Must not modify the_index.cache as it may be passed to realloc at some point.
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index fbb79d70a93dd0c6c46a1d24505e55f737d44189..71b340c5368fb287ce7d7b242aa51436ed5dda93 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -8,6 +8,7 @@
 #include "delta.h"
 #include "xdiff-interface.h"
 #include "color.h"
+#include "attr.h"
 
 #ifdef NO_FAST_WORKING_DIRECTORY
 #define FAST_WORKING_DIRECTORY 0
 #define FAST_WORKING_DIRECTORY 1
 #endif
 
-static int use_size_cache;
-
 static int diff_detect_rename_default;
-static int diff_rename_limit_default = -1;
+static int diff_rename_limit_default = 100;
 static int diff_use_color_default;
+int diff_auto_refresh_index = 1;
 
 static char diff_colors[][COLOR_MAXLEN] = {
        "\033[m",       /* reset */
@@ -51,6 +51,96 @@ static int parse_diff_color_slot(const char *var, int ofs)
        die("bad config variable '%s'", var);
 }
 
+static struct ll_diff_driver {
+       const char *name;
+       struct ll_diff_driver *next;
+       char *cmd;
+} *user_diff, **user_diff_tail;
+
+static void read_config_if_needed(void)
+{
+       if (!user_diff_tail) {
+               user_diff_tail = &user_diff;
+               git_config(git_diff_ui_config);
+       }
+}
+
+/*
+ * Currently there is only "diff.<drivername>.command" variable;
+ * because there are "diff.color.<slot>" variables, we are parsing
+ * this in a bit convoluted way to allow low level diff driver
+ * called "color".
+ */
+static int parse_lldiff_command(const char *var, const char *ep, const char *value)
+{
+       const char *name;
+       int namelen;
+       struct ll_diff_driver *drv;
+
+       name = var + 5;
+       namelen = ep - name;
+       for (drv = user_diff; drv; drv = drv->next)
+               if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
+                       break;
+       if (!drv) {
+               char *namebuf;
+               drv = xcalloc(1, sizeof(struct ll_diff_driver));
+               namebuf = xmalloc(namelen + 1);
+               memcpy(namebuf, name, namelen);
+               namebuf[namelen] = 0;
+               drv->name = namebuf;
+               drv->next = NULL;
+               if (!user_diff_tail)
+                       user_diff_tail = &user_diff;
+               *user_diff_tail = drv;
+               user_diff_tail = &(drv->next);
+       }
+
+       if (!value)
+               return error("%s: lacks value", var);
+       drv->cmd = strdup(value);
+       return 0;
+}
+
+/*
+ * 'diff.<what>.funcname' attribute can be specified in the configuration
+ * to define a customized regexp to find the beginning of a function to
+ * be used for hunk header lines of "diff -p" style output.
+ */
+static struct funcname_pattern {
+       char *name;
+       char *pattern;
+       struct funcname_pattern *next;
+} *funcname_pattern_list;
+
+static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
+{
+       const char *name;
+       int namelen;
+       struct funcname_pattern *pp;
+
+       name = var + 5; /* "diff." */
+       namelen = ep - name;
+
+       for (pp = funcname_pattern_list; pp; pp = pp->next)
+               if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
+                       break;
+       if (!pp) {
+               char *namebuf;
+               pp = xcalloc(1, sizeof(*pp));
+               namebuf = xmalloc(namelen + 1);
+               memcpy(namebuf, name, namelen);
+               namebuf[namelen] = 0;
+               pp->name = namebuf;
+               pp->next = funcname_pattern_list;
+               funcname_pattern_list = pp;
+       }
+       if (pp->pattern)
+               free(pp->pattern);
+       pp->pattern = xstrdup(value);
+       return 0;
+}
+
 /*
  * These are to give UI layer defaults.
  * The core-level commands such as git-diff-files should
@@ -77,11 +167,26 @@ int git_diff_ui_config(const char *var, const char *value)
                        diff_detect_rename_default = DIFF_DETECT_RENAME;
                return 0;
        }
+       if (!strcmp(var, "diff.autorefreshindex")) {
+               diff_auto_refresh_index = git_config_bool(var, value);
+               return 0;
+       }
+       if (!prefixcmp(var, "diff.")) {
+               const char *ep = strrchr(var, '.');
+
+               if (ep != var + 4) {
+                       if (!strcmp(ep, ".command"))
+                               return parse_lldiff_command(var, ep, value);
+                       if (!strcmp(ep, ".funcname"))
+                               return parse_funcname_pattern(var, ep, value);
+               }
+       }
        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
                int slot = parse_diff_color_slot(var, 11);
                color_parse(value, var, diff_colors[slot]);
                return 0;
        }
+
        return git_default_config(var, value);
 }
 
@@ -137,13 +242,11 @@ static const char *external_diff(void)
        return external_diff_cmd;
 }
 
-#define TEMPFILE_PATH_LEN              50
-
 static struct diff_tempfile {
        const char *name; /* filename external diff should read from */
        char hex[41];
        char mode[10];
-       char tmp_path[TEMPFILE_PATH_LEN];
+       char tmp_path[PATH_MAX];
 } diff_temp[2];
 
 static int count_lines(const char *data, int size)
@@ -343,6 +446,7 @@ static void diff_words_show(struct diff_words_data *diff_words)
        mmfile_t minus, plus;
        int i;
 
+       memset(&xecfg, 0, sizeof(xecfg));
        minus.size = diff_words->minus.text.size;
        minus.ptr = xmalloc(minus.size);
        memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
@@ -361,7 +465,6 @@ static void diff_words_show(struct diff_words_data *diff_words)
 
        xpp.flags = XDF_NEED_MINIMAL;
        xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
-       xecfg.flags = 0;
        ecb.outf = xdiff_outf;
        ecb.priv = diff_words;
        diff_words->xm.consume = fn_out_diff_words_aux;
@@ -1051,13 +1154,115 @@ static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
        emit_binary_diff_body(two, one);
 }
 
-#define FIRST_FEW_BYTES 8000
-static int mmfile_is_binary(mmfile_t *mf)
+static void setup_diff_attr_check(struct git_attr_check *check)
+{
+       static struct git_attr *attr_diff;
+
+       if (!attr_diff) {
+               attr_diff = git_attr("diff", 4);
+       }
+       check[0].attr = attr_diff;
+}
+
+static void diff_filespec_check_attr(struct diff_filespec *one)
 {
-       long sz = mf->size;
-       if (FIRST_FEW_BYTES < sz)
-               sz = FIRST_FEW_BYTES;
-       return !!memchr(mf->ptr, 0, sz);
+       struct git_attr_check attr_diff_check;
+       int check_from_data = 0;
+
+       if (one->checked_attr)
+               return;
+
+       setup_diff_attr_check(&attr_diff_check);
+       one->is_binary = 0;
+       one->funcname_pattern_ident = NULL;
+
+       if (!git_checkattr(one->path, 1, &attr_diff_check)) {
+               const char *value;
+
+               /* binaryness */
+               value = attr_diff_check.value;
+               if (ATTR_TRUE(value))
+                       ;
+               else if (ATTR_FALSE(value))
+                       one->is_binary = 1;
+               else
+                       check_from_data = 1;
+
+               /* funcname pattern ident */
+               if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
+                       ;
+               else
+                       one->funcname_pattern_ident = value;
+       }
+
+       if (check_from_data) {
+               if (!one->data && DIFF_FILE_VALID(one))
+                       diff_populate_filespec(one, 0);
+
+               if (one->data)
+                       one->is_binary = buffer_is_binary(one->data, one->size);
+       }
+}
+
+int diff_filespec_is_binary(struct diff_filespec *one)
+{
+       diff_filespec_check_attr(one);
+       return one->is_binary;
+}
+
+static const char *funcname_pattern(const char *ident)
+{
+       struct funcname_pattern *pp;
+
+       read_config_if_needed();
+       for (pp = funcname_pattern_list; pp; pp = pp->next)
+               if (!strcmp(ident, pp->name))
+                       return pp->pattern;
+       return NULL;
+}
+
+static struct builtin_funcname_pattern {
+       const char *name;
+       const char *pattern;
+} builtin_funcname_pattern[] = {
+       { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
+                       "new\\|return\\|switch\\|throw\\|while\\)\n"
+                       "^[     ]*\\(\\([       ]*"
+                       "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
+                       "[      ]*([^;]*$\\)" },
+       { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
+};
+
+static const char *diff_funcname_pattern(struct diff_filespec *one)
+{
+       const char *ident, *pattern;
+       int i;
+
+       diff_filespec_check_attr(one);
+       ident = one->funcname_pattern_ident;
+
+       if (!ident)
+               /*
+                * If the config file has "funcname.default" defined, that
+                * regexp is used; otherwise NULL is returned and xemit uses
+                * the built-in default.
+                */
+               return funcname_pattern("default");
+
+       /* Look up custom "funcname.$ident" regexp from config. */
+       pattern = funcname_pattern(ident);
+       if (pattern)
+               return pattern;
+
+       /*
+        * And define built-in fallback patterns here.  Note that
+        * these can be overriden by the user's config settings.
+        */
+       for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
+               if (!strcmp(ident, builtin_funcname_pattern[i].name))
+                       return builtin_funcname_pattern[i].pattern;
+
+       return NULL;
 }
 
 static void builtin_diff(const char *name_a,
@@ -1114,7 +1319,8 @@ static void builtin_diff(const char *name_a,
        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
                die("unable to read files to diff");
 
-       if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
+       if (!o->text &&
+           (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
                /* Quite common confusing case */
                if (mf1.size == mf2.size &&
                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
@@ -1133,7 +1339,13 @@ static void builtin_diff(const char *name_a,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
                struct emit_callback ecbdata;
+               const char *funcname_pattern;
 
+               funcname_pattern = diff_funcname_pattern(one);
+               if (!funcname_pattern)
+                       funcname_pattern = diff_funcname_pattern(two);
+
+               memset(&xecfg, 0, sizeof(xecfg));
                memset(&ecbdata, 0, sizeof(ecbdata));
                ecbdata.label_path = lbl;
                ecbdata.color_diff = o->color_diff;
@@ -1141,6 +1353,8 @@ static void builtin_diff(const char *name_a,
                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
                xecfg.ctxlen = o->context;
                xecfg.flags = XDL_EMIT_FUNCNAMES;
+               if (funcname_pattern)
+                       xdiff_set_find_func(&xecfg, funcname_pattern);
                if (!diffopts)
                        ;
                else if (!prefixcmp(diffopts, "--unified="))
@@ -1159,6 +1373,8 @@ static void builtin_diff(const char *name_a,
        }
 
  free_ab_and_return:
+       diff_free_filespec_data(one);
+       diff_free_filespec_data(two);
        free(a_one);
        free(b_two);
        return;
@@ -1185,12 +1401,12 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
                diff_populate_filespec(two, 0);
                data->deleted = count_lines(one->data, one->size);
                data->added = count_lines(two->data, two->size);
-               return;
+               goto free_and_return;
        }
        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
                die("unable to read files to diff");
 
-       if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2)) {
+       if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
                data->is_binary = 1;
                data->added = mf2.size;
                data->deleted = mf1.size;
@@ -1200,13 +1416,16 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
 
+               memset(&xecfg, 0, sizeof(xecfg));
                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
-               xecfg.ctxlen = 0;
-               xecfg.flags = 0;
                ecb.outf = xdiff_outf;
                ecb.priv = diffstat;
                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
        }
+
+ free_and_return:
+       diff_free_filespec_data(one);
+       diff_free_filespec_data(two);
 }
 
 static void builtin_checkdiff(const char *name_a, const char *name_b,
@@ -1228,21 +1447,23 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
                die("unable to read files to diff");
 
-       if (mmfile_is_binary(&mf2))
-               return;
+       if (diff_filespec_is_binary(two))
+               goto free_and_return;
        else {
                /* Crazy xdl interfaces.. */
                xpparam_t xpp;
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
 
+               memset(&xecfg, 0, sizeof(xecfg));
                xpp.flags = XDF_NEED_MINIMAL;
-               xecfg.ctxlen = 0;
-               xecfg.flags = 0;
                ecb.outf = xdiff_outf;
                ecb.priv = &data;
                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
        }
+ free_and_return:
+       diff_free_filespec_data(one);
+       diff_free_filespec_data(two);
 }
 
 struct diff_filespec *alloc_filespec(const char *path)
@@ -1267,7 +1488,7 @@ void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 }
 
 /*
- * Given a name and sha1 pair, if the dircache tells us the file in
+ * Given a name and sha1 pair, if the index 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.
  */
@@ -1322,61 +1543,12 @@ static int reuse_worktree_file(const char *name, const unsigned char *sha1, int
        return 1;
 }
 
-static struct sha1_size_cache {
-       unsigned char sha1[20];
-       unsigned long size;
-} **sha1_size_cache;
-static int sha1_size_cache_nr, sha1_size_cache_alloc;
-
-static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
-                                                int find_only,
-                                                unsigned long size)
-{
-       int first, last;
-       struct sha1_size_cache *e;
-
-       first = 0;
-       last = sha1_size_cache_nr;
-       while (last > first) {
-               int cmp, next = (last + first) >> 1;
-               e = sha1_size_cache[next];
-               cmp = hashcmp(e->sha1, sha1);
-               if (!cmp)
-                       return e;
-               if (cmp < 0) {
-                       last = next;
-                       continue;
-               }
-               first = next+1;
-       }
-       /* not found */
-       if (find_only)
-               return NULL;
-       /* insert to make it at "first" */
-       if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
-               sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
-               sha1_size_cache = xrealloc(sha1_size_cache,
-                                          sha1_size_cache_alloc *
-                                          sizeof(*sha1_size_cache));
-       }
-       sha1_size_cache_nr++;
-       if (first < sha1_size_cache_nr)
-               memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
-                       (sha1_size_cache_nr - first - 1) *
-                       sizeof(*sha1_size_cache));
-       e = xmalloc(sizeof(struct sha1_size_cache));
-       sha1_size_cache[first] = e;
-       hashcpy(e->sha1, sha1);
-       e->size = size;
-       return e;
-}
-
 static int populate_from_stdin(struct diff_filespec *s)
 {
 #define INCREMENT 1024
        char *buf;
        unsigned long size;
-       int got;
+       ssize_t got;
 
        size = 0;
        buf = NULL;
@@ -1397,6 +1569,22 @@ static int populate_from_stdin(struct diff_filespec *s)
        return 0;
 }
 
+static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
+{
+       int len;
+       char *data = xmalloc(100);
+       len = snprintf(data, 100,
+               "Subproject commit %s\n", sha1_to_hex(s->sha1));
+       s->data = data;
+       s->size = len;
+       s->should_free = 1;
+       if (size_only) {
+               s->data = NULL;
+               free(data);
+       }
+       return 0;
+}
+
 /*
  * 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.
@@ -1410,11 +1598,15 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
        if (S_ISDIR(s->mode))
                return -1;
 
-       if (!use_size_cache)
-               size_only = 0;
-
        if (s->data)
-               return err;
+               return 0;
+
+       if (size_only && 0 < s->size)
+               return 0;
+
+       if (S_ISGITLINK(s->mode))
+               return diff_populate_gitlink(s, size_only);
+
        if (!s->sha1_valid ||
            reuse_worktree_file(s->path, s->sha1, 0)) {
                struct stat st;
@@ -1461,9 +1653,9 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
                /*
                 * Convert from working tree format to canonical git format
                 */
-               buf = s->data;
                size = s->size;
-               if (convert_to_git(s->path, &buf, &size)) {
+               buf = convert_to_git(s->path, s->data, &size);
+               if (buf) {
                        munmap(s->data, s->size);
                        s->should_munmap = 0;
                        s->data = buf;
@@ -1473,18 +1665,8 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
        }
        else {
                enum object_type type;
-               struct sha1_size_cache *e;
-
-               if (size_only) {
-                       e = locate_size_cache(s->sha1, 1, 0);
-                       if (e) {
-                               s->size = e->size;
-                               return 0;
-                       }
+               if (size_only)
                        type = sha1_object_info(s->sha1, &s->size);
-                       if (type < 0)
-                               locate_size_cache(s->sha1, 0, s->size);
-               }
                else {
                        s->data = read_sha1_file(s->sha1, &type, &s->size);
                        s->should_free = 1;
@@ -1493,14 +1675,22 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
        return 0;
 }
 
-void diff_free_filespec_data(struct diff_filespec *s)
+void diff_free_filespec_blob(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;
+
+       if (s->should_free || s->should_munmap) {
+               s->should_free = s->should_munmap = 0;
+               s->data = NULL;
+       }
+}
+
+void diff_free_filespec_data(struct diff_filespec *s)
+{
+       diff_free_filespec_blob(s);
        free(s->cnt_data);
        s->cnt_data = NULL;
 }
@@ -1513,9 +1703,9 @@ static void prep_temp_blob(struct diff_tempfile *temp,
 {
        int fd;
 
-       fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
+       fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
        if (fd < 0)
-               die("unable to create temp-file");
+               die("unable to create temp-file: %s", strerror(errno));
        if (write_in_full(fd, blob, size) != size)
                die("unable to write temp-file");
        close(fd);
@@ -1700,6 +1890,27 @@ static void run_external_diff(const char *pgm,
        }
 }
 
+static const char *external_diff_attr(const char *name)
+{
+       struct git_attr_check attr_diff_check;
+
+       setup_diff_attr_check(&attr_diff_check);
+       if (!git_checkattr(name, 1, &attr_diff_check)) {
+               const char *value = attr_diff_check.value;
+               if (!ATTR_TRUE(value) &&
+                   !ATTR_FALSE(value) &&
+                   !ATTR_UNSET(value)) {
+                       struct ll_diff_driver *drv;
+
+                       read_config_if_needed();
+                       for (drv = user_diff; drv; drv = drv->next)
+                               if (!strcmp(drv->name, value))
+                                       return drv->cmd;
+               }
+       }
+       return NULL;
+}
+
 static void run_diff_cmd(const char *pgm,
                         const char *name,
                         const char *other,
@@ -1709,6 +1920,14 @@ static void run_diff_cmd(const char *pgm,
                         struct diff_options *o,
                         int complete_rewrite)
 {
+       if (!o->allow_external)
+               pgm = NULL;
+       else {
+               const char *cmd = external_diff_attr(name);
+               if (cmd)
+                       pgm = cmd;
+       }
+
        if (pgm) {
                run_external_diff(pgm, name, other, one, two, xfrm_msg,
                                  complete_rewrite);
@@ -1740,6 +1959,11 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
                hashclr(one->sha1);
 }
 
+static int similarity_index(struct diff_filepair *p)
+{
+       return p->score * 100 / MAX_SCORE;
+}
+
 static void run_diff(struct diff_filepair *p, struct diff_options *o)
 {
        const char *pgm = external_diff();
@@ -1774,23 +1998,20 @@ static void run_diff(struct diff_filepair *p, struct diff_options *o)
                                "similarity index %d%%\n"
                                "copy from %s\n"
                                "copy to %s\n",
-                               (int)(0.5 + p->score * 100.0/MAX_SCORE),
-                               name_munged, other_munged);
+                               similarity_index(p), name_munged, other_munged);
                break;
        case DIFF_STATUS_RENAMED:
                len += snprintf(msg + len, sizeof(msg) - len,
                                "similarity index %d%%\n"
                                "rename from %s\n"
                                "rename to %s\n",
-                               (int)(0.5 + p->score * 100.0/MAX_SCORE),
-                               name_munged, other_munged);
+                               similarity_index(p), name_munged, other_munged);
                break;
        case DIFF_STATUS_MODIFIED:
                if (p->score) {
                        len += snprintf(msg + len, sizeof(msg) - len,
                                        "dissimilarity index %d%%\n",
-                                       (int)(0.5 + p->score *
-                                             100.0/MAX_SCORE));
+                                       similarity_index(p));
                        complete_rewrite = 1;
                        break;
                }
@@ -1805,8 +2026,8 @@ static void run_diff(struct diff_filepair *p, struct diff_options *o)
 
                if (o->binary) {
                        mmfile_t mf;
-                       if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
-                           (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
+                       if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
+                           (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
                                abbrev = 40;
                }
                len += snprintf(msg + len, sizeof(msg) - len,
@@ -1960,8 +2181,6 @@ int diff_setup_done(struct diff_options *options)
                         */
                        read_cache();
        }
-       if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
-               use_size_cache = 1;
        if (options->abbrev <= 0 || 40 < options->abbrev)
                options->abbrev = 40; /* full */
 
@@ -2032,6 +2251,8 @@ static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *va
        return 1;
 }
 
+static int diff_scoreopt_parse(const char *opt);
+
 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 {
        const char *arg = av[0];
@@ -2128,6 +2349,8 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->detect_rename = DIFF_DETECT_RENAME;
        }
        else if (!prefixcmp(arg, "-C")) {
+               if (options->detect_rename == DIFF_DETECT_COPY)
+                       options->find_copies_harder = 1;
                if ((options->rename_score =
                     diff_scoreopt_parse(arg)) == -1)
                        return -1;
@@ -2135,6 +2358,8 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        }
        else if (!strcmp(arg, "--find-copies-harder"))
                options->find_copies_harder = 1;
+       else if (!strcmp(arg, "--follow"))
+               options->follow_renames = 1;
        else if (!strcmp(arg, "--abbrev"))
                options->abbrev = DEFAULT_ABBREV;
        else if (!prefixcmp(arg, "--abbrev=")) {
@@ -2162,6 +2387,10 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->exit_with_status = 1;
        else if (!strcmp(arg, "--quiet"))
                options->quiet = 1;
+       else if (!strcmp(arg, "--ext-diff"))
+               options->allow_external = 1;
+       else if (!strcmp(arg, "--no-ext-diff"))
+               options->allow_external = 0;
        else
                return 0;
        return 1;
@@ -2203,7 +2432,7 @@ static int parse_num(const char **cp_p)
        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
 }
 
-int diff_scoreopt_parse(const char *opt)
+static int diff_scoreopt_parse(const char *opt)
 {
        int opt1, opt2, cmd;
 
@@ -2310,8 +2539,7 @@ static void diff_flush_raw(struct diff_filepair *p,
        }
 
        if (p->score)
-               sprintf(status, "%c%03d", p->status,
-                       (int)(0.5 + p->score * 100.0/MAX_SCORE));
+               sprintf(status, "%c%03d", p->status, similarity_index(p));
        else {
                status[0] = p->status;
                status[1] = 0;
@@ -2336,7 +2564,8 @@ static void diff_flush_raw(struct diff_filepair *p,
                printf("%s ",
                       diff_unique_abbrev(p->two->sha1, abbrev));
        }
-       printf("%s%c%s", status, inter_name_termination, path_one);
+       printf("%s%c%s", status, inter_name_termination,
+                       two_paths || p->one->mode ?  path_one : path_two);
        if (two_paths)
                printf("%c%s", inter_name_termination, path_two);
        putchar(line_termination);
@@ -2593,8 +2822,7 @@ static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
 {
        char *names = pprint_rename(p->one->path, p->two->path);
 
-       printf(" %s %s (%d%%)\n", renamecopy, names,
-              (int)(0.5 + p->score * 100.0/MAX_SCORE));
+       printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
        free(names);
        show_mode_change(p, 0);
 }
@@ -2618,7 +2846,7 @@ static void diff_summary(struct diff_filepair *p)
                if (p->score) {
                        char *name = quote_one(p->two->path);
                        printf(" rewrite %s (%d%%)\n", name,
-                               (int)(0.5 + p->score * 100.0/MAX_SCORE));
+                              similarity_index(p));
                        free(name);
                        show_mode_change(p, 0);
                } else  show_mode_change(p, 1);
@@ -2682,6 +2910,7 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                struct diff_filepair *p = q->queue[i];
                int len1, len2;
 
+               memset(&xecfg, 0, sizeof(xecfg));
                if (p->status == 0)
                        return error("internal diff status error");
                if (p->status == DIFF_STATUS_UNKNOWN)
@@ -2700,10 +2929,6 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                                fill_mmfile(&mf2, p->two) < 0)
                        return error("unable to read files to diff");
 
-               /* Maybe hash p->two? into the patch id? */
-               if (mmfile_is_binary(&mf2))
-                       continue;
-
                len1 = remove_space(p->one->path, strlen(p->one->path));
                len2 = remove_space(p->two->path, strlen(p->two->path));
                if (p->one->mode == 0)
@@ -2924,10 +3149,77 @@ static void diffcore_apply_filter(const char *filter)
        *q = outq;
 }
 
+/* Check whether two filespecs with the same mode and size are identical */
+static int diff_filespec_is_identical(struct diff_filespec *one,
+                                     struct diff_filespec *two)
+{
+       if (S_ISGITLINK(one->mode)) {
+               diff_fill_sha1_info(one);
+               diff_fill_sha1_info(two);
+               return !hashcmp(one->sha1, two->sha1);
+       }
+       if (diff_populate_filespec(one, 0))
+               return 0;
+       if (diff_populate_filespec(two, 0))
+               return 0;
+       return !memcmp(one->data, two->data, one->size);
+}
+
+static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
+{
+       int i;
+       struct diff_queue_struct *q = &diff_queued_diff;
+       struct diff_queue_struct outq;
+       outq.queue = NULL;
+       outq.nr = outq.alloc = 0;
+
+       for (i = 0; i < q->nr; i++) {
+               struct diff_filepair *p = q->queue[i];
+
+               /*
+                * 1. Entries that come from stat info dirtyness
+                *    always have both sides (iow, not create/delete),
+                *    one side of the object name is unknown, with
+                *    the same mode and size.  Keep the ones that
+                *    do not match these criteria.  They have real
+                *    differences.
+                *
+                * 2. At this point, the file is known to be modified,
+                *    with the same mode and size, and the object
+                *    name of one side is unknown.  Need to inspect
+                *    the identical contents.
+                */
+               if (!DIFF_FILE_VALID(p->one) || /* (1) */
+                   !DIFF_FILE_VALID(p->two) ||
+                   (p->one->sha1_valid && p->two->sha1_valid) ||
+                   (p->one->mode != p->two->mode) ||
+                   diff_populate_filespec(p->one, 1) ||
+                   diff_populate_filespec(p->two, 1) ||
+                   (p->one->size != p->two->size) ||
+                   !diff_filespec_is_identical(p->one, p->two)) /* (2) */
+                       diff_q(&outq, p);
+               else {
+                       /*
+                        * The caller can subtract 1 from skip_stat_unmatch
+                        * to determine how many paths were dirty only
+                        * due to stat info mismatch.
+                        */
+                       if (!diffopt->no_index)
+                               diffopt->skip_stat_unmatch++;
+                       diff_free_filepair(p);
+               }
+       }
+       free(q->queue);
+       *q = outq;
+}
+
 void diffcore_std(struct diff_options *options)
 {
        if (options->quiet)
                return;
+
+       if (options->skip_stat_unmatch && !options->find_copies_harder)
+               diffcore_skip_stat_unmatch(options);
        if (options->break_opt != -1)
                diffcore_break(options->break_opt);
        if (options->detect_rename)
@@ -2960,7 +3252,7 @@ void diff_addremove(struct diff_options *options,
         * 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() 
+        * Feeding the same new and old to diff_change()
         * also has the same effect.
         * Before the final output happens, they are pruned after
         * merged into rename/copy pairs as appropriate.
@@ -2987,7 +3279,7 @@ void diff_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) 
+                const char *base, const char *path)
 {
        char concatpath[PATH_MAX];
        struct diff_filespec *one, *two;