index : cgit

/* ui-tree.c: functions for tree output
 *
 * Copyright (C) 2006-2017 cgit Development Team <cgit@lists.zx2c4.com>
 *
 * Licensed under GNU General Public License v2
 *   (see LICENSE.txt for full license text)
 */

 #define USE_THE_REPOSITORY_VARIABLE

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

/* Bytes shown per row of the binary hex dump. */
#define HEXDUMP_ROW_BYTES 32

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

struct walk_tree_context {
	char *curr_rev;
	char *match_path;
	int state;
	struct tree_ls_entry *entries;
	size_t entries_nr, entries_alloc;
};

static void print_text_buffer(const char *name, char *buf, unsigned long size)
{
	unsigned long lineno, idx;
	const char *numberfmt = "<a id='n%1$d' href='#n%1$d'>%1$d</a>\n";

	html("<table summary='blob content' class='blob'>\n");

	if (ctx.cfg.enable_tree_linenumbers) {
		struct strbuf numbers = STRBUF_INIT;

		html("<tr><td class='linenumbers'><pre>");
		idx = 0;
		lineno = 0;

		// Build the column in batches. A formatted write per line meant
		// a syscall and a temporary buffer for every line of the file,
		// which is most of what rendering a large blob cost, and the
		// whole column at once would come to several times the blob.
		if (size) {
			strbuf_addf(&numbers, numberfmt, ++lineno);
			while (idx < size - 1) { // skip absolute last newline
				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);
		html("</pre></td>\n");
	}
	else {
		html("<tr>\n");
	}

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

	/* No source filter is configured, so serve the text plain. Syntax
	 * highlighting ships as an optional source filter in custom/extensions/,
	 * keeping language knowledge out of the core. */
	html("<td class='lines'><pre><code>");
	html_txt(buf);
	html("</code></pre></td></tr></table>\n");
}

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

	html("<table summary='blob content' class='bin-blob'>\n");
	html("<tr><th>ofs</th><th>hex dump</th><th>ascii</th></tr>");
	for (ofs = 0; ofs < size; ofs += HEXDUMP_ROW_BYTES, buf += HEXDUMP_ROW_BYTES) {
		// One write per row rather than one per byte. At the default
		// blob limit the per-byte form spent almost all of its time in
		// the kernel, which made a single request for a large binary
		// blob far more expensive than the page it produced.
		strbuf_reset(&row);
		strbuf_addf(&row, "<tr><td class='right'>%04lx</td><td class='hex'>", ofs);
		for (idx = 0; idx < HEXDUMP_ROW_BYTES && ofs + idx < size; idx++)
			strbuf_addf(&row, "%*s%02x",
				    idx == 16 ? 4 : 1, "",
				    buf[idx] & 0xff);
		strbuf_addstr(&row, " </td><td class='hex'>");
		html_raw(row.buf, row.len);

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

/* Returns 1 if it opened the page layout for the caller to close, or 0 if
 * it emitted a complete standalone error page. */
static int print_object(const struct object_id *oid, const char *path, const char *basename, 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 0;
	}

	/* Reject an oversized object before reading it whole into memory. */
	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 0;
	}

	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 0;
	}
	is_binary = buffer_is_binary(buf, 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(basename, buf, size);

	free(buf);
	return 1;
}

struct single_tree_ctx {
	struct strbuf *path;
	struct object_id oid;
	char *name;
	size_t count;
};

static int single_tree_cb(const struct object_id *oid, struct strbuf *base,
			  const char *pathname, unsigned mode, void *cbdata)
{
	// Not named ctx: that is the global request context, and shadowing it
	// here would hide it from anything added to this function later.
	struct single_tree_ctx *tree_ctx = cbdata;

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

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

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

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

	oidcpy(&tree_ctx.oid, oid);

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

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

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

		read_tree(the_repository, tree, &paths, single_tree_cb, &tree_ctx);

		if (tree_ctx.count != 1)
			break;

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

	strbuf_setlen(fullpath, initial_length);
}

static void render_ls_item(const struct object_id *oid, const char *pathname,
			   unsigned mode, struct walk_tree_context *walk_tree_ctx)
{
	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) {
			htmlf("<tr><td colspan='3'>Bad object: %s %s</td></tr>",
			      name,
			      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)) {
		write_tree_link(oid, name, walk_tree_ctx->curr_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_tree_ctx->curr_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_tree_ctx->curr_rev, linkpath.buf);
		free(buf);
		strbuf_release(&linkpath);
	}
	htmlf("</td><td class='ls-size'>%lu</td>", size);

	html("<td>");
	cgit_log_link("log", NULL, "button", ctx.qry.head,
		      walk_tree_ctx->curr_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_tree_ctx->curr_rev, fullpath.buf);
	if (!S_ISDIR(mode) && ctx.repo->enable_blame)
		cgit_blame_link("blame", NULL, "button", ctx.qry.head,
				walk_tree_ctx->curr_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 *cbdata)
{
	struct walk_tree_context *walk_tree_ctx = cbdata;

	/* When grouping directories first, collect the entries now and render
	 * them once the whole level has been read (see ls_flush). */
	if (ctx.cfg.enable_tree_group_dirs) {
		struct tree_ls_entry *e;
		ALLOC_GROW(walk_tree_ctx->entries, walk_tree_ctx->entries_nr + 1,
			   walk_tree_ctx->entries_alloc);
		e = &walk_tree_ctx->entries[walk_tree_ctx->entries_nr++];
		oidcpy(&e->oid, oid);
		e->name = xstrdup(pathname);
		e->mode = mode;
		return 0;
	}

	render_ls_item(oid, pathname, mode, walk_tree_ctx);
	return 0;
}

/* Render any entries collected by ls_item, directories first and then the
 * rest, keeping git's ordering within each group. A no-op unless directory
 * grouping is enabled, in which case nothing was collected. */
static void ls_flush(struct walk_tree_context *walk_tree_ctx)
{
	size_t i;

	for (i = 0; i < walk_tree_ctx->entries_nr; i++) {
		struct tree_ls_entry *e = &walk_tree_ctx->entries[i];
		if (S_ISDIR(e->mode))
			render_ls_item(&e->oid, e->name, e->mode, walk_tree_ctx);
	}
	for (i = 0; i < walk_tree_ctx->entries_nr; i++) {
		struct tree_ls_entry *e = &walk_tree_ctx->entries[i];
		if (!S_ISDIR(e->mode))
			render_ls_item(&e->oid, e->name, e->mode, walk_tree_ctx);
	}

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

static void ls_head(void)
{
	cgit_print_layout_start();
	html("<table summary='tree listing' class='list'>\n");
	html("<tr class='nohover'>");
	html("<th class='left'>Mode</th>");
	html("<th class='left'>Name</th>");
	html("<th class='right'>Size</th>");
	html("<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_tree_ctx)
{
	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_tree_ctx);
	ls_flush(walk_tree_ctx);
	ls_tail();
}

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

	if (walk_tree_ctx->state == 0) {
		struct strbuf buffer = STRBUF_INIT;

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

		if (S_ISDIR(mode)) {
			walk_tree_ctx->state = 1;
			cgit_set_title_from_path(buffer.buf);
			strbuf_release(&buffer);
			ls_head();
			return READ_TREE_RECURSIVE;
		} else {
			/* state 2: layout left open for us to close; state 3:
			 * print_object already emitted a standalone error page. */
			walk_tree_ctx->state =
				print_object(oid, buffer.buf, pathname,
					     walk_tree_ctx->curr_rev) ? 2 : 3;
			strbuf_release(&buffer);
			return 0;
		}
	}
	ls_item(oid, base, pathname, mode, walk_tree_ctx);
	return 0;
}

/*
 * Show a tree or a blob
 *   rev:  the commit pointing at the root tree object
 *   path: path to tree or blob
 */
void cgit_print_tree(const char *rev, char *path)
{
	struct object_id oid;
	struct commit *commit;
	struct pathspec_item path_items = {
		.match = path,
		.len = path ? strlen(path) : 0
	};
	struct pathspec paths = {
		.nr = path ? 1 : 0,
		.items = &path_items
	};
	struct walk_tree_context walk_tree_ctx = {
		.match_path = path,
		.state = 0
	};

	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_tree_ctx.curr_rev = xstrdup(rev);

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

	read_tree(the_repository, repo_get_commit_tree(the_repository, commit),
		  &paths, walk_tree, &walk_tree_ctx);
	if (walk_tree_ctx.state == 1) {
		ls_flush(&walk_tree_ctx);
		ls_tail();
	} else if (walk_tree_ctx.state == 2)
		cgit_print_layout_end();
	else if (walk_tree_ctx.state == 0)
		cgit_print_error_page(404, "Not found", "Path not found");
	/* state 3: print_object already emitted a complete error page */

cleanup:
	free(walk_tree_ctx.curr_rev);
}