index : cgit

/*
 * The repository index, which is the page a cgit site opens on, and the site
 * readme the about page falls back to when a request names no repository.
 * Every repository cgitrc registered is a candidate row, cut down to those
 * matching the search terms and the url prefix the request carried, ordered
 * either by the column the reader asked for or by section, and split into
 * pages of max-repo-count rows.
 */

#include "cgit.h"
#include "filter.h"
#include "html.h"
#include "shared.h"
#include "ui-repolist.h"
#include "ui-shared.h"

// A section heading spans the whole table, so this has to stay in step with
// the headings print_header_row emits. Name, description and idle are always
// there, owner and links are added when cgitrc asks for them.
#define BASE_COLUMNS 3

struct sort_column {
	const char *name;
	int (*cmp)(const void *a, const void *b);
};

static time_t read_agefile(const char *path)
{
	time_t result;
	char *buf = NULL;
	struct strbuf date = STRBUF_INIT;

	if (cgit_read_first_line(path, &buf)) {
		free(buf);
		return 0;
	}

	// parse_date writes the timestamp followed by its timezone offset, so
	// only the seconds at the front are wanted here.
	if (parse_date(buf, &date) == 0)
		result = strtoul(date.buf, NULL, 10);
	else
		result = 0;
	free(buf);
	strbuf_release(&date);
	return result;
}

/*
 * The returned name can point into head, which the caller has to keep alive
 * for as long as it uses it.
 */
static const char *tip_branch(const struct cgit_repo *repo, struct strbuf *head)
{
	struct strbuf path = STRBUF_INIT;
	const char *branch = repo->defbranch;

	// A scanned repository sets its own defbranch, which is stat'ed below
	// the repository, so it is held to what a ref name may hold.
	if (branch)
		return check_refname_format(branch, REFNAME_ALLOW_ONELEVEL) ? "master" : branch;

	strbuf_addf(&path, "%s/HEAD", repo->path);
	if (strbuf_read_file(head, path.buf, 0) > 0) {
		strbuf_rtrim(head);
		if (!skip_prefix(head->buf, "ref: refs/heads/", &branch))
			branch = NULL;
		// HEAD belongs to the repository, so a crafted target such as
		// "ref: refs/heads/../../.." must not let the caller's stat
		// walk outside it. Git forbids ".." in a ref name anyway.
		if (branch && strstr(branch, ".."))
			branch = NULL;
	}
	strbuf_release(&path);

	return branch ? branch : "master";
}

static time_t get_repo_modtime(const struct cgit_repo *repo)
{
	struct strbuf path = STRBUF_INIT;
	struct strbuf head = STRBUF_INIT;
	struct stat st;
	const char *branch;
	// The comparators below are handed const repositories, but the answer
	// is kept in the repository itself so one page stats it only once.
	struct cgit_repo *writable = (struct cgit_repo *)repo;

	if (repo->mtime != -1)
		return repo->mtime;
	// A repository configured without a path has nothing on disk to date.
	if (!repo->path) {
		writable->mtime = 0;
		return 0;
	}
	strbuf_addf(&path, "%s/%s", repo->path, ctx.cfg.agefile);
	writable->mtime = read_agefile(path.buf);
	if (repo->mtime)
		goto end;

	branch = tip_branch(repo, &head);
	strbuf_reset(&path);
	strbuf_addf(&path, "%s/refs/heads/%s", repo->path, branch);
	if (stat(path.buf, &st) == 0) {
		writable->mtime = st.st_mtime;
		goto end;
	}

	strbuf_reset(&path);
	strbuf_addf(&path, "%s/packed-refs", repo->path);
	if (stat(path.buf, &st) == 0) {
		writable->mtime = st.st_mtime;
		goto end;
	}

	// A reftable repository has neither file, but rewrites this list on
	// every ref update.
	strbuf_reset(&path);
	strbuf_addf(&path, "%s/reftable/tables.list", repo->path);
	if (stat(path.buf, &st) == 0)
		writable->mtime = st.st_mtime;
end:
	strbuf_release(&path);
	strbuf_release(&head);
	return repo->mtime;
}

static int matches_search(struct cgit_repo *repo)
{
	if (!ctx.qry.search)
		return 1;
	if (repo->url && strcasestr(repo->url, ctx.qry.search))
		return 1;
	if (repo->name && strcasestr(repo->name, ctx.qry.search))
		return 1;
	if (repo->desc && strcasestr(repo->desc, ctx.qry.search))
		return 1;
	if (repo->owner && strcasestr(repo->owner, ctx.qry.search))
		return 1;
	return 0;
}

static int matches_url(struct cgit_repo *repo)
{
	if (!ctx.qry.url)
		return 1;
	if (repo->url && starts_with(repo->url, ctx.qry.url))
		return 1;
	return 0;
}

static int is_visible(struct cgit_repo *repo)
{
	if (repo->hide || repo->ignore)
		return 0;
	if (!(matches_search(repo) && matches_url(repo)))
		return 0;
	return 1;
}

static int any_repos_visible(void)
{
	int i;

	for (i = 0; i < cgit_repolist.count; i++) {
		if (is_visible(&cgit_repolist.repos[i]))
			return 1;
	}
	return 0;
}

/*
 * Whether any listed repository lies under the url the request carried,
 * whatever the search terms.
 */
static int any_repos_under_url(void)
{
	int i;

	for (i = 0; i < cgit_repolist.count; i++) {
		struct cgit_repo *repo = &cgit_repolist.repos[i];

		if (!repo->hide && !repo->ignore && matches_url(repo))
			return 1;
	}
	return 0;
}

/*
 * currenturl is passed in because it is the same for every heading and every
 * row, and working it out here would mean an allocation and a free per cell.
 */
static void print_column_header(const char *title, const char *column, const char *currenturl)
{
	htmlf("<th class='left col-%s' scope='col'", column);
	// Idle puts the latest change first, which is the shortest idle time,
	// so every column the index sorts by runs ascending.
	if (ctx.qry.sort && !strcmp(ctx.qry.sort, column))
		html(" aria-sort='ascending'");
	html("><a href='");
	html_attr(currenturl);
	htmlf("?s=%s", column);
	if (ctx.qry.search) {
		html("&amp;q=");
		html_url_arg(ctx.qry.search);
	}
	htmlf("'>%s</a></th>", title);
}

static void print_header_row(const char *currenturl)
{
	html("<tbody class='repos'>\n<tr class='nohover'>");
	print_column_header("Name", "name", currenturl);
	print_column_header("Description", "desc", currenturl);
	if (ctx.cfg.enable_index_owner)
		print_column_header("Owner", "owner", currenturl);
	print_column_header("Idle", "idle", currenturl);
	if (ctx.cfg.enable_index_links)
		html("<th class='left col-links' scope='col'>Links</th>");
	html("</tr>\n");
}

static int section_changed(const char *section, const char *last)
{
	if (!section && !last)
		return 0;
	if (!section || !last)
		return 1;
	return strcmp(section, last);
}

static void print_section_row(const char *section, int columns)
{
	htmlf("<tr class='nohover-highlight'><td colspan='%d' class='reposection'>", columns);
	html_txt(section);
	html("</td></tr>\n");
}

static void print_repo_row(const char *currenturl, int sublevel)
{
	char *repourl;

	htmlf("<tr class='repo'><td class='col-name %s'>",
		sublevel ? "sublevel-repo" : "toplevel-repo");
	cgit_summary_link(ctx.repo->name, NULL, NULL, NULL);
	html("</td><td class='col-desc'>");
	repourl = cgit_repourl(ctx.repo->url);
	html("<a href='");
	html_attr(repourl);
	html("'>");
	free(repourl);
	if (html_ntxt(ctx.repo->desc, ctx.cfg.max_repodesc_len) < 0)
		html("...");
	html("</a>");
	html("</td>");
	if (ctx.cfg.enable_index_owner) {
		html("<td class='col-owner'>");
		// A repo without an owner gets an empty cell, since a link
		// with no text cannot be clicked or read out.
		if (ctx.repo->owner && *ctx.repo->owner) {
			html("<a href='");
			html_attr(currenturl);
			html("?q=");
			html_url_arg(ctx.repo->owner);
			html("'>");
			html_txt(ctx.repo->owner);
			html("</a>");
		}
		html("</td>");
	}
	html("<td class='col-idle'>");
	cgit_print_age(get_repo_modtime(ctx.repo), 0, -1);
	html("</td>");
	if (ctx.cfg.enable_index_links) {
		html("<td class='col-links'>");
		cgit_summary_link("summary", NULL, "button", NULL);
		cgit_log_link("log", NULL, "button", NULL, NULL, NULL, 0, NULL, NULL, ctx.qry.showmsg, 0);
		cgit_tree_link("tree", NULL, "button", NULL, NULL, NULL);
		html("</td>");
	}
	html("</tr>\n");
}

static void print_pager(int total, int pagelen, char *search, char *sort)
{
	int i, ofs;
	const char *class;

	html("<nav class='pager' aria-label='Pager'><ul>\n");
	for (i = 0, ofs = 0; ofs < total; i++, ofs = i * pagelen) {
		class = (ctx.qry.ofs == ofs) ? "current" : NULL;
		html("<li>");
		cgit_index_link(cgit_fmt("[%d]", i + 1), cgit_fmt("Page %d", i + 1), class, search, sort,
			ofs, 0);
		html("</li>");
	}
	html("</ul></nav>\n");
}

static int cmp_str(const char *s1, const char *s2)
{
	if (s1 && s2) {
		if (ctx.cfg.case_sensitive_sort)
			return strcmp(s1, s2);
		else
			return strcasecmp(s1, s2);
	}
	if (s1 && !s2)
		return -1;
	if (s2 && !s1)
		return 1;
	return 0;
}

static int cmp_name(const void *a, const void *b)
{
	const struct cgit_repo *r1 = a;
	const struct cgit_repo *r2 = b;

	return cmp_str(r1->name, r2->name);
}

static int cmp_desc(const void *a, const void *b)
{
	const struct cgit_repo *r1 = a;
	const struct cgit_repo *r2 = b;

	return cmp_str(r1->desc, r2->desc);
}

static int cmp_owner(const void *a, const void *b)
{
	const struct cgit_repo *r1 = a;
	const struct cgit_repo *r2 = b;

	return cmp_str(r1->owner, r2->owner);
}

static int cmp_idle(const void *a, const void *b)
{
	const struct cgit_repo *r1 = a;
	const struct cgit_repo *r2 = b;
	time_t t1 = get_repo_modtime(r1), t2 = get_repo_modtime(r2);

	// Only the sign is returned, because a 64-bit difference truncated
	// into an int could come back with the wrong sign and leave the
	// ordering inconsistent.
	if (t2 > t1)
		return 1;
	if (t2 < t1)
		return -1;
	return 0;
}

static int cmp_section(const void *a, const void *b)
{
	const struct cgit_repo *r1 = a;
	const struct cgit_repo *r2 = b;
	int result;

	result = cmp_str(r1->section, r2->section);
	if (!result) {
		if (!strcmp(ctx.cfg.repository_sort, "age"))
			result = cmp_idle(r1, r2);
		if (!result)
			result = cmp_str(r1->name, r2->name);
	}
	return result;
}

/*
 * get_repo_modtime caches into the repository it is handed, but qsort moves
 * those structs as it works, so the cache is filled before sorting.
 */
static void resolve_modtimes(void)
{
	int i;

	for (i = 0; i < cgit_repolist.count; i++)
		get_repo_modtime(&cgit_repolist.repos[i]);
}

static const struct sort_column sort_columns[] = {
	{"section", cmp_section},
	{"name", cmp_name},
	{"desc", cmp_desc},
	{"owner", cmp_owner},
	{"idle", cmp_idle},
	{NULL, NULL}
};

static int sort_repolist(const char *field)
{
	const struct sort_column *column;

	for (column = &sort_columns[0]; column->name; column++) {
		if (strcmp(field, column->name))
			continue;
		if (column->cmp == cmp_idle || column->cmp == cmp_section)
			resolve_modtimes();
		QSORT(cgit_repolist.repos, cgit_repolist.count, column->cmp);
		return 1;
	}
	return 0;
}

void cgit_print_repolist(void)
{
	const char *last_section = NULL;
	char *currenturl;
	int columns = BASE_COLUMNS;
	int column_sorted = 0;
	int hits = 0;
	int shown = 0;
	int i;

	// Ahead of the search and the sort, which both look at descriptions.
	for (i = 0; i < cgit_repolist.count; i++)
		if (!cgit_repolist.repos[i].hide && !cgit_repolist.repos[i].ignore)
			cgit_read_repo_desc(&cgit_repolist.repos[i]);

	if (!any_repos_visible()) {
		// A url naming no repository or page lands here, as a prefix
		// that no repository's url starts with.
		if (ctx.qry.url && *ctx.qry.url && !any_repos_under_url())
			cgit_print_error_page(404, "Not Found", "Nothing matches this url");
		else
			cgit_print_error_page(404, "Not Found", "No repositories found");
		return;
	}

	if (ctx.cfg.enable_index_links)
		++columns;
	if (ctx.cfg.enable_index_owner)
		++columns;

	ctx.page.title = ctx.cfg.root_title;
	cgit_print_layout_start();

	if (ctx.qry.sort)
		column_sorted = sort_repolist(ctx.qry.sort);
	else if (ctx.cfg.section_sort)
		sort_repolist("section");

	// An offset of -1 gets through for the stats page, where it means
	// every author, and would start this page a row before the first.
	if (ctx.qry.ofs < 0)
		ctx.qry.ofs = 0;

	currenturl = cgit_currenturl();
	html("<table class='list repolist' aria-label='Repositories'>\n");
	for (i = 0; i < cgit_repolist.count; i++) {
		const char *section;

		ctx.repo = &cgit_repolist.repos[i];
		if (!is_visible(ctx.repo))
			continue;
		hits++;
		// Rows outside the page are stepped over rather than broken
		// out of, because the pager below is sized from the total.
		if (hits <= ctx.qry.ofs)
			continue;
		// Written as a subtraction because max-repo-count of zero
		// means unlimited and is held as INT_MAX, which any positive
		// offset would overflow if it were added to instead.
		if (hits - ctx.qry.ofs > ctx.cfg.max_repo_count)
			continue;
		if (!shown++)
			print_header_row(currenturl);
		section = ctx.repo->section;
		if (section && !strcmp(section, ""))
			section = NULL;
		if (!column_sorted && section_changed(section, last_section)) {
			print_section_row(section, columns);
			last_section = section;
		}
		print_repo_row(currenturl, !column_sorted && section);
	}
	if (shown)
		html("</tbody>\n");
	html("</table>\n");
	if (hits > ctx.cfg.max_repo_count)
		print_pager(hits, ctx.cfg.max_repo_count, ctx.qry.search, ctx.qry.sort);
	free(currenturl);
	cgit_print_layout_end();
}

void cgit_print_site_readme(void)
{
	cgit_print_layout_start();
	if (!ctx.cfg.root_readme)
		goto done;
	cgit_open_filter(ctx.cfg.about_filter, ctx.cfg.root_readme, "", "");
	html_include(ctx.cfg.root_readme);
	cgit_close_filter(ctx.cfg.about_filter);
done:
	cgit_print_layout_end();
}