index : cgit

/* ui-blame.c: functions for blame 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-blame.h"
#include "html.h"
#include "ui-shared.h"
#include "strvec.h"
#include "blame.h"


/* Blame coalesces neighbouring lines from one commit, but a commit that
 * touched several separate parts of the file still comes back once per part.
 * Each of those repeats a full commit parse, so the rendered detail is kept
 * and looked up by object id. */
static struct string_list suspect_details = STRING_LIST_INIT_DUP;

static void free_suspect_details(void)
{
	struct string_list_item *item;

	for_each_string_list_item(item, &suspect_details)
		free(item->util);
	string_list_clear(&suspect_details, 0);
}

/* Write out a run of one repeated character. A single blame entry can cover
 * the whole file, so the run is handed on in batches rather than built whole. */
static void emit_chars(char ch, unsigned long count)
{
	struct strbuf run = STRBUF_INIT;

	while (count) {
		unsigned long n = count < HTML_BATCH ? count : HTML_BATCH;

		strbuf_addchars(&run, ch, n);
		html_raw(run.buf, run.len);
		strbuf_reset(&run);
		count -= n;
	}
	strbuf_release(&run);
}

static char *emit_suspect_detail(struct blame_origin *suspect)
{
	struct commitinfo *info;
	struct strbuf detail = STRBUF_INIT;
	struct string_list_item *cached;
	char key[GIT_MAX_HEXSZ + 1];

	// A private copy, since oid_to_hex hands back a buffer it reuses and
	// the commit parse below is free to call it again.
	oid_to_hex_r(key, &suspect->commit->object.oid);

	cached = string_list_lookup(&suspect_details, key);
	if (cached)
		return cached->util;

	info = cgit_parse_commit(suspect->commit);

	strbuf_addf(&detail, "author  %s", info->author);
	if (!ctx.cfg.noplainemail)
		strbuf_addf(&detail, " %s", info->author_email);
	strbuf_addf(&detail, "  %s\n",
		    show_date(info->author_date, info->author_tz,
				    cgit_date_mode(DATE_ISO8601)));

	strbuf_addf(&detail, "committer  %s", info->committer);
	if (!ctx.cfg.noplainemail)
		strbuf_addf(&detail, " %s", info->committer_email);
	strbuf_addf(&detail, "  %s\n\n",
		    show_date(info->committer_date, info->committer_tz,
				    cgit_date_mode(DATE_ISO8601)));

	strbuf_addstr(&detail, info->subject);

	cgit_free_commitinfo(info);
	cached = string_list_insert(&suspect_details, key);
	cached->util = strbuf_detach(&detail, NULL);
	return cached->util;
}

static void emit_blame_entry_hash(struct blame_entry *ent)
{
	struct blame_origin *suspect = ent->suspect;
	struct object_id *oid = &suspect->commit->object.oid;

	const char *detail = emit_suspect_detail(suspect);
	html("<span class='oid'>");
	cgit_commit_link(repo_find_unique_abbrev(the_repository, oid, DEFAULT_ABBREV), detail,
			 NULL, ctx.qry.head, oid_to_hex(oid), suspect->path);
	html("</span>");

	if (!repo_parse_commit(the_repository, suspect->commit) && suspect->commit->parents) {
		struct commit *parent = suspect->commit->parents->item;

		html(" ");
		cgit_blame_link("^", "Blame the previous revision", NULL,
				ctx.qry.head, oid_to_hex(&parent->object.oid),
				suspect->path);
	}

	// Batched rather than one write per line of the entry.
	emit_chars('\n', ent->num_lines);
}

static void emit_blame_entry_linenumber(struct blame_entry *ent)
{
	const char *numberfmt = "<a id='n%1$d' href='#n%1$d'>%1$d</a>\n";
	struct strbuf numbers = STRBUF_INIT;
	int lineno = ent->lno;

	// Batched rather than a formatted write per line of the entry.
	while (lineno < ent->lno + ent->num_lines) {
		strbuf_addf(&numbers, numberfmt, ++lineno);
		if (numbers.len >= HTML_BATCH) {
			html_raw(numbers.buf, numbers.len);
			strbuf_reset(&numbers);
		}
	}
	html_raw(numbers.buf, numbers.len);
	strbuf_release(&numbers);
}

static void emit_blame_entry_line_background(struct blame_scoreboard *sb,
					     struct blame_entry *ent)
{
	int line;
	size_t len, maxlen = 2;
	const char* pos, *endpos;

	for (line = ent->lno; line < ent->lno + ent->num_lines; line++) {
		pos = blame_nth_line(sb, line);
		endpos = blame_nth_line(sb, line + 1);
		len = 0;
		while (pos < endpos) {
			len++;
			if (*pos++ == '\t')
				len = (len + 7) & ~7;
		}
		if (len > maxlen)
			maxlen = len;
	}

	// The widest line decides the padding, so the entry has to be measured
	// before any of it is written. The newlines do not depend on that
	// measurement, so both runs go out batched once it is known.
	emit_chars('\n', ent->num_lines);
	emit_chars(' ', maxlen - 1);
}

struct walk_tree_context {
	char *curr_rev;
	int match_baselen;
	int state;
};

static void 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;
	struct strvec rev_argv = STRVEC_INIT;
	struct rev_info revs;
	struct blame_scoreboard sb;
	struct blame_origin *o;
	struct blame_entry *ent = NULL;

	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;
	}
	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;
	}

	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;
	}

	strvec_push(&rev_argv, "blame");
	strvec_push(&rev_argv, rev);
	repo_init_revisions(the_repository, &revs, NULL);
	revs.diffopt.flags.allow_textconv = 1;
	setup_revisions(rev_argv.nr, rev_argv.v, &revs, NULL);
	init_scoreboard(&sb);
	sb.revs = &revs;
	sb.repo = the_repository;
	sb.path = path;
	setup_scoreboard(&sb, &o);
	o->suspects = blame_entry_prepend(NULL, 0, sb.num_lines, o);
	prio_queue_put(&sb.commits, o->commit);
	blame_origin_decref(o);
	sb.ent = NULL;
	sb.path = path;
	assign_blame(&sb, 0);
	blame_sort_final(&sb);
	blame_coalesce(&sb);

	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);
	html(") (");
	cgit_tree_link("tree", NULL, NULL, ctx.qry.head, rev, path);
	html(")\n");

	if (buffer_is_binary(buf, size)) {
		html("<div class='error'>blob is binary.</div>");
		goto cleanup;
	}
	html("<table class='blame blob'>\n<tr>\n");

	/* Commit hashes */
	html("<td class='hashes'>");
	for (ent = sb.ent; ent; ent = ent->next) {
		html("<div class='alt'><pre>");
		emit_blame_entry_hash(ent);
		html("</pre></div>");
	}
	html("</td>\n");

	/* Line numbers */
	if (ctx.cfg.enable_tree_linenumbers) {
		html("<td class='linenumbers'>");
		for (ent = sb.ent; ent; ent = ent->next) {
			html("<div class='alt'><pre>");
			emit_blame_entry_linenumber(ent);
			html("</pre></div>");
		}
		html("</td>\n");
	}

	html("<td class='lines'><div>");

	/* Colored bars behind lines */
	html("<div>");
	for (ent = sb.ent; ent; ) {
		struct blame_entry *e = ent->next;
		html("<div class='alt'><pre>");
		emit_blame_entry_line_background(&sb, ent);
		html("</pre></div>");
		free(ent);
		ent = e;
	}
	html("</div>");

	free((void *)sb.final_buf);

	/* Lines */
	html("<pre><code>");
	if (ctx.repo->source_filter) {
		char *filter_arg = xstrdup(basename);
		cgit_open_filter(ctx.repo->source_filter, filter_arg);
		html_raw(buf, size);
		cgit_close_filter(ctx.repo->source_filter);
		free(filter_arg);
	} else {
		html_txt(buf);
	}
	html("</code></pre>");

	html("</div></td>\n");

	html("</tr>\n</table>\n");

cleanup:
	/* The binary and oversized branches jump here with the layout still
	 * open, so close it on every path rather than only the normal one. */
	cgit_print_layout_end();
	free_suspect_details();
	free(buf);
}

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;

	// match_baselen is -1 when no path was given, which no length equals.
	if (walk_tree_ctx->match_baselen >= 0 &&
	    base->len == (size_t)walk_tree_ctx->match_baselen) {
		if (S_ISREG(mode)) {
			struct strbuf buffer = STRBUF_INIT;
			strbuf_addbuf(&buffer, base);
			strbuf_addstr(&buffer, pathname);
			print_object(oid, buffer.buf, pathname,
				     walk_tree_ctx->curr_rev);
			strbuf_release(&buffer);
			walk_tree_ctx->state = 1;
		} else if (S_ISDIR(mode)) {
			walk_tree_ctx->state = 2;
		}
	} else if (base->len < INT_MAX
			&& (int)base->len > walk_tree_ctx->match_baselen) {
		walk_tree_ctx->state = 2;
	} else if (S_ISDIR(mode)) {
		return READ_TREE_RECURSIVE;
	}
	return 0;
}

static int basedir_len(const char *path)
{
	const char *p = strrchr(path, '/');
	if (p)
		return p - path + 1;
	return 0;
}

void cgit_print_blame(void)
{
	const char *rev = ctx.qry.oid;
	struct object_id oid;
	struct commit *commit;
	struct pathspec_item path_items = {
		.match = ctx.qry.path,
		.len = ctx.qry.path ? strlen(ctx.qry.path) : 0
	};
	struct pathspec paths = {
		.nr = 1,
		.items = &path_items
	};
	struct walk_tree_context walk_tree_ctx = {
		.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);
	walk_tree_ctx.match_baselen = (path_items.match) ?
				       basedir_len(path_items.match) : -1;

	read_tree(the_repository, repo_get_commit_tree(the_repository, commit),
		  &paths, walk_tree, &walk_tree_ctx);
	if (!walk_tree_ctx.state)
		cgit_print_error_page(404, "Not found", "Not found");
	else if (walk_tree_ctx.state == 2)
		cgit_print_error_page(404, "No blame for folders",
			"Blame is not available for folders.");

	free(walk_tree_ctx.curr_rev);
}