
+	return error;
+}
+
+void git_pathspec_match_list_free(git_pathspec_match_list *m)
+{
+	if (m)
+		pathspec_match_free(m);
+}
+
+size_t git_pathspec_match_list_entrycount(
+	const git_pathspec_match_list *m)
+{
+	return m ? git_array_size(m->matches) : 0;
+}
+
+const char *git_pathspec_match_list_entry(
+	const git_pathspec_match_list *m, size_t pos)
+{
+	if (!m || m->datatype != PATHSPEC_DATATYPE_STRINGS ||
+		!git_array_valid_index(m->matches, pos))
+		return NULL;
+
+	return *((const char **)git_array_get(m->matches, pos));
+}
+
+const git_diff_delta *git_pathspec_match_list_diff_entry(
+	const git_pathspec_match_list *m, size_t pos)
+{
+	if (!m || m->datatype != PATHSPEC_DATATYPE_DIFF ||
+		!git_array_valid_index(m->matches, pos))
+		return NULL;
+
+	return *((const git_diff_delta **)git_array_get(m->matches, pos));
+}
+
+size_t git_pathspec_match_list_failed_entrycount(
+	const git_pathspec_match_list *m)
+{
+	return m ? git_array_size(m->failures) : 0;
+}
+
+const char * git_pathspec_match_list_failed_entry(
+	const git_pathspec_match_list *m, size_t pos)
+{
+	char **entry = m ? git_array_get(m->failures, pos) : NULL;
+
+	return entry ? *entry : NULL;
+}
+
diff --git a/src/pathspec.h b/src/pathspec.h
index f6509df4c..40cd21c3f 100644
--- a/src/pathspec.h
+++ b/src/pathspec.h
@@ -8,9 +8,35 @@
 #define INCLUDE_pathspec_h__
 
 #include "common.h"
+#include <git2/pathspec.h>
 #include "buffer.h"
 #include "vector.h"
 #include "pool.h"
+#include "array.h"
+
+/* public compiled pathspec */
+struct git_pathspec {
+	git_refcount rc;
+	char *prefix;
+	git_vector pathspec;
+	git_pool pool;
+};
+
+enum {
+	PATHSPEC_DATATYPE_STRINGS = 0,
+	PATHSPEC_DATATYPE_DIFF = 1,
+};
+
+typedef git_array_t(char *) git_pathspec_string_array_t;
+
+/* public interface to pathspec matching */
+struct git_pathspec_match_list {
+	git_pathspec *pathspec;
+	git_array_t(void *) matches;
+	git_pathspec_string_array_t failures;
+	git_pool pool;
+	int datatype;
+};
 
 /* what is the common non-wildcard prefix for all items in the pathspec */
 extern char *git_pathspec_prefix(const git_strarray *pathspec);
@@ -19,36 +45,31 @@ extern char *git_pathspec_prefix(const git_strarray *pathspec);
 extern bool git_pathspec_is_empty(const git_strarray *pathspec);
 
 /* build a vector of fnmatch patterns to evaluate efficiently */
-extern int git_pathspec_init(
+extern int git_pathspec__vinit(
 	git_vector *vspec, const git_strarray *strspec, git_pool *strpool);
 
 /* free data from the pathspec vector */
-extern void git_pathspec_free(git_vector *vspec);
+extern void git_pathspec__vfree(git_vector *vspec);
+
+#define GIT_PATHSPEC_NOMATCH ((size_t)-1)
 
 /*
  * Match a path against the vectorized pathspec.
  * The matched pathspec is passed back into the `matched_pathspec` parameter,
  * unless it is passed as NULL by the caller.
  */
-extern bool git_pathspec_match_path(
-	git_vector *vspec,
+extern bool git_pathspec__match(
+	const git_vector *vspec,
 	const char *path,
 	bool disable_fnmatch,
 	bool casefold,
-	const char **matched_pathspec);
+	const char **matched_pathspec,
+	size_t *matched_at);
 
 /* easy pathspec setup */
 
-typedef struct {
-	char *prefix;
-	git_vector pathspec;
-	git_pool pool;
-} git_pathspec_context;
-
-extern int git_pathspec_context_init(
-	git_pathspec_context *ctxt, const git_strarray *paths);
+extern int git_pathspec__init(git_pathspec *ps, const git_strarray *paths);
 
-extern void git_pathspec_context_free(
-	git_pathspec_context *ctxt);
+extern void git_pathspec__clear(git_pathspec *ps);
 
 #endif