Update tutorial a bit for scripted helpers.
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index ebec71aeee3d318b81c67d4a00687cd363145cf7..e9936016c6a2b40799317d6b2a1122396a5e59e6 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -12,6 +12,7 @@ static const char *diff_opts = "-pu";
 static unsigned char null_sha1[20] = { 0, };
 
 static int reverse_diff;
+static int use_size_cache;
 
 static const char *external_diff(void)
 {
@@ -131,10 +132,16 @@ static void builtin_diff(const char *name_a,
                            diff_arg, input_name_sq[0], input_name_sq[1]);
 
        printf("diff --git a/%s b/%s\n", name_a, name_b);
-       if (!path1[0][0])
+       if (!path1[0][0]) {
                printf("new file mode %s\n", temp[1].mode);
-       else if (!path1[1][0])
+               if (xfrm_msg && xfrm_msg[0])
+                       puts(xfrm_msg);
+       }
+       else if (!path1[1][0]) {
                printf("deleted file mode %s\n", temp[0].mode);
+               if (xfrm_msg && xfrm_msg[0])
+                       puts(xfrm_msg);
+       }
        else {
                if (strcmp(temp[0].mode, temp[1].mode)) {
                        printf("old mode %s\n", temp[0].mode);
@@ -222,12 +229,61 @@ static int work_tree_matches(const char *name, const unsigned char *sha1)
        return 1;
 }
 
+static struct sha1_size_cache {
+       unsigned char sha1[20];
+       unsigned long size;
+} **sha1_size_cache;
+static int sha1_size_cache_nr, sha1_size_cache_alloc;
+
+static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
+                                                int find_only,
+                                                unsigned long size)
+{
+       int first, last;
+       struct sha1_size_cache *e;
+
+       first = 0;
+       last = sha1_size_cache_nr;
+       while (last > first) {
+               int cmp, next = (last + first) >> 1;
+               e = sha1_size_cache[next];
+               cmp = memcmp(e->sha1, sha1, 20);
+               if (!cmp)
+                       return e;
+               if (cmp < 0) {
+                       last = next;
+                       continue;
+               }
+               first = next+1;
+       }
+       /* not found */
+       if (find_only)
+               return NULL;
+       /* insert to make it at "first" */
+       if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
+               sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
+               sha1_size_cache = xrealloc(sha1_size_cache,
+                                          sha1_size_cache_alloc *
+                                          sizeof(*sha1_size_cache));
+       }
+       sha1_size_cache_nr++;
+       if (first < sha1_size_cache_nr)
+               memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
+                       (sha1_size_cache_nr - first - 1) *
+                       sizeof(*sha1_size_cache));
+       e = xmalloc(sizeof(struct sha1_size_cache));
+       sha1_size_cache[first] = e;
+       memcpy(e->sha1, sha1, 20);
+       e->size = size;
+       return e;
+}
+
 /*
  * While doing rename detection and pickaxe operation, we may need to
  * grab the data for the blob (or file) for our own in-core comparison.
  * diff_filespec has data and size fields for this purpose.
  */
-int diff_populate_filespec(struct diff_filespec *s)
+int diff_populate_filespec(struct diff_filespec *s, int size_only)
 {
        int err = 0;
        if (!DIFF_FILE_VALID(s))
@@ -235,6 +291,9 @@ int diff_populate_filespec(struct diff_filespec *s)
        if (S_ISDIR(s->mode))
                return -1;
 
+       if (!use_size_cache)
+               size_only = 0;
+
        if (s->data)
                return err;
        if (!s->sha1_valid ||
@@ -254,6 +313,8 @@ int diff_populate_filespec(struct diff_filespec *s)
                s->size = st.st_size;
                if (!s->size)
                        goto empty;
+               if (size_only)
+                       return 0;
                if (S_ISLNK(st.st_mode)) {
                        int ret;
                        s->data = xmalloc(s->size);
@@ -274,8 +335,21 @@ int diff_populate_filespec(struct diff_filespec *s)
        }
        else {
                char type[20];
-               s->data = read_sha1_file(s->sha1, type, &s->size);
-               s->should_free = 1;
+               struct sha1_size_cache *e;
+
+               if (size_only) {
+                       e = locate_size_cache(s->sha1, 1, 0);
+                       if (e) {
+                               s->size = e->size;
+                               return 0;
+                       }
+                       if (!sha1_file_size(s->sha1, &s->size))
+                               locate_size_cache(s->sha1, 0, s->size);
+               }
+               else {
+                       s->data = read_sha1_file(s->sha1, type, &s->size);
+                       s->should_free = 1;
+               }
        }
        return 0;
 }
@@ -355,13 +429,18 @@ static void prepare_temp_file(const char *name,
                                strcpy(temp->hex, sha1_to_hex(null_sha1));
                        else
                                strcpy(temp->hex, sha1_to_hex(one->sha1));
-                       sprintf(temp->mode, "%06o",
-                               S_IFREG |ce_permissions(st.st_mode));
+                       /* Even though we may sometimes borrow the
+                        * contents from the work tree, we always want
+                        * one->mode.  mode is trustworthy even when
+                        * !(one->sha1_valid), as long as
+                        * DIFF_FILE_VALID(one).
+                        */
+                       sprintf(temp->mode, "%06o", one->mode);
                }
                return;
        }
        else {
-               if (diff_populate_filespec(one))
+               if (diff_populate_filespec(one, 0))
                        die("cannot read data blob for %s", one->path);
                prep_temp_blob(temp, one->data, one->size,
                               one->sha1, one->mode);
@@ -476,6 +555,7 @@ static void run_diff(const char *name,
 {
        const char *pgm = external_diff();
        if (!pgm &&
+           one && two &&
            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
                /* a filepair that changes between file and symlink
@@ -496,6 +576,70 @@ void diff_setup(int flags)
 {
        if (flags & DIFF_SETUP_REVERSE)
                reverse_diff = 1;
+       if (flags & DIFF_SETUP_USE_CACHE) {
+               if (!active_cache)
+                       /* read-cache does not die even when it fails
+                        * so it is safe for us to do this here.  Also
+                        * it does not smudge active_cache or active_nr
+                        * when it fails, so we do not have to worry about
+                        * cleaning it up oufselves either.
+                        */
+                       read_cache();
+       }
+       if (flags & DIFF_SETUP_USE_SIZE_CACHE)
+               use_size_cache = 1;
+       
+}
+
+static int parse_num(const char **cp_p)
+{
+       int num, scale, ch, cnt;
+       const char *cp = *cp_p;
+
+       cnt = num = 0;
+       scale = 1;
+       while ('0' <= (ch = *cp) && ch <= '9') {
+               if (cnt++ < 5) {
+                       /* We simply ignore more than 5 digits precision. */
+                       scale *= 10;
+                       num = num * 10 + ch - '0';
+               }
+               *cp++;
+       }
+       *cp_p = cp;
+
+       /* user says num divided by scale and we say internally that
+        * is MAX_SCORE * num / scale.
+        */
+       return (MAX_SCORE * num / scale);
+}
+
+int diff_scoreopt_parse(const char *opt)
+{
+       int opt1, opt2, cmd;
+
+       if (*opt++ != '-')
+               return -1;
+       cmd = *opt++;
+       if (cmd != 'M' && cmd != 'C' && cmd != 'B')
+               return -1; /* that is not a -M, -C nor -B option */
+
+       opt1 = parse_num(&opt);
+       if (cmd != 'B')
+               opt2 = 0;
+       else {
+               if (*opt == 0)
+                       opt2 = 0;
+               else if (*opt != '/')
+                       return -1; /* we expect -B80/99 or -B80 */
+               else {
+                       opt++;
+                       opt2 = parse_num(&opt);
+               }
+       }
+       if (*opt != 0)
+               return -1;
+       return opt1 | (opt2 << 16);
 }
 
 struct diff_queue_struct diff_queued_diff;
@@ -518,7 +662,9 @@ struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
        dp->one = one;
        dp->two = two;
        dp->score = 0;
+       dp->status = 0;
        dp->source_stays = 0;
+       dp->broken_pair = 0;
        diff_q(queue, dp);
        return dp;
 }
@@ -553,6 +699,16 @@ static void diff_flush_raw(struct diff_filepair *p,
                sprintf(status, "%c%03d", p->status,
                        (int)(0.5 + p->score * 100.0/MAX_SCORE));
                break;
+       case 'N': case 'D':
+               two_paths = 0;
+               if (p->score)
+                       sprintf(status, "%c%03d", p->status,
+                               (int)(0.5 + p->score * 100.0/MAX_SCORE));
+               else {
+                       status[0] = p->status;
+                       status[1] = 0;
+               }
+               break;
        default:
                two_paths = 0;
                status[0] = p->status;
@@ -632,12 +788,22 @@ static void diff_flush_patch(struct diff_filepair *p)
        case 'R':
                sprintf(msg_,
                        "similarity index %d%%\n"
-                       "rename old %s\n"
-                       "rename new %s",
+                       "rename from %s\n"
+                       "rename to %s",
                        (int)(0.5 + p->score * 100.0/MAX_SCORE),
                        p->one->path, p->two->path);
                msg = msg_;
                break;
+       case 'D': case 'N':
+               if (DIFF_PAIR_BROKEN(p)) {
+                       sprintf(msg_,
+                               "dissimilarity index %d%%",
+                               (int)(0.5 + p->score * 100.0/MAX_SCORE));
+                       msg = msg_;
+               }
+               else
+                       msg = NULL;
+               break;
        default:
                msg = NULL;
        }
@@ -676,8 +842,9 @@ void diff_debug_filepair(const struct diff_filepair *p, int i)
 {
        diff_debug_filespec(p->one, i, "one");
        diff_debug_filespec(p->two, i, "two");
-       fprintf(stderr, "score %d, status %c source_stays %d\n",
-               p->score, p->status ? : '?', p->source_stays);
+       fprintf(stderr, "score %d, status %c stays %d broken %d\n",
+               p->score, p->status ? : '?',
+               p->source_stays, p->broken_pair);
 }
 
 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
@@ -708,27 +875,8 @@ static void diff_resolve_rename_copy(void)
                        p->status = 'U';
                else if (!DIFF_FILE_VALID(p->one))
                        p->status = 'N';
-               else if (!DIFF_FILE_VALID(p->two)) {
-                       /* Deleted entry may have been picked up by
-                        * another rename-copy entry.  So we scan the
-                        * queue and if we find one that uses us as the
-                        * source we do not say delete for this entry.
-                        */
-                       for (j = 0; j < q->nr; j++) {
-                               pp = q->queue[j];
-                               if (!strcmp(p->one->path, pp->one->path) &&
-                                   pp->score) {
-                                       /* rename/copy are always valid
-                                        * so we do not say DIFF_FILE_VALID()
-                                        * on pp->one and pp->two.
-                                        */
-                                       p->status = 'X';
-                                       break;
-                               }
-                       }
-                       if (!p->status)
-                               p->status = 'D';
-               }
+               else if (!DIFF_FILE_VALID(p->two))
+                       p->status = 'D';
                else if (DIFF_PAIR_TYPE_CHANGED(p))
                        p->status = 'T';
 
@@ -736,7 +884,7 @@ static void diff_resolve_rename_copy(void)
                 * whose both sides are valid and of the same type, i.e.
                 * either in-place edit or rename/copy edit.
                 */
-               else if (p->score) {
+               else if (DIFF_PAIR_RENAME(p)) {
                        if (p->source_stays) {
                                p->status = 'C';
                                continue;
@@ -749,7 +897,7 @@ static void diff_resolve_rename_copy(void)
                                pp = q->queue[j];
                                if (strcmp(pp->one->path, p->one->path))
                                        continue; /* not us */
-                               if (!pp->score)
+                               if (!DIFF_PAIR_RENAME(pp))
                                        continue; /* not a rename/copy */
                                /* pp is a rename/copy from the same source */
                                p->status = 'C';
@@ -761,17 +909,19 @@ static void diff_resolve_rename_copy(void)
                else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
                         p->one->mode != p->two->mode)
                        p->status = 'M';
-               else
-                       /* this is a "no-change" entry.
-                        * should not happen anymore.
-                        * p->status = 'X';
+               else {
+                       /* This is a "no-change" entry and should not
+                        * happen anymore, but prepare for broken callers.
                         */
-                       die("internal error in diffcore: unmodified entry remains");
+                       error("feeding unmodified %s to diffcore",
+                             p->one->path);
+                       p->status = 'X';
+               }
        }
        diff_debug_queue("resolve-rename-copy done", q);
 }
 
-void diff_flush(int diff_output_style, int resolve_rename_copy)
+void diff_flush(int diff_output_style)
 {
        struct diff_queue_struct *q = &diff_queued_diff;
        int i;
@@ -780,8 +930,6 @@ void diff_flush(int diff_output_style, int resolve_rename_copy)
 
        if (diff_output_style == DIFF_FORMAT_MACHINE)
                line_termination = inter_name_termination = 0;
-       if (resolve_rename_copy)
-               diff_resolve_rename_copy();
 
        for (i = 0; i < q->nr; i++) {
                struct diff_filepair *p = q->queue[i];
@@ -808,6 +956,90 @@ void diff_flush(int diff_output_style, int resolve_rename_copy)
        q->nr = q->alloc = 0;
 }
 
+static void diffcore_apply_filter(const char *filter)
+{
+       int i;
+       struct diff_queue_struct *q = &diff_queued_diff;
+       struct diff_queue_struct outq;
+       outq.queue = NULL;
+       outq.nr = outq.alloc = 0;
+
+       if (!filter)
+               return;
+
+       if (strchr(filter, 'A')) {
+               /* All-or-none */
+               int found;
+               for (i = found = 0; !found && i < q->nr; i++) {
+                       struct diff_filepair *p = q->queue[i];
+                       if ((p->broken_pair && strchr(filter, 'B')) ||
+                           (!p->broken_pair && strchr(filter, p->status)))
+                               found++;
+               }
+               if (found)
+                       return;
+
+               /* otherwise we will clear the whole queue
+                * by copying the empty outq at the end of this
+                * function, but first clear the current entries
+                * in the queue.
+                */
+               for (i = 0; i < q->nr; i++)
+                       diff_free_filepair(q->queue[i]);
+       }
+       else {
+               /* Only the matching ones */
+               for (i = 0; i < q->nr; i++) {
+                       struct diff_filepair *p = q->queue[i];
+                       if ((p->broken_pair && strchr(filter, 'B')) ||
+                           (!p->broken_pair && strchr(filter, p->status)))
+                               diff_q(&outq, p);
+                       else
+                               diff_free_filepair(p);
+               }
+       }
+       free(q->queue);
+       *q = outq;
+}
+
+void diffcore_std(const char **paths,
+                 int detect_rename, int rename_score,
+                 const char *pickaxe, int pickaxe_opts,
+                 int break_opt,
+                 const char *orderfile,
+                 const char *filter)
+{
+       if (paths && paths[0])
+               diffcore_pathspec(paths);
+       if (break_opt != -1)
+               diffcore_break(break_opt);
+       if (detect_rename)
+               diffcore_rename(detect_rename, rename_score);
+       if (break_opt != -1)
+               diffcore_merge_broken();
+       if (pickaxe)
+               diffcore_pickaxe(pickaxe, pickaxe_opts);
+       if (orderfile)
+               diffcore_order(orderfile);
+       diff_resolve_rename_copy();
+       diffcore_apply_filter(filter);
+}
+
+
+void diffcore_std_no_resolve(const char **paths,
+                            const char *pickaxe, int pickaxe_opts,
+                            const char *orderfile,
+                            const char *filter)
+{
+       if (paths && paths[0])
+               diffcore_pathspec(paths);
+       if (pickaxe)
+               diffcore_pickaxe(pickaxe, pickaxe_opts);
+       if (orderfile)
+               diffcore_order(orderfile);
+       diffcore_apply_filter(filter);
+}
+
 void diff_addremove(int addremove, unsigned mode,
                    const unsigned char *sha1,
                    const char *base, const char *path)