Code

send-email: discard blank around address in extract_valid_address as well.
[git.git] / diff.c
diff --git a/diff.c b/diff.c
index 165d2520f3aec738e1b8a2b52f1a819e3edb9f33..cd6b0c4e0bb5efad828fa29e92e8729418401080 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
@@ -15,8 +16,6 @@
 #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_use_color_default;
@@ -51,6 +50,96 @@ 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 +166,22 @@ int git_diff_ui_config(const char *var, const char *value)
                        diff_detect_rename_default = DIFF_DETECT_RENAME;
                return 0;
        }
-       if (!strncmp(var, "diff.color.", 11) || !strncmp(var, "color.diff.", 11)) {
+       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 +237,11 @@ 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)
@@ -184,31 +282,43 @@ static void print_line_count(int count)
        }
 }
 
-static void copy_file(int prefix, const char *data, int size)
+static void copy_file(int prefix, const char *data, int size,
+               const char *set, const char *reset)
 {
        int ch, nl_just_seen = 1;
        while (0 < size--) {
                ch = *data++;
-               if (nl_just_seen)
+               if (nl_just_seen) {
+                       fputs(set, stdout);
                        putchar(prefix);
-               putchar(ch);
-               if (ch == '\n')
+               }
+               if (ch == '\n') {
                        nl_just_seen = 1;
-               else
+                       fputs(reset, stdout);
+               } else
                        nl_just_seen = 0;
+               putchar(ch);
        }
        if (!nl_just_seen)
-               printf("\n\\ No newline at end of file\n");
+               printf("%s\n\\ No newline at end of file\n", reset);
 }
 
 static void emit_rewrite_diff(const char *name_a,
                              const char *name_b,
                              struct diff_filespec *one,
-                             struct diff_filespec *two)
+                             struct diff_filespec *two,
+                             int color_diff)
 {
        int lc_a, lc_b;
        const char *name_a_tab, *name_b_tab;
-
+       const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
+       const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
+       const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
+       const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
+       const char *reset = diff_get_color(color_diff, DIFF_RESET);
+
+       name_a += (*name_a == '/');
+       name_b += (*name_b == '/');
        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 
@@ -216,17 +326,17 @@ static void emit_rewrite_diff(const char *name_a,
        diff_populate_filespec(two, 0);
        lc_a = count_lines(one->data, one->size);
        lc_b = count_lines(two->data, two->size);
-       printf("--- a/%s%s\n+++ b/%s%s\n@@ -",
-              name_a, name_a_tab,
-              name_b, name_b_tab);
+       printf("%s--- a/%s%s%s\n%s+++ b/%s%s%s\n%s@@ -",
+              metainfo, name_a, name_a_tab, reset,
+              metainfo, name_b, name_b_tab, reset, fraginfo);
        print_line_count(lc_a);
        printf(" +");
        print_line_count(lc_b);
-       printf(" @@\n");
+       printf(" @@%s\n", reset);
        if (lc_a)
-               copy_file('-', one->data, one->size);
+               copy_file('-', one->data, one->size, old, reset);
        if (lc_b)
-               copy_file('+', two->data, two->size);
+               copy_file('+', two->data, two->size, new, reset);
 }
 
 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
@@ -331,6 +441,7 @@ 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);
@@ -349,7 +460,6 @@ 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;
@@ -370,6 +480,7 @@ struct emit_callback {
        int nparents, color_diff;
        const char **label_path;
        struct diff_words_data *diff_words;
+       int *found_changesp;
 };
 
 static void free_diff_words_data(struct emit_callback *ecbdata)
@@ -405,22 +516,16 @@ static void emit_line(const char *set, const char *reset, const char *line, int
        puts(reset);
 }
 
-static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
+static void emit_line_with_ws(int nparents,
+               const char *set, const char *reset, const char *ws,
+               const char *line, int len)
 {
-       int col0 = ecbdata->nparents;
+       int col0 = nparents;
        int last_tab_in_indent = -1;
        int last_space_in_indent = -1;
        int i;
        int tail = len;
        int need_highlight_leading_space = 0;
-       const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
-       const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
-
-       if (!*ws) {
-               emit_line(set, reset, line, len);
-               return;
-       }
-
        /* The line is a newly added line.  Does it have funny leading
         * whitespaces?  In indent, SP should never precede a TAB.
         */
@@ -475,6 +580,18 @@ static void emit_add_line(const char *reset, struct emit_callback *ecbdata, cons
                emit_line(set, reset, line + i, len - i);
 }
 
+static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
+{
+       const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
+       const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
+
+       if (!*ws)
+               emit_line(set, reset, line, len);
+       else
+               emit_line_with_ws(ecbdata->nparents, set, reset, ws,
+                               line, len);
+}
+
 static void fn_out_consume(void *priv, char *line, unsigned long len)
 {
        int i;
@@ -483,6 +600,8 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
        const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 
+       *(ecbdata->found_changesp) = 1;
+
        if (ecbdata->label_path[0]) {
                const char *name_a_tab, *name_b_tab;
 
@@ -790,7 +909,12 @@ static void show_stats(struct diffstat_t* data, struct diff_options *options)
 
                if (data->files[i]->is_binary) {
                        show_name(prefix, name, len, reset, set);
-                       printf("  Bin\n");
+                       printf("  Bin ");
+                       printf("%s%d%s", del_c, deleted, reset);
+                       printf(" -> ");
+                       printf("%s%d%s", add_c, added, reset);
+                       printf(" bytes");
+                       printf("\n");
                        goto free_diffstat_file;
                }
                else if (data->files[i]->is_unmerged) {
@@ -884,30 +1008,44 @@ static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 struct checkdiff_t {
        struct xdiff_emit_state xm;
        const char *filename;
-       int lineno;
+       int lineno, color_diff;
 };
 
 static void checkdiff_consume(void *priv, char *line, unsigned long len)
 {
        struct checkdiff_t *data = priv;
+       const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
+       const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
+       const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
 
        if (line[0] == '+') {
-               int i, spaces = 0;
+               int i, spaces = 0, space_before_tab = 0, white_space_at_end = 0;
 
                /* check space before tab */
                for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
                        if (line[i] == ' ')
                                spaces++;
                if (line[i - 1] == '\t' && spaces)
-                       printf("%s:%d: space before tab:%.*s\n",
-                               data->filename, data->lineno, (int)len, line);
+                       space_before_tab = 1;
 
                /* check white space at line end */
                if (line[len - 1] == '\n')
                        len--;
                if (isspace(line[len - 1]))
-                       printf("%s:%d: white space at end: %.*s\n",
-                               data->filename, data->lineno, (int)len, line);
+                       white_space_at_end = 1;
+
+               if (space_before_tab || white_space_at_end) {
+                       printf("%s:%d: %s", data->filename, data->lineno, ws);
+                       if (space_before_tab) {
+                               printf("space before tab");
+                               if (white_space_at_end)
+                                       putchar(',');
+                       }
+                       if (white_space_at_end)
+                               printf("white space at end");
+                       printf(":%s ", reset);
+                       emit_line_with_ws(1, set, reset, ws, line, len);
+               }
 
                data->lineno++;
        } else if (line[0] == ' ')
@@ -1011,13 +1149,115 @@ 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)
+{
+       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)
 {
-       long sz = mf->size;
-       if (FIRST_FEW_BYTES < sz)
-               sz = FIRST_FEW_BYTES;
-       return !!memchr(mf->ptr, 0, sz);
+       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,
@@ -1034,8 +1274,8 @@ static void builtin_diff(const char *name_a,
        const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
        const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
 
-       a_one = quote_two("a/", name_a);
-       b_two = quote_two("b/", name_b);
+       a_one = quote_two("a/", name_a + (*name_a == '/'));
+       b_two = quote_two("b/", name_b + (*name_b == '/'));
        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
        printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
@@ -1064,7 +1304,9 @@ static void builtin_diff(const char *name_a,
                if ((one->mode ^ two->mode) & S_IFMT)
                        goto free_ab_and_return;
                if (complete_rewrite) {
-                       emit_rewrite_diff(name_a, name_b, one, two);
+                       emit_rewrite_diff(name_a, name_b, one, two,
+                                       o->color_diff);
+                       o->found_changes = 1;
                        goto free_ab_and_return;
                }
        }
@@ -1072,7 +1314,8 @@ 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))
@@ -1082,6 +1325,7 @@ static void builtin_diff(const char *name_a,
                else
                        printf("Binary files %s and %s differ\n",
                               lbl[0], lbl[1]);
+               o->found_changes = 1;
        }
        else {
                /* Crazy xdl interfaces.. */
@@ -1090,18 +1334,27 @@ 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;
+               ecbdata.found_changesp = &o->found_changes;
                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 (!strncmp(diffopts, "--unified=", 10))
+               else if (!prefixcmp(diffopts, "--unified="))
                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
-               else if (!strncmp(diffopts, "-u", 2))
+               else if (!prefixcmp(diffopts, "-u"))
                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
                ecb.outf = xdiff_outf;
                ecb.priv = &ecbdata;
@@ -1115,6 +1368,8 @@ 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;
@@ -1141,31 +1396,36 @@ 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;
-       else {
+               data->added = mf2.size;
+               data->deleted = mf1.size;
+       } else {
                /* Crazy xdl interfaces.. */
                xpparam_t xpp;
                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,
                             struct diff_filespec *one,
-                            struct diff_filespec *two)
+                            struct diff_filespec *two, struct diff_options *o)
 {
        mmfile_t mf1, mf2;
        struct checkdiff_t data;
@@ -1177,25 +1437,28 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
        data.xm.consume = checkdiff_consume;
        data.filename = name_b ? name_b : name_a;
        data.lineno = 0;
+       data.color_diff = o->color_diff;
 
        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)
@@ -1220,7 +1483,7 @@ 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.
  */
@@ -1275,53 +1538,46 @@ static int reuse_worktree_file(const char *name, const unsigned char *sha1, int
        return 1;
 }
 
-static struct sha1_size_cache {
-       unsigned char sha1[20];
+static int populate_from_stdin(struct diff_filespec *s)
+{
+#define INCREMENT 1024
+       char *buf;
        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;
+       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",
+                                    strerror(errno));
+               size += got;
+       }
+       s->should_munmap = 0;
+       s->data = buf;
+       s->size = size;
+       s->should_free = 1;
+       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);
        }
-       /* 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;
+       return 0;
 }
 
 /*
@@ -1337,15 +1593,25 @@ 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;
                int fd;
+               char *buf;
+               unsigned long size;
+
+               if (!strcmp(s->path, "-"))
+                       return populate_from_stdin(s);
+
                if (lstat(s->path, &st) < 0) {
                        if (errno == ENOENT) {
                        err_empty:
@@ -1356,7 +1622,7 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
                                return err;
                        }
                }
-               s->size = st.st_size;
+               s->size = xsize_t(st.st_size);
                if (!s->size)
                        goto empty;
                if (size_only)
@@ -1378,23 +1644,26 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
                close(fd);
                s->should_munmap = 1;
-               /* FIXME! CRLF -> LF conversion goes here, based on "s->path" */
-       }
-       else {
-               char type[20];
-               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 (!sha1_object_info(s->sha1, type, &s->size))
-                               locate_size_cache(s->sha1, 0, s->size);
+               /*
+                * Convert from working tree format to canonical git format
+                */
+               size = s->size;
+               buf = convert_to_git(s->path, s->data, &size);
+               if (buf) {
+                       munmap(s->data, s->size);
+                       s->should_munmap = 0;
+                       s->data = buf;
+                       s->size = size;
+                       s->should_free = 1;
                }
+       }
+       else {
+               enum object_type type;
+               if (size_only)
+                       type = sha1_object_info(s->sha1, &s->size);
                else {
-                       s->data = read_sha1_file(s->sha1, type, &s->size);
+                       s->data = read_sha1_file(s->sha1, &type, &s->size);
                        s->should_free = 1;
                }
        }
@@ -1407,8 +1676,11 @@ void diff_free_filespec_data(struct diff_filespec *s)
                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;
+       }
        free(s->cnt_data);
        s->cnt_data = NULL;
 }
@@ -1421,7 +1693,7 @@ 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");
        if (write_in_full(fd, blob, size) != size)
@@ -1459,12 +1731,13 @@ static void prepare_temp_file(const char *name,
                if (S_ISLNK(st.st_mode)) {
                        int ret;
                        char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
+                       size_t sz = xsize_t(st.st_size);
                        if (sizeof(buf) <= st.st_size)
                                die("symlink too long: %s", name);
-                       ret = readlink(name, buf, st.st_size);
+                       ret = readlink(name, buf, sz);
                        if (ret < 0)
                                die("readlink(%s)", name);
-                       prep_temp_blob(temp, buf, st.st_size,
+                       prep_temp_blob(temp, buf, sz,
                                       (one->sha1_valid ?
                                        one->sha1 : null_sha1),
                                       (one->sha1_valid ?
@@ -1607,6 +1880,27 @@ 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,
@@ -1616,6 +1910,14 @@ 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);
@@ -1633,6 +1935,10 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
        if (DIFF_FILE_VALID(one)) {
                if (!one->sha1_valid) {
                        struct stat st;
+                       if (!strcmp(one->path, "-")) {
+                               hashcpy(one->sha1, null_sha1);
+                               return;
+                       }
                        if (lstat(one->path, &st) < 0)
                                die("stat %s", one->path);
                        if (index_path(one->sha1, one->path, &st, 0))
@@ -1643,6 +1949,11 @@ 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();
@@ -1677,23 +1988,20 @@ 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;
                }
@@ -1708,8 +2016,8 @@ 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,
@@ -1787,7 +2095,7 @@ static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
        diff_fill_sha1_info(p->one);
        diff_fill_sha1_info(p->two);
 
-       builtin_checkdiff(name, other, p->one, p->two);
+       builtin_checkdiff(name, other, p->one, p->two, o);
 }
 
 void diff_setup(struct diff_options *options)
@@ -1863,11 +2171,26 @@ 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 */
 
+       /*
+        * It does not make sense to show the first hit we happened
+        * to have found.  It does not make sense not to return with
+        * exit code in such a case either.
+        */
+       if (options->quiet) {
+               options->output_format = DIFF_FORMAT_NO_OUTPUT;
+               options->exit_with_status = 1;
+       }
+
+       /*
+        * If we postprocess in diffcore, we cannot simply return
+        * upon the first hit.  We need to run diff as usual.
+        */
+       if (options->pickaxe || options->filter)
+               options->quiet = 0;
+
        return 0;
 }
 
@@ -1918,6 +2241,8 @@ 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];
@@ -1936,7 +2261,7 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        else if (!strcmp(arg, "--shortstat")) {
                options->output_format |= DIFF_FORMAT_SHORTSTAT;
        }
-       else if (!strncmp(arg, "--stat", 6)) {
+       else if (!prefixcmp(arg, "--stat")) {
                char *end;
                int width = options->stat_width;
                int name_width = options->stat_name_width;
@@ -1945,9 +2270,9 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 
                switch (*arg) {
                case '-':
-                       if (!strncmp(arg, "-width=", 7))
+                       if (!prefixcmp(arg, "-width="))
                                width = strtoul(arg + 7, &end, 10);
-                       else if (!strncmp(arg, "-name-width=", 12))
+                       else if (!prefixcmp(arg, "-name-width="))
                                name_width = strtoul(arg + 12, &end, 10);
                        break;
                case '=':
@@ -1972,7 +2297,7 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        }
        else if (!strcmp(arg, "-z"))
                options->line_termination = 0;
-       else if (!strncmp(arg, "-l", 2))
+       else if (!prefixcmp(arg, "-l"))
                options->rename_limit = strtoul(arg+2, NULL, 10);
        else if (!strcmp(arg, "--full-index"))
                options->full_index = 1;
@@ -1989,31 +2314,33 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->output_format |= DIFF_FORMAT_NAME_STATUS;
        else if (!strcmp(arg, "-R"))
                options->reverse_diff = 1;
-       else if (!strncmp(arg, "-S", 2))
+       else if (!prefixcmp(arg, "-S"))
                options->pickaxe = arg + 2;
        else if (!strcmp(arg, "-s")) {
                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
        }
-       else if (!strncmp(arg, "-O", 2))
+       else if (!prefixcmp(arg, "-O"))
                options->orderfile = arg + 2;
-       else if (!strncmp(arg, "--diff-filter=", 14))
+       else if (!prefixcmp(arg, "--diff-filter="))
                options->filter = arg + 14;
        else if (!strcmp(arg, "--pickaxe-all"))
                options->pickaxe_opts = DIFF_PICKAXE_ALL;
        else if (!strcmp(arg, "--pickaxe-regex"))
                options->pickaxe_opts = DIFF_PICKAXE_REGEX;
-       else if (!strncmp(arg, "-B", 2)) {
+       else if (!prefixcmp(arg, "-B")) {
                if ((options->break_opt =
                     diff_scoreopt_parse(arg)) == -1)
                        return -1;
        }
-       else if (!strncmp(arg, "-M", 2)) {
+       else if (!prefixcmp(arg, "-M")) {
                if ((options->rename_score =
                     diff_scoreopt_parse(arg)) == -1)
                        return -1;
                options->detect_rename = DIFF_DETECT_RENAME;
        }
-       else if (!strncmp(arg, "-C", 2)) {
+       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;
@@ -2021,9 +2348,11 @@ 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 (!strncmp(arg, "--abbrev=", 9)) {
+       else if (!prefixcmp(arg, "--abbrev=")) {
                options->abbrev = strtoul(arg + 9, NULL, 10);
                if (options->abbrev < MINIMUM_ABBREV)
                        options->abbrev = MINIMUM_ABBREV;
@@ -2038,10 +2367,20 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->xdl_opts |= XDF_IGNORE_WHITESPACE;
        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
                options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
+       else if (!strcmp(arg, "--ignore-space-at-eol"))
+               options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
        else if (!strcmp(arg, "--color-words"))
                options->color_diff = options->color_diff_words = 1;
        else if (!strcmp(arg, "--no-renames"))
                options->detect_rename = 0;
+       else if (!strcmp(arg, "--exit-code"))
+               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;
@@ -2080,10 +2419,10 @@ static int parse_num(const char **cp_p)
        /* user says num divided by scale and we say internally that
         * is MAX_SCORE * num / scale.
         */
-       return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
+       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;
 
@@ -2190,8 +2529,7 @@ 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;
@@ -2216,7 +2554,8 @@ 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);
@@ -2404,7 +2743,8 @@ static void diff_resolve_rename_copy(void)
                                p->status = DIFF_STATUS_RENAMED;
                }
                else if (hashcmp(p->one->sha1, p->two->sha1) ||
-                        p->one->mode != p->two->mode)
+                        p->one->mode != p->two->mode ||
+                        is_null_sha1(p->one->sha1))
                        p->status = DIFF_STATUS_MODIFIED;
                else {
                        /* This is a "no-change" entry and should not
@@ -2472,8 +2812,7 @@ 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);
 }
@@ -2497,7 +2836,7 @@ 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);
@@ -2530,7 +2869,7 @@ static void patch_id_consume(void *priv, char *line, unsigned long len)
        int new_len;
 
        /* Ignore line numbers when computing the SHA1 of the patch */
-       if (!strncmp(line, "@@ -", 4))
+       if (!prefixcmp(line, "@@ -"))
                return;
 
        new_len = remove_space(line, len);
@@ -2561,6 +2900,7 @@ 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)
@@ -2580,7 +2920,7 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                        return error("unable to read files to diff");
 
                /* Maybe hash p->two? into the patch id? */
-               if (mmfile_is_binary(&mf2))
+               if (diff_filespec_is_binary(p->two))
                        continue;
 
                len1 = remove_space(p->one->path, strlen(p->one->path));
@@ -2805,6 +3145,9 @@ static void diffcore_apply_filter(const char *filter)
 
 void diffcore_std(struct diff_options *options)
 {
+       if (options->quiet)
+               return;
+
        if (options->break_opt != -1)
                diffcore_break(options->break_opt);
        if (options->detect_rename)
@@ -2817,18 +3160,11 @@ void diffcore_std(struct diff_options *options)
                diffcore_order(options->orderfile);
        diff_resolve_rename_copy();
        diffcore_apply_filter(options->filter);
-}
-
 
-void diffcore_std_no_resolve(struct diff_options *options)
-{
-       if (options->pickaxe)
-               diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
-       if (options->orderfile)
-               diffcore_order(options->orderfile);
-       diffcore_apply_filter(options->filter);
+       options->has_changes = !!diff_queued_diff.nr;
 }
 
+
 void diff_addremove(struct diff_options *options,
                    int addremove, unsigned mode,
                    const unsigned char *sha1,
@@ -2844,7 +3180,7 @@ 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.
@@ -2864,13 +3200,14 @@ void diff_addremove(struct diff_options *options,
                fill_filespec(two, sha1, mode);
 
        diff_queue(&diff_queued_diff, one, two);
+       options->has_changes = 1;
 }
 
 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;
@@ -2889,6 +3226,7 @@ void diff_change(struct diff_options *options,
        fill_filespec(two, new_sha1, new_mode);
 
        diff_queue(&diff_queued_diff, one, two);
+       options->has_changes = 1;
 }
 
 void diff_unmerge(struct diff_options *options,