combine-diff: better hunk splitting.
[gitweb.git] / rev-list.c
index c4a61513438b707c097815b7e0acbcea400f6d79..e00e6fc76df8236c925e904fa270a2293099d6cf 100644 (file)
@@ -1,28 +1,40 @@
 #include "cache.h"
+#include "refs.h"
 #include "tag.h"
 #include "commit.h"
 #include "tree.h"
 #include "blob.h"
 #include "epoch.h"
+#include "diff.h"
 
 #define SEEN           (1u << 0)
 #define INTERESTING    (1u << 1)
 #define COUNTED                (1u << 2)
 #define SHOWN          (1u << 3)
-#define DUPCHECK       (1u << 4)
+#define TREECHANGE     (1u << 4)
 
 static const char rev_list_usage[] =
-       "usage: git-rev-list [OPTION] commit-id <commit-id>\n"
-                     "  --max-count=nr\n"
-                     "  --max-age=epoch\n"
-                     "  --min-age=epoch\n"
-                     "  --bisect\n"
-                     "  --objects\n"
-                     "  --unpacked\n"
-                     "  --header\n"
-                     "  --pretty\n"
-                     "  --merge-order [ --show-breaks ]";
-
+"git-rev-list [OPTION] <commit-id>... [ -- paths... ]\n"
+"  limiting output:\n"
+"    --max-count=nr\n"
+"    --max-age=epoch\n"
+"    --min-age=epoch\n"
+"    --sparse\n"
+"    --no-merges\n"
+"    --all\n"
+"  ordering output:\n"
+"    --merge-order [ --show-breaks ]\n"
+"    --topo-order\n"
+"  formatting output:\n"
+"    --parents\n"
+"    --objects\n"
+"    --unpacked\n"
+"    --header | --pretty\n"
+"  special purpose:\n"
+"    --bisect"
+;
+
+static int dense = 1;
 static int unpacked = 0;
 static int bisect_list = 0;
 static int tag_objects = 0;
@@ -31,7 +43,7 @@ static int blob_objects = 0;
 static int verbose_header = 0;
 static int show_parents = 0;
 static int hdr_termination = 0;
-static const char *prefix = "";
+static const char *commit_prefix = "";
 static unsigned long max_age = -1;
 static unsigned long min_age = -1;
 static int max_count = -1;
@@ -40,19 +52,21 @@ static int merge_order = 0;
 static int show_breaks = 0;
 static int stop_traversal = 0;
 static int topo_order = 0;
+static int no_merges = 0;
+static const char **paths = NULL;
 
 static void show_commit(struct commit *commit)
 {
        commit->object.flags |= SHOWN;
        if (show_breaks) {
-               prefix = "| ";
+               commit_prefix = "| ";
                if (commit->object.flags & DISCONTINUITY) {
-                       prefix = "^ ";     
+                       commit_prefix = "^ ";     
                } else if (commit->object.flags & BOUNDARY) {
-                       prefix = "= ";
+                       commit_prefix = "= ";
                } 
         }                      
-       printf("%s%s", prefix, sha1_to_hex(commit->object.sha1));
+       printf("%s%s", commit_prefix, sha1_to_hex(commit->object.sha1));
        if (show_parents) {
                struct commit_list *parents = commit->parents;
                while (parents) {
@@ -60,7 +74,11 @@ static void show_commit(struct commit *commit)
                        parents = parents->next;
                }
        }
-       putchar('\n');
+       if (commit_format == CMIT_FMT_ONELINE)
+               putchar(' ');
+       else
+               putchar('\n');
+
        if (verbose_header) {
                static char pretty_header[16384];
                pretty_print_commit(commit_format, commit->buffer, ~0, pretty_header, sizeof(pretty_header));
@@ -69,6 +87,31 @@ static void show_commit(struct commit *commit)
        fflush(stdout);
 }
 
+static int rewrite_one(struct commit **pp)
+{
+       for (;;) {
+               struct commit *p = *pp;
+               if (p->object.flags & (TREECHANGE | UNINTERESTING))
+                       return 0;
+               if (!p->parents)
+                       return -1;
+               *pp = p->parents->item;
+       }
+}
+
+static void rewrite_parents(struct commit *commit)
+{
+       struct commit_list **pp = &commit->parents;
+       while (*pp) {
+               struct commit_list *parent = *pp;
+               if (rewrite_one(&parent->item) < 0) {
+                       *pp = parent->next;
+                       continue;
+               }
+               pp = &parent->next;
+       }
+}
+
 static int filter_commit(struct commit * commit)
 {
        if (stop_traversal && (commit->object.flags & BOUNDARY))
@@ -79,10 +122,15 @@ static int filter_commit(struct commit * commit)
                return CONTINUE;
        if (max_age != -1 && (commit->date < max_age)) {
                stop_traversal=1;
-               return merge_order?CONTINUE:STOP;
+               return CONTINUE;
+       }
+       if (no_merges && (commit->parents && commit->parents->next))
+               return CONTINUE;
+       if (paths && dense) {
+               if (!(commit->object.flags & TREECHANGE))
+                       return CONTINUE;
+               rewrite_parents(commit);
        }
-       if (max_count != -1 && !max_count--)
-               return STOP;
        return DO;
 }
 
@@ -98,6 +146,9 @@ static int process_commit(struct commit * commit)
                return CONTINUE;
        }
 
+       if (max_count != -1 && !max_count--)
+               return STOP;
+
        show_commit(commit);
 
        return CONTINUE;
@@ -138,11 +189,16 @@ static struct object_list **process_tree(struct tree *tree, struct object_list *
                die("bad tree object %s", sha1_to_hex(obj->sha1));
        obj->flags |= SEEN;
        p = add_object(obj, p, name);
-       for (entry = tree->entries ; entry ; entry = entry->next) {
+       entry = tree->entries;
+       tree->entries = NULL;
+       while (entry) {
+               struct tree_entry_list *next = entry->next;
                if (entry->directory)
                        p = process_tree(entry->item.tree, p, entry->name);
                else
                        p = process_blob(entry->item.blob, p, entry->name);
+               free(entry);
+               entry = next;
        }
        return p;
 }
@@ -180,7 +236,17 @@ static void show_commit_list(struct commit_list *list)
                die("unknown pending object %s (%s)", sha1_to_hex(obj->sha1), name);
        }
        while (objects) {
-               printf("%s %s\n", sha1_to_hex(objects->item->sha1), objects->name);
+               /* An object with name "foo\n0000000000000000000000000000000000000000"
+                * can be used confuse downstream git-pack-objects very badly.
+                */
+               const char *ep = strchr(objects->name, '\n');
+               if (ep) {
+                       printf("%s %.*s\n", sha1_to_hex(objects->item->sha1),
+                              (int) (ep - objects->name),
+                              objects->name);
+               }
+               else
+                       printf("%s %s\n", sha1_to_hex(objects->item->sha1), objects->name);
                objects = objects->next;
        }
 }
@@ -204,15 +270,20 @@ static void mark_tree_uninteresting(struct tree *tree)
        if (obj->flags & UNINTERESTING)
                return;
        obj->flags |= UNINTERESTING;
+       if (!has_sha1_file(obj->sha1))
+               return;
        if (parse_tree(tree) < 0)
                die("bad tree %s", sha1_to_hex(obj->sha1));
        entry = tree->entries;
+       tree->entries = NULL;
        while (entry) {
+               struct tree_entry_list *next = entry->next;
                if (entry->directory)
                        mark_tree_uninteresting(entry->item.tree);
                else
                        mark_blob_uninteresting(entry->item.blob);
-               entry = entry->next;
+               free(entry);
+               entry = next;
        }
 }
 
@@ -220,17 +291,38 @@ static void mark_parents_uninteresting(struct commit *commit)
 {
        struct commit_list *parents = commit->parents;
 
-       if (tree_objects)
-               mark_tree_uninteresting(commit->tree);
        while (parents) {
                struct commit *commit = parents->item;
                commit->object.flags |= UNINTERESTING;
+
+               /*
+                * Normally we haven't parsed the parent
+                * yet, so we won't have a parent of a parent
+                * here. However, it may turn out that we've
+                * reached this commit some other way (where it
+                * wasn't uninteresting), in which case we need
+                * to mark its parents recursively too..
+                */
+               if (commit->parents)
+                       mark_parents_uninteresting(commit);
+
+               /*
+                * A missing commit is ok iff its parent is marked 
+                * uninteresting.
+                *
+                * We just mark such a thing parsed, so that when
+                * it is popped next time around, we won't be trying
+                * to parse it and get an error.
+                */
+               if (!has_sha1_file(commit->object.sha1))
+                       commit->object.parsed = 1;
                parents = parents->next;
        }
 }
 
-static int everybody_uninteresting(struct commit_list *list)
+static int everybody_uninteresting(struct commit_list *orig)
 {
+       struct commit_list *list = orig;
        while (list) {
                struct commit *commit = list->item;
                list = list->next;
@@ -258,7 +350,8 @@ static int count_distance(struct commit_list *entry)
 
                if (commit->object.flags & (UNINTERESTING | COUNTED))
                        break;
-               nr++;
+               if (!paths || (commit->object.flags & TREECHANGE))
+                       nr++;
                commit->object.flags |= COUNTED;
                p = commit->parents;
                entry = p;
@@ -270,6 +363,7 @@ static int count_distance(struct commit_list *entry)
                        }
                }
        }
+
        return nr;
 }
 
@@ -290,15 +384,20 @@ static struct commit_list *find_bisection(struct commit_list *list)
        nr = 0;
        p = list;
        while (p) {
-               nr++;
+               if (!paths || (p->item->object.flags & TREECHANGE))
+                       nr++;
                p = p->next;
        }
        closest = 0;
        best = list;
 
-       p = list;
-       while (p) {
-               int distance = count_distance(p);
+       for (p = list; p; p = p->next) {
+               int distance;
+
+               if (paths && !(p->item->object.flags & TREECHANGE))
+                       continue;
+
+               distance = count_distance(p);
                clear_distance(list);
                if (nr - distance < distance)
                        distance = nr - distance;
@@ -306,31 +405,217 @@ static struct commit_list *find_bisection(struct commit_list *list)
                        best = p;
                        closest = distance;
                }
-               p = p->next;
        }
        if (best)
                best->next = NULL;
        return best;
 }
 
+static void mark_edges_uninteresting(struct commit_list *list)
+{
+       for ( ; list; list = list->next) {
+               struct commit_list *parents = list->item->parents;
+
+               for ( ; parents; parents = parents->next) {
+                       struct commit *commit = parents->item;
+                       if (commit->object.flags & UNINTERESTING)
+                               mark_tree_uninteresting(commit->tree);
+               }
+       }
+}
+
+static int is_different = 0;
+
+static void file_add_remove(struct diff_options *options,
+                   int addremove, unsigned mode,
+                   const unsigned char *sha1,
+                   const char *base, const char *path)
+{
+       is_different = 1;
+}
+
+static void file_change(struct diff_options *options,
+                unsigned old_mode, unsigned new_mode,
+                const unsigned char *old_sha1,
+                const unsigned char *new_sha1,
+                const char *base, const char *path)
+{
+       is_different = 1;
+}
+
+static struct diff_options diff_opt = {
+       .recursive = 1,
+       .add_remove = file_add_remove,
+       .change = file_change,
+};
+
+static int same_tree(struct tree *t1, struct tree *t2)
+{
+       is_different = 0;
+       if (diff_tree_sha1(t1->object.sha1, t2->object.sha1, "", &diff_opt) < 0)
+               return 0;
+       return !is_different;
+}
+
+static int same_tree_as_empty(struct tree *t1)
+{
+       int retval;
+       void *tree;
+       struct tree_desc empty, real;
+
+       if (!t1)
+               return 0;
+
+       tree = read_object_with_reference(t1->object.sha1, "tree", &real.size, NULL);
+       if (!tree)
+               return 0;
+       real.buf = tree;
+
+       empty.buf = "";
+       empty.size = 0;
+
+       is_different = 0;
+       retval = diff_tree(&empty, &real, "", &diff_opt);
+       free(tree);
+
+       return retval >= 0 && !is_different;
+}
+
+static struct commit *try_to_simplify_merge(struct commit *commit, struct commit_list *parent)
+{
+       if (!commit->tree)
+               return NULL;
+
+       while (parent) {
+               struct commit *p = parent->item;
+               parent = parent->next;
+               parse_commit(p);
+               if (!p->tree)
+                       continue;
+               if (same_tree(commit->tree, p->tree))
+                       return p;
+       }
+       return NULL;
+}
+
+static void add_parents_to_list(struct commit *commit, struct commit_list **list)
+{
+       struct commit_list *parent = commit->parents;
+
+       /*
+        * If the commit is uninteresting, don't try to
+        * prune parents - we want the maximal uninteresting
+        * set.
+        *
+        * Normally we haven't parsed the parent
+        * yet, so we won't have a parent of a parent
+        * here. However, it may turn out that we've
+        * reached this commit some other way (where it
+        * wasn't uninteresting), in which case we need
+        * to mark its parents recursively too..
+        */
+       if (commit->object.flags & UNINTERESTING) {
+               while (parent) {
+                       struct commit *p = parent->item;
+                       parent = parent->next;
+                       parse_commit(p);
+                       p->object.flags |= UNINTERESTING;
+                       if (p->parents)
+                               mark_parents_uninteresting(p);
+                       if (p->object.flags & SEEN)
+                               continue;
+                       p->object.flags |= SEEN;
+                       insert_by_date(p, list);
+               }
+               return;
+       }
+
+       /*
+        * Ok, the commit wasn't uninteresting. If it
+        * is a merge, try to find the parent that has
+        * no differences in the path set if one exists.
+        */
+       if (paths && parent && parent->next) {
+               struct commit *preferred;
+
+               preferred = try_to_simplify_merge(commit, parent);
+               if (preferred) {
+                       parent->item = preferred;
+                       parent->next = NULL;
+               }
+       }
+
+       while (parent) {
+               struct commit *p = parent->item;
+
+               parent = parent->next;
+
+               parse_commit(p);
+               if (p->object.flags & SEEN)
+                       continue;
+               p->object.flags |= SEEN;
+               insert_by_date(p, list);
+       }
+}
+
+static void compress_list(struct commit_list *list)
+{
+       while (list) {
+               struct commit *commit = list->item;
+               struct commit_list *parent = commit->parents;
+               list = list->next;
+
+               if (!parent) {
+                       if (!same_tree_as_empty(commit->tree))
+                               commit->object.flags |= TREECHANGE;
+                       continue;
+               }
+
+               /*
+                * Exactly one parent? Check if it leaves the tree
+                * unchanged
+                */
+               if (!parent->next) {
+                       struct tree *t1 = commit->tree;
+                       struct tree *t2 = parent->item->tree;
+                       if (!t1 || !t2 || same_tree(t1, t2))
+                               continue;
+               }
+               commit->object.flags |= TREECHANGE;
+       }
+}
+
 static struct commit_list *limit_list(struct commit_list *list)
 {
        struct commit_list *newlist = NULL;
        struct commit_list **p = &newlist;
        while (list) {
-               struct commit *commit = pop_most_recent_commit(&list, SEEN);
+               struct commit_list *entry = list;
+               struct commit *commit = list->item;
                struct object *obj = &commit->object;
 
+               list = list->next;
+               free(entry);
+
+               if (max_age != -1 && (commit->date < max_age))
+                       obj->flags |= UNINTERESTING;
                if (unpacked && has_sha1_pack(obj->sha1))
                        obj->flags |= UNINTERESTING;
+               add_parents_to_list(commit, &list);
                if (obj->flags & UNINTERESTING) {
                        mark_parents_uninteresting(commit);
                        if (everybody_uninteresting(list))
                                break;
                        continue;
                }
+               if (min_age != -1 && (commit->date > min_age))
+                       continue;
                p = &commit_list_insert(commit, p)->next;
        }
+       if (tree_objects)
+               mark_edges_uninteresting(newlist);
+       if (paths && dense)
+               compress_list(newlist);
        if (bisect_list)
                newlist = find_bisection(newlist);
        return newlist;
@@ -341,13 +626,10 @@ static void add_pending_object(struct object *obj, const char *name)
        add_object(obj, &pending_objects, name);
 }
 
-static struct commit *get_commit_reference(const char *name, unsigned int flags)
+static struct commit *get_commit_reference(const char *name, const unsigned char *sha1, unsigned int flags)
 {
-       unsigned char sha1[20];
        struct object *object;
 
-       if (get_sha1(name, sha1))
-               usage(rev_list_usage);
        object = parse_object(sha1);
        if (!object)
                die("bad object %s", name);
@@ -355,12 +637,14 @@ static struct commit *get_commit_reference(const char *name, unsigned int flags)
        /*
         * Tag object? Look what it points to..
         */
-       if (object->type == tag_type) {
+       while (object->type == tag_type) {
                struct tag *tag = (struct tag *) object;
                object->flags |= flags;
                if (tag_objects && !(object->flags & UNINTERESTING))
                        add_pending_object(object, tag->tag);
-               object = tag->tagged;
+               object = parse_object(tag->tagged->sha1);
+               if (!object)
+                       die("bad object %s", sha1_to_hex(tag->tagged->sha1));
        }
 
        /*
@@ -372,6 +656,8 @@ static struct commit *get_commit_reference(const char *name, unsigned int flags)
                object->flags |= flags;
                if (parse_commit(commit) < 0)
                        die("unable to parse commit %s", name);
+               if (flags & UNINTERESTING)
+                       mark_parents_uninteresting(commit);
                return commit;
        }
 
@@ -408,17 +694,43 @@ static struct commit *get_commit_reference(const char *name, unsigned int flags)
        die("%s is unknown object", name);
 }
 
-int main(int argc, char **argv)
+static void handle_one_commit(struct commit *com, struct commit_list **lst)
+{
+       if (!com || com->object.flags & SEEN)
+               return;
+       com->object.flags |= SEEN;
+       commit_list_insert(com, lst);
+}
+
+/* for_each_ref() callback does not allow user data -- Yuck. */
+static struct commit_list **global_lst;
+
+static int include_one_commit(const char *path, const unsigned char *sha1)
+{
+       struct commit *com = get_commit_reference(path, sha1, 0);
+       handle_one_commit(com, global_lst);
+       return 0;
+}
+
+static void handle_all(struct commit_list **lst)
 {
+       global_lst = lst;
+       for_each_ref(include_one_commit);
+       global_lst = NULL;
+}
+
+int main(int argc, const char **argv)
+{
+       const char *prefix = setup_git_directory();
        struct commit_list *list = NULL;
-       struct commit_list *(*insert)(struct commit *, struct commit_list **);
        int i, limited = 0;
 
-       insert = insert_by_date;
        for (i = 1 ; i < argc; i++) {
                int flags;
-               char *arg = argv[i];
+               const char *arg = argv[i];
+               char *dotdot;
                struct commit *commit;
+               unsigned char sha1[20];
 
                if (!strncmp(arg, "--max-count=", 12)) {
                        max_count = atoi(arg + 12);
@@ -426,10 +738,12 @@ int main(int argc, char **argv)
                }
                if (!strncmp(arg, "--max-age=", 10)) {
                        max_age = atoi(arg + 10);
+                       limited = 1;
                        continue;
                }
                if (!strncmp(arg, "--min-age=", 10)) {
                        min_age = atoi(arg + 10);
+                       limited = 1;
                        continue;
                }
                if (!strcmp(arg, "--header")) {
@@ -440,7 +754,14 @@ int main(int argc, char **argv)
                        commit_format = get_commit_format(arg+8);
                        verbose_header = 1;
                        hdr_termination = '\n';
-                       prefix = "commit ";
+                       if (commit_format == CMIT_FMT_ONELINE)
+                               commit_prefix = "";
+                       else
+                               commit_prefix = "commit ";
+                       continue;
+               }
+               if (!strncmp(arg, "--no-merges", 11)) {
+                       no_merges = 1;
                        continue;
                }
                if (!strcmp(arg, "--parents")) {
@@ -451,6 +772,10 @@ int main(int argc, char **argv)
                        bisect_list = 1;
                        continue;
                }
+               if (!strcmp(arg, "--all")) {
+                       handle_all(&list);
+                       continue;
+               }
                if (!strcmp(arg, "--objects")) {
                        tag_objects = 1;
                        tree_objects = 1;
@@ -464,7 +789,6 @@ int main(int argc, char **argv)
                }
                if (!strcmp(arg, "--merge-order")) {
                        merge_order = 1;
-                       insert = commit_list_insert;
                        continue;
                }
                if (!strcmp(arg, "--show-breaks")) {
@@ -473,36 +797,96 @@ int main(int argc, char **argv)
                }
                if (!strcmp(arg, "--topo-order")) {
                        topo_order = 1;
+                       limited = 1;
+                       continue;
+               }
+               if (!strcmp(arg, "--dense")) {
+                       dense = 1;
+                       continue;
+               }
+               if (!strcmp(arg, "--sparse")) {
+                       dense = 0;
                        continue;
                }
+               if (!strcmp(arg, "--")) {
+                       i++;
+                       break;
+               }
+
+               if (show_breaks && !merge_order)
+                       usage(rev_list_usage);
 
                flags = 0;
+               dotdot = strstr(arg, "..");
+               if (dotdot) {
+                       unsigned char from_sha1[20];
+                       char *next = dotdot + 2;
+                       *dotdot = 0;
+                       if (!*next)
+                               next = "HEAD";
+                       if (!get_sha1(arg, from_sha1) && !get_sha1(next, sha1)) {
+                               struct commit *exclude;
+                               struct commit *include;
+                               
+                               exclude = get_commit_reference(arg, from_sha1, UNINTERESTING);
+                               include = get_commit_reference(next, sha1, 0);
+                               if (!exclude || !include)
+                                       die("Invalid revision range %s..%s", arg, next);
+                               limited = 1;
+                               handle_one_commit(exclude, &list);
+                               handle_one_commit(include, &list);
+                               continue;
+                       }
+                       *dotdot = '.';
+               }
                if (*arg == '^') {
                        flags = UNINTERESTING;
                        arg++;
                        limited = 1;
                }
-               if (show_breaks && !merge_order)
-                       usage(rev_list_usage);
-               commit = get_commit_reference(arg, flags);
-               if (!commit)
-                       continue;
-               if (commit->object.flags & DUPCHECK)
-                       continue;
-               commit->object.flags |= DUPCHECK;
-               insert(commit, &list);
+               if (get_sha1(arg, sha1) < 0) {
+                       struct stat st;
+                       if (lstat(arg, &st) < 0)
+                               die("'%s': %s", arg, strerror(errno));
+                       break;
+               }
+               commit = get_commit_reference(arg, sha1, flags);
+               handle_one_commit(commit, &list);
        }
 
+       if (!list &&
+           (!(tag_objects||tree_objects||blob_objects) && !pending_objects))
+               usage(rev_list_usage);
+
+       paths = get_pathspec(prefix, argv + i);
+       if (paths) {
+               limited = 1;
+               diff_tree_setup_paths(paths);
+       }
+
+       save_commit_buffer = verbose_header;
+       track_object_refs = 0;
+
        if (!merge_order) {             
+               sort_by_date(&list);
+               if (list && !limited && max_count == 1 &&
+                   !tag_objects && !tree_objects && !blob_objects) {
+                       show_commit(list->item);
+                       return 0;
+               }
                if (limited)
                        list = limit_list(list);
                if (topo_order)
                        sort_in_topological_order(&list);
                show_commit_list(list);
        } else {
+#ifndef NO_OPENSSL
                if (sort_list_in_merge_order(list, &process_commit)) {
-                         die("merge order sort failed\n");
+                       die("merge order sort failed\n");
                }
+#else
+               die("merge order sort unsupported, OpenSSL not linked");
+#endif
        }
 
        return 0;