[PATCH] Rename/copy detection fix.
[git.git] / diff.c
diff --git a/diff.c b/diff.c
index c5ecc43..ababcaa 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -4,14 +4,18 @@
 #include <sys/types.h>
 #include <sys/wait.h>
 #include <signal.h>
-#include <limits.h>
 #include "cache.h"
 #include "diff.h"
-#include "delta.h"
+#include "diffcore.h"
 
 static const char *diff_opts = "-pu";
 static unsigned char null_sha1[20] = { 0, };
 
+static int reverse_diff;
+static int generate_patch;
+static int line_termination = '\n';
+static int inter_name_termination = '\t';
+
 static const char *external_diff(void)
 {
        static const char *external_diff_cmd = NULL;
@@ -55,7 +59,7 @@ static char *sq_expand(const char *src)
        const char *cp;
        char *bp;
 
-       /* count bytes needed to store the quoted string. */ 
+       /* count bytes needed to store the quoted string. */
        for (cnt = 1, cp = src; *cp; cnt++, cp++)
                if (*cp == '\'')
                        cnt += 3;
@@ -75,27 +79,18 @@ static char *sq_expand(const char *src)
 }
 
 static struct diff_tempfile {
-       const char *name;
+       const char *name; /* filename external diff should read from */
        char hex[41];
        char mode[10];
        char tmp_path[50];
 } diff_temp[2];
 
-struct diff_spec {
-       unsigned char blob_sha1[20];
-       unsigned short mode;     /* file mode */
-       unsigned sha1_valid : 1; /* if true, use blob_sha1 and trust mode;
-                                 * if false, use the name and read from
-                                 * the filesystem.
-                                 */
-       unsigned file_valid : 1; /* if false the file does not exist */
-};
-
 static void builtin_diff(const char *name_a,
                         const char *name_b,
-                        struct diff_tempfile *temp)
+                        struct diff_tempfile *temp,
+                        const char *xfrm_msg)
 {
-       int i, next_at;
+       int i, next_at, cmd_size;
        const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
        const char *diff_arg  = "'%s' '%s'||:"; /* "||:" is to return 0 */
        const char *input_name_sq[2];
@@ -112,7 +107,7 @@ static void builtin_diff(const char *name_a,
         * we use 2 spaces around diff-opts, and we need to count
         * terminating NUL, so we subtract 9 here.
         */
-       int cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
+       cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
                        strlen(diff_arg) - 9);
        for (i = 0; i < 2; i++) {
                input_name_sq[i] = sq_expand(temp[i].name);
@@ -148,10 +143,9 @@ static void builtin_diff(const char *name_a,
                        printf("old mode %s\n", temp[0].mode);
                        printf("new mode %s\n", temp[1].mode);
                }
-               if (strcmp(name_a, name_b)) {
-                       printf("rename old %s\n", name_a);
-                       printf("rename new %s\n", name_b);
-               }
+               if (xfrm_msg && xfrm_msg[0])
+                       fputs(xfrm_msg, stdout);
+
                if (strncmp(temp[0].mode, temp[1].mode, 3))
                        /* we do not run diff between different kind
                         * of objects.
@@ -162,6 +156,31 @@ static void builtin_diff(const char *name_a,
        execlp("/bin/sh","sh", "-c", cmd, NULL);
 }
 
+struct diff_filespec *alloc_filespec(const char *path)
+{
+       int namelen = strlen(path);
+       struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
+       spec->path = (char *)(spec + 1);
+       strcpy(spec->path, path);
+       spec->should_free = spec->should_munmap = 0;
+       spec->xfrm_flags = 0;
+       spec->size = 0;
+       spec->data = NULL;
+       spec->mode = 0;
+       memset(spec->sha1, 0, 20);
+       return spec;
+}
+
+void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
+                  unsigned short mode)
+{
+       if (mode) { /* just playing defensive */
+               spec->mode = mode;
+               memcpy(spec->sha1, sha1, 20);
+               spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
+       }
+}
+
 /*
  * Given a name and sha1 pair, if the dircache tells us the file in
  * the work tree has that object contents, return true, so that
@@ -184,7 +203,7 @@ static int work_tree_matches(const char *name, const unsigned char *sha1)
         * file.  Practically, this code only helps when we are used
         * by diff-cache --cached, which does read the cache before
         * calling us.
-        */ 
+        */
        if (!active_cache)
                return 0;
 
@@ -194,13 +213,86 @@ static int work_tree_matches(const char *name, const unsigned char *sha1)
                return 0;
        ce = active_cache[pos];
        if ((lstat(name, &st) < 0) ||
-           !S_ISREG(st.st_mode) ||
+           !S_ISREG(st.st_mode) || /* careful! */
            ce_match_stat(ce, &st) ||
            memcmp(sha1, ce->sha1, 20))
                return 0;
+       /* we return 1 only when we can stat, it is a regular file,
+        * stat information matches, and sha1 recorded in the cache
+        * matches.  I.e. we know the file in the work tree really is
+        * the same as the <name, sha1> pair.
+        */
        return 1;
 }
 
+/*
+ * 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 err = 0;
+       if (!DIFF_FILE_VALID(s))
+               die("internal error: asking to populate invalid file.");
+       if (S_ISDIR(s->mode))
+               return -1;
+
+       if (s->data)
+               return err;
+       if (!s->sha1_valid ||
+           work_tree_matches(s->path, s->sha1)) {
+               struct stat st;
+               int fd;
+               if (lstat(s->path, &st) < 0) {
+                       if (errno == ENOENT) {
+                       err_empty:
+                               err = -1;
+                       empty:
+                               s->data = "";
+                               s->size = 0;
+                               return err;
+                       }
+               }
+               s->size = st.st_size;
+               if (!s->size)
+                       goto empty;
+               if (S_ISLNK(st.st_mode)) {
+                       int ret;
+                       s->data = xmalloc(s->size);
+                       s->should_free = 1;
+                       ret = readlink(s->path, s->data, s->size);
+                       if (ret < 0) {
+                               free(s->data);
+                               goto err_empty;
+                       }
+                       return 0;
+               }
+               fd = open(s->path, O_RDONLY);
+               if (fd < 0)
+                       goto err_empty;
+               s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
+               s->should_munmap = 1;
+               close(fd);
+       }
+       else {
+               char type[20];
+               s->data = read_sha1_file(s->sha1, type, &s->size);
+               s->should_free = 1;
+       }
+       return 0;
+}
+
+void diff_free_filespec_data(struct diff_filespec *s)
+{
+       if (s->should_free)
+               free(s->data);
+       else if (s->should_munmap)
+               munmap(s->data, s->size);
+       s->should_free = s->should_munmap = 0;
+       s->data = NULL;
+}
+
 static void prep_temp_blob(struct diff_tempfile *temp,
                           void *blob,
                           unsigned long size,
@@ -224,9 +316,9 @@ static void prep_temp_blob(struct diff_tempfile *temp,
 
 static void prepare_temp_file(const char *name,
                              struct diff_tempfile *temp,
-                             struct diff_spec *one)
+                             struct diff_filespec *one)
 {
-       if (!one->file_valid) {
+       if (!DIFF_FILE_VALID(one)) {
        not_a_valid_file:
                /* A '-' entry produces this for file-2, and
                 * a '+' entry produces this for file-1.
@@ -238,13 +330,12 @@ static void prepare_temp_file(const char *name,
        }
 
        if (!one->sha1_valid ||
-           work_tree_matches(name, one->blob_sha1)) {
+           work_tree_matches(name, one->sha1)) {
                struct stat st;
-               temp->name = name;
-               if (lstat(temp->name, &st) < 0) {
+               if (lstat(name, &st) < 0) {
                        if (errno == ENOENT)
                                goto not_a_valid_file;
-                       die("stat(%s): %s", temp->name, strerror(errno));
+                       die("stat(%s): %s", name, strerror(errno));
                }
                if (S_ISLNK(st.st_mode)) {
                        int ret;
@@ -256,31 +347,27 @@ static void prepare_temp_file(const char *name,
                                die("readlink(%s)", name);
                        prep_temp_blob(temp, buf, st.st_size,
                                       (one->sha1_valid ?
-                                       one->blob_sha1 : null_sha1),
+                                       one->sha1 : null_sha1),
                                       (one->sha1_valid ?
                                        one->mode : S_IFLNK));
                }
                else {
+                       /* we can borrow from the file in the work tree */
+                       temp->name = name;
                        if (!one->sha1_valid)
                                strcpy(temp->hex, sha1_to_hex(null_sha1));
                        else
-                               strcpy(temp->hex, sha1_to_hex(one->blob_sha1));
+                               strcpy(temp->hex, sha1_to_hex(one->sha1));
                        sprintf(temp->mode, "%06o",
                                S_IFREG |ce_permissions(st.st_mode));
                }
                return;
        }
        else {
-               void *blob;
-               char type[20];
-               unsigned long size;
-
-               blob = read_sha1_file(one->blob_sha1, type, &size);
-               if (!blob || strcmp(type, "blob"))
-                       die("unable to read blob object for %s (%s)",
-                           name, sha1_to_hex(one->blob_sha1));
-               prep_temp_blob(temp, blob, size, one->blob_sha1, one->mode);
-               free(blob);
+               if (diff_populate_filespec(one))
+                       die("cannot read data blob for %s", one->path);
+               prep_temp_blob(temp, one->data, one->size,
+                              one->sha1, one->mode);
        }
 }
 
@@ -300,31 +387,6 @@ static void remove_tempfile_on_signal(int signo)
        remove_tempfile();
 }
 
-static int detect_rename;
-static int reverse_diff;
-static const char **pathspec;
-static int speccnt;
-static int diff_rename_minimum_score;
-
-static int matches_pathspec(const char *name)
-{
-       int i;
-       int namelen;
-
-       if (speccnt == 0)
-               return 1;
-
-       namelen = strlen(name);
-       for (i = 0; i < speccnt; i++) {
-               int speclen = strlen(pathspec[i]);
-               if (! strncmp(pathspec[i], name, speclen) &&
-                   speclen <= namelen &&
-                   (name[speclen] == 0 || name[speclen] == '/'))
-                       return 1;
-       }
-       return 0;
-}
-
 /* An external diff command takes:
  *
  * diff-cmd name infile1 infile1-sha1 infile1-mode \
@@ -333,26 +395,15 @@ static int matches_pathspec(const char *name)
  */
 static void run_external_diff(const char *name,
                              const char *other,
-                             struct diff_spec *one,
-                             struct diff_spec *two)
+                             struct diff_filespec *one,
+                             struct diff_filespec *two,
+                             const char *xfrm_msg)
 {
        struct diff_tempfile *temp = diff_temp;
        pid_t pid;
        int status;
        static int atexit_asked = 0;
 
-       if (reverse_diff) {
-               struct diff_spec *tmp_spec;
-               tmp_spec = one; one = two; two = tmp_spec;
-               if (other) {
-                       const char *tmp;
-                       tmp = name; name = other; other = tmp;
-               }
-       }
-
-       if (!matches_pathspec(name) && (!other || !matches_pathspec(other)))
-               return;
-
        if (one && two) {
                prepare_temp_file(name, &temp[0], one);
                prepare_temp_file(other ? : name, &temp[1], two);
@@ -373,7 +424,7 @@ static void run_external_diff(const char *name,
                const char *pgm = external_diff();
                if (pgm) {
                        if (one && two) {
-                               const char *exec_arg[9];
+                               const char *exec_arg[10];
                                const char **arg = &exec_arg[0];
                                *arg++ = pgm;
                                *arg++ = name;
@@ -383,9 +434,11 @@ static void run_external_diff(const char *name,
                                *arg++ = temp[1].name;
                                *arg++ = temp[1].hex;
                                *arg++ = temp[1].mode;
-                               if (other)
+                               if (other) {
                                        *arg++ = other;
-                               *arg = 0;
+                                       *arg++ = xfrm_msg;
+                               }
+                               *arg = NULL;
                                execvp(pgm, (char *const*) exec_arg);
                        }
                        else
@@ -395,7 +448,7 @@ static void run_external_diff(const char *name,
                 * otherwise we use the built-in one.
                 */
                if (one && two)
-                       builtin_diff(name, other ? : name, temp);
+                       builtin_diff(name, other ? : name, temp, xfrm_msg);
                else
                        printf("* Unmerged path %s\n", name);
                exit(0);
@@ -418,303 +471,244 @@ static void run_external_diff(const char *name,
        remove_tempfile();
 }
 
-/*
- * We do not detect circular renames.  Just hold created and deleted
- * entries and later attempt to match them up.  If they do not match,
- * then spit them out as deletes or creates as original.
- */
-
-static struct diff_spec_hold {
-       struct diff_spec_hold *next;
-       struct diff_spec it;
-       unsigned long size;
-       int flags;
-#define MATCHED 1
-#define SHOULD_FREE 2
-#define SHOULD_MUNMAP 4
-       void *data;
-       char path[1];
-} *createdfile, *deletedfile;
-
-static void hold_diff(const char *name,
-                     struct diff_spec *one,
-                     struct diff_spec *two)
+void diff_setup(int reverse_diff_)
 {
-       struct diff_spec_hold **list, *elem;
-
-       if (one->file_valid && two->file_valid)
-               die("internal error");
-
-       if (!detect_rename) {
-               run_external_diff(name, NULL, one, two);
-               return;
-       }
-       elem = xmalloc(sizeof(*elem) + strlen(name));
-       strcpy(elem->path, name);
-       elem->size = 0;
-       elem->data = NULL;
-       elem->flags = 0;
-       if (one->file_valid) {
-               list = &deletedfile;
-               elem->it = *one;
-       }
-       else {
-               list = &createdfile;
-               elem->it = *two;
-       }
-       elem->next = *list;
-       *list = elem;
+       reverse_diff = reverse_diff_;
 }
 
-static int populate_data(struct diff_spec_hold *s)
-{
-       char type[20];
+struct diff_queue_struct diff_queued_diff;
 
-       if (s->data)
-               return 0;
-       if (s->it.sha1_valid) {
-               s->data = read_sha1_file(s->it.blob_sha1, type, &s->size);
-               s->flags |= SHOULD_FREE;
-       }
-       else {
-               struct stat st;
-               int fd;
-               fd = open(s->path, O_RDONLY);
-               if (fd < 0)
-                       return -1;
-               if (fstat(fd, &st)) {
-                       close(fd);
-                       return -1;
-               }
-               s->size = st.st_size;
-               s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
-               close(fd);
-               if (!s->size)
-                       s->data = "";
-               else
-                       s->flags |= SHOULD_MUNMAP;
+void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
+{
+       if (queue->alloc <= queue->nr) {
+               queue->alloc = alloc_nr(queue->alloc);
+               queue->queue = xrealloc(queue->queue,
+                                       sizeof(dp) * queue->alloc);
        }
-       return 0;
+       queue->queue[queue->nr++] = dp;
 }
 
-static void free_data(struct diff_spec_hold *s)
+struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
+                                struct diff_filespec *one,
+                                struct diff_filespec *two)
 {
-       if (s->flags & SHOULD_FREE)
-               free(s->data);
-       else if (s->flags & SHOULD_MUNMAP)
-               munmap(s->data, s->size);
-       s->flags &= ~(SHOULD_FREE|SHOULD_MUNMAP);
-       s->data = 0;
+       struct diff_filepair *dp = xmalloc(sizeof(*dp));
+       dp->one = one;
+       dp->two = two;
+       dp->score = 0;
+       dp->orig_order = queue->nr;
+       dp->rename_rank = 0;
+       diff_q(queue, dp);
+       return dp;
 }
 
-static void flush_remaining_diff(struct diff_spec_hold *elem,
-                                int on_created_list)
+static void diff_flush_raw(struct diff_filepair *p)
 {
-       static struct diff_spec null_file_spec;
-
-       null_file_spec.file_valid = 0;
-       for ( ; elem ; elem = elem->next) {
-               free_data(elem);
-               if (elem->flags & MATCHED)
-                       continue;
-               if (on_created_list)
-                       run_external_diff(elem->path, NULL,
-                                         &null_file_spec, &elem->it);
-               else
-                       run_external_diff(elem->path, NULL,
-                                         &elem->it, &null_file_spec);
+       if (DIFF_PAIR_UNMERGED(p)) {
+               printf("U %s%c", p->one->path, line_termination);
+               return;
        }
+       printf(":%06o %06o %s ",
+              p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
+       printf("%s%c%s%c%s%c",
+              sha1_to_hex(p->two->sha1), inter_name_termination,
+              p->one->path, inter_name_termination,
+              p->two->path, line_termination);
 }
 
-static int is_exact_match(struct diff_spec_hold *src,
-                         struct diff_spec_hold *dst)
+int diff_unmodified_pair(struct diff_filepair *p)
 {
-       if (src->it.sha1_valid && dst->it.sha1_valid &&
-           !memcmp(src->it.blob_sha1, dst->it.blob_sha1, 20))
-               return 1;
-       if (populate_data(src) || populate_data(dst))
-               /* this is an error but will be caught downstream */
+       /* This function is written stricter than necessary to support
+        * the currently implemented transformers, but the idea is to
+        * let transformers to produce diff_filepairs any way they want,
+        * and filter and clean them up here before producing the output.
+        */
+       struct diff_filespec *one, *two;
+
+       if (DIFF_PAIR_UNMERGED(p))
+               return 0; /* unmerged is interesting */
+
+       one = p->one;
+       two = p->two;
+
+       /* deletion, addition, mode change and renames are all interesting. */
+       if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
+           (one->mode != two->mode) ||
+           strcmp(one->path, two->path))
                return 0;
-       if (src->size == dst->size &&
-           !memcmp(src->data, dst->data, src->size))
-               return 1;
+
+       /* both are valid and point at the same path.  that is, we are
+        * dealing with a change.
+        */
+       if (one->sha1_valid && two->sha1_valid &&
+           !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
+               return 1; /* no change */
+       if (!one->sha1_valid && !two->sha1_valid)
+               return 1; /* both look at the same file on the filesystem. */
        return 0;
 }
 
-#define MINIMUM_SCORE 5000
-int estimate_similarity(struct diff_spec_hold *src, struct diff_spec_hold *dst)
+static void diff_flush_patch(struct diff_filepair *p, const char *msg)
 {
-       /* src points at a deleted file and dst points at a created
-        * file.  They may be quite similar, in which case we want to
-        * say src is renamed to dst.
-        *
-        * Compare them and return how similar they are, representing
-        * the score as an integer between 0 and 10000.  10000 is
-        * reserved for the case where they match exactly.
+       const char *name, *other;
+
+       /* diffcore_prune() keeps "stay" entries for diff-raw
+        * copy/rename detection, but when we are generating
+        * patches we do not need them.
         */
-       void *delta;
-       unsigned long delta_size;
+       if (diff_unmodified_pair(p))
+               return;
 
-       delta_size = ((src->size < dst->size) ?
-                     (dst->size - src->size) : (src->size - dst->size));
+       name = p->one->path;
+       other = (strcmp(name, p->two->path) ? p->two->path : NULL);
+       if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
+           (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
+               return; /* no tree diffs in patch format */ 
 
-       /* We would not consider rename followed by more than
-        * 20% edits; that is, delta_size must be smaller than
-        * (src->size + dst->size)/2 * 0.2, which means...
+       if (DIFF_PAIR_UNMERGED(p))
+               run_external_diff(name, NULL, NULL, NULL, NULL);
+       else
+               run_external_diff(name, other, p->one, p->two, msg);
+}
+
+int diff_needs_to_stay(struct diff_queue_struct *q, int i,
+                      struct diff_filespec *it)
+{
+       /* If it will be used in later entry (either stay or used
+        * as the source of rename/copy), we need to copy, not rename.
         */
-       if ((src->size + dst->size) < delta_size * 10)
-               return 0;
+       while (i < q->nr) {
+               struct diff_filepair *p = q->queue[i++];
+               if (!DIFF_FILE_VALID(p->two))
+                       continue; /* removed is fine */
+               if (strcmp(p->one->path, it->path))
+                       continue; /* not relevant */
+
+               /* p has its src set to *it and it is not a delete;
+                * it will be used for in-place change, rename/copy,
+                * or just stays there.  We cannot rename it out.
+                */
+               return 1;
+       }
+       return 0;
+}
 
-       delta = diff_delta(src->data, src->size,
-                          dst->data, dst->size,
-                          &delta_size);
-       free(delta);
+static int diff_used_as_source(struct diff_queue_struct *q, int lim,
+                              struct diff_filespec *it)
+{
+       int i;
+       for (i = 0; i < lim; i++) {
+               struct diff_filepair *p = q->queue[i++];
+               if (!strcmp(p->one->path, it->path))
+                       return 1;
+       }
+       return 0;
+}
 
-       /* This "delta" is really xdiff with adler32 and all the
-        * overheads but it is a quick and dirty approximation.
-        *
-        * Now we will give some score to it.  Let's say 20% edit gets
-        * 5000 points and 0% edit gets 9000 points.  That is, every
-        * 1/20000 edit gets 1 point penalty.  The amount of penalty is:
+void diffcore_prune(void)
+{
+       /*
+        * Although rename/copy detection wants to have "no-change"
+        * entries fed into them, the downstream do not need to see
+        * them, unless we had rename/copy for the same path earlier.
+        * This function removes such entries.
         *
-        * (delta_size * 2 / (src->size + dst->size)) * 20000
+        * The applications that use rename/copy should:
         *
+        * (1) feed change and "no-change" entries via diff_queue().
+        * (2) call diffcore_rename, and any other future diffcore_xxx
+        *     that would benefit by still having "no-change" entries.
+        * (3) call diffcore_prune
+        * (4) call other diffcore_xxx that do not need to see
+        *     "no-change" entries.
+        * (5) call diff_flush().
         */
-       return 9000 - (40000 * delta_size / (src->size+dst->size));
-}
+       struct diff_queue_struct *q = &diff_queued_diff;
+       struct diff_queue_struct outq;
+       int i;
 
-struct diff_score {
-       struct diff_spec_hold *src;
-       struct diff_spec_hold *dst;
-       int score;
-};
+       outq.queue = NULL;
+       outq.nr = outq.alloc = 0;
 
-static int score_compare(const void *a_, const void *b_)
-{
-       const struct diff_score *a = a_, *b = b_;
-       return b->score - a->score;
+       for (i = 0; i < q->nr; i++) {
+               struct diff_filepair *p = q->queue[i];
+               if (!diff_unmodified_pair(p) ||
+                   diff_used_as_source(q, i, p->one))
+                       diff_q(&outq, p);
+               else
+                       free(p);
+       }
+       free(q->queue);
+       *q = outq;
+       return;
 }
 
-static void flush_rename_pair(struct diff_spec_hold *src,
-                             struct diff_spec_hold *dst)
+static void diff_flush_one(struct diff_filepair *p, const char *msg)
 {
-       src->flags |= MATCHED;
-       dst->flags |= MATCHED;
-       free_data(src);
-       free_data(dst);
-       run_external_diff(src->path, dst->path,
-                         &src->it, &dst->it);
+       if (generate_patch)
+               diff_flush_patch(p, msg);
+       else
+               diff_flush_raw(p);
 }
 
-static void free_held_diff(struct diff_spec_hold *list)
+int diff_queue_is_empty(void)
 {
-       struct diff_spec_hold *h;
-       for (h = list; list; list = h) {
-               h = list->next;
-               free_data(list);
-               free(list);
-       }
+       struct diff_queue_struct *q = &diff_queued_diff;
+       return q->nr == 0;
 }
 
-void diff_flush(void)
+void diff_flush(int diff_output_style)
 {
-       int num_create, num_delete, c, d;
-       struct diff_spec_hold *elem, *src, *dst;
-       struct diff_score *mx;
+       struct diff_queue_struct *q = &diff_queued_diff;
+       int i;
 
-       /* We really want to cull the candidates list early
-        * with cheap tests in order to avoid doing deltas.
-        *
-        * With the current callers, we should not have already
-        * matched entries at this point, but it is nonetheless
-        * checked for sanity.
-        */
-       for (dst = createdfile; dst; dst = dst->next) {
-               if (dst->flags & MATCHED)
-                       continue;
-               for (src = deletedfile; src; src = src->next) {
-                       if (src->flags & MATCHED)
-                               continue;
-                       if (! is_exact_match(src, dst))
-                               continue;
-                       flush_rename_pair(src, dst);
-                       break;
-               }
+       generate_patch = 0;
+       switch (diff_output_style) {
+       case DIFF_FORMAT_HUMAN:
+               line_termination = '\n';
+               inter_name_termination = '\t';
+               break;
+       case DIFF_FORMAT_MACHINE:
+               line_termination = inter_name_termination = 0;
+               break;
+       case DIFF_FORMAT_PATCH:
+               generate_patch = 1;
+               break;
        }
-
-       /* Count surviving candidates */
-       for (num_create = 0, elem = createdfile; elem; elem = elem->next)
-               if (!(elem->flags & MATCHED))
-                       num_create++;
-
-       for (num_delete = 0, elem = deletedfile; elem; elem = elem->next)
-               if (!(elem->flags & MATCHED))
-                       num_delete++;
-
-       if (num_create == 0 ||  num_delete == 0)
-               goto exit_path;
-
-       mx = xmalloc(sizeof(*mx) * num_create * num_delete);
-       for (c = 0, dst = createdfile; dst; dst = dst->next) {
-               int base = c * num_delete;
-               if (dst->flags & MATCHED)
-                       continue;
-               for (d = 0, src = deletedfile; src; src = src->next) {
-                       struct diff_score *m = &mx[base+d];
-                       if (src->flags & MATCHED)
-                               continue;
-                       m->src = src;
-                       m->dst = dst;
-                       m->score = estimate_similarity(src, dst);
-                       d++;
+       for (i = 0; i < q->nr; i++) {
+               char msg_[PATH_MAX*2+200], *msg = NULL;
+               struct diff_filepair *p = q->queue[i];
+               if (strcmp(p->one->path, p->two->path)) {
+                       /* This is rename or copy.  Which one is it? */
+                       if (diff_needs_to_stay(q, i+1, p->one)) {
+                               sprintf(msg_,
+                                       "similarity index %d%%\n"
+                                       "copy from %s\n"
+                                       "copy to %s\n",
+                                       (int)(0.5 + p->score * 100/MAX_SCORE),
+                                       p->one->path, p->two->path);
+                       }
+                       else
+                               sprintf(msg_,
+                                       "similarity index %d%%\n"
+                                       "rename old %s\n"
+                                       "rename new %s\n",
+                                       (int)(0.5 + p->score * 100/MAX_SCORE),
+                                       p->one->path, p->two->path);
+                       msg = msg_;
                }
-               c++;
-       }
-       qsort(mx, num_create*num_delete, sizeof(*mx), score_compare); 
-
-#if 0
-       for (c = 0; c < num_create * num_delete; c++) {
-               src = mx[c].src;
-               dst = mx[c].dst;
-               if ((src->flags & MATCHED) || (dst->flags & MATCHED))
-                       continue;
-               fprintf(stderr,
-                       "**score ** %d %s %s\n",
-                       mx[c].score, src->path, dst->path);
-       }
-#endif
-
-       for (c = 0; c < num_create * num_delete; c++) {
-               src = mx[c].src;
-               dst = mx[c].dst;
-               if ((src->flags & MATCHED) || (dst->flags & MATCHED))
-                       continue;
-               if (mx[c].score < diff_rename_minimum_score)
-                       break;
-               flush_rename_pair(src, dst);
+               diff_flush_one(p, msg);
        }
-       free(mx);
-
- exit_path:
-       flush_remaining_diff(createdfile, 1);
-       flush_remaining_diff(deletedfile, 0);
-       free_held_diff(createdfile);
-       free_held_diff(deletedfile);
-       createdfile = deletedfile = NULL;
-}
-
-void diff_setup(int detect_rename_, int minimum_score_, int reverse_diff_,
-               const char **pathspec_, int speccnt_)
-{
-       free_held_diff(createdfile);
-       free_held_diff(deletedfile);
-       createdfile = deletedfile = NULL;
 
-       detect_rename = detect_rename_;
-       reverse_diff = reverse_diff_;
-       pathspec = pathspec_;
-       speccnt = speccnt_;
-       diff_rename_minimum_score = minimum_score_ ? : MINIMUM_SCORE;
+       for (i = 0; i < q->nr; i++) {
+               struct diff_filepair *p = q->queue[i];
+               diff_free_filespec_data(p->one);
+               diff_free_filespec_data(p->two);
+               free(p);
+       }
+       free(q->queue);
+       q->queue = NULL;
+       q->nr = q->alloc = 0;
 }
 
 void diff_addremove(int addremove, unsigned mode,
@@ -722,54 +716,89 @@ void diff_addremove(int addremove, unsigned mode,
                    const char *base, const char *path)
 {
        char concatpath[PATH_MAX];
-       struct diff_spec spec[2], *one, *two;
-
-       memcpy(spec[0].blob_sha1, sha1, 20);
-       spec[0].mode = mode;
-       spec[0].sha1_valid = !!memcmp(sha1, null_sha1, 20);
-       spec[0].file_valid = 1;
-       spec[1].file_valid = 0;
-
-       if (addremove == '+') {
-               one = spec + 1; two = spec;
-       } else {
-               one = spec; two = one + 1;
-       }
+       struct diff_filespec *one, *two;
+
+       /* This may look odd, but it is a preparation for
+        * feeding "there are unchanged files which should
+        * not produce diffs, but when you are doing copy
+        * detection you would need them, so here they are"
+        * entries to the diff-core.  They will be prefixed
+        * with something like '=' or '*' (I haven't decided
+        * which but should not make any difference).
+        * Feeding the same new and old to diff_change() 
+        * also has the same effect.  diffcore_prune() should
+        * be used to filter uninteresting ones out before the
+        * final output happens.
+        */
+       if (reverse_diff)
+               addremove = (addremove == '+' ? '-' :
+                            addremove == '-' ? '+' : addremove);
+
+       if (!path) path = "";
+       sprintf(concatpath, "%s%s", base, path);
+       one = alloc_filespec(concatpath);
+       two = alloc_filespec(concatpath);
 
-       if (path) {
-               strcpy(concatpath, base);
-               strcat(concatpath, path);
+       if (addremove != '+')
+               fill_filespec(one, sha1, mode);
+       if (addremove != '-')
+               fill_filespec(two, sha1, mode);
+
+       diff_queue(&diff_queued_diff, one, two);
+}
+
+void diff_guif(unsigned old_mode,
+              unsigned new_mode,
+              const unsigned char *old_sha1,
+              const unsigned char *new_sha1,
+              const char *old_path,
+              const char *new_path)
+{
+       struct diff_filespec *one, *two;
+
+       if (reverse_diff) {
+               unsigned tmp;
+               const unsigned char *tmp_c;
+               tmp = old_mode; old_mode = new_mode; new_mode = tmp;
+               tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
        }
-       hold_diff(path ? concatpath : base, one, two);
+       one = alloc_filespec(old_path);
+       two = alloc_filespec(new_path);
+       if (old_mode)
+               fill_filespec(one, old_sha1, old_mode);
+       if (new_mode)
+               fill_filespec(two, new_sha1, new_mode);
+       diff_queue(&diff_queued_diff, one, two);
 }
 
 void diff_change(unsigned old_mode, unsigned new_mode,
                 const unsigned char *old_sha1,
                 const unsigned char *new_sha1,
-                const char *base, const char *path) {
+                const char *base, const char *path) 
+{
        char concatpath[PATH_MAX];
-       struct diff_spec spec[2];
+       struct diff_filespec *one, *two;
 
-       if (path) {
-               strcpy(concatpath, base);
-               strcat(concatpath, path);
+       if (reverse_diff) {
+               unsigned tmp;
+               const unsigned char *tmp_c;
+               tmp = old_mode; old_mode = new_mode; new_mode = tmp;
+               tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
        }
-
-       memcpy(spec[0].blob_sha1, old_sha1, 20);
-       spec[0].mode = old_mode;
-       memcpy(spec[1].blob_sha1, new_sha1, 20);
-       spec[1].mode = new_mode;
-       spec[0].sha1_valid = !!memcmp(old_sha1, null_sha1, 20);
-       spec[1].sha1_valid = !!memcmp(new_sha1, null_sha1, 20);
-       spec[1].file_valid = spec[0].file_valid = 1;
-
-       /* We do not look at changed files as candidate for
-        * rename detection ever.
-        */
-       run_external_diff(path ? concatpath : base, NULL, &spec[0], &spec[1]);
+       if (!path) path = "";
+       sprintf(concatpath, "%s%s", base, path);
+       one = alloc_filespec(concatpath);
+       two = alloc_filespec(concatpath);
+       fill_filespec(one, old_sha1, old_mode);
+       fill_filespec(two, new_sha1, new_mode);
+
+       diff_queue(&diff_queued_diff, one, two);
 }
 
 void diff_unmerge(const char *path)
 {
-       run_external_diff(path, NULL, NULL, NULL);
+       struct diff_filespec *one, *two;
+       one = alloc_filespec(path);
+       two = alloc_filespec(path);
+       diff_queue(&diff_queued_diff, one, two);
 }