diff: use new iterator pathlist handling

When using literal pathspecs in diff with `GIT_DIFF_DISABLE_PATHSPEC_MATCH`
turn on the faster iterator pathlist handling.

Updates iterator pathspecs to include directory prefixes (eg, `foo/`)
for compatibility with `GIT_DIFF_DISABLE_PATHSPEC_MATCH`.

src/diff.c

@@ -80,14 +80,13 @@ static bool diff_pathspec_match(
 	/* The iterator has filtered out paths for us, so the fact that we're
 	 * seeing this patch means that it must match the given path list.
 	 */
-	if (DIFF_FLAG_IS_SET(diff, GIT_DIFF_ENABLE_FILELIST_MATCH)) {
+	if (DIFF_FLAG_IS_SET(diff, GIT_DIFF_DISABLE_PATHSPEC_MATCH)) {
 		*matched_pathspec = path;
 		return true;
 	}
 
 	return git_pathspec__match(
-		&diff->pathspec, path,
-		DIFF_FLAG_IS_SET(diff, GIT_DIFF_DISABLE_PATHSPEC_MATCH),
+		&diff->pathspec, path, false,
 		DIFF_FLAG_IS_SET(diff, GIT_DIFF_IGNORE_CASE),
 		matched_pathspec, NULL);
 }
@@ -1063,6 +1062,12 @@ static int handle_unmatched_new_item(
 					&info->nitem, &untracked_state, info->new_iter)) < 0)
 				return error;
 
+			/* if we found nothing that matched our pathlist filter, exclude */
+			if (untracked_state == GIT_ITERATOR_STATUS_FILTERED) {
+				git_vector_pop(&diff->deltas);
+				git__free(last);
+			}
+
 			/* if we found nothing or just ignored items, update the record */
 			if (untracked_state == GIT_ITERATOR_STATUS_IGNORED ||
 				untracked_state == GIT_ITERATOR_STATUS_EMPTY) {
@@ -1276,8 +1281,7 @@ cleanup:
 
 #define DIFF_FROM_ITERATORS(MAKE_FIRST, FLAGS_FIRST, MAKE_SECOND, FLAGS_SECOND) do { \
 	git_iterator *a = NULL, *b = NULL; \
-	git_vector pathlist = GIT_VECTOR_INIT; \
-	char *pfx = (opts && !(opts->flags & GIT_DIFF_ENABLE_FILELIST_MATCH)) ? \
+	char *pfx = (opts && !(opts->flags & GIT_DIFF_DISABLE_PATHSPEC_MATCH)) ? \
 		git_pathspec_prefix(&opts->pathspec) : NULL; \
 	git_iterator_options a_opts = GIT_ITERATOR_OPTIONS_INIT, \
 		b_opts = GIT_ITERATOR_OPTIONS_INIT; \
@@ -1288,19 +1292,15 @@ cleanup:
 	b_opts.start = pfx; \
 	b_opts.end = pfx; \
 	GITERR_CHECK_VERSION(opts, GIT_DIFF_OPTIONS_VERSION, "git_diff_options"); \
-	if (opts && (opts->flags & GIT_DIFF_ENABLE_FILELIST_MATCH) && opts->pathspec.count) { \
-		size_t __i; \
-		error = git_vector_init(&pathlist, opts->pathspec.count, NULL); \
-		for (__i = 0; !error && __i < opts->pathspec.count; __i++) { \
-			error = git_vector_insert(&pathlist, opts->pathspec.strings[__i]); \
-		} \
-		a_opts.pathlist = &pathlist; \
-		b_opts.pathlist = &pathlist; \
+	if (opts && (opts->flags & GIT_DIFF_DISABLE_PATHSPEC_MATCH)) { \
+		a_opts.pathlist.strings = opts->pathspec.strings; \
+		a_opts.pathlist.count = opts->pathspec.count; \
+		b_opts.pathlist.strings = opts->pathspec.strings; \
+		b_opts.pathlist.count = opts->pathspec.count; \
 	} \
 	if (!error && !(error = MAKE_FIRST) && !(error = MAKE_SECOND)) \
 		error = git_diff__from_iterators(diff, repo, a, b, opts); \
 	git__free(pfx); git_iterator_free(a); git_iterator_free(b); \
-	git_vector_free(&pathlist); \
 } while (0)
 
 int git_diff_tree_to_tree(

src/iterator.h

@@ -38,15 +38,14 @@ typedef enum {
 	GIT_ITERATOR_INCLUDE_CONFLICTS = (1u << 5),
 } git_iterator_flag_t;
 
-
 typedef struct {
 	const char *start;
 	const char *end;
 
-	/* paths to include in the iterator (literal).  any paths not listed
-	 * will be excluded.  note that this vector may be resorted!
+	/* paths to include in the iterator (literal).  if set, any paths not
+	 * listed here will be excluded from iteration.
 	 */
-	git_vector *pathlist;
+	git_strarray pathlist;
 
 	/* flags, from above */
 	unsigned int flags;
@@ -70,8 +69,9 @@ struct git_iterator {
 	git_repository *repo;
 	char *start;
 	char *end;
-	git_vector *pathlist;
-	size_t pathlist_idx;
+	git_vector pathlist;
+	int (*strcomp)(const char *a, const char *b);
+	int (*strncomp)(const char *a, const char *b, size_t n);
 	int (*prefixcomp)(const char *str, const char *prefix);
 	size_t stat_calls;
 	unsigned int flags;
@@ -277,7 +277,8 @@ extern git_index *git_iterator_get_index(git_iterator *iter);
 typedef enum {
 	GIT_ITERATOR_STATUS_NORMAL = 0,
 	GIT_ITERATOR_STATUS_IGNORED = 1,
-	GIT_ITERATOR_STATUS_EMPTY = 2
+	GIT_ITERATOR_STATUS_EMPTY = 2,
+	GIT_ITERATOR_STATUS_FILTERED = 3
 } git_iterator_status_t;
 
 /* Advance over a directory and check if it contains no files or just

src/merge.c

@@ -2357,7 +2357,8 @@ static int merge_check_index(size_t *conflicts, git_repository *repo, git_index
 	}
 
 	iter_opts.flags = GIT_ITERATOR_DONT_IGNORE_CASE;
-	iter_opts.pathlist = &staged_paths;
+	iter_opts.pathlist.strings = (char **)staged_paths.contents;
+	iter_opts.pathlist.count = staged_paths.length;
 
 	if ((error = git_iterator_for_index(&iter_repo, index_repo, &iter_opts)) < 0 ||
 		(error = git_iterator_for_index(&iter_new, index_new, &iter_opts)) < 0 ||

tests/diff/workdir.c

@@ -486,7 +486,7 @@ void test_diff_workdir__to_index_with_pathlist_disabling_fnmatch(void)
 	/* ensure that a single NULL pathspec is filtered out (like when using
 	 * fnmatch filtering)
 	 */
-	opts.pathspec.strings = &pathspec;
+
 	opts.pathspec.count   = 1;
 
 	cl_git_pass(git_diff_index_to_workdir(&diff, g_repo, NULL, &opts));
@@ -581,6 +581,30 @@ void test_diff_workdir__to_index_with_pathlist_disabling_fnmatch(void)
 
 	git_diff_free(diff);
 
+	/* ensure that multiple trailing slashes are ignored */
+	pathspec = "subdir//////";
+
+	cl_git_pass(git_diff_index_to_workdir(&diff, g_repo, NULL, &opts));
+
+	for (use_iterator = 0; use_iterator <= 1; use_iterator++) {
+		memset(&exp, 0, sizeof(exp));
+
+		if (use_iterator)
+			cl_git_pass(diff_foreach_via_iterator(
+				diff, diff_file_cb, NULL, NULL, NULL, &exp));
+		else
+			cl_git_pass(git_diff_foreach(diff, diff_file_cb, NULL, NULL, NULL, &exp));
+
+		cl_assert_equal_i(3, exp.files);
+		cl_assert_equal_i(0, exp.file_status[GIT_DELTA_ADDED]);
+		cl_assert_equal_i(1, exp.file_status[GIT_DELTA_DELETED]);
+		cl_assert_equal_i(1, exp.file_status[GIT_DELTA_MODIFIED]);
+		cl_assert_equal_i(0, exp.file_status[GIT_DELTA_IGNORED]);
+		cl_assert_equal_i(1, exp.file_status[GIT_DELTA_UNTRACKED]);
+	}
+
+	git_diff_free(diff);
+
 	/* ensure that fnmatching is completely disabled */
 	pathspec = "subdir/*";
 

tests/repo/iterator.c

@@ -26,7 +26,7 @@ static void expect_iterator_items(
 	const git_index_entry *entry;
 	int count, error;
 	int no_trees = !(git_iterator_flags(i) & GIT_ITERATOR_INCLUDE_TREES);
-	bool v = true;
+	bool v = false;
 
 	if (expected_flat < 0) { v = true; expected_flat = -expected_flat; }
 	if (expected_total < 0) { v = true; expected_total = -expected_total; }
@@ -1099,7 +1099,8 @@ void test_repo_iterator__indexfilelist(void)
 	/* In this test we DO NOT force a case setting on the index. */
 	default_icase = ((git_index_caps(index) & GIT_INDEXCAP_IGNORE_CASE) != 0);
 
-	i_opts.pathlist = &filelist;
+	i_opts.pathlist.strings = (char **)filelist.contents;
+	i_opts.pathlist.count = filelist.length;
 
 	/* All indexfilelist iterator tests are "autoexpand with no tree entries" */
 
@@ -1147,7 +1148,8 @@ void test_repo_iterator__indexfilelist_2(void)
 	cl_git_pass(git_vector_insert(&filelist, "e"));
 	cl_git_pass(git_vector_insert(&filelist, "k/a"));
 
-	i_opts.pathlist = &filelist;
+	i_opts.pathlist.strings = (char **)filelist.contents;
+	i_opts.pathlist.count = filelist.length;
 
 	i_opts.start = "b";
 	i_opts.end = "k/D";
@@ -1188,7 +1190,8 @@ void test_repo_iterator__indexfilelist_icase(void)
 
 	/* All indexfilelist iterator tests are "autoexpand with no tree entries" */
 
-	i_opts.pathlist = &filelist;
+	i_opts.pathlist.strings = (char **)filelist.contents;
+	i_opts.pathlist.count = filelist.length;
 
 	i_opts.start = "c";
 	i_opts.end = "k/D";
@@ -1248,7 +1251,8 @@ void test_repo_iterator__workdirfilelist(void)
 	/* All indexfilelist iterator tests are "autoexpand with no tree entries" */
 	/* In this test we DO NOT force a case on the iteratords and verify default behavior. */
 
-	i_opts.pathlist = &filelist;
+	i_opts.pathlist.strings = (char **)filelist.contents;
+	i_opts.pathlist.count = filelist.length;
 
 	cl_git_pass(git_iterator_for_workdir(&i, g_repo, NULL, NULL, &i_opts));
 	expect_iterator_items(i, 8, NULL, 8, NULL);
@@ -1297,7 +1301,8 @@ void test_repo_iterator__workdirfilelist_icase(void)
 	g_repo = cl_git_sandbox_init("icase");
 
 	i_opts.flags = GIT_ITERATOR_DONT_IGNORE_CASE;
-	i_opts.pathlist = &filelist;
+	i_opts.pathlist.strings = (char **)filelist.contents;
+	i_opts.pathlist.count = filelist.length;
 
 	i_opts.start = "c";
 	i_opts.end = "k/D";