Make read_patch_file work on a strbuf.
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index dc6da5e253c4a64440a2da9e2f785f3ea06bea61..ab575191d11801f023655fe6cf718bfa5fa04c9b 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -17,8 +17,9 @@
 #endif
 
 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 */
@@ -56,6 +57,14 @@ static struct ll_diff_driver {
        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
@@ -74,13 +83,8 @@ static int parse_lldiff_command(const char *var, const char *ep, const char *val
                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;
+               drv->name = xmemdupz(name, namelen);
                if (!user_diff_tail)
                        user_diff_tail = &user_diff;
                *user_diff_tail = drv;
@@ -93,6 +97,41 @@ static int parse_lldiff_command(const char *var, const char *ep, const char *val
        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) {
+               pp = xcalloc(1, sizeof(*pp));
+               pp->name = xmemdupz(name, namelen);
+               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
@@ -119,11 +158,19 @@ 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 && !strcmp(ep, ".command"))
-                       return parse_lldiff_command(var, ep, value);
+               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);
@@ -134,44 +181,23 @@ int git_diff_ui_config(const char *var, const char *value)
        return git_default_config(var, value);
 }
 
-static char *quote_one(const char *str)
-{
-       int needlen;
-       char *xp;
-
-       if (!str)
-               return NULL;
-       needlen = quote_c_style(str, NULL, NULL, 0);
-       if (!needlen)
-               return xstrdup(str);
-       xp = xmalloc(needlen + 1);
-       quote_c_style(str, xp, NULL, 0);
-       return xp;
-}
-
 static char *quote_two(const char *one, const char *two)
 {
        int need_one = quote_c_style(one, NULL, NULL, 1);
        int need_two = quote_c_style(two, NULL, NULL, 1);
-       char *xp;
+       struct strbuf res;
 
+       strbuf_init(&res, 0);
        if (need_one + need_two) {
-               if (!need_one) need_one = strlen(one);
-               if (!need_two) need_one = strlen(two);
-
-               xp = xmalloc(need_one + need_two + 3);
-               xp[0] = '"';
-               quote_c_style(one, xp + 1, NULL, 1);
-               quote_c_style(two, xp + need_one + 1, NULL, 1);
-               strcpy(xp + need_one + need_two + 1, "\"");
-               return xp;
+               strbuf_addch(&res, '"');
+               quote_c_style(one, &res, NULL, 1);
+               quote_c_style(two, &res, NULL, 1);
+               strbuf_addch(&res, '"');
+       } else {
+               strbuf_addstr(&res, one);
+               strbuf_addstr(&res, two);
        }
-       need_one = strlen(one);
-       need_two = strlen(two);
-       xp = xmalloc(need_one + need_two + 1);
-       strcpy(xp, one);
-       strcpy(xp + need_one, two);
-       return xp;
+       return strbuf_detach(&res, NULL);
 }
 
 static const char *external_diff(void)
@@ -390,6 +416,7 @@ 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);
@@ -408,7 +435,6 @@ 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;
@@ -623,27 +649,20 @@ static char *pprint_rename(const char *a, const char *b)
 {
        const char *old = a;
        const char *new = b;
-       char *name = NULL;
+       struct strbuf name;
        int pfx_length, sfx_length;
        int len_a = strlen(a);
        int len_b = strlen(b);
+       int a_midlen, b_midlen;
        int qlen_a = quote_c_style(a, NULL, NULL, 0);
        int qlen_b = quote_c_style(b, NULL, NULL, 0);
 
+       strbuf_init(&name, 0);
        if (qlen_a || qlen_b) {
-               if (qlen_a) len_a = qlen_a;
-               if (qlen_b) len_b = qlen_b;
-               name = xmalloc( len_a + len_b + 5 );
-               if (qlen_a)
-                       quote_c_style(a, name, NULL, 0);
-               else
-                       memcpy(name, a, len_a);
-               memcpy(name + len_a, " => ", 4);
-               if (qlen_b)
-                       quote_c_style(b, name + len_a + 4, NULL, 0);
-               else
-                       memcpy(name + len_a + 4, b, len_b + 1);
-               return name;
+               quote_c_style(a, &name, NULL, 0);
+               strbuf_addstr(&name, " => ");
+               quote_c_style(b, &name, NULL, 0);
+               return strbuf_detach(&name, NULL);
        }
 
        /* Find common prefix */
@@ -672,24 +691,26 @@ static char *pprint_rename(const char *a, const char *b)
         * pfx{sfx-a => sfx-b}
         * name-a => name-b
         */
+       a_midlen = len_a - pfx_length - sfx_length;
+       b_midlen = len_b - pfx_length - sfx_length;
+       if (a_midlen < 0)
+               a_midlen = 0;
+       if (b_midlen < 0)
+               b_midlen = 0;
+
+       strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
        if (pfx_length + sfx_length) {
-               int a_midlen = len_a - pfx_length - sfx_length;
-               int b_midlen = len_b - pfx_length - sfx_length;
-               if (a_midlen < 0) a_midlen = 0;
-               if (b_midlen < 0) b_midlen = 0;
-
-               name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
-               sprintf(name, "%.*s{%.*s => %.*s}%s",
-                       pfx_length, a,
-                       a_midlen, a + pfx_length,
-                       b_midlen, b + pfx_length,
-                       a + len_a - sfx_length);
+               strbuf_add(&name, a, pfx_length);
+               strbuf_addch(&name, '{');
        }
-       else {
-               name = xmalloc(len_a + len_b + 5);
-               sprintf(name, "%s => %s", a, b);
+       strbuf_add(&name, a + pfx_length, a_midlen);
+       strbuf_addstr(&name, " => ");
+       strbuf_add(&name, b + pfx_length, b_midlen);
+       if (pfx_length + sfx_length) {
+               strbuf_addch(&name, '}');
+               strbuf_add(&name, a + len_a - sfx_length, sfx_length);
        }
-       return name;
+       return strbuf_detach(&name, NULL);
 }
 
 struct diffstat_t {
@@ -802,12 +823,13 @@ static void show_stats(struct diffstat_t* data, struct diff_options *options)
                int change = file->added + file->deleted;
 
                if (!file->is_renamed) {  /* renames are already quoted by pprint_rename */
-                       len = quote_c_style(file->name, NULL, NULL, 0);
-                       if (len) {
-                               char *qname = xmalloc(len + 1);
-                               quote_c_style(file->name, qname, NULL, 0);
+                       struct strbuf buf;
+                       strbuf_init(&buf, 0);
+                       if (quote_c_style(file->name, &buf, NULL, 0)) {
                                free(file->name);
-                               file->name = qname;
+                               file->name = strbuf_detach(&buf, NULL);
+                       } else {
+                               strbuf_release(&buf);
                        }
                }
 
@@ -945,12 +967,12 @@ static void show_numstat(struct diffstat_t* data, struct diff_options *options)
                        printf("-\t-\t");
                else
                        printf("%d\t%d\t", file->added, file->deleted);
-               if (options->line_termination && !file->is_renamed &&
-                   quote_c_style(file->name, NULL, NULL, 0))
-                       quote_c_style(file->name, NULL, stdout, 0);
-               else
+               if (!file->is_renamed) {
+                       write_name_quoted(file->name, stdout, options->line_termination);
+               } else {
                        fputs(file->name, stdout);
-               putchar(options->line_termination);
+                       putchar(options->line_termination);
+               }
        }
 }
 
@@ -1102,30 +1124,111 @@ static void setup_diff_attr_check(struct git_attr_check *check)
 {
        static struct git_attr *attr_diff;
 
-       if (!attr_diff)
+       if (!attr_diff) {
                attr_diff = git_attr("diff", 4);
-       check->attr = attr_diff;
+       }
+       check[0].attr = attr_diff;
 }
 
-static int file_is_binary(struct diff_filespec *one)
+static void diff_filespec_check_attr(struct diff_filespec *one)
 {
        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 = attr_diff_check.value;
+               const char *value;
+
+               /* binaryness */
+               value = attr_diff_check.value;
                if (ATTR_TRUE(value))
-                       return 0;
+                       ;
                else if (ATTR_FALSE(value))
-                       return 1;
+                       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 (!one->data) {
-               if (!DIFF_FILE_VALID(one))
-                       return 0;
-               diff_populate_filespec(one, 0);
+       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);
        }
-       return 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,
@@ -1182,7 +1285,8 @@ 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 && (file_is_binary(one) || file_is_binary(two))) {
+       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))
@@ -1201,7 +1305,13 @@ 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;
@@ -1209,6 +1319,8 @@ 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="))
@@ -1260,7 +1372,7 @@ static void builtin_diffstat(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 (file_is_binary(one) || file_is_binary(two)) {
+       if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
                data->is_binary = 1;
                data->added = mf2.size;
                data->deleted = mf1.size;
@@ -1270,9 +1382,8 @@ 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);
@@ -1302,7 +1413,7 @@ 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 (file_is_binary(two))
+       if (diff_filespec_is_binary(two))
                goto free_and_return;
        else {
                /* Crazy xdl interfaces.. */
@@ -1310,9 +1421,8 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
                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);
@@ -1401,26 +1511,15 @@ static int reuse_worktree_file(const char *name, const unsigned char *sha1, int
 
 static int populate_from_stdin(struct diff_filespec *s)
 {
-#define INCREMENT 1024
-       char *buf;
-       unsigned long size;
-       ssize_t got;
-
-       size = 0;
-       buf = NULL;
-       while (1) {
-               buf = xrealloc(buf, size + INCREMENT);
-               got = xread(0, buf + size, INCREMENT);
-               if (!got)
-                       break; /* EOF */
-               if (got < 0)
-                       return error("error while reading from stdin %s",
+       struct strbuf buf;
+
+       strbuf_init(&buf, 0);
+       if (strbuf_read(&buf, 0, 0) < 0)
+               return error("error while reading from stdin %s",
                                     strerror(errno));
-               size += got;
-       }
+
        s->should_munmap = 0;
-       s->data = buf;
-       s->size = size;
+       s->data = strbuf_detach(&buf, &s->size);
        s->should_free = 1;
        return 0;
 }
@@ -1460,15 +1559,14 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
        if (size_only && 0 < s->size)
                return 0;
 
-       if (S_ISDIRLNK(s->mode))
+       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 strbuf buf;
                struct stat st;
                int fd;
-               char *buf;
-               unsigned long size;
 
                if (!strcmp(s->path, "-"))
                        return populate_from_stdin(s);
@@ -1509,13 +1607,11 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
                /*
                 * Convert from working tree format to canonical git format
                 */
-               size = s->size;
-               buf = convert_to_git(s->path, s->data, &size);
-               if (buf) {
+               strbuf_init(&buf, 0);
+               if (convert_to_git(s->path, s->data, s->size, &buf)) {
                        munmap(s->data, s->size);
                        s->should_munmap = 0;
-                       s->data = buf;
-                       s->size = size;
+                       s->data = strbuf_detach(&buf, &s->size);
                        s->should_free = 1;
                }
        }
@@ -1556,7 +1652,7 @@ static void prep_temp_blob(struct diff_tempfile *temp,
 
        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);
@@ -1753,10 +1849,7 @@ static const char *external_diff_attr(const char *name)
                    !ATTR_UNSET(value)) {
                        struct ll_diff_driver *drv;
 
-                       if (!user_diff_tail) {
-                               user_diff_tail = &user_diff;
-                               git_config(git_diff_ui_config);
-                       }
+                       read_config_if_needed();
                        for (drv = user_diff; drv; drv = drv->next)
                                if (!strcmp(drv->name, value))
                                        return drv->cmd;
@@ -1813,57 +1906,55 @@ 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();
-       char msg[PATH_MAX*2+300], *xfrm_msg;
-       struct diff_filespec *one;
-       struct diff_filespec *two;
+       struct strbuf msg;
+       char *xfrm_msg;
+       struct diff_filespec *one = p->one;
+       struct diff_filespec *two = p->two;
        const char *name;
        const char *other;
-       char *name_munged, *other_munged;
        int complete_rewrite = 0;
-       int len;
+
 
        if (DIFF_PAIR_UNMERGED(p)) {
-               /* unmerged */
                run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
                return;
        }
 
-       name = p->one->path;
+       name  = p->one->path;
        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
-       name_munged = quote_one(name);
-       other_munged = quote_one(other);
-       one = p->one; two = p->two;
-
        diff_fill_sha1_info(one);
        diff_fill_sha1_info(two);
 
-       len = 0;
+       strbuf_init(&msg, PATH_MAX * 2 + 300);
        switch (p->status) {
        case DIFF_STATUS_COPIED:
-               len += snprintf(msg + len, sizeof(msg) - len,
-                               "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);
+               strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
+               strbuf_addstr(&msg, "\ncopy from ");
+               quote_c_style(name, &msg, NULL, 0);
+               strbuf_addstr(&msg, "\ncopy to ");
+               quote_c_style(other, &msg, NULL, 0);
+               strbuf_addch(&msg, '\n');
                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);
+               strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
+               strbuf_addstr(&msg, "\nrename from ");
+               quote_c_style(name, &msg, NULL, 0);
+               strbuf_addstr(&msg, "\nrename to ");
+               quote_c_style(other, &msg, NULL, 0);
+               strbuf_addch(&msg, '\n');
                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));
+                       strbuf_addf(&msg, "dissimilarity index %d%%\n",
+                                       similarity_index(p));
                        complete_rewrite = 1;
                        break;
                }
@@ -1878,23 +1969,21 @@ static void run_diff(struct diff_filepair *p, struct diff_options *o)
 
                if (o->binary) {
                        mmfile_t mf;
-                       if ((!fill_mmfile(&mf, one) && file_is_binary(one)) ||
-                           (!fill_mmfile(&mf, two) && file_is_binary(two)))
+                       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,
-                               "index %.*s..%.*s",
+               strbuf_addf(&msg, "index %.*s..%.*s",
                                abbrev, sha1_to_hex(one->sha1),
                                abbrev, sha1_to_hex(two->sha1));
                if (one->mode == two->mode)
-                       len += snprintf(msg + len, sizeof(msg) - len,
-                                       " %06o", one->mode);
-               len += snprintf(msg + len, sizeof(msg) - len, "\n");
+                       strbuf_addf(&msg, " %06o", one->mode);
+               strbuf_addch(&msg, '\n');
        }
 
-       if (len)
-               msg[--len] = 0;
-       xfrm_msg = len ? msg : NULL;
+       if (msg.len)
+               strbuf_setlen(&msg, msg.len - 1);
+       xfrm_msg = msg.len ? msg.buf : NULL;
 
        if (!pgm &&
            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
@@ -1913,8 +2002,7 @@ static void run_diff(struct diff_filepair *p, struct diff_options *o)
                run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
                             complete_rewrite);
 
-       free(name_munged);
-       free(other_munged);
+       strbuf_release(&msg);
 }
 
 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
@@ -2103,6 +2191,8 @@ 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];
@@ -2199,6 +2289,8 @@ 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;
@@ -2206,6 +2298,8 @@ 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=")) {
@@ -2233,6 +2327,10 @@ 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;
@@ -2274,7 +2372,7 @@ 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;
 
@@ -2360,72 +2458,30 @@ const char *diff_unique_abbrev(const unsigned char *sha1, int len)
        return sha1_to_hex(sha1);
 }
 
-static void diff_flush_raw(struct diff_filepair *p,
-                          struct diff_options *options)
+static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
 {
-       int two_paths;
-       char status[10];
-       int abbrev = options->abbrev;
-       const char *path_one, *path_two;
-       int inter_name_termination = '\t';
-       int line_termination = options->line_termination;
+       int line_termination = opt->line_termination;
+       int inter_name_termination = line_termination ? '\t' : '\0';
 
-       if (!line_termination)
-               inter_name_termination = 0;
-
-       path_one = p->one->path;
-       path_two = p->two->path;
-       if (line_termination) {
-               path_one = quote_one(path_one);
-               path_two = quote_one(path_two);
-       }
-
-       if (p->score)
-               sprintf(status, "%c%03d", p->status,
-                       (int)(0.5 + p->score * 100.0/MAX_SCORE));
-       else {
-               status[0] = p->status;
-               status[1] = 0;
-       }
-       switch (p->status) {
-       case DIFF_STATUS_COPIED:
-       case DIFF_STATUS_RENAMED:
-               two_paths = 1;
-               break;
-       case DIFF_STATUS_ADDED:
-       case DIFF_STATUS_DELETED:
-               two_paths = 0;
-               break;
-       default:
-               two_paths = 0;
-               break;
+       if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
+               printf(":%06o %06o %s ", p->one->mode, p->two->mode,
+                      diff_unique_abbrev(p->one->sha1, opt->abbrev));
+               printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
        }
-       if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
-               printf(":%06o %06o %s ",
-                      p->one->mode, p->two->mode,
-                      diff_unique_abbrev(p->one->sha1, abbrev));
-               printf("%s ",
-                      diff_unique_abbrev(p->two->sha1, abbrev));
+       if (p->score) {
+               printf("%c%03d%c", p->status, similarity_index(p),
+                          inter_name_termination);
+       } else {
+               printf("%c%c", p->status, inter_name_termination);
        }
-       printf("%s%c%s", status, inter_name_termination, path_one);
-       if (two_paths)
-               printf("%c%s", inter_name_termination, path_two);
-       putchar(line_termination);
-       if (path_one != p->one->path)
-               free((void*)path_one);
-       if (path_two != p->two->path)
-               free((void*)path_two);
-}
-
-static void diff_flush_name(struct diff_filepair *p, struct diff_options *opt)
-{
-       char *path = p->two->path;
 
-       if (opt->line_termination)
-               path = quote_one(p->two->path);
-       printf("%s%c", path, opt->line_termination);
-       if (p->two->path != path)
-               free(path);
+       if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
+               write_name_quoted(p->one->path, stdout, inter_name_termination);
+               write_name_quoted(p->two->path, stdout, line_termination);
+       } else {
+               const char *path = p->one->mode ? p->one->path : p->two->path;
+               write_name_quoted(path, stdout, line_termination);
+       }
 }
 
 int diff_unmodified_pair(struct diff_filepair *p)
@@ -2435,14 +2491,11 @@ int diff_unmodified_pair(struct diff_filepair *p)
         * 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;
+       struct diff_filespec *one = p->one, *two = p->two;
 
        if (DIFF_PAIR_UNMERGED(p))
                return 0; /* unmerged is interesting */
 
-       one = p->one;
-       two = p->two;
-
        /* deletion, addition, mode or type change
         * and rename are all interesting.
         */
@@ -2631,32 +2684,27 @@ static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
                diff_flush_raw(p, opt);
        else if (fmt & DIFF_FORMAT_NAME)
-               diff_flush_name(p, opt);
+               write_name_quoted(p->two->path, stdout, opt->line_termination);
 }
 
 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
 {
-       char *name = quote_one(fs->path);
        if (fs->mode)
-               printf(" %s mode %06o %s\n", newdelete, fs->mode, name);
+               printf(" %s mode %06o ", newdelete, fs->mode);
        else
-               printf(" %s %s\n", newdelete, name);
-       free(name);
+               printf(" %s ", newdelete);
+       write_name_quoted(fs->path, stdout, '\n');
 }
 
 
 static void show_mode_change(struct diff_filepair *p, int show_name)
 {
        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
+               printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
+                       show_name ? ' ' : '\n');
                if (show_name) {
-                       char *name = quote_one(p->two->path);
-                       printf(" mode change %06o => %06o %s\n",
-                              p->one->mode, p->two->mode, name);
-                       free(name);
+                       write_name_quoted(p->two->path, stdout, '\n');
                }
-               else
-                       printf(" mode change %06o => %06o\n",
-                              p->one->mode, p->two->mode);
        }
 }
 
@@ -2664,8 +2712,7 @@ 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);
 }
@@ -2687,12 +2734,11 @@ static void diff_summary(struct diff_filepair *p)
                break;
        default:
                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));
-                       free(name);
-                       show_mode_change(p, 0);
-               } else  show_mode_change(p, 1);
+                       puts(" rewrite ");
+                       write_name_quoted(p->two->path, stdout, ' ');
+                       printf("(%d%%)\n", similarity_index(p));
+               }
+               show_mode_change(p, !p->score);
                break;
        }
 }
@@ -2706,14 +2752,14 @@ struct patch_id_t {
 static int remove_space(char *line, int len)
 {
        int i;
-        char *dst = line;
-        unsigned char c;
+       char *dst = line;
+       unsigned char c;
 
-        for (i = 0; i < len; i++)
-                if (!isspace((c = line[i])))
-                        *dst++ = c;
+       for (i = 0; i < len; i++)
+               if (!isspace((c = line[i])))
+                       *dst++ = c;
 
-        return dst - line;
+       return dst - line;
 }
 
 static void patch_id_consume(void *priv, char *line, unsigned long len)
@@ -2753,6 +2799,7 @@ 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)
@@ -2771,10 +2818,6 @@ 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 (file_is_binary(p->two))
-                       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)
@@ -2995,10 +3038,77 @@ 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)
@@ -3031,7 +3141,7 @@ 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.
@@ -3058,7 +3168,7 @@ 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;