Fix problems in diff iterator record chaining

There is a bug in building the linked list of line records in the
diff iterator and also an off by one element error in the hunk
counts.  This fixes both of these, adds some test data with more
complex sets of hunk and line diffs to exercise this code better.

21 files changed, 299 insertions, 13 deletions

diff --git a/src/diff_output.c b/src/diff_output.c
index ea40c3355..50e3cc1de 100644
--- a/src/diff_output.c
+++ b/src/diff_output.c
@@ -1204,13 +1204,17 @@ static int diffiter_hunk_cb(
 	if (info->last_hunk)
 		info->last_hunk->next = hunk;
 	info->last_hunk = hunk;
+	info->last_line = NULL;
 
 	memcpy(&hunk->range, range, sizeof(hunk->range));
 
 	iter->hunk_count++;
 
-	if (iter->hunk_head == NULL)
-		iter->hunk_curr = iter->hunk_head = hunk;
+	/* adding first hunk to list */
+	if (iter->hunk_head == NULL) {
+		iter->hunk_head = hunk;
+		iter->hunk_curr = NULL;
+	}
 
 	return 0;
 }
@@ -1345,9 +1349,14 @@ int git_diff_iterator_num_hunks_in_file(git_diff_iterator *iter)
 int git_diff_iterator_num_lines_in_hunk(git_diff_iterator *iter)
 {
 	int error = diffiter_do_diff_file(iter);
-	if (!error && iter->hunk_curr)
-		error = iter->hunk_curr->line_count;
-	return error;
+	if (error)
+		return error;
+
+	if (iter->hunk_curr)
+		return iter->hunk_curr->line_count;
+	if (iter->hunk_head)
+		return iter->hunk_head->line_count;
+	return 0;
 }
 
 int git_diff_iterator_next_file(
@@ -1386,7 +1395,7 @@ int git_diff_iterator_next_file(
 	}
 
 	if (iter->ctxt.delta == NULL) {
-		iter->hunk_curr = NULL;
+		iter->hunk_curr = iter->hunk_head = NULL;
 		iter->line_curr = NULL;
 	}
 
@@ -1409,11 +1418,13 @@ int git_diff_iterator_next_hunk(
 		return error;
 
 	if (iter->hunk_curr == NULL) {
-		if (range_ptr) *range_ptr = NULL;
-		if (header) *header = NULL;
-		if (header_len) *header_len = 0;
-		iter->line_curr = NULL;
-		return GIT_ITEROVER;
+		if (iter->hunk_head == NULL)
+			goto no_more_hunks;
+		iter->hunk_curr = iter->hunk_head;
+	} else {
+		if (iter->hunk_curr->next == NULL)
+			goto no_more_hunks;
+		iter->hunk_curr = iter->hunk_curr->next;
 	}
 
 	range = &iter->hunk_curr->range;
@@ -1436,9 +1447,16 @@ int git_diff_iterator_next_hunk(
 	}
 
 	iter->line_curr = iter->hunk_curr->line_head;
-	iter->hunk_curr = iter->hunk_curr->next;
 
 	return error;
+
+no_more_hunks:
+	if (range_ptr) *range_ptr = NULL;
+	if (header) *header = NULL;
+	if (header_len) *header_len = 0;
+	iter->line_curr = NULL;
+
+	return GIT_ITEROVER;
 }
 
 int git_diff_iterator_next_line(
@@ -1453,7 +1471,7 @@ int git_diff_iterator_next_line(
 		return error;
 
 	/* if the user has not called next_hunk yet, call it implicitly (OK?) */
-	if (iter->hunk_curr == iter->hunk_head) {
+	if (iter->hunk_curr == NULL) {
 		error = git_diff_iterator_next_hunk(NULL, NULL, NULL, iter);
 		if (error)
 			return error;
diff --git a/tests-clar/diff/tree.c b/tests-clar/diff/tree.c
index 3003374a5..f5e72cadc 100644
--- a/tests-clar/diff/tree.c
+++ b/tests-clar/diff/tree.c
@@ -256,3 +256,85 @@ void test_diff_tree__merge(void)
 
 	git_diff_list_free(diff1);
 }
+
+void test_diff_tree__larger_hunks(void)
+{
+	const char *a_commit = "d70d245ed97ed2aa596dd1af6536e4bfdb047b69";
+	const char *b_commit = "7a9e0b02e63179929fed24f0a3e0f19168114d10";
+	git_tree *a, *b;
+	git_diff_options opts = {0};
+	git_diff_list *diff = NULL;
+	git_diff_iterator *iter = NULL;
+	git_diff_delta *delta;
+	diff_expects exp;
+	int error, num_files = 0;
+
+	g_repo = cl_git_sandbox_init("diff");
+
+	cl_assert((a = resolve_commit_oid_to_tree(g_repo, a_commit)) != NULL);
+	cl_assert((b = resolve_commit_oid_to_tree(g_repo, b_commit)) != NULL);
+
+	opts.context_lines = 1;
+	opts.interhunk_lines = 0;
+
+	memset(&exp, 0, sizeof(exp));
+
+	cl_git_pass(git_diff_tree_to_tree(g_repo, &opts, a, b, &diff));
+	cl_git_pass(git_diff_iterator_new(&iter, diff));
+
+	/* this should be exact */
+	cl_assert(git_diff_iterator_progress(iter) == 0.0f);
+
+	/* You wouldn't actually structure an iterator loop this way, but
+	 * I have here for testing purposes of the return value
+	 */
+	while (!(error = git_diff_iterator_next_file(&delta, iter))) {
+		git_diff_range *range;
+		const char *header;
+		size_t header_len;
+		int actual_hunks = 0, num_hunks;
+		float expected_progress;
+
+		num_files++;
+
+		expected_progress = (float)num_files / 2.0f;
+		cl_assert(expected_progress == git_diff_iterator_progress(iter));
+
+		num_hunks = git_diff_iterator_num_hunks_in_file(iter);
+
+		while (!(error = git_diff_iterator_next_hunk(
+					&range, &header, &header_len, iter)))
+		{
+			int actual_lines = 0;
+			int num_lines = git_diff_iterator_num_lines_in_hunk(iter);
+			char origin;
+			const char *line;
+			size_t line_len;
+
+			while (!(error = git_diff_iterator_next_line(
+						&origin, &line, &line_len, iter)))
+			{
+				actual_lines++;
+			}
+
+			cl_assert_equal_i(GIT_ITEROVER, error);
+			cl_assert_equal_i(actual_lines, num_lines);
+
+			actual_hunks++;
+		}
+
+		cl_assert_equal_i(GIT_ITEROVER, error);
+		cl_assert_equal_i(actual_hunks, num_hunks);
+	}
+
+	cl_assert_equal_i(GIT_ITEROVER, error);
+	cl_assert_equal_i(2, num_files);
+	cl_assert(git_diff_iterator_progress(iter) == 1.0f);
+
+	git_diff_iterator_free(iter);
+	git_diff_list_free(diff);
+	diff = NULL;
+
+	git_tree_free(a);
+	git_tree_free(b);
+}
diff --git a/tests-clar/diff/workdir.c b/tests-clar/diff/workdir.c
index eac7eb87d..40a888544 100644
--- a/tests-clar/diff/workdir.c
+++ b/tests-clar/diff/workdir.c
@@ -670,3 +670,94 @@ void test_diff_workdir__eof_newline_changes(void)
  *
  * Expect 13 files, 0 ADD, 4 DEL, 4 MOD, 1 IGN, 4 UNTR
  */
+
+
+void test_diff_workdir__larger_hunks(void)
+{
+	const char *a_commit = "d70d245ed97ed2aa596dd1af6536e4bfdb047b69";
+	const char *b_commit = "7a9e0b02e63179929fed24f0a3e0f19168114d10";
+	git_tree *a, *b;
+	git_diff_options opts = {0};
+	int i, error;
+
+	g_repo = cl_git_sandbox_init("diff");
+
+	cl_assert((a = resolve_commit_oid_to_tree(g_repo, a_commit)) != NULL);
+	cl_assert((b = resolve_commit_oid_to_tree(g_repo, b_commit)) != NULL);
+
+	opts.context_lines = 1;
+	opts.interhunk_lines = 0;
+
+	for (i = 0; i <= 2; ++i) {
+		git_diff_list *diff = NULL;
+		git_diff_iterator *iter = NULL;
+		git_diff_delta *delta;
+		int num_files = 0;
+
+		/* okay, this is a bit silly, but oh well */
+		switch (i) {
+		case 0:
+			cl_git_pass(git_diff_workdir_to_index(g_repo, &opts, &diff));
+			break;
+		case 1:
+			cl_git_pass(git_diff_workdir_to_tree(g_repo, &opts, a, &diff));
+			break;
+		case 2:
+			cl_git_pass(git_diff_workdir_to_tree(g_repo, &opts, b, &diff));
+			break;
+		}
+
+		cl_git_pass(git_diff_iterator_new(&iter, diff));
+
+		cl_assert(git_diff_iterator_progress(iter) == 0.0f);
+
+		while (!(error = git_diff_iterator_next_file(&delta, iter))) {
+			git_diff_range *range;
+			const char *header;
+			size_t header_len;
+			int actual_hunks = 0, num_hunks;
+			float expected_progress;
+
+			num_files++;
+
+			expected_progress = (float)num_files / 2.0f;
+			cl_assert(expected_progress == git_diff_iterator_progress(iter));
+
+			num_hunks = git_diff_iterator_num_hunks_in_file(iter);
+
+			while (!(error = git_diff_iterator_next_hunk(
+						 &range, &header, &header_len, iter)))
+			{
+				int actual_lines = 0;
+				int num_lines = git_diff_iterator_num_lines_in_hunk(iter);
+				char origin;
+				const char *line;
+				size_t line_len;
+
+				while (!(error = git_diff_iterator_next_line(
+							 &origin, &line, &line_len, iter)))
+				{
+					actual_lines++;
+				}
+
+				cl_assert_equal_i(GIT_ITEROVER, error);
+				cl_assert_equal_i(actual_lines, num_lines);
+
+				actual_hunks++;
+			}
+
+			cl_assert_equal_i(GIT_ITEROVER, error);
+			cl_assert_equal_i(actual_hunks, num_hunks);
+		}
+
+		cl_assert_equal_i(GIT_ITEROVER, error);
+		cl_assert_equal_i(2, num_files);
+		cl_assert(git_diff_iterator_progress(iter) == 1.0f);
+
+		git_diff_iterator_free(iter);
+		git_diff_list_free(diff);
+	}
+
+	git_tree_free(a);
+	git_tree_free(b);
+}
diff --git a/tests-clar/resources/diff/.gitted/HEAD b/tests-clar/resources/diff/.gitted/HEAD
new file mode 100644
index 000000000..cb089cd89
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/HEAD
@@ -0,0 +1 @@
+ref: refs/heads/master
diff --git a/tests-clar/resources/diff/.gitted/config b/tests-clar/resources/diff/.gitted/config
new file mode 100644
index 000000000..77a27ef1d
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/config
@@ -0,0 +1,6 @@
+[core]
+	repositoryformatversion = 0
+	filemode = true
+	bare = false
+	logallrefupdates = true
+	ignorecase = false
diff --git a/tests-clar/resources/diff/.gitted/description b/tests-clar/resources/diff/.gitted/description
new file mode 100644
index 000000000..498b267a8
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/description
@@ -0,0 +1 @@
+Unnamed repository; edit this file 'description' to name the repository.
diff --git a/tests-clar/resources/diff/.gitted/index b/tests-clar/resources/diff/.gitted/index
new file mode 100644
index 000000000..e1071874e
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/index
diff --git a/tests-clar/resources/diff/.gitted/info/exclude b/tests-clar/resources/diff/.gitted/info/exclude
new file mode 100644
index 000000000..a5196d1be
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/info/exclude
@@ -0,0 +1,6 @@
+# git ls-files --others --exclude-from=.git/info/exclude
+# Lines that start with '#' are comments.
+# For a project mostly in C, the following would be a good set of
+# exclude patterns (uncomment them if you want to use them):
+# *.[oa]
+# *~
diff --git a/tests-clar/resources/diff/.gitted/logs/HEAD b/tests-clar/resources/diff/.gitted/logs/HEAD
new file mode 100644
index 000000000..8c6f6fd18
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/logs/HEAD
@@ -0,0 +1,2 @@
+0000000000000000000000000000000000000000 d70d245ed97ed2aa596dd1af6536e4bfdb047b69 Russell Belfer <rb@github.com> 1347559804 -0700	commit (initial): initial commit
+d70d245ed97ed2aa596dd1af6536e4bfdb047b69 7a9e0b02e63179929fed24f0a3e0f19168114d10 Russell Belfer <rb@github.com> 1347560491 -0700	commit: some changes
diff --git a/tests-clar/resources/diff/.gitted/logs/refs/heads/master b/tests-clar/resources/diff/.gitted/logs/refs/heads/master
new file mode 100644
index 000000000..8c6f6fd18
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/logs/refs/heads/master
@@ -0,0 +1,2 @@
+0000000000000000000000000000000000000000 d70d245ed97ed2aa596dd1af6536e4bfdb047b69 Russell Belfer <rb@github.com> 1347559804 -0700	commit (initial): initial commit
+d70d245ed97ed2aa596dd1af6536e4bfdb047b69 7a9e0b02e63179929fed24f0a3e0f19168114d10 Russell Belfer <rb@github.com> 1347560491 -0700	commit: some changes
diff --git a/tests-clar/resources/diff/.gitted/objects/29/ab7053bb4dde0298e03e2c179e890b7dd465a7 b/tests-clar/resources/diff/.gitted/objects/29/ab7053bb4dde0298e03e2c179e890b7dd465a7
new file mode 100644
index 000000000..94f9a676d
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/29/ab7053bb4dde0298e03e2c179e890b7dd465a7
diff --git a/tests-clar/resources/diff/.gitted/objects/3e/5bcbad2a68e5bc60a53b8388eea53a1a7ab847 b/tests-clar/resources/diff/.gitted/objects/3e/5bcbad2a68e5bc60a53b8388eea53a1a7ab847
new file mode 100644
index 000000000..9fed523dc
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/3e/5bcbad2a68e5bc60a53b8388eea53a1a7ab847
diff --git a/tests-clar/resources/diff/.gitted/objects/54/6c735f16a3b44d9784075c2c0dab2ac9bf1989 b/tests-clar/resources/diff/.gitted/objects/54/6c735f16a3b44d9784075c2c0dab2ac9bf1989
new file mode 100644
index 000000000..d7df4d6a1
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/54/6c735f16a3b44d9784075c2c0dab2ac9bf1989
diff --git a/tests-clar/resources/diff/.gitted/objects/7a/9e0b02e63179929fed24f0a3e0f19168114d10 b/tests-clar/resources/diff/.gitted/objects/7a/9e0b02e63179929fed24f0a3e0f19168114d10
new file mode 100644
index 000000000..9bc25eb34
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/7a/9e0b02e63179929fed24f0a3e0f19168114d10
diff --git a/tests-clar/resources/diff/.gitted/objects/7b/808f723a8ca90df319682c221187235af76693 b/tests-clar/resources/diff/.gitted/objects/7b/808f723a8ca90df319682c221187235af76693
new file mode 100644
index 000000000..2fd266be6
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/7b/808f723a8ca90df319682c221187235af76693
diff --git a/tests-clar/resources/diff/.gitted/objects/88/789109439c1e1c3cd45224001edee5304ed53c b/tests-clar/resources/diff/.gitted/objects/88/789109439c1e1c3cd45224001edee5304ed53c
new file mode 100644
index 000000000..7598b5914
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/88/789109439c1e1c3cd45224001edee5304ed53c
@@ -0,0 +1 @@
+x
+)JMU07g040031QH��/�H-�+�(a�)�[��wz{��j���%;�ʊRSrS��4W����4そ���N��+��
�a
\ No newline at end of file
diff --git a/tests-clar/resources/diff/.gitted/objects/cb/8294e696339863df760b2ff5d1e275bee72455 b/tests-clar/resources/diff/.gitted/objects/cb/8294e696339863df760b2ff5d1e275bee72455
new file mode 100644
index 000000000..86ebe04fe
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/cb/8294e696339863df760b2ff5d1e275bee72455
diff --git a/tests-clar/resources/diff/.gitted/objects/d7/0d245ed97ed2aa596dd1af6536e4bfdb047b69 b/tests-clar/resources/diff/.gitted/objects/d7/0d245ed97ed2aa596dd1af6536e4bfdb047b69
new file mode 100644
index 000000000..99304c4aa
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/objects/d7/0d245ed97ed2aa596dd1af6536e4bfdb047b69
@@ -0,0 +1 @@
+x
���	!�m��_��RB:X�kVpWp��� ���9��{� ,^z#�����7JygԚ��A��� i1Y��2�V��yR)𢒨����'�m�[�����;���-�l��O�_��#���%��v8�
\ No newline at end of file
diff --git a/tests-clar/resources/diff/.gitted/refs/heads/master b/tests-clar/resources/diff/.gitted/refs/heads/master
new file mode 100644
index 000000000..a83afc38b
--- /dev/null
+++ b/tests-clar/resources/diff/.gitted/refs/heads/master
@@ -0,0 +1 @@
+7a9e0b02e63179929fed24f0a3e0f19168114d10
diff --git a/tests-clar/resources/diff/another.txt b/tests-clar/resources/diff/another.txt
new file mode 100644
index 000000000..d0e0bae4d
--- /dev/null
+++ b/tests-clar/resources/diff/another.txt
@@ -0,0 +1,38 @@
+Git is fast. With Git, nearly all operations are performed locally, giving
+it an huge speed advantage on centralized systems that constantly have to
+communicate with a server somewh3r3.
+
+For testing, large AWS instances were set up in the same availability
+zone. Git and SVN were installed on both machines, the Ruby repository was
+copied to both Git and SVN servers, and common operations were performed on
+both.
+
+In some cases the commands don't match up exactly. Here, matching on the
+lowest common denominator was attempted. For example, the 'commit' tests
+also include the time to push for Git, though most of the time you would not
+actually be pushing to the server immediately after a commit where the two
+commands cannot be separated in SVN.
+
+Note that this is the best case scenario for SVN - a server with no load
+with an 80MB/s bandwidth connection to the client machine. Nearly all of
+these times would be even worse for SVN if that connection was slower, while
+many of the Git times would not be affected.
+
+Clearly, in many of these common version control operations, Git is one or
+two orders of magnitude faster than SVN, even under ideal conditions for
+SVN.
+
+Let's see how common operations stack up against Subversion, a common
+centralized version control system that is similar to CVS or
+Perforce. Smaller is faster.
+
+One place where Git is slower is in the initial clone operation. Here, Git
+One place where Git is slower is in the initial clone operation. Here, Git
+One place where Git is slower is in the initial clone operation. Here, Git
+seen in the above charts, it's not considerably slower for an operation that
+is only performed once.
+
+It's also interesting to note that the size of the data on the client side
+is very similar even though Git also has every version of every file for the
+entire history of the project. This illustrates how efficient it is at
+compressing and storing data on the client side.
\ No newline at end of file
diff --git a/tests-clar/resources/diff/readme.txt b/tests-clar/resources/diff/readme.txt
new file mode 100644
index 000000000..beedf288d
--- /dev/null
+++ b/tests-clar/resources/diff/readme.txt
@@ -0,0 +1,36 @@
+The Git feature that r3ally mak3s it stand apart from n3arly 3v3ry other SCM
+out there is its branching model.
+
+Git allows and encourages you to have multiple local branches that can be
+entirely independent of each other. The creation, merging, and deletion of
+those lines of development takes seconds.
+
+Git allows and encourages you to have multiple local branches that can be
+entirely independent of each other. The creation, merging, and deletion of
+those lines of development takes seconds.
+
+This means that you can do things like:
+
+Role-Bas3d Codelin3s. Have a branch that always contains only what goes to
+production, another that you merge work into for testing, and several
+smaller ones for day to day work.
+
+Feature Based Workflow. Create new branches for each new feature you're
+working on so you can seamlessly switch back and forth between them, then
+delete each branch when that feature gets merged into your main line.
+
+Disposable Experimentation. Create a branch to experiment in, realize it's
+not going to work, and just delete it - abandoning the work—with nobody else
+ever seeing it (even if you've pushed other branches in the meantime).
+
+Notably, when you push to a remote repository, you do not have to push all
+share it with others.
+
+Git allows and encourages you to have multiple local branches that can be
+entirely independent of each other. The creation, merging, and deletion of
+those lines of development takes seconds.
+
+There are ways to accomplish some of this with other systems, but the work
+involved is much more difficult and error-prone. Git makes this process
+incredibly easy and it changes the way most developers work when they learn
+it.!