toposort: rename "lifo" field
[gitweb.git] / commit.c
index 47027f68ec8dfa9652cdb6d7bbac465f901b85be..11b9635df27ae2d6379cf70f5737c33d9d49e701 100644 (file)
--- a/commit.c
+++ b/commit.c
@@ -7,10 +7,15 @@
 #include "revision.h"
 #include "notes.h"
 #include "gpg-interface.h"
+#include "mergesort.h"
+#include "commit-slab.h"
+
+static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
 
 int save_commit_buffer = 1;
 
 const char *commit_type = "commit";
+static int commit_count;
 
 static struct commit *check_commit(struct object *obj,
                                   const unsigned char *sha1,
@@ -55,8 +60,11 @@ struct commit *lookup_commit_or_die(const unsigned char *sha1, const char *ref_n
 struct commit *lookup_commit(const unsigned char *sha1)
 {
        struct object *obj = lookup_object(sha1);
-       if (!obj)
-               return create_object(sha1, OBJ_COMMIT, alloc_commit_node());
+       if (!obj) {
+               struct commit *c = alloc_commit_node();
+               c->index = commit_count++;
+               return create_object(sha1, OBJ_COMMIT, c);
+       }
        if (!obj->type)
                obj->type = OBJ_COMMIT;
        return check_commit(obj, sha1, 0);
@@ -67,7 +75,7 @@ struct commit *lookup_commit_reference_by_name(const char *name)
        unsigned char sha1[20];
        struct commit *commit;
 
-       if (get_sha1(name, sha1))
+       if (get_sha1_committish(name, sha1))
                return NULL;
        commit = lookup_commit_reference(sha1);
        if (!commit || parse_commit(commit))
@@ -390,15 +398,31 @@ struct commit_list * commit_list_insert_by_date(struct commit *item, struct comm
        return commit_list_insert(item, pp);
 }
 
+static int commit_list_compare_by_date(const void *a, const void *b)
+{
+       unsigned long a_date = ((const struct commit_list *)a)->item->date;
+       unsigned long b_date = ((const struct commit_list *)b)->item->date;
+       if (a_date < b_date)
+               return 1;
+       if (a_date > b_date)
+               return -1;
+       return 0;
+}
+
+static void *commit_list_get_next(const void *a)
+{
+       return ((const struct commit_list *)a)->next;
+}
+
+static void commit_list_set_next(void *a, void *next)
+{
+       ((struct commit_list *)a)->next = next;
+}
 
 void commit_list_sort_by_date(struct commit_list **list)
 {
-       struct commit_list *ret = NULL;
-       while (*list) {
-               commit_list_insert_by_date((*list)->item, &ret);
-               *list = (*list)->next;
-       }
-       *list = ret;
+       *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
+                               commit_list_compare_by_date);
 }
 
 struct commit *pop_most_recent_commit(struct commit_list **list,
@@ -422,7 +446,8 @@ struct commit *pop_most_recent_commit(struct commit_list **list,
        return ret;
 }
 
-void clear_commit_marks(struct commit *commit, unsigned int mark)
+static void clear_commit_marks_1(struct commit_list **plist,
+                                struct commit *commit, unsigned int mark)
 {
        while (commit) {
                struct commit_list *parents;
@@ -437,12 +462,20 @@ void clear_commit_marks(struct commit *commit, unsigned int mark)
                        return;
 
                while ((parents = parents->next))
-                       clear_commit_marks(parents->item, mark);
+                       commit_list_insert(parents->item, plist);
 
                commit = commit->parents->item;
        }
 }
 
+void clear_commit_marks(struct commit *commit, unsigned int mark)
+{
+       struct commit_list *list = NULL;
+       commit_list_insert(commit, &list);
+       while (list)
+               clear_commit_marks_1(&list, pop_commit(&list), mark);
+}
+
 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 {
        struct object *object;
@@ -469,23 +502,33 @@ struct commit *pop_commit(struct commit_list **stack)
        return item;
 }
 
+/*
+ * Topological sort support
+ */
+
+/* count number of children that have not been emitted */
+define_commit_slab(indegree_slab, int);
+
 /*
  * Performs an in-place topological sort on the list supplied.
  */
-void sort_in_topological_order(struct commit_list ** list, int lifo)
+void sort_in_topological_order(struct commit_list ** list, enum rev_sort_order sort_order)
 {
        struct commit_list *next, *orig = *list;
        struct commit_list *work, **insert;
        struct commit_list **pptr;
+       struct indegree_slab indegree;
 
        if (!orig)
                return;
        *list = NULL;
 
+       init_indegree_slab(&indegree);
+
        /* Mark them and clear the indegree */
        for (next = orig; next; next = next->next) {
                struct commit *commit = next->item;
-               commit->indegree = 1;
+               *(indegree_slab_at(&indegree, commit)) = 1;
        }
 
        /* update the indegree */
@@ -493,9 +536,10 @@ void sort_in_topological_order(struct commit_list ** list, int lifo)
                struct commit_list * parents = next->item->parents;
                while (parents) {
                        struct commit *parent = parents->item;
+                       int *pi = indegree_slab_at(&indegree, parent);
 
-                       if (parent->indegree)
-                               parent->indegree++;
+                       if (*pi)
+                               (*pi)++;
                        parents = parents->next;
                }
        }
@@ -512,12 +556,12 @@ void sort_in_topological_order(struct commit_list ** list, int lifo)
        for (next = orig; next; next = next->next) {
                struct commit *commit = next->item;
 
-               if (commit->indegree == 1)
+               if (*(indegree_slab_at(&indegree, commit)) == 1)
                        insert = &commit_list_insert(commit, insert)->next;
        }
 
        /* process the list in topological order */
-       if (!lifo)
+       if (sort_order != REV_SORT_IN_GRAPH_ORDER)
                commit_list_sort_by_date(&work);
 
        pptr = list;
@@ -533,8 +577,9 @@ void sort_in_topological_order(struct commit_list ** list, int lifo)
                commit = work_item->item;
                for (parents = commit->parents; parents ; parents = parents->next) {
                        struct commit *parent = parents->item;
+                       int *pi = indegree_slab_at(&indegree, parent);
 
-                       if (!parent->indegree)
+                       if (!*pi)
                                continue;
 
                        /*
@@ -542,21 +587,27 @@ void sort_in_topological_order(struct commit_list ** list, int lifo)
                         * when all their children have been emitted thereby
                         * guaranteeing topological order.
                         */
-                       if (--parent->indegree == 1) {
-                               if (!lifo)
+                       if (--(*pi) == 1) {
+                               switch (sort_order) {
+                               case REV_SORT_BY_COMMIT_DATE:
                                        commit_list_insert_by_date(parent, &work);
-                               else
+                                       break;
+                               default: /* REV_SORT_IN_GRAPH_ORDER */
                                        commit_list_insert(parent, &work);
+                                       break;
+                               }
                        }
                }
                /*
                 * work_item is a commit all of whose children
                 * have already been emitted. we can emit it now.
                 */
-               commit->indegree = 0;
+               *(indegree_slab_at(&indegree, commit)) = 0;
                *pptr = work_item;
                pptr = &work_item->next;
        }
+
+       clear_indegree_slab(&indegree);
 }
 
 /* merge-base stuff */
@@ -806,7 +857,7 @@ struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
 }
 
 /*
- * Is "commit" a decendant of one of the elements on the "with_commit" list?
+ * Is "commit" a descendant of one of the elements on the "with_commit" list?
  */
 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 {
@@ -1052,8 +1103,9 @@ static int excluded_header_field(const char *field, size_t len, const char **exc
        return 0;
 }
 
-struct commit_extra_header *read_commit_extra_header_lines(const char *buffer, size_t size,
-                                                          const char **exclude)
+static struct commit_extra_header *read_commit_extra_header_lines(
+       const char *buffer, size_t size,
+       const char **exclude)
 {
        struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
        const char *line, *next, *eof, *eob;
@@ -1120,8 +1172,92 @@ int commit_tree(const struct strbuf *msg, unsigned char *tree,
        return result;
 }
 
+static int find_invalid_utf8(const char *buf, int len)
+{
+       int offset = 0;
+
+       while (len) {
+               unsigned char c = *buf++;
+               int bytes, bad_offset;
+
+               len--;
+               offset++;
+
+               /* Simple US-ASCII? No worries. */
+               if (c < 0x80)
+                       continue;
+
+               bad_offset = offset-1;
+
+               /*
+                * Count how many more high bits set: that's how
+                * many more bytes this sequence should have.
+                */
+               bytes = 0;
+               while (c & 0x40) {
+                       c <<= 1;
+                       bytes++;
+               }
+
+               /* Must be between 1 and 5 more bytes */
+               if (bytes < 1 || bytes > 5)
+                       return bad_offset;
+
+               /* Do we *have* that many bytes? */
+               if (len < bytes)
+                       return bad_offset;
+
+               offset += bytes;
+               len -= bytes;
+
+               /* And verify that they are good continuation bytes */
+               do {
+                       if ((*buf++ & 0xc0) != 0x80)
+                               return bad_offset;
+               } while (--bytes);
+
+               /* We could/should check the value and length here too */
+       }
+       return -1;
+}
+
+/*
+ * This verifies that the buffer is in proper utf8 format.
+ *
+ * If it isn't, it assumes any non-utf8 characters are Latin1,
+ * and does the conversion.
+ *
+ * Fixme: we should probably also disallow overlong forms and
+ * invalid characters. But we don't do that currently.
+ */
+static int verify_utf8(struct strbuf *buf)
+{
+       int ok = 1;
+       long pos = 0;
+
+       for (;;) {
+               int bad;
+               unsigned char c;
+               unsigned char replace[2];
+
+               bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
+               if (bad < 0)
+                       return ok;
+               pos += bad;
+               ok = 0;
+               c = buf->buf[pos];
+               strbuf_remove(buf, pos, 1);
+
+               /* We know 'c' must be in the range 128-255 */
+               replace[0] = 0xc0 + (c >> 6);
+               replace[1] = 0x80 + (c & 0x3f);
+               strbuf_insert(buf, pos, replace, 2);
+               pos += 2;
+       }
+}
+
 static const char commit_utf8_warn[] =
-"Warning: commit message does not conform to UTF-8.\n"
+"Warning: commit message did not conform to UTF-8.\n"
 "You may want to amend it after fixing the message, or set the config\n"
 "variable i18n.commitencoding to the encoding your project uses.\n";
 
@@ -1162,9 +1298,9 @@ int commit_tree_extended(const struct strbuf *msg, unsigned char *tree,
 
        /* Person/date information */
        if (!author)
-               author = git_author_info(IDENT_ERROR_ON_NO_NAME);
+               author = git_author_info(IDENT_STRICT);
        strbuf_addf(&buffer, "author %s\n", author);
-       strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_ERROR_ON_NO_NAME));
+       strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
        if (!encoding_is_utf8)
                strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
 
@@ -1178,7 +1314,7 @@ int commit_tree_extended(const struct strbuf *msg, unsigned char *tree,
        strbuf_addbuf(&buffer, msg);
 
        /* And check the encoding */
-       if (encoding_is_utf8 && !is_utf8(buffer.buf))
+       if (encoding_is_utf8 && !verify_utf8(&buffer))
                fprintf(stderr, commit_utf8_warn);
 
        if (sign_commit && do_sign_commit(&buffer, sign_commit))
@@ -1207,3 +1343,40 @@ struct commit *get_merge_parent(const char *name)
        }
        return commit;
 }
+
+/*
+ * Append a commit to the end of the commit_list.
+ *
+ * next starts by pointing to the variable that holds the head of an
+ * empty commit_list, and is updated to point to the "next" field of
+ * the last item on the list as new commits are appended.
+ *
+ * Usage example:
+ *
+ *     struct commit_list *list;
+ *     struct commit_list **next = &list;
+ *
+ *     next = commit_list_append(c1, next);
+ *     next = commit_list_append(c2, next);
+ *     assert(commit_list_count(list) == 2);
+ *     return list;
+ */
+struct commit_list **commit_list_append(struct commit *commit,
+                                       struct commit_list **next)
+{
+       struct commit_list *new = xmalloc(sizeof(struct commit_list));
+       new->item = commit;
+       *next = new;
+       new->next = NULL;
+       return &new->next;
+}
+
+void print_commit_list(struct commit_list *list,
+                      const char *format_cur,
+                      const char *format_last)
+{
+       for ( ; list; list = list->next) {
+               const char *format = list->next ? format_cur : format_last;
+               printf(format, sha1_to_hex(list->item->object.sha1));
+       }
+}