Teach git list-objects logic not to follow gitlinks
[gitweb.git] / builtin-rev-list.c
index 8437454fbe4e3237aaa6aaecefe54caefb621f1b..09774f9559b81050d89bd6663b8b672438da4342 100644 (file)
@@ -7,6 +7,7 @@
 #include "tree-walk.h"
 #include "diff.h"
 #include "revision.h"
+#include "list-objects.h"
 #include "builtin.h"
 
 /* bits #0-15 in revision.h */
@@ -23,6 +24,7 @@ static const char rev_list_usage[] =
 "    --no-merges\n"
 "    --remove-empty\n"
 "    --all\n"
+"    --stdin\n"
 "  ordering output:\n"
 "    --topo-order\n"
 "    --date-order\n"
@@ -33,8 +35,10 @@ static const char rev_list_usage[] =
 "    --header | --pretty\n"
 "    --abbrev=nr | --no-abbrev\n"
 "    --abbrev-commit\n"
+"    --left-right\n"
 "  special purpose:\n"
-"    --bisect"
+"    --bisect\n"
+"    --bisect-vars"
 ;
 
 static struct rev_info revs;
@@ -52,6 +56,12 @@ static void show_commit(struct commit *commit)
                fputs(header_prefix, stdout);
        if (commit->object.flags & BOUNDARY)
                putchar('-');
+       else if (revs.left_right) {
+               if (commit->object.flags & SYMMETRIC_LEFT)
+                       putchar('<');
+               else
+                       putchar('>');
+       }
        if (revs.abbrev_commit && revs.abbrev)
                fputs(find_unique_abbrev(commit->object.sha1, revs.abbrev),
                      stdout);
@@ -97,104 +107,24 @@ static void show_commit(struct commit *commit)
        commit->buffer = NULL;
 }
 
-static void process_blob(struct blob *blob,
-                        struct object_array *p,
-                        struct name_path *path,
-                        const char *name)
+static void show_object(struct object_array_entry *p)
 {
-       struct object *obj = &blob->object;
-
-       if (!revs.blob_objects)
-               return;
-       if (obj->flags & (UNINTERESTING | SEEN))
-               return;
-       obj->flags |= SEEN;
-       name = xstrdup(name);
-       add_object(obj, p, path, name);
-}
-
-static void process_tree(struct tree *tree,
-                        struct object_array *p,
-                        struct name_path *path,
-                        const char *name)
-{
-       struct object *obj = &tree->object;
-       struct tree_desc desc;
-       struct name_entry entry;
-       struct name_path me;
-
-       if (!revs.tree_objects)
-               return;
-       if (obj->flags & (UNINTERESTING | SEEN))
-               return;
-       if (parse_tree(tree) < 0)
-               die("bad tree object %s", sha1_to_hex(obj->sha1));
-       obj->flags |= SEEN;
-       name = xstrdup(name);
-       add_object(obj, p, path, name);
-       me.up = path;
-       me.elem = name;
-       me.elem_len = strlen(name);
-
-       desc.buf = tree->buffer;
-       desc.size = tree->size;
-
-       while (tree_entry(&desc, &entry)) {
-               if (S_ISDIR(entry.mode))
-                       process_tree(lookup_tree(entry.sha1), p, &me, entry.path);
-               else
-                       process_blob(lookup_blob(entry.sha1), p, &me, entry.path);
+       /* An object with name "foo\n0000000..." can be used to
+        * confuse downstream git-pack-objects very badly.
+        */
+       const char *ep = strchr(p->name, '\n');
+       if (ep) {
+               printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
+                      (int) (ep - p->name),
+                      p->name);
        }
-       free(tree->buffer);
-       tree->buffer = NULL;
+       else
+               printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
 }
 
-static void show_commit_list(struct rev_info *revs)
+static void show_edge(struct commit *commit)
 {
-       int i;
-       struct commit *commit;
-       struct object_array objects = { 0, 0, NULL };
-
-       while ((commit = get_revision(revs)) != NULL) {
-               process_tree(commit->tree, &objects, NULL, "");
-               show_commit(commit);
-       }
-       for (i = 0; i < revs->pending.nr; i++) {
-               struct object_array_entry *pending = revs->pending.objects + i;
-               struct object *obj = pending->item;
-               const char *name = pending->name;
-               if (obj->flags & (UNINTERESTING | SEEN))
-                       continue;
-               if (obj->type == OBJ_TAG) {
-                       obj->flags |= SEEN;
-                       add_object_array(obj, name, &objects);
-                       continue;
-               }
-               if (obj->type == OBJ_TREE) {
-                       process_tree((struct tree *)obj, &objects, NULL, name);
-                       continue;
-               }
-               if (obj->type == OBJ_BLOB) {
-                       process_blob((struct blob *)obj, &objects, NULL, name);
-                       continue;
-               }
-               die("unknown pending object %s (%s)", sha1_to_hex(obj->sha1), name);
-       }
-       for (i = 0; i < objects.nr; i++) {
-               struct object_array_entry *p = objects.objects + i;
-
-               /* An object with name "foo\n0000000..." can be used to
-                * confuse downstream git-pack-objects very badly.
-                */
-               const char *ep = strchr(p->name, '\n');
-               if (ep) {
-                       printf("%s %.*s\n", sha1_to_hex(p->item->sha1),
-                              (int) (ep - p->name),
-                              p->name);
-               }
-               else
-                       printf("%s %s\n", sha1_to_hex(p->item->sha1), p->name);
-       }
+       printf("-%s\n", sha1_to_hex(commit->object.sha1));
 }
 
 /*
@@ -240,67 +170,290 @@ static void clear_distance(struct commit_list *list)
        }
 }
 
-static struct commit_list *find_bisection(struct commit_list *list)
+#define DEBUG_BISECT 0
+
+static inline int weight(struct commit_list *elem)
 {
-       int nr, closest;
-       struct commit_list *p, *best;
+       return *((int*)(elem->item->util));
+}
 
-       nr = 0;
-       p = list;
-       while (p) {
-               if (!revs.prune_fn || (p->item->object.flags & TREECHANGE))
-                       nr++;
-               p = p->next;
+static inline void weight_set(struct commit_list *elem, int weight)
+{
+       *((int*)(elem->item->util)) = weight;
+}
+
+static int count_interesting_parents(struct commit *commit)
+{
+       struct commit_list *p;
+       int count;
+
+       for (count = 0, p = commit->parents; p; p = p->next) {
+               if (p->item->object.flags & UNINTERESTING)
+                       continue;
+               count++;
        }
-       closest = 0;
-       best = list;
+       return count;
+}
+
+static inline int halfway(struct commit_list *p, int distance, int nr)
+{
+       /*
+        * Don't short-cut something we are not going to return!
+        */
+       if (revs.prune_fn && !(p->item->object.flags & TREECHANGE))
+               return 0;
+       if (DEBUG_BISECT)
+               return 0;
+       /*
+        * 2 and 3 are halfway of 5.
+        * 3 is halfway of 6 but 2 and 4 are not.
+        */
+       distance *= 2;
+       switch (distance - nr) {
+       case -1: case 0: case 1:
+               return 1;
+       default:
+               return 0;
+       }
+}
+
+#if !DEBUG_BISECT
+#define show_list(a,b,c,d) do { ; } while (0)
+#else
+static void show_list(const char *debug, int counted, int nr,
+                     struct commit_list *list)
+{
+       struct commit_list *p;
+
+       fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);
 
        for (p = list; p; p = p->next) {
-               int distance;
+               struct commit_list *pp;
+               struct commit *commit = p->item;
+               unsigned flags = commit->object.flags;
+               enum object_type type;
+               unsigned long size;
+               char *buf = read_sha1_file(commit->object.sha1, &type, &size);
+               char *ep, *sp;
+
+               fprintf(stderr, "%c%c%c ",
+                       (flags & TREECHANGE) ? 'T' : ' ',
+                       (flags & UNINTERESTING) ? 'U' : ' ',
+                       (flags & COUNTED) ? 'C' : ' ');
+               if (commit->util)
+                       fprintf(stderr, "%3d", weight(p));
+               else
+                       fprintf(stderr, "---");
+               fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
+               for (pp = commit->parents; pp; pp = pp->next)
+                       fprintf(stderr, " %.*s", 8,
+                               sha1_to_hex(pp->item->object.sha1));
+
+               sp = strstr(buf, "\n\n");
+               if (sp) {
+                       sp += 2;
+                       for (ep = sp; *ep && *ep != '\n'; ep++)
+                               ;
+                       fprintf(stderr, " %.*s", (int)(ep - sp), sp);
+               }
+               fprintf(stderr, "\n");
+       }
+}
+#endif /* DEBUG_BISECT */
 
-               if (revs.prune_fn && !(p->item->object.flags & TREECHANGE))
+/*
+ * zero or positive weight is the number of interesting commits it can
+ * reach, including itself.  Especially, weight = 0 means it does not
+ * reach any tree-changing commits (e.g. just above uninteresting one
+ * but traversal is with pathspec).
+ *
+ * weight = -1 means it has one parent and its distance is yet to
+ * be computed.
+ *
+ * weight = -2 means it has more than one parent and its distance is
+ * unknown.  After running count_distance() first, they will get zero
+ * or positive distance.
+ */
+
+static struct commit_list *find_bisection(struct commit_list *list,
+                                         int *reaches, int *all)
+{
+       int n, nr, on_list, counted, distance;
+       struct commit_list *p, *best, *next, *last;
+       int *weights;
+
+       show_list("bisection 2 entry", 0, 0, list);
+
+       /*
+        * Count the number of total and tree-changing items on the
+        * list, while reversing the list.
+        */
+       for (nr = on_list = 0, last = NULL, p = list;
+            p;
+            p = next) {
+               unsigned flags = p->item->object.flags;
+
+               next = p->next;
+               if (flags & UNINTERESTING)
                        continue;
+               p->next = last;
+               last = p;
+               if (!revs.prune_fn || (flags & TREECHANGE))
+                       nr++;
+               on_list++;
+       }
+       list = last;
+       show_list("bisection 2 sorted", 0, nr, list);
+
+       *all = nr;
+       weights = xcalloc(on_list, sizeof(int*));
+       counted = 0;
+
+       for (n = 0, p = list; p; p = p->next) {
+               struct commit *commit = p->item;
+               unsigned flags = commit->object.flags;
+
+               p->item->util = &weights[n++];
+               switch (count_interesting_parents(commit)) {
+               case 0:
+                       if (!revs.prune_fn || (flags & TREECHANGE)) {
+                               weight_set(p, 1);
+                               counted++;
+                               show_list("bisection 2 count one",
+                                         counted, nr, list);
+                       }
+                       /*
+                        * otherwise, it is known not to reach any
+                        * tree-changing commit and gets weight 0.
+                        */
+                       break;
+               case 1:
+                       weight_set(p, -1);
+                       break;
+               default:
+                       weight_set(p, -2);
+                       break;
+               }
+       }
 
+       show_list("bisection 2 initialize", counted, nr, list);
+
+       /*
+        * If you have only one parent in the resulting set
+        * then you can reach one commit more than that parent
+        * can reach.  So we do not have to run the expensive
+        * count_distance() for single strand of pearls.
+        *
+        * However, if you have more than one parents, you cannot
+        * just add their distance and one for yourself, since
+        * they usually reach the same ancestor and you would
+        * end up counting them twice that way.
+        *
+        * So we will first count distance of merges the usual
+        * way, and then fill the blanks using cheaper algorithm.
+        */
+       for (p = list; p; p = p->next) {
+               if (p->item->object.flags & UNINTERESTING)
+                       continue;
+               n = weight(p);
+               if (n != -2)
+                       continue;
                distance = count_distance(p);
                clear_distance(list);
+               weight_set(p, distance);
+
+               /* Does it happen to be at exactly half-way? */
+               if (halfway(p, distance, nr)) {
+                       p->next = NULL;
+                       *reaches = distance;
+                       free(weights);
+                       return p;
+               }
+               counted++;
+       }
+
+       show_list("bisection 2 count_distance", counted, nr, list);
+
+       while (counted < nr) {
+               for (p = list; p; p = p->next) {
+                       struct commit_list *q;
+                       unsigned flags = p->item->object.flags;
+
+                       if (0 <= weight(p))
+                               continue;
+                       for (q = p->item->parents; q; q = q->next) {
+                               if (q->item->object.flags & UNINTERESTING)
+                                       continue;
+                               if (0 <= weight(q))
+                                       break;
+                       }
+                       if (!q)
+                               continue;
+
+                       /*
+                        * weight for p is unknown but q is known.
+                        * add one for p itself if p is to be counted,
+                        * otherwise inherit it from q directly.
+                        */
+                       if (!revs.prune_fn || (flags & TREECHANGE)) {
+                               weight_set(p, weight(q)+1);
+                               counted++;
+                               show_list("bisection 2 count one",
+                                         counted, nr, list);
+                       }
+                       else
+                               weight_set(p, weight(q));
+
+                       /* Does it happen to be at exactly half-way? */
+                       distance = weight(p);
+                       if (halfway(p, distance, nr)) {
+                               p->next = NULL;
+                               *reaches = distance;
+                               free(weights);
+                               return p;
+                       }
+               }
+       }
+
+       show_list("bisection 2 counted all", counted, nr, list);
+
+       /* Then find the best one */
+       counted = -1;
+       best = list;
+       for (p = list; p; p = p->next) {
+               unsigned flags = p->item->object.flags;
+
+               if (revs.prune_fn && !(flags & TREECHANGE))
+                       continue;
+               distance = weight(p);
                if (nr - distance < distance)
                        distance = nr - distance;
-               if (distance > closest) {
+               if (distance > counted) {
                        best = p;
-                       closest = distance;
+                       counted = distance;
+                       *reaches = weight(p);
                }
        }
        if (best)
                best->next = NULL;
+       free(weights);
        return best;
 }
 
-static void mark_edge_parents_uninteresting(struct commit *commit)
+static void read_revisions_from_stdin(struct rev_info *revs)
 {
-       struct commit_list *parents;
-
-       for (parents = commit->parents; parents; parents = parents->next) {
-               struct commit *parent = parents->item;
-               if (!(parent->object.flags & UNINTERESTING))
-                       continue;
-               mark_tree_uninteresting(parent->tree);
-               if (revs.edge_hint && !(parent->object.flags & SHOWN)) {
-                       parent->object.flags |= SHOWN;
-                       printf("-%s\n", sha1_to_hex(parent->object.sha1));
-               }
-       }
-}
-
-static void mark_edges_uninteresting(struct commit_list *list)
-{
-       for ( ; list; list = list->next) {
-               struct commit *commit = list->item;
+       char line[1000];
 
-               if (commit->object.flags & UNINTERESTING) {
-                       mark_tree_uninteresting(commit->tree);
-                       continue;
-               }
-               mark_edge_parents_uninteresting(commit);
+       while (fgets(line, sizeof(line), stdin) != NULL) {
+               int len = strlen(line);
+               if (line[len - 1] == '\n')
+                       line[--len] = 0;
+               if (!len)
+                       break;
+               if (line[0] == '-')
+                       die("options not supported in --stdin mode");
+               if (handle_revision_arg(line, revs, 0, 1))
+                       die("bad revision '%s'", line);
        }
 }
 
@@ -308,7 +461,10 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
 {
        struct commit_list *list;
        int i;
+       int read_from_stdin = 0;
+       int bisect_show_vars = 0;
 
+       git_config(git_default_config);
        init_revisions(&revs, prefix);
        revs.abbrev = 0;
        revs.commit_format = CMIT_FMT_UNSPECIFIED;
@@ -329,6 +485,17 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
                        bisect_list = 1;
                        continue;
                }
+               if (!strcmp(arg, "--bisect-vars")) {
+                       bisect_list = 1;
+                       bisect_show_vars = 1;
+                       continue;
+               }
+               if (!strcmp(arg, "--stdin")) {
+                       if (read_from_stdin++)
+                               die("--stdin given twice?");
+                       read_revisions_from_stdin(&revs);
+                       continue;
+               }
                usage(rev_list_usage);
 
        }
@@ -352,19 +519,50 @@ int cmd_rev_list(int argc, const char **argv, const char *prefix)
            revs.diff)
                usage(rev_list_usage);
 
-       save_commit_buffer = revs.verbose_header;
+       save_commit_buffer = revs.verbose_header || revs.grep_filter;
        track_object_refs = 0;
        if (bisect_list)
                revs.limited = 1;
 
        prepare_revision_walk(&revs);
        if (revs.tree_objects)
-               mark_edges_uninteresting(revs.commits);
-
-       if (bisect_list)
-               revs.commits = find_bisection(revs.commits);
+               mark_edges_uninteresting(revs.commits, &revs, show_edge);
+
+       if (bisect_list) {
+               int reaches = reaches, all = all;
+
+               revs.commits = find_bisection(revs.commits, &reaches, &all);
+               if (bisect_show_vars) {
+                       int cnt;
+                       if (!revs.commits)
+                               return 1;
+                       /*
+                        * revs.commits can reach "reaches" commits among
+                        * "all" commits.  If it is good, then there are
+                        * (all-reaches) commits left to be bisected.
+                        * On the other hand, if it is bad, then the set
+                        * to bisect is "reaches".
+                        * A bisect set of size N has (N-1) commits further
+                        * to test, as we already know one bad one.
+                        */
+                       cnt = all-reaches;
+                       if (cnt < reaches)
+                               cnt = reaches;
+                       printf("bisect_rev=%s\n"
+                              "bisect_nr=%d\n"
+                              "bisect_good=%d\n"
+                              "bisect_bad=%d\n"
+                              "bisect_all=%d\n",
+                              sha1_to_hex(revs.commits->item->object.sha1),
+                              cnt - 1,
+                              all - reaches - 1,
+                              reaches - 1,
+                              all);
+                       return 0;
+               }
+       }
 
-       show_commit_list(&revs);
+       traverse_commit_list(&revs, show_commit, show_object);
 
        return 0;
 }