diff-tree: fix up comparison of "interesting" sub-trees
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 4a54688cbcd559c0e37db2bc1524fd8c25b3d3fa..74004e5a3f9fa491b30ab3d5f231826593e4eae4 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -4,14 +4,15 @@
 #include <sys/types.h>
 #include <sys/wait.h>
 #include <signal.h>
+#include <limits.h>
 #include "cache.h"
 #include "diff.h"
 
-static char *diff_opts = "-pu";
+static const char *diff_opts = "-pu";
 
 static const char *external_diff(void)
 {
-       static char *external_diff_cmd = NULL;
+       static const char *external_diff_cmd = NULL;
        static int done_preparing = 0;
 
        if (done_preparing)
@@ -25,11 +26,11 @@ static const char *external_diff(void)
         *
         * GIT_DIFF_OPTS="-c";
         */
-       if (getenv("GIT_EXTERNAL_DIFF"))
-               external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
+       if (gitenv("GIT_EXTERNAL_DIFF"))
+               external_diff_cmd = gitenv("GIT_EXTERNAL_DIFF");
 
        /* In case external diff fails... */
-       diff_opts = getenv("GIT_DIFF_OPTS") ? : diff_opts;
+       diff_opts = gitenv("GIT_DIFF_OPTS") ? : diff_opts;
 
        done_preparing = 1;
        return external_diff_cmd;
@@ -78,7 +79,8 @@ static struct diff_tempfile {
        char tmp_path[50];
 } diff_temp[2];
 
-static void builtin_diff(const char *name,
+static void builtin_diff(const char *name_a,
+                        const char *name_b,
                         struct diff_tempfile *temp)
 {
        int i, next_at;
@@ -87,9 +89,12 @@ static void builtin_diff(const char *name,
        const char *input_name_sq[2];
        const char *path0[2];
        const char *path1[2];
-       const char *name_sq = sq_expand(name);
+       const char *name_sq[2];
        char *cmd;
-       
+
+       name_sq[0] = sq_expand(name_a);
+       name_sq[1] = sq_expand(name_b);
+
        /* diff_cmd and diff_arg have 6 %s in total which makes
         * the sum of these strings 12 bytes larger than required.
         * we use 2 spaces around diff-opts, and we need to count
@@ -104,7 +109,7 @@ static void builtin_diff(const char *name,
                        path1[i] = "";
                } else {
                        path0[i] = i ? "b/" : "a/";
-                       path1[i] = name_sq;
+                       path1[i] = name_sq[i];
                }
                cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
                             strlen(input_name_sq[i]));
@@ -121,22 +126,96 @@ static void builtin_diff(const char *name,
        next_at += snprintf(cmd+next_at, cmd_size-next_at,
                            diff_arg, input_name_sq[0], input_name_sq[1]);
 
+       printf("diff --git a/%s b/%s\n", name_a, name_b);
        if (!path1[0][0])
-               printf("Created: %s (mode:%s)\n", name, temp[1].mode);
+               printf("new file mode %s\n", temp[1].mode);
        else if (!path1[1][0])
-               printf("Deleted: %s\n", name);
-       else if (strcmp(temp[0].mode, temp[1].mode))
-               printf("Mode changed: %s (%s->%s)\n", name,
-                      temp[0].mode, temp[1].mode);
+               printf("deleted file mode %s\n", temp[0].mode);
+       else {
+               if (strcmp(temp[0].mode, temp[1].mode)) {
+                       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 (strncmp(temp[0].mode, temp[1].mode, 3))
+                       /* we do not run diff between different kind
+                        * of objects.
+                        */
+                       exit(0);
+       }
        fflush(NULL);
        execlp("/bin/sh","sh", "-c", cmd, NULL);
 }
 
+/*
+ * 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
+ * prepare_temp_file() does not have to inflate and extract.
+ */
+static int work_tree_matches(const char *name, const unsigned char *sha1)
+{
+       struct cache_entry *ce;
+       struct stat st;
+       int pos, len;
+       
+       /* We do not read the cache ourselves here, because the
+        * benchmark with my previous version that always reads cache
+        * shows that it makes things worse for diff-tree comparing
+        * two linux-2.6 kernel trees in an already checked out work
+        * tree.  This is because most diff-tree comparisons deal with
+        * only a small number of files, while reading the cache is
+        * expensive for a large project, and its cost outweighs the
+        * savings we get by not inflating the object to a temporary
+        * 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;
+
+       len = strlen(name);
+       pos = cache_name_pos(name, len);
+       if (pos < 0)
+               return 0;
+       ce = active_cache[pos];
+       if ((lstat(name, &st) < 0) ||
+           !S_ISREG(st.st_mode) ||
+           ce_match_stat(ce, &st) ||
+           memcmp(sha1, ce->sha1, 20))
+               return 0;
+       return 1;
+}
+
+static void prep_temp_blob(struct diff_tempfile *temp,
+                          void *blob,
+                          unsigned long size,
+                          unsigned char *sha1,
+                          int mode)
+{
+       int fd;
+
+       strcpy(temp->tmp_path, ".diff_XXXXXX");
+       fd = mkstemp(temp->tmp_path);
+       if (fd < 0)
+               die("unable to create temp-file");
+       if (write(fd, blob, size) != size)
+               die("unable to write temp-file");
+       close(fd);
+       temp->name = temp->tmp_path;
+       strcpy(temp->hex, sha1_to_hex(sha1));
+       temp->hex[40] = 0;
+       sprintf(temp->mode, "%06o", mode);
+}
+
 static void prepare_temp_file(const char *name,
                              struct diff_tempfile *temp,
                              struct diff_spec *one)
 {
        static unsigned char null_sha1[20] = { 0, };
+       int use_work_tree = 0;
 
        if (!one->file_valid) {
        not_a_valid_file:
@@ -150,46 +229,53 @@ static void prepare_temp_file(const char *name,
        }
 
        if (one->sha1_valid &&
-           !memcmp(one->u.sha1, null_sha1, sizeof(null_sha1))) {
-               one->sha1_valid = 0;
-               one->u.name = name;
-       }
+           (!memcmp(one->blob_sha1, null_sha1, sizeof(null_sha1)) ||
+            work_tree_matches(name, one->blob_sha1)))
+               use_work_tree = 1;
 
-       if (!one->sha1_valid) {
+       if (!one->sha1_valid || use_work_tree) {
                struct stat st;
-               temp->name = one->u.name;
-               if (stat(temp->name, &st) < 0) {
+               temp->name = name;
+               if (lstat(temp->name, &st) < 0) {
                        if (errno == ENOENT)
                                goto not_a_valid_file;
                        die("stat(%s): %s", temp->name, strerror(errno));
                }
-               strcpy(temp->hex, sha1_to_hex(null_sha1));
-               sprintf(temp->mode, "%06o",
-                       S_IFREG |ce_permissions(st.st_mode));
+               if (S_ISLNK(st.st_mode)) {
+                       int ret;
+                       char *buf, buf_[1024];
+                       buf = ((sizeof(buf_) < st.st_size) ?
+                              xmalloc(st.st_size) : buf_);
+                       ret = readlink(name, buf, st.st_size);
+                       if (ret < 0)
+                               die("readlink(%s)", name);
+                       prep_temp_blob(temp, buf, st.st_size,
+                                      (one->sha1_valid ?
+                                       one->blob_sha1 : null_sha1),
+                                      (one->sha1_valid ?
+                                       one->mode : S_IFLNK));
+               }
+               else {
+                       if (!one->sha1_valid)
+                               strcpy(temp->hex, sha1_to_hex(null_sha1));
+                       else
+                               strcpy(temp->hex, sha1_to_hex(one->blob_sha1));
+                       sprintf(temp->mode, "%06o",
+                               S_IFREG |ce_permissions(st.st_mode));
+               }
+               return;
        }
        else {
-               int fd;
                void *blob;
                char type[20];
                unsigned long size;
 
-               blob = read_sha1_file(one->u.sha1, type, &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->u.sha1));
-
-               strcpy(temp->tmp_path, ".diff_XXXXXX");
-               fd = mkstemp(temp->tmp_path);
-               if (fd < 0)
-                       die("unable to create temp-file");
-               if (write(fd, blob, size) != size)
-                       die("unable to write temp-file");
-               close(fd);
+                           name, sha1_to_hex(one->blob_sha1));
+               prep_temp_blob(temp, blob, size, one->blob_sha1, one->mode);
                free(blob);
-               temp->name = temp->tmp_path;
-               strcpy(temp->hex, sha1_to_hex(one->u.sha1));
-               temp->hex[40] = 0;
-               sprintf(temp->mode, "%06o", one->mode);
        }
 }
 
@@ -216,6 +302,7 @@ static void remove_tempfile_on_signal(int signo)
  *
  */
 void run_external_diff(const char *name,
+                      const char *other,
                       struct diff_spec *one,
                       struct diff_spec *two)
 {
@@ -226,7 +313,7 @@ void run_external_diff(const char *name,
 
        if (one && two) {
                prepare_temp_file(name, &temp[0], one);
-               prepare_temp_file(name, &temp[1], two);
+               prepare_temp_file(other ? : name, &temp[1], two);
                if (! atexit_asked &&
                    (temp[0].name == temp[0].tmp_path ||
                     temp[1].name == temp[1].tmp_path)) {
@@ -242,7 +329,8 @@ void run_external_diff(const char *name,
                die("unable to fork");
        if (!pid) {
                const char *pgm = external_diff();
-               if (pgm) {
+               /* not passing rename patch to external ones */
+               if (!other && pgm) {
                        if (one && two)
                                execlp(pgm, pgm,
                                       name,
@@ -256,7 +344,7 @@ void run_external_diff(const char *name,
                 * otherwise we use the built-in one.
                 */
                if (one && two)
-                       builtin_diff(name, temp);
+                       builtin_diff(name, other ? : name, temp);
                else
                        printf("* Unmerged path %s\n", name);
                exit(0);
@@ -286,7 +374,7 @@ void diff_addremove(int addremove, unsigned mode,
        char concatpath[PATH_MAX];
        struct diff_spec spec[2], *one, *two;
 
-       memcpy(spec[0].u.sha1, sha1, 20);
+       memcpy(spec[0].blob_sha1, sha1, 20);
        spec[0].mode = mode;
        spec[0].sha1_valid = spec[0].file_valid = 1;
        spec[1].file_valid = 0;
@@ -301,7 +389,7 @@ void diff_addremove(int addremove, unsigned mode,
                strcpy(concatpath, base);
                strcat(concatpath, path);
        }
-       run_external_diff(path ? concatpath : base, one, two);
+       run_external_diff(path ? concatpath : base, NULL, one, two);
 }
 
 void diff_change(unsigned old_mode, unsigned new_mode,
@@ -311,9 +399,9 @@ void diff_change(unsigned old_mode, unsigned new_mode,
        char concatpath[PATH_MAX];
        struct diff_spec spec[2];
 
-       memcpy(spec[0].u.sha1, old_sha1, 20);
+       memcpy(spec[0].blob_sha1, old_sha1, 20);
        spec[0].mode = old_mode;
-       memcpy(spec[1].u.sha1, new_sha1, 20);
+       memcpy(spec[1].blob_sha1, new_sha1, 20);
        spec[1].mode = new_mode;
        spec[0].sha1_valid = spec[0].file_valid = 1;
        spec[1].sha1_valid = spec[1].file_valid = 1;
@@ -322,10 +410,10 @@ void diff_change(unsigned old_mode, unsigned new_mode,
                strcpy(concatpath, base);
                strcat(concatpath, path);
        }
-       run_external_diff(path ? concatpath : base, &spec[0], &spec[1]);
+       run_external_diff(path ? concatpath : base, NULL, &spec[0], &spec[1]);
 }
 
 void diff_unmerge(const char *path)
 {
-       run_external_diff(path, NULL, NULL);
+       run_external_diff(path, NULL, NULL, NULL);
 }