/* Distribute collected unsorted blames to the respected sorted lists
* in the various origins.
*/
static void distribute_blame(struct blame_scoreboard *sb, struct blame_entry *blamed)
{
blamed = llist_mergesort(blamed, get_next_blame, set_next_blame,
compare_blame_suspect);
while (blamed)
{
struct blame_origin *porigin = blamed->suspect;
struct blame_entry *suspects = NULL;
do {
struct blame_entry *next = blamed->next;
blamed->next = suspects;
suspects = blamed;
blamed = next;
} while (blamed && blamed->suspect == porigin);
suspects = reverse_blame(suspects, NULL);
queue_blames(sb, porigin, suspects);
}
}
/*
* See if lines currently target is suspected for can be attributed to
* parent.
*/
static void find_move_in_parent(struct blame_scoreboard *sb,
struct blame_entry ***blamed,
struct blame_entry **toosmall,
struct blame_origin *target,
struct blame_origin *parent)
{
struct blame_entry *e, split[3];
struct blame_entry *unblamed = target->suspects;
struct blame_entry *leftover = NULL;
mmfile_t file_p;
if (!unblamed)
return; /* nothing remains for this target */
fill_origin_blob(&sb->revs->diffopt, parent, &file_p, &sb->num_read_blob);
if (!file_p.ptr)
return;
/* At each iteration, unblamed has a NULL-terminated list of
* entries that have not yet been tested for blame. leftover
* contains the reversed list of entries that have been tested
* without being assignable to the parent.
*/
do {
struct blame_entry **unblamedtail = &unblamed;
struct blame_entry *next;
for (e = unblamed; e; e = next) {
next = e->next;
find_copy_in_blob(sb, e, parent, split, &file_p);
if (split[1].suspect &&
sb->move_score < blame_entry_score(sb, &split[1])) {
split_blame(blamed, &unblamedtail, split, e);
} else {
e->next = leftover;
leftover = e;
}
decref_split(split);
}
*unblamedtail = NULL;
toosmall = filter_small(sb, toosmall, &unblamed, sb->move_score);
} while (unblamed);
target->suspects = reverse_blame(leftover, NULL);
}
/*
* Process one hunk from the patch between the current suspect for
* blame_entry e and its parent. This first blames any unfinished
* entries before the chunk (which is where target and parent start
* differing) on the parent, and then splits blame entries at the
* start and at the end of the difference region. Since use of -M and
* -C options may lead to overlapping/duplicate source line number
* ranges, all we can rely on from sorting/merging is the order of the
* first suspect line number.
*/
static void blame_chunk(struct blame_entry ***dstq, struct blame_entry ***srcq,
int tlno, int offset, int same,
struct blame_origin *parent)
{
struct blame_entry *e = **srcq;
struct blame_entry *samep = NULL, *diffp = NULL;
while (e && e->s_lno < tlno) {
struct blame_entry *next = e->next;
/*
* current record starts before differing portion. If
* it reaches into it, we need to split it up and
* examine the second part separately.
*/
if (e->s_lno + e->num_lines > tlno) {
/* Move second half to a new record */
int len = tlno - e->s_lno;
struct blame_entry *n = xcalloc(1, sizeof (struct blame_entry));
n->suspect = e->suspect;
n->lno = e->lno + len;
n->s_lno = e->s_lno + len;
n->num_lines = e->num_lines - len;
e->num_lines = len;
e->score = 0;
/* Push new record to diffp */
n->next = diffp;
diffp = n;
} else
blame_origin_decref(e->suspect);
/* Pass blame for everything before the differing
* chunk to the parent */
e->suspect = blame_origin_incref(parent);
e->s_lno += offset;
e->next = samep;
samep = e;
e = next;
}
/*
* As we don't know how much of a common stretch after this
* diff will occur, the currently blamed parts are all that we
* can assign to the parent for now.
*/
if (samep) {
**dstq = reverse_blame(samep, **dstq);
*dstq = &samep->next;
}
/*
* Prepend the split off portions: everything after e starts
* after the blameable portion.
*/
e = reverse_blame(diffp, e);
/*
* Now retain records on the target while parts are different
* from the parent.
*/
samep = NULL;
diffp = NULL;
while (e && e->s_lno < same) {
struct blame_entry *next = e->next;
/*
* If current record extends into sameness, need to split.
*/
if (e->s_lno + e->num_lines > same) {
/*
* Move second half to a new record to be
* processed by later chunks
*/
int len = same - e->s_lno;
struct blame_entry *n = xcalloc(1, sizeof (struct blame_entry));
n->suspect = blame_origin_incref(e->suspect);
n->lno = e->lno + len;
n->s_lno = e->s_lno + len;
n->num_lines = e->num_lines - len;
e->num_lines = len;
e->score = 0;
/* Push new record to samep */
n->next = samep;
samep = n;
}
e->next = diffp;
diffp = e;
e = next;
}
**srcq = reverse_blame(diffp, reverse_blame(samep, e));
/* Move across elements that are in the unblamable portion */
if (diffp)
*srcq = &diffp->next;
}
/*
* For lines target is suspected for, see if we can find code movement
* across file boundary from the parent commit. porigin is the path
* in the parent we already tried.
*/
static void find_copy_in_parent(struct blame_scoreboard *sb,
struct blame_entry ***blamed,
struct blame_entry **toosmall,
struct blame_origin *target,
struct commit *parent,
struct blame_origin *porigin,
int opt)
{
struct diff_options diff_opts;
int i, j;
struct blame_list *blame_list;
int num_ents;
struct blame_entry *unblamed = target->suspects;
struct blame_entry *leftover = NULL;
if (!unblamed)
return; /* nothing remains for this target */
diff_setup(&diff_opts);
diff_opts.flags.recursive = 1;
diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
diff_setup_done(&diff_opts);
/* Try "find copies harder" on new path if requested;
* we do not want to use diffcore_rename() actually to
* match things up; find_copies_harder is set only to
* force diff_tree_oid() to feed all filepairs to diff_queue,
* and this code needs to be after diff_setup_done(), which
* usually makes find-copies-harder imply copy detection.
*/
if ((opt & PICKAXE_BLAME_COPY_HARDEST)
|| ((opt & PICKAXE_BLAME_COPY_HARDER)
&& (!porigin || strcmp(target->path, porigin->path))))
diff_opts.flags.find_copies_harder = 1;
if (is_null_oid(&target->commit->object.oid))
do_diff_cache(get_commit_tree_oid(parent), &diff_opts);
else
diff_tree_oid(get_commit_tree_oid(parent),
get_commit_tree_oid(target->commit),
"", &diff_opts);
if (!diff_opts.flags.find_copies_harder)
diffcore_std(&diff_opts);
do {
struct blame_entry **unblamedtail = &unblamed;
blame_list = setup_blame_list(unblamed, &num_ents);
for (i = 0; i < diff_queued_diff.nr; i++) {
struct diff_filepair *p = diff_queued_diff.queue[i];
struct blame_origin *norigin;
mmfile_t file_p;
struct blame_entry potential[3];
if (!DIFF_FILE_VALID(p->one))
continue; /* does not exist in parent */
if (S_ISGITLINK(p->one->mode))
continue; /* ignore git links */
if (porigin && !strcmp(p->one->path, porigin->path))
/* find_move already dealt with this path */
continue;
norigin = get_origin(parent, p->one->path);
oidcpy(&norigin->blob_oid, &p->one->oid);
norigin->mode = p->one->mode;
fill_origin_blob(&sb->revs->diffopt, norigin, &file_p, &sb->num_read_blob);
if (!file_p.ptr)
continue;
for (j = 0; j < num_ents; j++) {
find_copy_in_blob(sb, blame_list[j].ent,
norigin, potential, &file_p);
copy_split_if_better(sb, blame_list[j].split,
potential);
decref_split(potential);
}
blame_origin_decref(norigin);
}
for (j = 0; j < num_ents; j++) {
struct blame_entry *split = blame_list[j].split;
if (split[1].suspect &&
sb->copy_score < blame_entry_score(sb, &split[1])) {
split_blame(blamed, &unblamedtail, split,
blame_list[j].ent);
} else {
blame_list[j].ent->next = leftover;
leftover = blame_list[j].ent;
}
decref_split(split);
}
free(blame_list);
*unblamedtail = NULL;
toosmall = filter_small(sb, toosmall, &unblamed, sb->copy_score);
} while (unblamed);
target->suspects = reverse_blame(leftover, NULL);
diff_flush(&diff_opts);
clear_pathspec(&diff_opts.pathspec);
}