Reduce the number of connects when fetching
[gitweb.git] / merge-tree.c
index 00ae4e8d3208e09f2cf7a38202a126f728cadb48..e08324686cc090fa9bd94d7f069b025454c7acdf 100644 (file)
 #include "cache.h"
+#include "tree-walk.h"
+#include "xdiff-interface.h"
+#include "blob.h"
 
-struct tree_entry {
-       unsigned mode;
-       unsigned char *sha1;
-       char *path;
-       struct tree_entry *next;
+static const char merge_tree_usage[] = "git-merge-tree <base-tree> <branch1> <branch2>";
+static int resolve_directories = 1;
+
+struct merge_list {
+       struct merge_list *next;
+       struct merge_list *link;        /* other stages for this object */
+
+       unsigned int stage : 2,
+                    flags : 30;
+       unsigned int mode;
+       const char *path;
+       struct blob *blob;
 };
 
-static struct tree_entry *read_tree(unsigned char *sha1)
+static struct merge_list *merge_result, **merge_result_end = &merge_result;
+
+static void add_merge_entry(struct merge_list *entry)
 {
-       char type[20];
-       unsigned long size;
-       void *buf = read_sha1_file(sha1, type, &size);
-       struct tree_entry *ret = NULL, **tp = &ret;
-
-       if (!buf || strcmp(type, "tree"))
-               die("unable to read 'tree' object %s", sha1_to_hex(sha1));
-       while (size) {
-               int len = strlen(buf)+1;
-               struct tree_entry * entry = malloc(sizeof(struct tree_entry));
-               if (size < len + 20 || sscanf(buf, "%o", &entry->mode) != 1)
-                       die("corrupt 'tree' object %s", sha1_to_hex(sha1));
-               entry->path = strchr(buf, ' ')+1;
-               entry->sha1 = buf + len;
-               entry->next = NULL;
-               *tp = entry;
-               tp = &entry->next;
-               len += 20;
-               buf += len;
-               size -= len;
+       *merge_result_end = entry;
+       merge_result_end = &entry->next;
+}
+
+static void merge_trees(struct tree_desc t[3], const char *base);
+
+static const char *explanation(struct merge_list *entry)
+{
+       switch (entry->stage) {
+       case 0:
+               return "merged";
+       case 3:
+               return "added in remote";
+       case 2:
+               if (entry->link)
+                       return "added in both";
+               return "added in local";
        }
-       return ret;
+
+       /* Existed in base */
+       entry = entry->link;
+       if (!entry)
+               return "removed in both";
+
+       if (entry->link)
+               return "changed in both";
+
+       if (entry->stage == 3)
+               return "removed in local";
+       return "removed in remote";
 }
 
-static void show(const struct tree_entry *a, const char *path)
+extern void *merge_file(struct blob *, struct blob *, struct blob *, unsigned long *);
+
+static void *result(struct merge_list *entry, unsigned long *size)
 {
-       printf("select %o %s %s%c", a->mode, sha1_to_hex(a->sha1), path, 0);
+       enum object_type type;
+       struct blob *base, *our, *their;
+
+       if (!entry->stage)
+               return read_sha1_file(entry->blob->object.sha1, &type, size);
+       base = NULL;
+       if (entry->stage == 1) {
+               base = entry->blob;
+               entry = entry->link;
+       }
+       our = NULL;
+       if (entry && entry->stage == 2) {
+               our = entry->blob;
+               entry = entry->link;
+       }
+       their = NULL;
+       if (entry)
+               their = entry->blob;
+       return merge_file(base, our, their, size);
 }
 
-static void merge(const struct tree_entry *a, const struct tree_entry *b, const struct tree_entry *c, const char *path)
+static void *origin(struct merge_list *entry, unsigned long *size)
 {
-       char hex_a[60], hex_b[60], hex_c[60];
-       strcpy(hex_a, sha1_to_hex(a->sha1));
-       strcpy(hex_b, sha1_to_hex(b->sha1));
-       strcpy(hex_c, sha1_to_hex(c->sha1));
-       printf("merge %o->%o,%o %s->%s,%s %s%c",
-               a->mode, b->mode, c->mode,
-               hex_a, hex_b, hex_c, path, 0);
+       enum object_type type;
+       while (entry) {
+               if (entry->stage == 2)
+                       return read_sha1_file(entry->blob->object.sha1, &type, size);
+               entry = entry->link;
+       }
+       return NULL;
 }
 
-static int same(const struct tree_entry *a, const struct tree_entry *b)
+static int show_outf(void *priv_, mmbuffer_t *mb, int nbuf)
 {
-       return a->mode == b->mode && !memcmp(a->sha1, b->sha1, 20);
+       int i;
+       for (i = 0; i < nbuf; i++)
+               printf("%.*s", (int) mb[i].size, mb[i].ptr);
+       return 0;
 }
 
-static void merge_entry(const struct tree_entry *src, const struct tree_entry *dst1, const struct tree_entry *dst2)
+static void show_diff(struct merge_list *entry)
 {
-       static unsigned char nullsha1[20];
-       static const struct tree_entry none = { 0, nullsha1, "", NULL };
-       const char *path = NULL;
-       const struct tree_entry *a, *b, *c;
+       unsigned long size;
+       mmfile_t src, dst;
+       xpparam_t xpp;
+       xdemitconf_t xecfg;
+       xdemitcb_t ecb;
 
-       a = &none;
-       b = &none;
-       c = &none;
-       if (src) { a = src; path = src->path; }
-       if (dst1) { b = dst1; path = dst1->path; }
-       if (dst2) { c = dst2; path = dst2->path; }
-       if (same(b, c)) {
-               show(b, path);
-               return;
+       xpp.flags = XDF_NEED_MINIMAL;
+       memset(&xecfg, 0, sizeof(xecfg));
+       xecfg.ctxlen = 3;
+       ecb.outf = show_outf;
+       ecb.priv = NULL;
+
+       src.ptr = origin(entry, &size);
+       if (!src.ptr)
+               size = 0;
+       src.size = size;
+       dst.ptr = result(entry, &size);
+       if (!dst.ptr)
+               size = 0;
+       dst.size = size;
+       xdi_diff(&src, &dst, &xpp, &xecfg, &ecb);
+       free(src.ptr);
+       free(dst.ptr);
+}
+
+static void show_result_list(struct merge_list *entry)
+{
+       printf("%s\n", explanation(entry));
+       do {
+               struct merge_list *link = entry->link;
+               static const char *desc[4] = { "result", "base", "our", "their" };
+               printf("  %-6s %o %s %s\n", desc[entry->stage], entry->mode, sha1_to_hex(entry->blob->object.sha1), entry->path);
+               entry = link;
+       } while (entry);
+}
+
+static void show_result(void)
+{
+       struct merge_list *walk;
+
+       walk = merge_result;
+       while (walk) {
+               show_result_list(walk);
+               show_diff(walk);
+               walk = walk->next;
        }
-       if (same(a, b)) {
-               show(c, path);
+}
+
+/* An empty entry never compares same, not even to another empty entry */
+static int same_entry(struct name_entry *a, struct name_entry *b)
+{
+       return  a->sha1 &&
+               b->sha1 &&
+               !hashcmp(a->sha1, b->sha1) &&
+               a->mode == b->mode;
+}
+
+static struct merge_list *create_entry(unsigned stage, unsigned mode, const unsigned char *sha1, const char *path)
+{
+       struct merge_list *res = xmalloc(sizeof(*res));
+
+       memset(res, 0, sizeof(*res));
+       res->stage = stage;
+       res->path = path;
+       res->mode = mode;
+       res->blob = lookup_blob(sha1);
+       return res;
+}
+
+static void resolve(const char *base, struct name_entry *branch1, struct name_entry *result)
+{
+       struct merge_list *orig, *final;
+       const char *path;
+
+       /* If it's already branch1, don't bother showing it */
+       if (!branch1)
                return;
+
+       path = xstrdup(mkpath("%s%s", base, result->path));
+       orig = create_entry(2, branch1->mode, branch1->sha1, path);
+       final = create_entry(0, result->mode, result->sha1, path);
+
+       final->link = orig;
+
+       add_merge_entry(final);
+}
+
+static int unresolved_directory(const char *base, struct name_entry n[3])
+{
+       int baselen, pathlen;
+       char *newbase;
+       struct name_entry *p;
+       struct tree_desc t[3];
+       void *buf0, *buf1, *buf2;
+
+       if (!resolve_directories)
+               return 0;
+       p = n;
+       if (!p->mode) {
+               p++;
+               if (!p->mode)
+                       p++;
        }
-       if (same(a, c)) {
-               show(b, path);
+       if (!S_ISDIR(p->mode))
+               return 0;
+       baselen = strlen(base);
+       pathlen = tree_entry_len(p->path, p->sha1);
+       newbase = xmalloc(baselen + pathlen + 2);
+       memcpy(newbase, base, baselen);
+       memcpy(newbase + baselen, p->path, pathlen);
+       memcpy(newbase + baselen + pathlen, "/", 2);
+
+       buf0 = fill_tree_descriptor(t+0, n[0].sha1);
+       buf1 = fill_tree_descriptor(t+1, n[1].sha1);
+       buf2 = fill_tree_descriptor(t+2, n[2].sha1);
+       merge_trees(t, newbase);
+
+       free(buf0);
+       free(buf1);
+       free(buf2);
+       free(newbase);
+       return 1;
+}
+
+
+static struct merge_list *link_entry(unsigned stage, const char *base, struct name_entry *n, struct merge_list *entry)
+{
+       const char *path;
+       struct merge_list *link;
+
+       if (!n->mode)
+               return entry;
+       if (entry)
+               path = entry->path;
+       else
+               path = xstrdup(mkpath("%s%s", base, n->path));
+       link = create_entry(stage, n->mode, n->sha1, path);
+       link->link = entry;
+       return link;
+}
+
+static void unresolved(const char *base, struct name_entry n[3])
+{
+       struct merge_list *entry = NULL;
+
+       if (unresolved_directory(base, n))
                return;
-       }
-       merge(a, b, c, path);
+
+       /*
+        * Do them in reverse order so that the resulting link
+        * list has the stages in order - link_entry adds new
+        * links at the front.
+        */
+       entry = link_entry(3, base, n + 2, entry);
+       entry = link_entry(2, base, n + 1, entry);
+       entry = link_entry(1, base, n + 0, entry);
+
+       add_merge_entry(entry);
 }
 
-/* For two entries, select the smaller one, clear the bigger one */
-static void smaller(struct tree_entry **ap, struct tree_entry **bp)
+/*
+ * Merge two trees together (t[1] and t[2]), using a common base (t[0])
+ * as the origin.
+ *
+ * This walks the (sorted) trees in lock-step, checking every possible
+ * name. Note that directories automatically sort differently from other
+ * files (see "base_name_compare"), so you'll never see file/directory
+ * conflicts, because they won't ever compare the same.
+ *
+ * IOW, if a directory changes to a filename, it will automatically be
+ * seen as the directory going away, and the filename being created.
+ *
+ * Think of this as a three-way diff.
+ *
+ * The output will be either:
+ *  - successful merge
+ *      "0 mode sha1 filename"
+ *    NOTE NOTE NOTE! FIXME! We really really need to walk the index
+ *    in parallel with this too!
+ *
+ *  - conflict:
+ *     "1 mode sha1 filename"
+ *     "2 mode sha1 filename"
+ *     "3 mode sha1 filename"
+ *    where not all of the 1/2/3 lines may exist, of course.
+ *
+ * The successful merge rules are the same as for the three-way merge
+ * in git-read-tree.
+ */
+static void threeway_callback(int n, unsigned long mask, struct name_entry *entry, const char *base)
 {
-       struct tree_entry *a = *ap, *b = *bp;
-       if (a && b) {
-               int cmp = cache_name_compare(a->path, strlen(a->path), b->path, strlen(b->path));
-               if (cmp) {
-                       if (cmp < 0)
-                               *bp = NULL;
-                       else
-                               *ap = NULL;
+       /* Same in both? */
+       if (same_entry(entry+1, entry+2)) {
+               if (entry[0].sha1) {
+                       resolve(base, NULL, entry+1);
+                       return;
                }
        }
-}
 
-static void merge_tree(struct tree_entry *src, struct tree_entry *dst1, struct tree_entry *dst2)
-{
-       while (src || dst1 || dst2) {
-               struct tree_entry *a, *b, *c;
-               a = src;
-               b = dst1;
-               c = dst2;
-               smaller(&a,&b);
-               smaller(&a,&c);
-               smaller(&b,&c);
-               if (a) src = a->next;
-               if (b) dst1 = b->next;
-               if (c) dst2 = c->next;
-               merge_entry(a,b,c);
+       if (same_entry(entry+0, entry+1)) {
+               if (entry[2].sha1 && !S_ISDIR(entry[2].mode)) {
+                       resolve(base, entry+1, entry+2);
+                       return;
+               }
        }
+
+       if (same_entry(entry+0, entry+2)) {
+               if (entry[1].sha1 && !S_ISDIR(entry[1].mode)) {
+                       resolve(base, NULL, entry+1);
+                       return;
+               }
+       }
+
+       unresolved(base, entry);
+}
+
+static void merge_trees(struct tree_desc t[3], const char *base)
+{
+       traverse_trees(3, t, base, threeway_callback);
+}
+
+static void *get_tree_descriptor(struct tree_desc *desc, const char *rev)
+{
+       unsigned char sha1[20];
+       void *buf;
+
+       if (get_sha1(rev, sha1))
+               die("unknown rev %s", rev);
+       buf = fill_tree_descriptor(desc, sha1);
+       if (!buf)
+               die("%s is not a tree", rev);
+       return buf;
 }
 
 int main(int argc, char **argv)
 {
-       unsigned char src[20], dst1[20], dst2[20];
+       struct tree_desc t[3];
+       void *buf1, *buf2, *buf3;
+
+       if (argc != 4)
+               usage(merge_tree_usage);
+
+       setup_git_directory();
+
+       buf1 = get_tree_descriptor(t+0, argv[1]);
+       buf2 = get_tree_descriptor(t+1, argv[2]);
+       buf3 = get_tree_descriptor(t+2, argv[3]);
+       merge_trees(t, "");
+       free(buf1);
+       free(buf2);
+       free(buf3);
 
-       if (argc != 4 ||
-           get_sha1_hex(argv[1], src) ||
-           get_sha1_hex(argv[2], dst1) ||
-           get_sha1_hex(argv[3], dst2))
-               usage("merge-tree <src> <dst1> <dst2>");
-       merge_tree(read_tree(src), read_tree(dst1), read_tree(dst2));
+       show_result();
        return 0;
 }