Add test for diffs with submodules and bug fixes

The adds a test for the submodule diff capabilities and then
fixes a few bugs with how the output is generated.  It improves
the accuracy of OIDs in the diff delta object and makes the
submodule output more closely mirror the OIDs that will be used
by core git.

include/git2/submodule.h

@@ -120,8 +120,15 @@ typedef enum {
 } git_submodule_status_t;
 
 #define GIT_SUBMODULE_STATUS_IS_UNMODIFIED(S) \
-	(((S) & ~(GIT_SUBMODULE_STATUS_IN_HEAD | GIT_SUBMODULE_STATUS_IN_INDEX | \
-	GIT_SUBMODULE_STATUS_IN_CONFIG | GIT_SUBMODULE_STATUS_IN_WD)) == 0)
+	(((S) & ~(GIT_SUBMODULE_STATUS_IN_HEAD | \
+	GIT_SUBMODULE_STATUS_IN_INDEX | \
+	GIT_SUBMODULE_STATUS_IN_CONFIG | \
+	GIT_SUBMODULE_STATUS_IN_WD)) == 0)
+
+#define GIT_SUBMODULE_STATUS_IS_WD_DIRTY(S) \
+	(((S) & (GIT_SUBMODULE_STATUS_WD_INDEX_MODIFIED | \
+	GIT_SUBMODULE_STATUS_WD_WD_MODIFIED | \
+	GIT_SUBMODULE_STATUS_WD_UNTRACKED)) != 0)
 
 /**
  * Lookup submodule information by name or path.

src/diff.c

@@ -224,6 +224,9 @@ static int diff_delta__from_one(
 	}
 
 	delta->old_file.flags |= GIT_DIFF_FILE_VALID_OID;
+
+	if (delta->status == GIT_DELTA_DELETED ||
+		!git_oid_iszero(&delta->new_file.oid))
 		delta->new_file.flags |= GIT_DIFF_FILE_VALID_OID;
 
 	if (git_vector_insert(&diff->deltas, delta) < 0) {
@@ -441,17 +444,28 @@ static int oid_for_workdir_item(
 	const git_index_entry *item,
 	git_oid *oid)
 {
-	int result;
+	int result = 0;
 	git_buf full_path = GIT_BUF_INIT;
 
-	if (git_buf_joinpath(&full_path, git_repository_workdir(repo), item->path) < 0)
+	if (git_buf_joinpath(
+		&full_path, git_repository_workdir(repo), item->path) < 0)
 		return -1;
 
-	/* calculate OID for file if possible*/
+	/* calculate OID for file if possible */
 	if (S_ISGITLINK(item->mode)) {
-		/* Don't bother to figure out an oid for a submodule. We won't use it anyway. */
+		git_submodule *sm;
+		const git_oid *sm_oid;
+
+		if (!git_submodule_lookup(&sm, repo, item->path) &&
+			(sm_oid = git_submodule_wd_oid(sm)) != NULL)
+			git_oid_cpy(oid, sm_oid);
+		else {
+			/* if submodule lookup failed probably just in an intermediate
+			 * state where some init hasn't happened, so ignore the error
+			 */
+			giterr_clear();
 			memset(oid, 0, sizeof(*oid));
-		result = 0;
+		}
 	} else if (S_ISLNK(item->mode))
 		result = git_odb__hashlink(oid, full_path.ptr);
 	else if (!git__is_sizet(item->file_size)) {
@@ -570,6 +584,15 @@ static int maybe_modified(
 					return -1;
 				status = GIT_SUBMODULE_STATUS_IS_UNMODIFIED(sm_status)
 						 ? GIT_DELTA_UNMODIFIED : GIT_DELTA_MODIFIED;
+
+				/* grab OID while we are here */
+				if (git_oid_iszero(&nitem->oid)) {
+					const git_oid *sm_oid = git_submodule_wd_oid(sub);
+					if (sub != NULL) {
+						git_oid_cpy(&noid, sm_oid);
+						use_noid = &noid;
+					}
+				}
 			}
 		}
 	}

src/diff_output.c

@@ -275,30 +275,34 @@ static int get_workdir_sm_content(
 	int error = 0;
 	git_buf content = GIT_BUF_INIT;
 	git_submodule* sm = NULL;
-	const git_oid* sm_head = NULL;
 	unsigned int sm_status = 0;
 	const char* sm_status_text = "";
 	char oidstr[GIT_OID_HEXSZ+1];
 
-	if ((error = git_submodule_lookup(&sm, ctxt->repo, file->path)) < 0) {
+	if ((error = git_submodule_lookup(&sm, ctxt->repo, file->path)) < 0 ||
+		(error = git_submodule_status(&sm_status, sm)) < 0)
 		return error;
-	}
 
-	if ((sm_head = git_submodule_head_oid(sm)) == NULL) {
-		giterr_set(GITERR_SUBMODULE, "Cannot find head of submodule '%s'", file->path);
-		return -1;
-	}
+	/* update OID if we didn't have it previously */
+	if ((file->flags & GIT_DIFF_FILE_VALID_OID) == 0) {
+		const git_oid* sm_head;
 
-	if ((error = git_submodule_status(&sm_status, sm)) < 0) {
-		return -1;
+		if ((sm_head = git_submodule_wd_oid(sm)) != NULL ||
+			(sm_head = git_submodule_head_oid(sm)) != NULL)
+		{
+			git_oid_cpy(&file->oid, sm_head);
+			file->flags |= GIT_DIFF_FILE_VALID_OID;
 		}
-	if (!GIT_SUBMODULE_STATUS_IS_UNMODIFIED(sm_status)) {
-		sm_status_text = "-dirty";
 	}
 
-	git_oid_fmt(oidstr, sm_head);
-	oidstr[GIT_OID_HEXSZ] = 0;
-	git_buf_printf(&content, "Subproject commit %s%s\n", oidstr, sm_status_text );
+	git_oid_fmt(oidstr, &file->oid);
+	oidstr[GIT_OID_HEXSZ] = '\0';
+
+	if (GIT_SUBMODULE_STATUS_IS_WD_DIRTY(sm_status))
+		sm_status_text = "-dirty";
+
+	git_buf_printf(&content, "Subproject commit %s%s\n",
+				   oidstr, sm_status_text);
 
 	map->data = git_buf_detach(&content);
 	map->len = strlen(map->data);
@@ -318,7 +322,7 @@ static int get_workdir_content(
 	git_buf path = GIT_BUF_INIT;
 	const char *wd = git_repository_workdir(ctxt->repo);
 
-	if (file->mode == GIT_FILEMODE_COMMIT)
+	if (S_ISGITLINK(file->mode))
 		return get_workdir_sm_content(ctxt, file, map);
 
 	if (S_ISDIR(file->mode))

tests-clar/diff/workdir.c

@@ -744,3 +744,77 @@ void test_diff_workdir__larger_hunks(void)
 	git_tree_free(a);
 	git_tree_free(b);
 }
+
+/* Set up a test that exercises this code. The easiest test using existing
+ * test data is probably to create a sandbox of submod2 and then run a
+ * git_diff_workdir_to_tree against tree
+ * 873585b94bdeabccea991ea5e3ec1a277895b698. As for what you should actually
+ * test, you can start by just checking that the number of lines of diff
+ * content matches the actual output of git diff. That will at least
+ * demonstrate that the submodule content is being used to generate somewhat
+ * comparable outputs. It is a test that would fail without this code and
+ * will succeed with it.
+ */
+
+#include "../submodule/submodule_helpers.h"
+
+void test_diff_workdir__submodules(void)
+{
+	const char *a_commit = "873585b94bdeabccea991ea5e3ec1a277895b698";
+	git_tree *a;
+	git_diff_options opts = {0};
+	git_diff_list *diff = NULL;
+	diff_expects exp;
+
+	g_repo = cl_git_sandbox_init("submod2");
+
+	cl_fixture_sandbox("submod2_target");
+	p_rename("submod2_target/.gitted", "submod2_target/.git");
+
+	rewrite_gitmodules(git_repository_workdir(g_repo));
+	p_rename("submod2/not_submodule/.gitted", "submod2/not_submodule/.git");
+
+	cl_fixture_cleanup("submod2_target");
+
+	a = resolve_commit_oid_to_tree(g_repo, a_commit);
+
+	opts.flags =
+		GIT_DIFF_INCLUDE_UNTRACKED |
+		GIT_DIFF_RECURSE_UNTRACKED_DIRS |
+		GIT_DIFF_INCLUDE_UNTRACKED_CONTENT;
+
+	cl_git_pass(git_diff_workdir_to_tree(g_repo, &opts, a, &diff));
+
+	/* diff_print(stderr, diff); */
+
+	/* essentially doing: git diff 873585b94bdeabccea991ea5e3ec1a277895b698 */
+
+	memset(&exp, 0, sizeof(exp));
+	cl_git_pass(git_diff_foreach(
+		diff, &exp, diff_file_fn, diff_hunk_fn, diff_line_fn));
+
+	/* the following differs from "git diff 873585" by one "untracked" file
+	 * because the diff list includes the "not_submodule/" directory which
+	 * is not displayed in the text diff.
+	 */
+
+	cl_assert_equal_i(10, exp.files);
+
+	cl_assert_equal_i(0, exp.file_adds);
+	cl_assert_equal_i(0, exp.file_dels);
+	cl_assert_equal_i(1, exp.file_mods);
+	cl_assert_equal_i(0, exp.file_ignored);
+	cl_assert_equal_i(9, exp.file_untracked);
+
+	/* the following numbers match "git diff 873585" exactly */
+
+	cl_assert_equal_i(9, exp.hunks);
+
+	cl_assert_equal_i(33, exp.lines);
+	cl_assert_equal_i(2, exp.line_ctxt);
+	cl_assert_equal_i(30, exp.line_adds);
+	cl_assert_equal_i(1, exp.line_dels);
+
+	git_diff_list_free(diff);
+	git_tree_free(a);
+}