index : cgit

/*
 * The diff page, which shows what one revision changed against another as a
 * summary table of the files it touched followed by the body of each change,
 * either unified or side by side. Both halves come out of a single walk over
 * the tree, because the summary has to be printed first while the bodies are
 * only cheap to produce while each file is already open. A commit or a file
 * whose diff would be too large stops at a link to a narrower view, and the
 * raw form of the same request is handed to git and served as a plain patch.
 */

#define USE_THE_REPOSITORY_VARIABLE

#include "cgit.h"
#include "html.h"
#include "shared.h"
#include "ui-diff.h"
#include "ui-shared.h"
#include "ui-ssdiff.h"

// A file's body is collected while the walk still has that file open,
// since the stat table above it has to be printed first. This bounds what
// one request may hold that way. Past it the collected bodies are dropped
// and a second walk renders the page instead.
#define BODY_BUDGET (8 * 1024 * 1024)

struct fileinfo {
	char status;
	struct object_id old_oid[1];
	struct object_id new_oid[1];
	unsigned short old_mode;
	unsigned short new_mode;
	char *old_path;
	char *new_path;
	unsigned int added;
	unsigned int removed;
	size_t old_size;
	size_t new_size;
	unsigned int binary:1;
	struct strbuf body;
};

struct object_id old_rev_oid[1];
struct object_id new_rev_oid[1];

static struct fileinfo *items;
static int files, slots;

static int total_adds, total_rems, max_changes;
static int lines_added, lines_removed;

static size_t body_bytes;
static int bodies_usable;

static linediff_fn render_line_fn;
static int render_suppressed;

static struct diff_filepair *current_filepair;
static const char *current_prefix;

static int use_ssdiff;

// Caps apply only to a whole commit view, since a single file diff is where
// the capped views send the reader.
static int cap_diffs;

static int item_idx;

static void release_bodies(void)
{
	int i;

	for (i = 0; i < files; i++)
		strbuf_release(&items[i].body);
}

static void print_fileinfo(struct fileinfo *info)
{
	int scale = max_changes ? max_changes : 1;
	int add_span, rem_span;
	const char *class;

	switch (info->status) {
	case DIFF_STATUS_ADDED:
		class = "add";
		break;
	case DIFF_STATUS_COPIED:
		class = "cpy";
		break;
	case DIFF_STATUS_DELETED:
		class = "del";
		break;
	case DIFF_STATUS_MODIFIED:
		class = "upd";
		break;
	case DIFF_STATUS_RENAMED:
		class = "mov";
		break;
	case DIFF_STATUS_TYPE_CHANGED:
		class = "typ";
		break;
	case DIFF_STATUS_UNKNOWN:
		class = "unk";
		break;
	case DIFF_STATUS_UNMERGED:
		class = "stg";
		break;
	default:
		BUG("unhandled diff status %c", info->status);
	}

	html("<tr>");
	html("<td class='mode'>");
	if (is_null_oid(info->new_oid)) {
		cgit_print_filemode(info->old_mode);
	} else {
		cgit_print_filemode(info->new_mode);
	}

	if (
		info->old_mode != info->new_mode &&
		!is_null_oid(info->old_oid)      &&
		!is_null_oid(info->new_oid)
	) {
		html("<span class='modechange'>[");
		cgit_print_filemode(info->old_mode);
		html("]</span>");
	}
	htmlf("</td><td class='%s'>", class);
	cgit_diff_link(info->new_path, NULL, NULL, ctx.qry.head, ctx.qry.oid, ctx.qry.oid2,
		info->new_path);
	if (info->status == DIFF_STATUS_COPIED || info->status == DIFF_STATUS_RENAMED) {
		htmlf(" (%s from ", info->status == DIFF_STATUS_COPIED ? "copied" : "renamed");
		html_txt(info->old_path);
		html(")");
	}
	html("</td><td class='right'>");
	if (info->binary) {
		htmlf("bin</td><td class='graph'>%zu -> %zu bytes</td></tr>\n", info->old_size,
			info->new_size);
		return;
	}
	htmlf("%u", info->added + info->removed);
	html("</td><td class='graph'>");
	add_span = (int)(info->added * 1000.0 / scale + 0.5);
	rem_span = (int)(info->removed * 1000.0 / scale + 0.5);
	// The bar is drawn on a canvas 1000 units wide that the cell stretches,
	// so each segment takes its width from an attribute rather than from an
	// inline style a CSP would have to allow.
	html("<svg viewBox='0 0 1000 1' preserveAspectRatio='none' aria-hidden='true'>");
	if (add_span > 0)
		htmlf("<rect class='add' width='%d' height='1'/>", add_span);
	if (rem_span > 0)
		htmlf("<rect class='rem' x='%d' width='%d' height='1'/>", add_span, rem_span);
	html("</svg></td></tr>\n");
}

/*
 * Counting is only half of what this does. It also renders each line until
 * max-diff-lines is passed, so the two cannot be split into separate passes.
 */
static void count_diff_lines(char *line, int len)
{
	if (line && (len > 0)) {
		if (line[0] == '+')
			lines_added++;
		else if (line[0] == '-')
			lines_removed++;
	}
	if (!render_line_fn || render_suppressed)
		return;
	if (
		cap_diffs                  &&
		ctx.cfg.max_diff_lines > 0 &&
		lines_added + lines_removed > ctx.cfg.max_diff_lines
	) {
		render_suppressed = 1;
		return;
	}
	render_line_fn(line, len);
}

static int show_filepair(struct diff_filepair *pair)
{
	if (!current_prefix)
		return 1;
	return starts_with(pair->one->path, current_prefix) ||
		starts_with(pair->two->path, current_prefix);
}

/*
 * The line xdiff produces ends in its newline, which is swapped for a
 * terminator while the text is escaped and then put back, because the line
 * points into a buffer xdiff still owns.
 */
static void print_line(char *line, int len)
{
	const char *class = "ctx";
	char c = line[len - 1];

	if (line[0] == '+')
		class = "add";
	else if (line[0] == '-')
		class = "del";
	else if (line[0] == '@')
		class = "hunk";

	htmlf("<div class='%s'>", class);
	line[len - 1] = '\0';
	html_txt(line);
	html("</div>");
	line[len - 1] = c;
}

/*
 * The buffer repo_find_unique_abbrev returns is reused, so a caller printing
 * two abbreviated names together keeps a copy of each.
 */
static char *abbrev_oid(const struct object_id *oid)
{
	return xstrdup(repo_find_unique_abbrev(the_repository, oid, DEFAULT_ABBREV));
}

static void print_file_header(const struct object_id *old_oid, const char *old_path, unsigned old_mode,
	const struct object_id *new_oid, const char *new_path, unsigned new_mode)
{
	char *old_abbrev, *new_abbrev;
	int subproject;

	subproject = (S_ISGITLINK(old_mode) || S_ISGITLINK(new_mode));
	// The side by side view opens its file group before the header row,
	// so only the unified view opens one here.
	if (!use_ssdiff) {
		html("<div class='file' data-path='");
		html_attr(new_path);
		html("'>");
	}
	html("<div class='head'>");
	html("diff --git a/");
	html_txt(old_path);
	html(" b/");
	html_txt(new_path);

	if (old_mode == 0)
		htmlf("<br>new file mode %.6o", new_mode);

	if (new_mode == 0)
		htmlf("<br>deleted file mode %.6o", old_mode);

	if (!subproject) {
		old_abbrev = abbrev_oid(old_oid);
		new_abbrev = abbrev_oid(new_oid);
		htmlf("<br>index %s..%s", old_abbrev, new_abbrev);
		free(old_abbrev);
		free(new_abbrev);
		if (old_mode != 0 && new_mode != 0) {
			htmlf(" %.6o", old_mode);
			if (new_mode != old_mode)
				htmlf("..%.6o", new_mode);
		}
		if (is_null_oid(old_oid)) {
			old_path = "dev/null";
			html("<br>--- /");
		} else {
			html("<br>--- a/");
		}
		if (old_mode != 0)
			cgit_tree_link(old_path, NULL, NULL, ctx.qry.head, oid_to_hex(old_rev_oid), old_path);
		else
			html_txt(old_path);
		if (is_null_oid(new_oid)) {
			new_path = "dev/null";
			html("<br>+++ /");
		} else {
			html("<br>+++ b/");
		}
		if (new_mode != 0)
			cgit_tree_link(new_path, NULL, NULL, ctx.qry.head, oid_to_hex(new_rev_oid), new_path);
		else
			html_txt(new_path);
	}
	html("</div>");
}

/*
 * Closes the group print_file_header or cgit_ssdiff_header_begin opened. The
 * side by side footer closes its own, so the unified view is the only one with
 * work to do here.
 */
static void end_file(void)
{
	// The unified table keeps its whitespace, so no newline follows.
	if (use_ssdiff)
		cgit_ssdiff_footer();
	else
		html("</div>");
}

/*
 * The length passed counts the terminator, because a renderer strips the last
 * byte of every line it is given.
 */
static void print_subproject_lines(struct diff_filepair *pair, linediff_fn line_fn)
{
	if (S_ISGITLINK(pair->one->mode)) {
		char *text = cgit_fmt("-Subproject %s", oid_to_hex(&pair->one->oid));
		line_fn(text, strlen(text) + 1);
	}
	if (S_ISGITLINK(pair->two->mode)) {
		char *text = cgit_fmt("+Subproject %s", oid_to_hex(&pair->two->oid));
		line_fn(text, strlen(text) + 1);
	}
}

static void print_binary_differs(void)
{
	if (use_ssdiff)
		html("<tr><td colspan='4'>Binary files differ</td></tr>");
	else
		html("Binary files differ");
}

static void print_truncated(const char *path)
{
	if (use_ssdiff)
		html("<tr><td colspan='4' class='truncated'>");
	else
		html("<div class='truncated'>");
	html("This diff is too large to be rendered inline. ");
	cgit_diff_link("View it on its own page", NULL, NULL, ctx.qry.head, ctx.qry.oid, ctx.qry.oid2,
		path);
	html(".");
	if (use_ssdiff)
		html("</td></tr>");
	else
		html("</div>");
	end_file();
}

static struct fileinfo *reserve_item(void)
{
	files++;
	if (files >= slots) {
		if (slots == 0)
			slots = 4;
		else
			slots = slots * 2;
		items = xrealloc(items, slots * sizeof(struct fileinfo));
	}
	memset(&items[files - 1], 0, sizeof(items[files - 1]));
	strbuf_init(&items[files - 1].body, 0);
	return &items[files - 1];
}

/*
 * Renders one file's body into its own buffer, the way filepair_cb would have
 * written it straight out on a second walk.
 */
static void collect_body(struct diff_filepair *pair, struct fileinfo *item, int *binary,
	size_t *old_size, size_t *new_size)
{
	struct strbuf *body = &item->body;
	linediff_fn line_fn = use_ssdiff ? cgit_ssdiff_line_cb : print_line;
	size_t header_len;

	current_filepair = pair;

	html_capture_begin(body);
	if (use_ssdiff)
		cgit_ssdiff_header_begin(pair->two->path);
	print_file_header(&pair->one->oid, pair->one->path, pair->one->mode,
		&pair->two->oid, pair->two->path, pair->two->mode);
	if (use_ssdiff)
		cgit_ssdiff_header_end();
	header_len = body->len;

	// Cleared for the whole file rather than in the branch below, because
	// the submodule branch never runs count_diff_lines with a render
	// function. A submodule following a file that tripped the cap would
	// otherwise inherit the flag and report itself as too large to render.
	render_suppressed = 0;

	if (S_ISGITLINK(pair->one->mode) || S_ISGITLINK(pair->two->mode)) {
		// The stat counts what a diff of the pair produces rather
		// than the two lines the body shows, so run that diff for
		// the count alone.
		render_line_fn = NULL;
		cgit_diff_files(&pair->one->oid, &pair->two->oid, old_size, new_size,
			binary, 0, ctx.qry.ignorews, count_diff_lines);
		print_subproject_lines(pair, line_fn);
	} else {
		render_line_fn = line_fn;
		if (cgit_diff_files(&pair->one->oid, &pair->two->oid, old_size, new_size,
				binary, ctx.qry.context, ctx.qry.ignorews, count_diff_lines))
			cgit_print_error("Error running diff");
		render_line_fn = NULL;
		if (*binary)
			print_binary_differs();
	}
	end_file();

	if (render_suppressed) {
		// Setting the length back would keep the grown allocation,
		// which the budget below cannot see because it only counts
		// what is kept, so rebuild the buffer at the kept size.
		char *header_text = xmemdupz(body->buf, header_len);

		strbuf_release(body);
		strbuf_attach(body, header_text, header_len, header_len + 1);
		print_truncated(pair->two->path);
	}
	html_capture_end();

	body_bytes += body->len;
	if (body_bytes > BODY_BUDGET) {
		bodies_usable = 0;
		release_bodies();
	}
}

static void inspect_filepair(struct diff_filepair *pair)
{
	struct fileinfo *item;
	int binary = 0;
	size_t old_size = 0;
	size_t new_size = 0;

	if (!show_filepair(pair))
		return;

	item = reserve_item();
	lines_added = 0;
	lines_removed = 0;

	if (bodies_usable)
		collect_body(pair, item, &binary, &old_size, &new_size);
	else
		cgit_diff_files(&pair->one->oid, &pair->two->oid, &old_size, &new_size,
			&binary, 0, ctx.qry.ignorews, count_diff_lines);

	item->status = pair->status;
	oidcpy(item->old_oid, &pair->one->oid);
	oidcpy(item->new_oid, &pair->two->oid);
	item->old_mode = pair->one->mode;
	item->new_mode = pair->two->mode;
	item->old_path = xstrdup(pair->one->path);
	item->new_path = xstrdup(pair->two->path);
	item->added = lines_added;
	item->removed = lines_removed;
	item->old_size = old_size;
	item->new_size = new_size;
	item->binary = binary;
	if (lines_added + lines_removed > max_changes)
		max_changes = lines_added + lines_removed;
	total_adds += lines_added;
	total_rems += lines_removed;
}

static void print_diffstat(const struct object_id *old_oid, const struct object_id *new_oid,
	const char *prefix)
{
	int i;

	html("<div class='diffstat-header'>");
	cgit_diff_link("Diffstat", NULL, NULL, ctx.qry.head, ctx.qry.oid, ctx.qry.oid2, NULL);
	if (prefix) {
		html(" (limited to '");
		html_txt(prefix);
		html("')");
	}
	html("</div>\n");
	max_changes = 0;
	cgit_diff_tree(old_oid, new_oid, inspect_filepair, prefix, ctx.qry.ignorews);
	// A change to no files would leave the table an empty frame.
	if (files) {
		html("<table class='diffstat'>\n");
		for (i = 0; i < files; i++)
			print_fileinfo(&items[i]);
		html("</table>\n");
	}
	html("<div class='diffstat-summary'>");
	htmlf("%d file%s changed, %d insertion%s, %d deletion%s",
		files, files == 1 ? "" : "s",
		total_adds, total_adds == 1 ? "" : "s",
		total_rems, total_rems == 1 ? "" : "s");
	html("</div>\n");
}

static int over_line_cap(int idx)
{
	return cap_diffs && ctx.cfg.max_diff_lines > 0 &&
		idx < files && !items[idx].binary &&
		items[idx].added + items[idx].removed > (unsigned int)ctx.cfg.max_diff_lines;
}

static void filepair_cb(struct diff_filepair *pair)
{
	size_t old_size = 0;
	size_t new_size = 0;
	int binary = 0;
	int idx;
	linediff_fn line_fn = print_line;

	if (!show_filepair(pair))
		return;

	idx = item_idx++;
	current_filepair = pair;
	if (use_ssdiff) {
		cgit_ssdiff_header_begin(pair->two->path);
		line_fn = cgit_ssdiff_line_cb;
	}
	print_file_header(&pair->one->oid, pair->one->path, pair->one->mode,
		&pair->two->oid, pair->two->path, pair->two->mode);
	if (use_ssdiff)
		cgit_ssdiff_header_end();

	if (over_line_cap(idx)) {
		print_truncated(items[idx].new_path);
		return;
	}
	if (S_ISGITLINK(pair->one->mode) || S_ISGITLINK(pair->two->mode)) {
		print_subproject_lines(pair, line_fn);
		end_file();
		return;
	}
	if (cgit_diff_files(&pair->one->oid, &pair->two->oid, &old_size, &new_size,
			&binary, ctx.qry.context, ctx.qry.ignorews, line_fn))
		cgit_print_error("Error running diff");
	if (binary)
		print_binary_differs();
	end_file();
}

static void print_raw_patch(const struct object_id *old_tree_oid, const struct object_id *new_tree_oid,
	const char *prefix)
{
	struct diff_options diffopt;

	// As on the patch page, a blob past max-blob-size is shown as binary
	// rather than inflated.
	if (ctx.cfg.max_blob_size)
		repo_settings_set_big_file_threshold(the_repository, (unsigned long)ctx.cfg.max_blob_size * 1024);
	repo_diff_setup(the_repository, &diffopt);
	diffopt.output_format = DIFF_FORMAT_PATCH;
	diffopt.flags.recursive = 1;
	if (prefix)
		cgit_diff_limit_path(&diffopt, prefix);
	diff_setup_done(&diffopt);

	ctx.page.mimetype = "text/plain";
	cgit_print_http_headers();
	if (old_tree_oid)
		diff_tree_oid(old_tree_oid, new_tree_oid, "", &diffopt);
	else
		diff_root_tree_oid(new_tree_oid, "", &diffopt);
	diffcore_std(&diffopt);
	diff_flush(&diffopt);
}

struct diff_filespec *cgit_get_current_old_file(void)
{
	return current_filepair->one;
}

struct diff_filespec *cgit_get_current_new_file(void)
{
	return current_filepair->two;
}

void cgit_print_diff_ctrls(void)
{
	int i, selected;

	html("<div class='cgit-panel'>\n");
	html("<strong>Diff options</strong>");
	html("<form method='get' aria-label='Diff options'>");
	cgit_add_hidden_formfields(1, 0, ctx.qry.page);
	html("<div class='fields'>");
	html("<label for='diff-context'>context</label>");
	html("<select id='diff-context' name='context'>");
	selected = ctx.qry.context;
	if (!selected)
		selected = DEFAULT_DIFF_CONTEXT;
	for (i = 1; i <= 10; i++)
		html_intoption(i, cgit_fmt("%d", i), selected);
	for (i = 15; i <= MAX_DIFF_CONTEXT_LINES; i += 5)
		html_intoption(i, cgit_fmt("%d", i), selected);
	html("</select>");
	html("<label for='diff-space'>space</label>");
	html("<select id='diff-space' name='ignorews'>");
	html_intoption(0, "include", ctx.qry.ignorews);
	html_intoption(1, "ignore", ctx.qry.ignorews);
	html("</select>");
	html("<label for='diff-mode'>mode</label>");
	html("<select id='diff-mode' name='dt'>");
	selected = ctx.qry.has_difftype ? ctx.qry.difftype : ctx.cfg.difftype;
	html_intoption(0, "unified", selected);
	html_intoption(1, "ssdiff", selected);
	html_intoption(2, "stat only", selected);
	html("</select>");
	html("<input type='submit' value='reload'>");
	html("</div>");
	html("</form>\n");
	html("</div>\n");
}

/*
 * Whether the path names a directory in either tree. The diff below it can
 * then still span many files, so the caps stay on.
 */
static int prefix_is_dir(const struct object_id *old_tree, const struct object_id *new_tree,
	const char *prefix)
{
	struct object_id oid;
	unsigned short mode;
	char *path = xstrdup(prefix);
	size_t len = strlen(path);
	int is_dir = 0;

	if (len && path[len - 1] == '/')
		path[len - 1] = '\0';
	if (new_tree && !get_tree_entry(the_repository, new_tree, path, &oid, &mode))
		is_dir = S_ISDIR(mode);
	else if (old_tree && !get_tree_entry(the_repository, old_tree, path, &oid, &mode))
		is_dir = S_ISDIR(mode);
	free(path);
	return is_dir;
}

void cgit_print_diff(const char *new_rev, const char *old_rev, const char *prefix,
	int show_ctrls, int raw)
{
	struct commit *new_commit, *old_commit;
	const struct object_id *old_tree_oid, *new_tree_oid;
	const char *path = prefix;
	diff_type difftype;

	// Detecting renames needs the diff machinery to examine the whole
	// commit, so with follow set the prefix is applied in show_filepair
	// instead of being passed down.
	if (ctx.qry.follow && ctx.repo->enable_follow_links) {
		current_prefix = prefix;
		prefix = "";
	} else {
		current_prefix = NULL;
	}

	if (!new_rev)
		new_rev = ctx.qry.head;
	if (repo_get_oid(the_repository, new_rev, new_rev_oid)) {
		cgit_print_error_page(404, "Not Found", "Bad object id: %s", new_rev);
		return;
	}
	new_commit = lookup_commit_reference(the_repository, new_rev_oid);
	if (!new_commit || repo_parse_commit(the_repository, new_commit)) {
		cgit_print_error_page(404, "Not Found", "Not a commit: %s", oid_to_hex(new_rev_oid));
		return;
	}
	new_tree_oid = get_commit_tree_oid(new_commit);

	if (old_rev) {
		if (repo_get_oid(the_repository, old_rev, old_rev_oid)) {
			cgit_print_error_page(404, "Not Found", "Bad object id: %s", old_rev);
			return;
		}
	} else if (new_commit->parents && new_commit->parents->item) {
		oidcpy(old_rev_oid, &new_commit->parents->item->object.oid);
	} else {
		oidclr(old_rev_oid, the_repository->hash_algo);
	}

	if (!is_null_oid(old_rev_oid)) {
		old_commit = lookup_commit_reference(the_repository, old_rev_oid);
		if (!old_commit || repo_parse_commit(the_repository, old_commit)) {
			cgit_print_error_page(404, "Not Found", "Not a commit: %s", oid_to_hex(old_rev_oid));
			return;
		}
		old_tree_oid = get_commit_tree_oid(old_commit);
	} else {
		old_tree_oid = NULL;
	}

	// A path narrows the diff to one file, whose own page always renders
	// in full. Decided from the caller's path, since the follow logic
	// above has emptied the prefix.
	cap_diffs = !path || prefix_is_dir(old_tree_oid, new_tree_oid, path);

	if (raw) {
		print_raw_patch(old_tree_oid, new_tree_oid, path);
		return;
	}

	difftype = ctx.qry.has_difftype ? ctx.qry.difftype : ctx.cfg.difftype;
	use_ssdiff = difftype == DIFF_SSDIFF;
	// Without max-diff-lines to stop it a single body grows with the file
	// it came from, which for a side by side diff is several times the blob
	// itself, and the budget cannot help there because it is only reached
	// once a body is complete.
	bodies_usable = difftype != DIFF_STATONLY && cap_diffs && ctx.cfg.max_diff_lines > 0;

	if (show_ctrls) {
		cgit_print_layout_start();
		cgit_print_diff_ctrls();
	}

	// Every link from here on leads to a single file, and a stat limited
	// to one file is useless, so the difftype those links carry is reset
	// rather than propagating DIFF_STATONLY to them.
	if (difftype == DIFF_STATONLY)
		ctx.qry.difftype = ctx.cfg.difftype;

	print_diffstat(old_rev_oid, new_rev_oid, prefix);

	if (difftype == DIFF_STATONLY) {
		if (show_ctrls)
			cgit_print_layout_end();
		return;
	}

	if (cap_diffs && ctx.cfg.max_diff_files > 0 && files > ctx.cfg.max_diff_files) {
		html("<div class='truncated'>");
		html("This diff is too large to be rendered inline. Follow a file above, or the ");
		cgit_patch_link("patch", NULL, NULL, NULL, ctx.qry.oid, NULL);
		html(" link for the whole commit.");
		html("</div>");
		if (show_ctrls)
			cgit_print_layout_end();
		return;
	}

	html("<div class='diff-scroll'>\n");
	if (use_ssdiff) {
		html("<table class='ssdiff'>\n");
	} else {
		html("<table class='diff'>\n");
		html("<tr><td>");
	}
	if (bodies_usable) {
		int i;

		for (i = 0; i < files; i++)
			html_raw(items[i].body.buf, items[i].body.len);
		release_bodies();
	} else {
		item_idx = 0;
		cgit_diff_tree(old_rev_oid, new_rev_oid, filepair_cb, prefix, ctx.qry.ignorews);
	}
	if (!use_ssdiff)
		html("</td></tr>\n");
	html("</table>\n");
	html("</div>\n");

	if (show_ctrls)
		cgit_print_layout_end();
}