index : cgit

/*
 * The tree page, which renders one level of a repository at a revision as a
 * table of the folders and files in it, or the contents of the file when the
 * path names one. A file is shown as numbered source, passed through the
 * repository's source filter when it has one, or as a hex dump when its bytes
 * look binary. A folder whose only entry is another folder is followed, and
 * both are named in the same row, so a long chain of single folders does not
 * cost a page each.
 */

#define USE_THE_REPOSITORY_VARIABLE

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

#define HEXDUMP_ROW_BYTES 32
#define HEXDUMP_ROW_HALF (HEXDUMP_ROW_BYTES / 2)
#define HEXDUMP_GAP_WIDTH 4

enum walk_state {
	WALK_LOOKING,
	WALK_LISTING,
	WALK_BLOB_SHOWN,
	WALK_ERROR_SHOWN,
};

struct ls_entry {
	struct object_id oid;
	char *name;
	unsigned mode;
};

struct walk_tree_context {
	char *rev;
	char *match_path;
	enum walk_state state;
	struct ls_entry *entries;
	size_t entries_nr, entries_alloc;
};

/*
 * The count runs past one as soon as a second entry or a file turns up, which
 * ends the descent.
 */
struct only_child {
	struct strbuf *path;
	struct object_id oid;
	char *name;
	size_t count;
};

/*
 * A formatted write per line meant a syscall and a temporary buffer for every
 * line of the file, so the anchors are handed over in batches. Building the
 * column whole was rejected because it would come to several times the size of
 * the blob.
 */
static void print_linenumbers(const char *buf, unsigned long size)
{
	const char *numberfmt = "<a id='n%1$d' href='#n%1$d'>%1$d</a>\n";
	struct strbuf numbers = STRBUF_INIT;
	unsigned long lineno = 0, idx = 0;

	if (size) {
		strbuf_addf(&numbers, numberfmt, ++lineno);

		// The newline that ends the last line must not open a line of
		// its own, so the final byte is left out of the scan.
		while (idx < size - 1) {
			if (buf[idx] == '\n') {
				strbuf_addf(&numbers, numberfmt, ++lineno);
				if (numbers.len >= HTML_BATCH) {
					html_raw(numbers.buf, numbers.len);
					strbuf_reset(&numbers);
				}
			}
			idx++;
		}
		html_raw(numbers.buf, numbers.len);
	}
	strbuf_release(&numbers);
}

static void print_text_buffer(const char *filename, char *buf,
			      unsigned long size)
{
	html("<table class='blob'>\n");

	if (ctx.cfg.enable_tree_linenumbers) {
		html("<tr><td class='linenumbers'><pre>");
		print_linenumbers(buf, size);
		html("</pre></td>\n");
	} else {
		html("<tr>\n");
	}

	if (ctx.repo->source_filter) {
		char *filter_arg = xstrdup(filename);

		html("<td class='lines'><pre><code>");
		cgit_open_filter(ctx.repo->source_filter, filter_arg);
		html_raw(buf, size);
		cgit_close_filter(ctx.repo->source_filter);
		free(filter_arg);
		html("</code></pre></td></tr></table>\n");
		return;
	}

	// Syntax highlighting ships as one of the filters under
	// custom/extensions, which keeps knowledge of languages out of cgit.
	html("<td class='lines'><pre><code>");
	html_txt(buf);
	html("</code></pre></td></tr></table>\n");
}

/*
 * git's ctype macros are locale free, but there is no isgraph among them, so
 * the dump works one out for itself.
 */
static int is_graphic(unsigned char ch)
{
	return isprint(ch) && !isspace(ch);
}

static void print_binary_buffer(char *buf, unsigned long size)
{
	unsigned long offset, idx;
	char ascii[HEXDUMP_ROW_BYTES + 1];
	struct strbuf row = STRBUF_INIT;

	html("<table class='bin-blob'>\n");
	html("<tr><th>ofs</th><th>hex dump</th><th>ascii</th></tr>\n");

	// At the default blob size limit a write per byte spent almost all of
	// its time in the kernel, so a row goes out in one write.
	for (offset = 0; offset < size;
	     offset += HEXDUMP_ROW_BYTES, buf += HEXDUMP_ROW_BYTES) {
		strbuf_reset(&row);
		strbuf_addf(&row, "<tr><td class='right'>%04lx</td><td class='hex'>", offset);
		for (idx = 0; idx < HEXDUMP_ROW_BYTES && offset + idx < size; idx++) {
			int gap = idx == HEXDUMP_ROW_HALF ? HEXDUMP_GAP_WIDTH : 1;

			strbuf_addf(&row, "%*s%02x", gap, "", buf[idx] & 0xff);
		}
		strbuf_addstr(&row, " </td><td class='hex'>");
		html_raw(row.buf, row.len);

		for (idx = 0; idx < HEXDUMP_ROW_BYTES && offset + idx < size; idx++)
			ascii[idx] = is_graphic(buf[idx]) ? buf[idx] : '.';
		ascii[idx] = '\0';
		html_txt(ascii);
		html("</td></tr>\n");
	}
	strbuf_release(&row);
	html("</table>\n");
}

/*
 * Answer whether the page layout was left open for the caller to close. A
 * false means a complete error page went out in place of the blob, so nothing
 * may be added to it.
 */
static bool print_object(const struct object_id *oid, const char *path,
			 const char *filename, const char *rev)
{
	enum object_type type;
	char *buf;
	unsigned long size;
	bool is_binary;

	type = odb_read_object_info(the_repository->objects, oid, &size);
	if (type == OBJ_BAD) {
		cgit_print_error_page(404, "Not found",
			"Bad object name: %s", oid_to_hex(oid));
		return false;
	}

	if (ctx.cfg.max_blob_size && size / 1024 > (unsigned long)ctx.cfg.max_blob_size) {
		cgit_print_error_page(413, "Too large",
			"blob size (%luKB) exceeds display size limit (%dKB)",
			size / 1024, ctx.cfg.max_blob_size);
		return false;
	}

	buf = odb_read_object(the_repository->objects, oid, &type, &size);
	if (!buf) {
		cgit_print_error_page(500, "Internal server error",
			"Error reading object %s", oid_to_hex(oid));
		return false;
	}
	// buffer_is_binary only sniffs the front of the blob, and a NUL past
	// that window would end the escaped text render early while the line
	// number column still counts the whole file, so check the rest too.
	is_binary = buffer_is_binary(buf, size) || memchr(buf, 0, size);

	cgit_set_title_from_path(path);

	cgit_print_layout_start();
	htmlf("blob: %s (", oid_to_hex(oid));
	cgit_plain_link("plain", NULL, NULL, ctx.qry.head,
		        rev, path);
	if (ctx.repo->enable_blame && !is_binary) {
		html(") (");
		cgit_blame_link("blame", NULL, NULL, ctx.qry.head,
			        rev, path);
	}
	html(")\n");

	if (is_binary)
		print_binary_buffer(buf, size);
	else
		print_text_buffer(filename, buf, size);

	free(buf);
	return true;
}

static int only_child_cb(const struct object_id *oid, struct strbuf *base,
			 const char *pathname, unsigned mode, void *data)
{
	struct only_child *child = data;

	if (++child->count > 1)
		return -1;

	if (!S_ISDIR(mode)) {
		child->count = 2;
		return -1;
	}

	child->name = xstrdup(pathname);
	oidcpy(&child->oid, oid);
	strbuf_addf(child->path, "/%s", pathname);
	return 0;
}

static void print_dir_chain(const struct object_id *oid, char *name,
			    char *rev, struct strbuf *fullpath)
{
	size_t initial_length = fullpath->len;
	struct tree *tree;
	struct only_child child = {
		.path = fullpath,
		.count = 1,
	};
	struct pathspec paths = {
		.nr = 0
	};

	oidcpy(&child.oid, oid);

	while (child.count == 1) {
		cgit_tree_link(name, NULL, "ls-dir", ctx.qry.head, rev,
			       fullpath->buf);

		tree = lookup_tree(the_repository, &child.oid);
		if (!tree)
			return;

		free(child.name);
		child.name = NULL;
		child.count = 0;

		read_tree(the_repository, tree, &paths, only_child_cb, &child);

		if (child.count != 1)
			break;

		html(" / ");
		name = child.name;
	}

	strbuf_setlen(fullpath, initial_length);
}

static void print_ls_row(const struct object_id *oid, const char *pathname,
			 unsigned mode, struct walk_tree_context *walk)
{
	char *name;
	struct strbuf fullpath = STRBUF_INIT;
	struct strbuf linkpath = STRBUF_INIT;
	struct strbuf class = STRBUF_INIT;
	enum object_type type;
	unsigned long size = 0;
	char *buf;

	name = xstrdup(pathname);
	strbuf_addf(&fullpath, "%s%s%s", ctx.qry.path ? ctx.qry.path : "",
		    ctx.qry.path ? "/" : "", name);

	if (!S_ISGITLINK(mode)) {
		type = odb_read_object_info(the_repository->objects, oid, &size);
		if (type == OBJ_BAD) {
			// The name comes from the tree, so it can hold
			// anything a commit was allowed to record.
			html("<tr><td colspan='4'>Bad object: ");
			html_txt(name);
			htmlf(" %s</td></tr>\n", oid_to_hex(oid));
			goto cleanup;
		}
	}

	html("<tr><td class='ls-mode'>");
	cgit_print_filemode(mode);
	html("</td><td>");
	if (S_ISGITLINK(mode)) {
		cgit_submodule_link("ls-mod", fullpath.buf, oid_to_hex(oid));
	} else if (S_ISDIR(mode)) {
		print_dir_chain(oid, name, walk->rev, &fullpath);
	} else {
		char *ext = strrchr(name, '.');
		strbuf_addstr(&class, "ls-blob");
		if (ext)
			strbuf_addf(&class, " %s", ext + 1);
		cgit_tree_link(name, NULL, class.buf, ctx.qry.head,
			       walk->rev, fullpath.buf);
	}
	if (S_ISLNK(mode)) {
		html(" -> ");
		buf = odb_read_object(the_repository->objects, oid, &type, &size);
		if (!buf) {
			htmlf("Error reading object: %s", oid_to_hex(oid));
			goto cleanup;
		}
		strbuf_addbuf(&linkpath, &fullpath);
		strbuf_addf(&linkpath, "/../%s", buf);
		strbuf_normalize_path(&linkpath);
		cgit_tree_link(buf, NULL, class.buf, ctx.qry.head,
			walk->rev, linkpath.buf);
		free(buf);
		strbuf_release(&linkpath);
	}
	if (S_ISGITLINK(mode))
		html("</td><td class='ls-size'>-</td>");
	else
		htmlf("</td><td class='ls-size'>%lu</td>", size);

	html("<td class='ls-links'>");
	cgit_log_link("log", NULL, "button", ctx.qry.head,
		      walk->rev, fullpath.buf, 0, NULL, NULL,
		      ctx.qry.showmsg, 0);
	if (ctx.repo->enable_stats)
		cgit_stats_link("stats", NULL, "button", ctx.qry.head,
				fullpath.buf);
	if (!S_ISGITLINK(mode))
		cgit_plain_link("plain", NULL, "button", ctx.qry.head,
				walk->rev, fullpath.buf);
	if (!S_ISDIR(mode) && !S_ISGITLINK(mode) && ctx.repo->enable_blame)
		cgit_blame_link("blame", NULL, "button", ctx.qry.head,
				walk->rev, fullpath.buf);
	html("</td></tr>\n");

cleanup:
	free(name);
	strbuf_release(&fullpath);
	strbuf_release(&class);
}

static int ls_item(const struct object_id *oid, struct strbuf *base,
		const char *pathname, unsigned mode, void *data)
{
	struct walk_tree_context *walk = data;

	// With folders grouped first, a row cannot go out as it arrives,
	// because git hands the level over in name order.
	if (ctx.cfg.enable_tree_group_dirs) {
		struct ls_entry *entry;

		ALLOC_GROW(walk->entries, walk->entries_nr + 1,
			   walk->entries_alloc);
		entry = &walk->entries[walk->entries_nr++];
		oidcpy(&entry->oid, oid);
		entry->name = xstrdup(pathname);
		entry->mode = mode;
		return 0;
	}

	print_ls_row(oid, pathname, mode, walk);
	return 0;
}

static void ls_flush(struct walk_tree_context *walk)
{
	size_t i;

	for (i = 0; i < walk->entries_nr; i++) {
		struct ls_entry *entry = &walk->entries[i];

		if (S_ISDIR(entry->mode))
			print_ls_row(&entry->oid, entry->name, entry->mode, walk);
	}
	for (i = 0; i < walk->entries_nr; i++) {
		struct ls_entry *entry = &walk->entries[i];

		if (!S_ISDIR(entry->mode))
			print_ls_row(&entry->oid, entry->name, entry->mode, walk);
	}

	for (i = 0; i < walk->entries_nr; i++)
		free(walk->entries[i].name);
	free(walk->entries);
	walk->entries = NULL;
	walk->entries_nr = walk->entries_alloc = 0;
}

static void ls_head(void)
{
	cgit_print_layout_start();
	html("<table class='list'>\n");
	html("<tr class='nohover'>");
	html("<th class='left ls-mode'>Mode</th>");
	html("<th class='left'>Name</th>");
	html("<th class='right'>Size</th>");
	html("<th class='ls-links'></th>");
	html("</tr>\n");
}

static void ls_tail(void)
{
	html("</table>\n");
	cgit_print_layout_end();
}

static void ls_tree(const struct object_id *oid, const char *path,
		    struct walk_tree_context *walk)
{
	struct tree *tree;
	struct pathspec paths = {
		.nr = 0
	};

	tree = parse_tree_indirect(oid);
	if (!tree) {
		cgit_print_error_page(404, "Not found",
			"Not a tree object: %s", oid_to_hex(oid));
		return;
	}

	ls_head();
	read_tree(the_repository, tree, &paths, ls_item, walk);
	ls_flush(walk);
	ls_tail();
}

static int walk_tree(const struct object_id *oid, struct strbuf *base,
		const char *pathname, unsigned mode, void *data)
{
	struct walk_tree_context *walk = data;

	if (walk->state == WALK_LOOKING) {
		struct strbuf fullpath = STRBUF_INIT;

		strbuf_addbuf(&fullpath, base);
		strbuf_addstr(&fullpath, pathname);
		if (strcmp(walk->match_path, fullpath.buf))
			return READ_TREE_RECURSIVE;

		if (S_ISDIR(mode)) {
			walk->state = WALK_LISTING;
			cgit_set_title_from_path(fullpath.buf);
			strbuf_release(&fullpath);
			ls_head();
			return READ_TREE_RECURSIVE;
		} else {
			bool shown = print_object(oid, fullpath.buf,
						  pathname, walk->rev);

			walk->state = shown ? WALK_BLOB_SHOWN : WALK_ERROR_SHOWN;
			strbuf_release(&fullpath);
			return 0;
		}
	}
	ls_item(oid, base, pathname, mode, walk);
	return 0;
}

// Either argument may be null, in which case the head of the current query
// stands in for the revision and the listing starts at the root of the tree.
void cgit_print_tree(const char *rev, char *path)
{
	struct object_id oid;
	struct commit *commit;
	int path_len = path ? strlen(path) : 0;
	// nowildcard_len matches len so git treats the path as literal rather
	// than as a glob.
	struct pathspec_item path_items = {
		.match = path,
		.len = path_len,
		.nowildcard_len = path_len
	};
	struct pathspec paths = {
		.nr = path ? 1 : 0,
		.items = &path_items
	};
	struct walk_tree_context walk = {
		.match_path = path,
		.state = WALK_LOOKING
	};

	if (!rev)
		rev = ctx.qry.head;

	if (repo_get_oid(the_repository, rev, &oid)) {
		cgit_print_error_page(404, "Not found",
			"Invalid revision name: %s", rev);
		return;
	}
	commit = lookup_commit_reference(the_repository, &oid);
	if (!commit || repo_parse_commit(the_repository, commit)) {
		cgit_print_error_page(404, "Not found",
			"Invalid commit reference: %s", rev);
		return;
	}

	walk.rev = xstrdup(rev);

	if (path == NULL) {
		ls_tree(get_commit_tree_oid(commit), NULL, &walk);
		goto cleanup;
	}

	read_tree(the_repository, repo_get_commit_tree(the_repository, commit),
		  &paths, walk_tree, &walk);
	if (walk.state == WALK_LISTING) {
		ls_flush(&walk);
		ls_tail();
	} else if (walk.state == WALK_BLOB_SHOWN)
		cgit_print_layout_end();
	else if (walk.state == WALK_LOOKING)
		cgit_print_error_page(404, "Not found", "Path not found");
	// WALK_ERROR_SHOWN is left alone, since the error page print_object
	// put out is already complete.

cleanup:
	free(walk.rev);
}