diff options
Diffstat (limited to 'builtin/blame.c')
-rw-r--r-- | builtin/blame.c | 2130 |
1 files changed, 73 insertions, 2057 deletions
diff --git a/builtin/blame.c b/builtin/blame.c index ca9ebe40e7..67adaef4d8 100644 --- a/builtin/blame.c +++ b/builtin/blame.c @@ -6,21 +6,14 @@ */ #include "cache.h" -#include "refs.h" +#include "config.h" #include "builtin.h" -#include "blob.h" #include "commit.h" -#include "tag.h" -#include "tree-walk.h" #include "diff.h" -#include "diffcore.h" #include "revision.h" #include "quote.h" -#include "xdiff-interface.h" -#include "cache-tree.h" #include "string-list.h" #include "mailmap.h" -#include "mergesort.h" #include "parse-options.h" #include "prio-queue.h" #include "utf8.h" @@ -29,6 +22,7 @@ #include "line-log.h" #include "dir.h" #include "progress.h" +#include "blame.h" static char blame_usage[] = N_("git blame [<options>] [<rev-opts>] [<rev>] [--] <file>"); @@ -62,1497 +56,21 @@ static struct string_list mailmap = STRING_LIST_INIT_NODUP; #define DEBUG 0 #endif -/* stats */ -static int num_read_blob; -static int num_get_patch; -static int num_commits; - -#define PICKAXE_BLAME_MOVE 01 -#define PICKAXE_BLAME_COPY 02 -#define PICKAXE_BLAME_COPY_HARDER 04 -#define PICKAXE_BLAME_COPY_HARDEST 010 - -/* - * blame for a blame_entry with score lower than these thresholds - * is not passed to the parent using move/copy logic. - */ static unsigned blame_move_score; static unsigned blame_copy_score; -#define BLAME_DEFAULT_MOVE_SCORE 20 -#define BLAME_DEFAULT_COPY_SCORE 40 /* Remember to update object flag allocation in object.h */ #define METAINFO_SHOWN (1u<<12) #define MORE_THAN_ONE_PATH (1u<<13) -/* - * One blob in a commit that is being suspected - */ -struct origin { - int refcnt; - /* Record preceding blame record for this blob */ - struct origin *previous; - /* origins are put in a list linked via `next' hanging off the - * corresponding commit's util field in order to make finding - * them fast. The presence in this chain does not count - * towards the origin's reference count. It is tempting to - * let it count as long as the commit is pending examination, - * but even under circumstances where the commit will be - * present multiple times in the priority queue of unexamined - * commits, processing the first instance will not leave any - * work requiring the origin data for the second instance. An - * interspersed commit changing that would have to be - * preexisting with a different ancestry and with the same - * commit date in order to wedge itself between two instances - * of the same commit in the priority queue _and_ produce - * blame entries relevant for it. While we don't want to let - * us get tripped up by this case, it certainly does not seem - * worth optimizing for. - */ - struct origin *next; - struct commit *commit; - /* `suspects' contains blame entries that may be attributed to - * this origin's commit or to parent commits. When a commit - * is being processed, all suspects will be moved, either by - * assigning them to an origin in a different commit, or by - * shipping them to the scoreboard's ent list because they - * cannot be attributed to a different commit. - */ - struct blame_entry *suspects; - mmfile_t file; - struct object_id blob_oid; - unsigned mode; - /* guilty gets set when shipping any suspects to the final - * blame list instead of other commits - */ - char guilty; - char path[FLEX_ARRAY]; -}; - struct progress_info { struct progress *progress; int blamed_lines; }; -static int diff_hunks(mmfile_t *file_a, mmfile_t *file_b, - xdl_emit_hunk_consume_func_t hunk_func, void *cb_data) -{ - xpparam_t xpp = {0}; - xdemitconf_t xecfg = {0}; - xdemitcb_t ecb = {NULL}; - - xpp.flags = xdl_opts; - xecfg.hunk_func = hunk_func; - ecb.priv = cb_data; - return xdi_diff(file_a, file_b, &xpp, &xecfg, &ecb); -} - -/* - * Prepare diff_filespec and convert it using diff textconv API - * if the textconv driver exists. - * Return 1 if the conversion succeeds, 0 otherwise. - */ -int textconv_object(const char *path, - unsigned mode, - const struct object_id *oid, - int oid_valid, - char **buf, - unsigned long *buf_size) -{ - struct diff_filespec *df; - struct userdiff_driver *textconv; - - df = alloc_filespec(path); - fill_filespec(df, oid->hash, oid_valid, mode); - textconv = get_textconv(df); - if (!textconv) { - free_filespec(df); - return 0; - } - - *buf_size = fill_textconv(textconv, df, buf); - free_filespec(df); - return 1; -} - -/* - * Given an origin, prepare mmfile_t structure to be used by the - * diff machinery - */ -static void fill_origin_blob(struct diff_options *opt, - struct origin *o, mmfile_t *file) -{ - if (!o->file.ptr) { - enum object_type type; - unsigned long file_size; - - num_read_blob++; - if (DIFF_OPT_TST(opt, ALLOW_TEXTCONV) && - textconv_object(o->path, o->mode, &o->blob_oid, 1, &file->ptr, &file_size)) - ; - else - file->ptr = read_sha1_file(o->blob_oid.hash, &type, - &file_size); - file->size = file_size; - - if (!file->ptr) - die("Cannot read blob %s for path %s", - oid_to_hex(&o->blob_oid), - o->path); - o->file = *file; - } - else - *file = o->file; -} - -/* - * Origin is refcounted and usually we keep the blob contents to be - * reused. - */ -static inline struct origin *origin_incref(struct origin *o) -{ - if (o) - o->refcnt++; - return o; -} - -static void origin_decref(struct origin *o) -{ - if (o && --o->refcnt <= 0) { - struct origin *p, *l = NULL; - if (o->previous) - origin_decref(o->previous); - free(o->file.ptr); - /* Should be present exactly once in commit chain */ - for (p = o->commit->util; p; l = p, p = p->next) { - if (p == o) { - if (l) - l->next = p->next; - else - o->commit->util = p->next; - free(o); - return; - } - } - die("internal error in blame::origin_decref"); - } -} - -static void drop_origin_blob(struct origin *o) -{ - if (o->file.ptr) { - free(o->file.ptr); - o->file.ptr = NULL; - } -} - -/* - * Each group of lines is described by a blame_entry; it can be split - * as we pass blame to the parents. They are arranged in linked lists - * kept as `suspects' of some unprocessed origin, or entered (when the - * blame origin has been finalized) into the scoreboard structure. - * While the scoreboard structure is only sorted at the end of - * processing (according to final image line number), the lists - * attached to an origin are sorted by the target line number. - */ -struct blame_entry { - struct blame_entry *next; - - /* the first line of this group in the final image; - * internally all line numbers are 0 based. - */ - int lno; - - /* how many lines this group has */ - int num_lines; - - /* the commit that introduced this group into the final image */ - struct origin *suspect; - - /* the line number of the first line of this group in the - * suspect's file; internally all line numbers are 0 based. - */ - int s_lno; - - /* how significant this entry is -- cached to avoid - * scanning the lines over and over. - */ - unsigned score; -}; - -/* - * Any merge of blames happens on lists of blames that arrived via - * different parents in a single suspect. In this case, we want to - * sort according to the suspect line numbers as opposed to the final - * image line numbers. The function body is somewhat longish because - * it avoids unnecessary writes. - */ - -static struct blame_entry *blame_merge(struct blame_entry *list1, - struct blame_entry *list2) -{ - struct blame_entry *p1 = list1, *p2 = list2, - **tail = &list1; - - if (!p1) - return p2; - if (!p2) - return p1; - - if (p1->s_lno <= p2->s_lno) { - do { - tail = &p1->next; - if ((p1 = *tail) == NULL) { - *tail = p2; - return list1; - } - } while (p1->s_lno <= p2->s_lno); - } - for (;;) { - *tail = p2; - do { - tail = &p2->next; - if ((p2 = *tail) == NULL) { - *tail = p1; - return list1; - } - } while (p1->s_lno > p2->s_lno); - *tail = p1; - do { - tail = &p1->next; - if ((p1 = *tail) == NULL) { - *tail = p2; - return list1; - } - } while (p1->s_lno <= p2->s_lno); - } -} - -static void *get_next_blame(const void *p) -{ - return ((struct blame_entry *)p)->next; -} - -static void set_next_blame(void *p1, void *p2) -{ - ((struct blame_entry *)p1)->next = p2; -} - -/* - * Final image line numbers are all different, so we don't need a - * three-way comparison here. - */ - -static int compare_blame_final(const void *p1, const void *p2) -{ - return ((struct blame_entry *)p1)->lno > ((struct blame_entry *)p2)->lno - ? 1 : -1; -} - -static int compare_blame_suspect(const void *p1, const void *p2) -{ - const struct blame_entry *s1 = p1, *s2 = p2; - /* - * to allow for collating suspects, we sort according to the - * respective pointer value as the primary sorting criterion. - * The actual relation is pretty unimportant as long as it - * establishes a total order. Comparing as integers gives us - * that. - */ - if (s1->suspect != s2->suspect) - return (intptr_t)s1->suspect > (intptr_t)s2->suspect ? 1 : -1; - if (s1->s_lno == s2->s_lno) - return 0; - return s1->s_lno > s2->s_lno ? 1 : -1; -} - -static struct blame_entry *blame_sort(struct blame_entry *head, - int (*compare_fn)(const void *, const void *)) -{ - return llist_mergesort (head, get_next_blame, set_next_blame, compare_fn); -} - -static int compare_commits_by_reverse_commit_date(const void *a, - const void *b, - void *c) -{ - return -compare_commits_by_commit_date(a, b, c); -} - -/* - * The current state of the blame assignment. - */ -struct scoreboard { - /* the final commit (i.e. where we started digging from) */ - struct commit *final; - /* Priority queue for commits with unassigned blame records */ - struct prio_queue commits; - struct rev_info *revs; - const char *path; - - /* - * The contents in the final image. - * Used by many functions to obtain contents of the nth line, - * indexed with scoreboard.lineno[blame_entry.lno]. - */ - const char *final_buf; - unsigned long final_buf_size; - - /* linked list of blames */ - struct blame_entry *ent; - - /* look-up a line in the final buffer */ - int num_lines; - int *lineno; -}; - -static void sanity_check_refcnt(struct scoreboard *); - -/* - * If two blame entries that are next to each other came from - * contiguous lines in the same origin (i.e. <commit, path> pair), - * merge them together. - */ -static void coalesce(struct scoreboard *sb) -{ - struct blame_entry *ent, *next; - - for (ent = sb->ent; ent && (next = ent->next); ent = next) { - if (ent->suspect == next->suspect && - ent->s_lno + ent->num_lines == next->s_lno) { - ent->num_lines += next->num_lines; - ent->next = next->next; - origin_decref(next->suspect); - free(next); - ent->score = 0; - next = ent; /* again */ - } - } - - if (DEBUG) /* sanity */ - sanity_check_refcnt(sb); -} - -/* - * Merge the given sorted list of blames into a preexisting origin. - * If there were no previous blames to that commit, it is entered into - * the commit priority queue of the score board. - */ - -static void queue_blames(struct scoreboard *sb, struct origin *porigin, - struct blame_entry *sorted) -{ - if (porigin->suspects) - porigin->suspects = blame_merge(porigin->suspects, sorted); - else { - struct origin *o; - for (o = porigin->commit->util; o; o = o->next) { - if (o->suspects) { - porigin->suspects = sorted; - return; - } - } - porigin->suspects = sorted; - prio_queue_put(&sb->commits, porigin->commit); - } -} - -/* - * Given a commit and a path in it, create a new origin structure. - * The callers that add blame to the scoreboard should use - * get_origin() to obtain shared, refcounted copy instead of calling - * this function directly. - */ -static struct origin *make_origin(struct commit *commit, const char *path) -{ - struct origin *o; - FLEX_ALLOC_STR(o, path, path); - o->commit = commit; - o->refcnt = 1; - o->next = commit->util; - commit->util = o; - return o; -} - -/* - * Locate an existing origin or create a new one. - * This moves the origin to front position in the commit util list. - */ -static struct origin *get_origin(struct scoreboard *sb, - struct commit *commit, - const char *path) -{ - struct origin *o, *l; - - for (o = commit->util, l = NULL; o; l = o, o = o->next) { - if (!strcmp(o->path, path)) { - /* bump to front */ - if (l) { - l->next = o->next; - o->next = commit->util; - commit->util = o; - } - return origin_incref(o); - } - } - return make_origin(commit, path); -} - -/* - * Fill the blob_sha1 field of an origin if it hasn't, so that later - * call to fill_origin_blob() can use it to locate the data. blob_sha1 - * for an origin is also used to pass the blame for the entire file to - * the parent to detect the case where a child's blob is identical to - * that of its parent's. - * - * This also fills origin->mode for corresponding tree path. - */ -static int fill_blob_sha1_and_mode(struct origin *origin) -{ - if (!is_null_oid(&origin->blob_oid)) - return 0; - if (get_tree_entry(origin->commit->object.oid.hash, - origin->path, - origin->blob_oid.hash, &origin->mode)) - goto error_out; - if (sha1_object_info(origin->blob_oid.hash, NULL) != OBJ_BLOB) - goto error_out; - return 0; - error_out: - oidclr(&origin->blob_oid); - origin->mode = S_IFINVALID; - return -1; -} - -/* - * We have an origin -- check if the same path exists in the - * parent and return an origin structure to represent it. - */ -static struct origin *find_origin(struct scoreboard *sb, - struct commit *parent, - struct origin *origin) -{ - struct origin *porigin; - struct diff_options diff_opts; - const char *paths[2]; - - /* First check any existing origins */ - for (porigin = parent->util; porigin; porigin = porigin->next) - if (!strcmp(porigin->path, origin->path)) { - /* - * The same path between origin and its parent - * without renaming -- the most common case. - */ - return origin_incref (porigin); - } - - /* See if the origin->path is different between parent - * and origin first. Most of the time they are the - * same and diff-tree is fairly efficient about this. - */ - diff_setup(&diff_opts); - DIFF_OPT_SET(&diff_opts, RECURSIVE); - diff_opts.detect_rename = 0; - diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; - paths[0] = origin->path; - paths[1] = NULL; - - parse_pathspec(&diff_opts.pathspec, - PATHSPEC_ALL_MAGIC & ~PATHSPEC_LITERAL, - PATHSPEC_LITERAL_PATH, "", paths); - diff_setup_done(&diff_opts); - - if (is_null_oid(&origin->commit->object.oid)) - do_diff_cache(parent->tree->object.oid.hash, &diff_opts); - else - diff_tree_sha1(parent->tree->object.oid.hash, - origin->commit->tree->object.oid.hash, - "", &diff_opts); - diffcore_std(&diff_opts); - - if (!diff_queued_diff.nr) { - /* The path is the same as parent */ - porigin = get_origin(sb, parent, origin->path); - oidcpy(&porigin->blob_oid, &origin->blob_oid); - porigin->mode = origin->mode; - } else { - /* - * Since origin->path is a pathspec, if the parent - * commit had it as a directory, we will see a whole - * bunch of deletion of files in the directory that we - * do not care about. - */ - int i; - struct diff_filepair *p = NULL; - for (i = 0; i < diff_queued_diff.nr; i++) { - const char *name; - p = diff_queued_diff.queue[i]; - name = p->one->path ? p->one->path : p->two->path; - if (!strcmp(name, origin->path)) - break; - } - if (!p) - die("internal error in blame::find_origin"); - switch (p->status) { - default: - die("internal error in blame::find_origin (%c)", - p->status); - case 'M': - porigin = get_origin(sb, parent, origin->path); - oidcpy(&porigin->blob_oid, &p->one->oid); - porigin->mode = p->one->mode; - break; - case 'A': - case 'T': - /* Did not exist in parent, or type changed */ - break; - } - } - diff_flush(&diff_opts); - clear_pathspec(&diff_opts.pathspec); - return porigin; -} - -/* - * We have an origin -- find the path that corresponds to it in its - * parent and return an origin structure to represent it. - */ -static struct origin *find_rename(struct scoreboard *sb, - struct commit *parent, - struct origin *origin) -{ - struct origin *porigin = NULL; - struct diff_options diff_opts; - int i; - - diff_setup(&diff_opts); - DIFF_OPT_SET(&diff_opts, RECURSIVE); - diff_opts.detect_rename = DIFF_DETECT_RENAME; - diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT; - diff_opts.single_follow = origin->path; - diff_setup_done(&diff_opts); - - if (is_null_oid(&origin->commit->object.oid)) - do_diff_cache(parent->tree->object.oid.hash, &diff_opts); - else - diff_tree_sha1(parent->tree->object.oid.hash, - origin->commit->tree->object.oid.hash, - "", &diff_opts); - diffcore_std(&diff_opts); - - for (i = 0; i < diff_queued_diff.nr; i++) { - struct diff_filepair *p = diff_queued_diff.queue[i]; - if ((p->status == 'R' || p->status == 'C') && - !strcmp(p->two->path, origin->path)) { - porigin = get_origin(sb, parent, p->one->path); - oidcpy(&porigin->blob_oid, &p->one->oid); - porigin->mode = p->one->mode; - break; - } - } - diff_flush(&diff_opts); - clear_pathspec(&diff_opts.pathspec); - return porigin; -} - -/* - * Append a new blame entry to a given output queue. - */ -static void add_blame_entry(struct blame_entry ***queue, - const struct blame_entry *src) -{ - struct blame_entry *e = xmalloc(sizeof(*e)); - memcpy(e, src, sizeof(*e)); - origin_incref(e->suspect); - - e->next = **queue; - **queue = e; - *queue = &e->next; -} - -/* - * src typically is on-stack; we want to copy the information in it to - * a malloced blame_entry that gets added to the given queue. The - * origin of dst loses a refcnt. - */ -static void dup_entry(struct blame_entry ***queue, - struct blame_entry *dst, struct blame_entry *src) -{ - origin_incref(src->suspect); - origin_decref(dst->suspect); - memcpy(dst, src, sizeof(*src)); - dst->next = **queue; - **queue = dst; - *queue = &dst->next; -} - -static const char *nth_line(struct scoreboard *sb, long lno) -{ - return sb->final_buf + sb->lineno[lno]; -} - static const char *nth_line_cb(void *data, long lno) { - return nth_line((struct scoreboard *)data, lno); -} - -/* - * It is known that lines between tlno to same came from parent, and e - * has an overlap with that range. it also is known that parent's - * line plno corresponds to e's line tlno. - * - * <---- e -----> - * <------> - * <------------> - * <------------> - * <------------------> - * - * Split e into potentially three parts; before this chunk, the chunk - * to be blamed for the parent, and after that portion. - */ -static void split_overlap(struct blame_entry *split, - struct blame_entry *e, - int tlno, int plno, int same, - struct origin *parent) -{ - int chunk_end_lno; - memset(split, 0, sizeof(struct blame_entry [3])); - - if (e->s_lno < tlno) { - /* there is a pre-chunk part not blamed on parent */ - split[0].suspect = origin_incref(e->suspect); - split[0].lno = e->lno; - split[0].s_lno = e->s_lno; - split[0].num_lines = tlno - e->s_lno; - split[1].lno = e->lno + tlno - e->s_lno; - split[1].s_lno = plno; - } - else { - split[1].lno = e->lno; - split[1].s_lno = plno + (e->s_lno - tlno); - } - - if (same < e->s_lno + e->num_lines) { - /* there is a post-chunk part not blamed on parent */ - split[2].suspect = origin_incref(e->suspect); - split[2].lno = e->lno + (same - e->s_lno); - split[2].s_lno = e->s_lno + (same - e->s_lno); - split[2].num_lines = e->s_lno + e->num_lines - same; - chunk_end_lno = split[2].lno; - } - else - chunk_end_lno = e->lno + e->num_lines; - split[1].num_lines = chunk_end_lno - split[1].lno; - - /* - * if it turns out there is nothing to blame the parent for, - * forget about the splitting. !split[1].suspect signals this. - */ - if (split[1].num_lines < 1) - return; - split[1].suspect = origin_incref(parent); -} - -/* - * split_overlap() divided an existing blame e into up to three parts - * in split. Any assigned blame is moved to queue to - * reflect the split. - */ -static void split_blame(struct blame_entry ***blamed, - struct blame_entry ***unblamed, - struct blame_entry *split, - struct blame_entry *e) -{ - if (split[0].suspect && split[2].suspect) { - /* The first part (reuse storage for the existing entry e) */ - dup_entry(unblamed, e, &split[0]); - - /* The last part -- me */ - add_blame_entry(unblamed, &split[2]); - - /* ... and the middle part -- parent */ - add_blame_entry(blamed, &split[1]); - } - else if (!split[0].suspect && !split[2].suspect) - /* - * The parent covers the entire area; reuse storage for - * e and replace it with the parent. - */ - dup_entry(blamed, e, &split[1]); - else if (split[0].suspect) { - /* me and then parent */ - dup_entry(unblamed, e, &split[0]); - add_blame_entry(blamed, &split[1]); - } - else { - /* parent and then me */ - dup_entry(blamed, e, &split[1]); - add_blame_entry(unblamed, &split[2]); - } -} - -/* - * After splitting the blame, the origins used by the - * on-stack blame_entry should lose one refcnt each. - */ -static void decref_split(struct blame_entry *split) -{ - int i; - - for (i = 0; i < 3; i++) - origin_decref(split[i].suspect); -} - -/* - * reverse_blame reverses the list given in head, appending tail. - * That allows us to build lists in reverse order, then reverse them - * afterwards. This can be faster than building the list in proper - * order right away. The reason is that building in proper order - * requires writing a link in the _previous_ element, while building - * in reverse order just requires placing the list head into the - * _current_ element. - */ - -static struct blame_entry *reverse_blame(struct blame_entry *head, - struct blame_entry *tail) -{ - while (head) { - struct blame_entry *next = head->next; - head->next = tail; - tail = head; - head = next; - } - return tail; -} - -/* - * 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 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 - origin_decref(e->suspect); - /* Pass blame for everything before the differing - * chunk to the parent */ - e->suspect = 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 = 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; -} - -struct blame_chunk_cb_data { - struct origin *parent; - long offset; - struct blame_entry **dstq; - struct blame_entry **srcq; -}; - -/* diff chunks are from parent to target */ -static int blame_chunk_cb(long start_a, long count_a, - long start_b, long count_b, void *data) -{ - struct blame_chunk_cb_data *d = data; - if (start_a - start_b != d->offset) - die("internal error in blame::blame_chunk_cb"); - blame_chunk(&d->dstq, &d->srcq, start_b, start_a - start_b, - start_b + count_b, d->parent); - d->offset = start_a + count_a - (start_b + count_b); - return 0; -} - -/* - * We are looking at the origin 'target' and aiming to pass blame - * for the lines it is suspected to its parent. Run diff to find - * which lines came from parent and pass blame for them. - */ -static void pass_blame_to_parent(struct scoreboard *sb, - struct origin *target, - struct origin *parent) -{ - mmfile_t file_p, file_o; - struct blame_chunk_cb_data d; - struct blame_entry *newdest = NULL; - - if (!target->suspects) - return; /* nothing remains for this target */ - - d.parent = parent; - d.offset = 0; - d.dstq = &newdest; d.srcq = &target->suspects; - - fill_origin_blob(&sb->revs->diffopt, parent, &file_p); - fill_origin_blob(&sb->revs->diffopt, target, &file_o); - num_get_patch++; - - if (diff_hunks(&file_p, &file_o, blame_chunk_cb, &d)) - die("unable to generate diff (%s -> %s)", - oid_to_hex(&parent->commit->object.oid), - oid_to_hex(&target->commit->object.oid)); - /* The rest are the same as the parent */ - blame_chunk(&d.dstq, &d.srcq, INT_MAX, d.offset, INT_MAX, parent); - *d.dstq = NULL; - queue_blames(sb, parent, newdest); - - return; -} - -/* - * The lines in blame_entry after splitting blames many times can become - * very small and trivial, and at some point it becomes pointless to - * blame the parents. E.g. "\t\t}\n\t}\n\n" appears everywhere in any - * ordinary C program, and it is not worth to say it was copied from - * totally unrelated file in the parent. - * - * Compute how trivial the lines in the blame_entry are. - */ -static unsigned ent_score(struct scoreboard *sb, struct blame_entry *e) -{ - unsigned score; - const char *cp, *ep; - - if (e->score) - return e->score; - - score = 1; - cp = nth_line(sb, e->lno); - ep = nth_line(sb, e->lno + e->num_lines); - while (cp < ep) { - unsigned ch = *((unsigned char *)cp); - if (isalnum(ch)) - score++; - cp++; - } - e->score = score; - return score; -} - -/* - * best_so_far[] and this[] are both a split of an existing blame_entry - * that passes blame to the parent. Maintain best_so_far the best split - * so far, by comparing this and best_so_far and copying this into - * bst_so_far as needed. - */ -static void copy_split_if_better(struct scoreboard *sb, - struct blame_entry *best_so_far, - struct blame_entry *this) -{ - int i; - - if (!this[1].suspect) - return; - if (best_so_far[1].suspect) { - if (ent_score(sb, &this[1]) < ent_score(sb, &best_so_far[1])) - return; - } - - for (i = 0; i < 3; i++) - origin_incref(this[i].suspect); - decref_split(best_so_far); - memcpy(best_so_far, this, sizeof(struct blame_entry [3])); -} - -/* - * We are looking at a part of the final image represented by - * ent (tlno and same are offset by ent->s_lno). - * tlno is where we are looking at in the final image. - * up to (but not including) same match preimage. - * plno is where we are looking at in the preimage. - * - * <-------------- final image ----------------------> - * <------ent------> - * ^tlno ^same - * <---------preimage-----> - * ^plno - * - * All line numbers are 0-based. - */ -static void handle_split(struct scoreboard *sb, - struct blame_entry *ent, - int tlno, int plno, int same, - struct origin *parent, - struct blame_entry *split) -{ - if (ent->num_lines <= tlno) - return; - if (tlno < same) { - struct blame_entry this[3]; - tlno += ent->s_lno; - same += ent->s_lno; - split_overlap(this, ent, tlno, plno, same, parent); - copy_split_if_better(sb, split, this); - decref_split(this); - } -} - -struct handle_split_cb_data { - struct scoreboard *sb; - struct blame_entry *ent; - struct origin *parent; - struct blame_entry *split; - long plno; - long tlno; -}; - -static int handle_split_cb(long start_a, long count_a, - long start_b, long count_b, void *data) -{ - struct handle_split_cb_data *d = data; - handle_split(d->sb, d->ent, d->tlno, d->plno, start_b, d->parent, - d->split); - d->plno = start_a + count_a; - d->tlno = start_b + count_b; - return 0; -} - -/* - * Find the lines from parent that are the same as ent so that - * we can pass blames to it. file_p has the blob contents for - * the parent. - */ -static void find_copy_in_blob(struct scoreboard *sb, - struct blame_entry *ent, - struct origin *parent, - struct blame_entry *split, - mmfile_t *file_p) -{ - const char *cp; - mmfile_t file_o; - struct handle_split_cb_data d; - - memset(&d, 0, sizeof(d)); - d.sb = sb; d.ent = ent; d.parent = parent; d.split = split; - /* - * Prepare mmfile that contains only the lines in ent. - */ - cp = nth_line(sb, ent->lno); - file_o.ptr = (char *) cp; - file_o.size = nth_line(sb, ent->lno + ent->num_lines) - cp; - - /* - * file_o is a part of final image we are annotating. - * file_p partially may match that image. - */ - memset(split, 0, sizeof(struct blame_entry [3])); - if (diff_hunks(file_p, &file_o, handle_split_cb, &d)) - die("unable to generate diff (%s)", - oid_to_hex(&parent->commit->object.oid)); - /* remainder, if any, all match the preimage */ - handle_split(sb, ent, d.tlno, d.plno, ent->num_lines, parent, split); -} - -/* Move all blame entries from list *source that have a score smaller - * than score_min to the front of list *small. - * Returns a pointer to the link pointing to the old head of the small list. - */ - -static struct blame_entry **filter_small(struct scoreboard *sb, - struct blame_entry **small, - struct blame_entry **source, - unsigned score_min) -{ - struct blame_entry *p = *source; - struct blame_entry *oldsmall = *small; - while (p) { - if (ent_score(sb, p) <= score_min) { - *small = p; - small = &p->next; - p = *small; - } else { - *source = p; - source = &p->next; - p = *source; - } - } - *small = oldsmall; - *source = NULL; - return small; -} - -/* - * See if lines currently target is suspected for can be attributed to - * parent. - */ -static void find_move_in_parent(struct scoreboard *sb, - struct blame_entry ***blamed, - struct blame_entry **toosmall, - struct origin *target, - struct 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); - 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 && - blame_move_score < ent_score(sb, &split[1])) { |