Allow cloning to an existing empty directory
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 9053d4d1f6e886835813e8ba62a7d7767eefe1c8..d23548292ad319defd691655409098f824ec9e94 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -11,6 +11,7 @@
 #include "attr.h"
 #include "run-command.h"
 #include "utf8.h"
+#include "userdiff.h"
 
 #ifdef NO_FAST_WORKING_DIRECTORY
 #define FAST_WORKING_DIRECTORY 0
@@ -37,6 +38,9 @@ static char diff_colors[][COLOR_MAXLEN] = {
        "\033[41m",     /* WHITESPACE (red background) */
 };
 
+static void diff_filespec_load_driver(struct diff_filespec *one);
+static char *run_textconv(const char *, struct diff_filespec *, size_t *);
+
 static int parse_diff_color_slot(const char *var, int ofs)
 {
        if (!strcasecmp(var+ofs, "plain"))
@@ -56,75 +60,6 @@ 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;
-       const char *cmd;
-} *user_diff, **user_diff_tail;
-
-/*
- * 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) {
-               drv = xcalloc(1, sizeof(struct ll_diff_driver));
-               drv->name = xmemdupz(name, namelen);
-               if (!user_diff_tail)
-                       user_diff_tail = &user_diff;
-               *user_diff_tail = drv;
-               user_diff_tail = &(drv->next);
-       }
-
-       return git_config_string(&(drv->cmd), var, value);
-}
-
-/*
- * '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;
-       }
-       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
@@ -157,12 +92,6 @@ int git_diff_ui_config(const char *var, const char *value, void *cb)
        }
        if (!strcmp(var, "diff.external"))
                return git_config_string(&external_diff_cmd_cfg, var, value);
-       if (!prefixcmp(var, "diff.")) {
-               const char *ep = strrchr(var, '.');
-
-               if (ep != var + 4 && !strcmp(ep, ".command"))
-                       return parse_lldiff_command(var, ep, value);
-       }
 
        return git_diff_basic_config(var, value, cb);
 }
@@ -174,6 +103,12 @@ int git_diff_basic_config(const char *var, const char *value, void *cb)
                return 0;
        }
 
+       switch (userdiff_config(var, value)) {
+               case 0: break;
+               case -1: return -1;
+               default: return 0;
+       }
+
        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
                int slot = parse_diff_color_slot(var, 11);
                if (!value)
@@ -188,17 +123,6 @@ int git_diff_basic_config(const char *var, const char *value, void *cb)
                return 0;
        }
 
-       if (!prefixcmp(var, "diff.")) {
-               const char *ep = strrchr(var, '.');
-               if (ep != var + 4) {
-                       if (!strcmp(ep, ".funcname")) {
-                               if (!value)
-                                       return config_error_nonbool(var);
-                               return parse_funcname_pattern(var, ep, value);
-                       }
-               }
-       }
-
        return git_color_default_config(var, value, cb);
 }
 
@@ -206,9 +130,8 @@ 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);
-       struct strbuf res;
+       struct strbuf res = STRBUF_INIT;
 
-       strbuf_init(&res, 0);
        if (need_one + need_two) {
                strbuf_addch(&res, '"');
                quote_c_style(one, &res, NULL, 1);
@@ -306,6 +229,8 @@ static void emit_rewrite_diff(const char *name_a,
                              const char *name_b,
                              struct diff_filespec *one,
                              struct diff_filespec *two,
+                             const char *textconv_one,
+                             const char *textconv_two,
                              struct diff_options *o)
 {
        int lc_a, lc_b;
@@ -318,6 +243,8 @@ static void emit_rewrite_diff(const char *name_a,
        const char *reset = diff_get_color(color_diff, DIFF_RESET);
        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
        const char *a_prefix, *b_prefix;
+       const char *data_one, *data_two;
+       size_t size_one, size_two;
 
        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
                a_prefix = o->b_prefix;
@@ -339,8 +266,27 @@ static void emit_rewrite_diff(const char *name_a,
 
        diff_populate_filespec(one, 0);
        diff_populate_filespec(two, 0);
-       lc_a = count_lines(one->data, one->size);
-       lc_b = count_lines(two->data, two->size);
+       if (textconv_one) {
+               data_one = run_textconv(textconv_one, one, &size_one);
+               if (!data_one)
+                       die("unable to read files to diff");
+       }
+       else {
+               data_one = one->data;
+               size_one = one->size;
+       }
+       if (textconv_two) {
+               data_two = run_textconv(textconv_two, two, &size_two);
+               if (!data_two)
+                       die("unable to read files to diff");
+       }
+       else {
+               data_two = two->data;
+               size_two = two->size;
+       }
+
+       lc_a = count_lines(data_one, size_one);
+       lc_b = count_lines(data_two, size_two);
        fprintf(o->file,
                "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
                metainfo, a_name.buf, name_a_tab, reset,
@@ -350,9 +296,9 @@ static void emit_rewrite_diff(const char *name_a,
        print_line_count(o->file, lc_b);
        fprintf(o->file, " @@%s\n", reset);
        if (lc_a)
-               copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
+               copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
        if (lc_b)
-               copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
+               copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
 }
 
 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
@@ -364,6 +310,7 @@ static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
        }
        else if (diff_populate_filespec(one, 0))
                return -1;
+
        mf->ptr = one->data;
        mf->size = one->size;
        return 0;
@@ -460,6 +407,7 @@ static void diff_words_show(struct diff_words_data *diff_words)
        mmfile_t minus, plus;
        int i;
 
+       memset(&xpp, 0, sizeof(xpp));
        memset(&xecfg, 0, sizeof(xecfg));
        minus.size = diff_words->minus.text.size;
        minus.ptr = xmalloc(minus.size);
@@ -672,7 +620,7 @@ static char *pprint_rename(const char *a, const char *b)
 {
        const char *old = a;
        const char *new = b;
-       struct strbuf name;
+       struct strbuf name = STRBUF_INIT;
        int pfx_length, sfx_length;
        int len_a = strlen(a);
        int len_b = strlen(b);
@@ -680,7 +628,6 @@ static char *pprint_rename(const char *a, const char *b)
        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) {
                quote_c_style(a, &name, NULL, 0);
                strbuf_addstr(&name, " => ");
@@ -823,8 +770,7 @@ static void fill_print_name(struct diffstat_file *file)
                return;
 
        if (!file->is_renamed) {
-               struct strbuf buf;
-               strbuf_init(&buf, 0);
+               struct strbuf buf = STRBUF_INIT;
                if (quote_c_style(file->name, &buf, NULL, 0)) {
                        pname = strbuf_detach(&buf, NULL);
                } else {
@@ -1131,9 +1077,13 @@ static void show_dirstat(struct diff_options *options)
                /*
                 * Original minus copied is the removed material,
                 * added is the new material.  They are both damages
-                * made to the preimage.
+                * made to the preimage. In --dirstat-by-file mode, count
+                * damaged files, not damaged lines. This is done by
+                * counting only a single damaged line per file.
                 */
                damage = (p->one->size - copied) + added;
+               if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
+                       damage = 1;
 
                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
                dir.files[dir.nr].name = name;
@@ -1340,125 +1290,37 @@ static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
        emit_binary_diff_body(file, two, one);
 }
 
-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)
+static void diff_filespec_load_driver(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);
-       }
+       if (!one->driver)
+               one->driver = userdiff_find_by_path(one->path);
+       if (!one->driver)
+               one->driver = userdiff_find_by_name("default");
 }
 
 int diff_filespec_is_binary(struct diff_filespec *one)
 {
-       diff_filespec_check_attr(one);
+       if (one->is_binary == -1) {
+               diff_filespec_load_driver(one);
+               if (one->driver->binary != -1)
+                       one->is_binary = one->driver->binary;
+               else {
+                       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);
+                       if (one->is_binary == -1)
+                               one->is_binary = 0;
+               }
+       }
        return one->is_binary;
 }
 
-static const char *funcname_pattern(const char *ident)
+static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
 {
-       struct funcname_pattern *pp;
-
-       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[] = {
-       { "bibtex", "\\(@[a-zA-Z]\\{1,\\}[ \t]*{\\{0,1\\}[ \t]*[^ \t\"@',\\#}{~%]*\\).*$" },
-       { "html", "^\\s*\\(<[Hh][1-6]\\s.*>.*\\)$" },
-       { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
-                       "new\\|return\\|switch\\|throw\\|while\\)\n"
-                       "^[     ]*\\(\\([       ]*"
-                       "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
-                       "[      ]*([^;]*\\)$" },
-       { "pascal", "^\\(\\(procedure\\|function\\|constructor\\|"
-                       "destructor\\|interface\\|implementation\\|"
-                       "initialization\\|finalization\\)[ \t]*.*\\)$"
-                       "\\|"
-                       "^\\(.*=[ \t]*\\(class\\|record\\).*\\)$"
-                       },
-       { "php", "^[\t ]*\\(\\(function\\|class\\).*\\)" },
-       { "python", "^\\s*\\(\\(class\\|def\\)\\s.*\\)$" },
-       { "ruby", "^\\s*\\(\\(class\\|module\\|def\\)\\s.*\\)$" },
-       { "tex", "^\\(\\\\\\(\\(sub\\)*section\\|chapter\\|part\\)\\*\\{0,1\\}{.*\\)$" },
-};
-
-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 overridden 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;
+       diff_filespec_load_driver(one);
+       return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
 }
 
 void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
@@ -1469,6 +1331,16 @@ void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const
                options->b_prefix = b;
 }
 
+static const char *get_textconv(struct diff_filespec *one)
+{
+       if (!DIFF_FILE_VALID(one))
+               return NULL;
+       if (!S_ISREG(one->mode))
+               return NULL;
+       diff_filespec_load_driver(one);
+       return one->driver->textconv;
+}
+
 static void builtin_diff(const char *name_a,
                         const char *name_b,
                         struct diff_filespec *one,
@@ -1483,6 +1355,12 @@ static void builtin_diff(const char *name_a,
        const char *set = diff_get_color_opt(o, DIFF_METAINFO);
        const char *reset = diff_get_color_opt(o, DIFF_RESET);
        const char *a_prefix, *b_prefix;
+       const char *textconv_one = NULL, *textconv_two = NULL;
+
+       if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
+               textconv_one = get_textconv(one);
+               textconv_two = get_textconv(two);
+       }
 
        diff_set_mnemonic_prefix(o, "a/", "b/");
        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
@@ -1493,6 +1371,10 @@ static void builtin_diff(const char *name_a,
                b_prefix = o->b_prefix;
        }
 
+       /* Never use a non-valid filename anywhere if at all possible */
+       name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
+       name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
+
        a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
        b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
@@ -1522,8 +1404,11 @@ 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, o);
+               if (complete_rewrite &&
+                   (textconv_one || !diff_filespec_is_binary(one)) &&
+                   (textconv_two || !diff_filespec_is_binary(two))) {
+                       emit_rewrite_diff(name_a, name_b, one, two,
+                                               textconv_one, textconv_two, o);
                        o->found_changes = 1;
                        goto free_ab_and_return;
                }
@@ -1533,7 +1418,8 @@ static void builtin_diff(const char *name_a,
                die("unable to read files to diff");
 
        if (!DIFF_OPT_TST(o, TEXT) &&
-           (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
+           ( (diff_filespec_is_binary(one) && !textconv_one) ||
+             (diff_filespec_is_binary(two) && !textconv_two) )) {
                /* Quite common confusing case */
                if (mf1.size == mf2.size &&
                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
@@ -1552,12 +1438,28 @@ static void builtin_diff(const char *name_a,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
                struct emit_callback ecbdata;
-               const char *funcname_pattern;
+               const struct userdiff_funcname *pe;
+
+               if (textconv_one) {
+                       size_t size;
+                       mf1.ptr = run_textconv(textconv_one, one, &size);
+                       if (!mf1.ptr)
+                               die("unable to read files to diff");
+                       mf1.size = size;
+               }
+               if (textconv_two) {
+                       size_t size;
+                       mf2.ptr = run_textconv(textconv_two, two, &size);
+                       if (!mf2.ptr)
+                               die("unable to read files to diff");
+                       mf2.size = size;
+               }
 
-               funcname_pattern = diff_funcname_pattern(one);
-               if (!funcname_pattern)
-                       funcname_pattern = diff_funcname_pattern(two);
+               pe = diff_funcname_pattern(one);
+               if (!pe)
+                       pe = diff_funcname_pattern(two);
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                memset(&ecbdata, 0, sizeof(ecbdata));
                ecbdata.label_path = lbl;
@@ -1567,9 +1469,10 @@ static void builtin_diff(const char *name_a,
                ecbdata.file = o->file;
                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
                xecfg.ctxlen = o->context;
+               xecfg.interhunkctxlen = o->interhunkcontext;
                xecfg.flags = XDL_EMIT_FUNCNAMES;
-               if (funcname_pattern)
-                       xdiff_set_find_func(&xecfg, funcname_pattern);
+               if (pe)
+                       xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
                if (!diffopts)
                        ;
                else if (!prefixcmp(diffopts, "--unified="))
@@ -1585,6 +1488,10 @@ static void builtin_diff(const char *name_a,
                              &xpp, &xecfg, &ecb);
                if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
                        free_diff_words_data(&ecbdata);
+               if (textconv_one)
+                       free(mf1.ptr);
+               if (textconv_two)
+                       free(mf2.ptr);
        }
 
  free_ab_and_return:
@@ -1631,6 +1538,7 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
                xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
@@ -1677,6 +1585,7 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                xecfg.ctxlen = 1; /* at least one context line */
                xpp.flags = XDF_NEED_MINIMAL;
@@ -1706,6 +1615,7 @@ struct diff_filespec *alloc_filespec(const char *path)
        spec->path = (char *)(spec + 1);
        memcpy(spec->path, path, namelen+1);
        spec->count = 1;
+       spec->is_binary = -1;
        return spec;
 }
 
@@ -1790,10 +1700,9 @@ static int reuse_worktree_file(const char *name, const unsigned char *sha1, int
 
 static int populate_from_stdin(struct diff_filespec *s)
 {
-       struct strbuf buf;
+       struct strbuf buf = STRBUF_INIT;
        size_t size = 0;
 
-       strbuf_init(&buf, 0);
        if (strbuf_read(&buf, 0, 0) < 0)
                return error("error while reading from stdin %s",
                                     strerror(errno));
@@ -1845,7 +1754,7 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
 
        if (!s->sha1_valid ||
            reuse_worktree_file(s->path, s->sha1, 0)) {
-               struct strbuf buf;
+               struct strbuf buf = STRBUF_INIT;
                struct stat st;
                int fd;
 
@@ -1865,19 +1774,18 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
                s->size = xsize_t(st.st_size);
                if (!s->size)
                        goto empty;
-               if (size_only)
-                       return 0;
                if (S_ISLNK(st.st_mode)) {
-                       int ret;
-                       s->data = xmalloc(s->size);
-                       s->should_free = 1;
-                       ret = readlink(s->path, s->data, s->size);
-                       if (ret < 0) {
-                               free(s->data);
+                       struct strbuf sb = STRBUF_INIT;
+
+                       if (strbuf_readlink(&sb, s->path, s->size))
                                goto err_empty;
-                       }
+                       s->size = sb.len;
+                       s->data = strbuf_detach(&sb, NULL);
+                       s->should_free = 1;
                        return 0;
                }
+               if (size_only)
+                       return 0;
                fd = open(s->path, O_RDONLY);
                if (fd < 0)
                        goto err_empty;
@@ -1888,7 +1796,6 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
                /*
                 * Convert from working tree format to canonical git format
                 */
-               strbuf_init(&buf, 0);
                if (convert_to_git(s->path, s->data, s->size, &buf, safe_crlf)) {
                        size_t size = 0;
                        munmap(s->data, s->size);
@@ -1976,13 +1883,12 @@ 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, sz);
+                       ret = readlink(name, buf, sizeof(buf));
                        if (ret < 0)
                                die("readlink(%s)", name);
-                       prep_temp_blob(temp, buf, sz,
+                       if (ret == sizeof(buf))
+                               die("symlink too long: %s", name);
+                       prep_temp_blob(temp, buf, ret,
                                       (one->sha1_valid ?
                                        one->sha1 : null_sha1),
                                       (one->sha1_valid ?
@@ -2092,29 +1998,6 @@ static void run_external_diff(const char *pgm,
        }
 }
 
-static const char *external_diff_attr(const char *name)
-{
-       struct git_attr_check attr_diff_check;
-
-       if (!name)
-               return NULL;
-
-       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;
-
-                       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,
@@ -2128,9 +2011,9 @@ static void run_diff_cmd(const char *pgm,
        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
                pgm = NULL;
        else {
-               const char *cmd = external_diff_attr(attr_path);
-               if (cmd)
-                       pgm = cmd;
+               struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
+               if (drv && drv->external)
+                       pgm = drv->external;
        }
 
        if (pgm) {
@@ -2157,7 +2040,7 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
                        if (lstat(one->path, &st) < 0)
                                die("stat %s", one->path);
                        if (index_path(one->sha1, one->path, &st, 0))
-                               die("cannot hash %s\n", one->path);
+                               die("cannot hash %s", one->path);
                }
        }
        else
@@ -2511,6 +2394,10 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        else if (!strcmp(arg, "--cumulative")) {
                options->output_format |= DIFF_FORMAT_DIRSTAT;
                DIFF_OPT_SET(options, DIRSTAT_CUMULATIVE);
+       } else if (opt_arg(arg, 0, "dirstat-by-file",
+                          &options->dirstat_percent)) {
+               options->output_format |= DIFF_FORMAT_DIRSTAT;
+               DIFF_OPT_SET(options, DIRSTAT_BY_FILE);
        }
        else if (!strcmp(arg, "--check"))
                options->output_format |= DIFF_FORMAT_CHECKDIFF;
@@ -2615,6 +2502,10 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
        else if (!strcmp(arg, "--no-ext-diff"))
                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
+       else if (!strcmp(arg, "--textconv"))
+               DIFF_OPT_SET(options, ALLOW_TEXTCONV);
+       else if (!strcmp(arg, "--no-textconv"))
+               DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
        else if (!strcmp(arg, "--ignore-submodules"))
                DIFF_OPT_SET(options, IGNORE_SUBMODULES);
 
@@ -2648,6 +2539,9 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->b_prefix = arg + 13;
        else if (!strcmp(arg, "--no-prefix"))
                options->a_prefix = options->b_prefix = "";
+       else if (opt_arg(arg, '\0', "inter-hunk-context",
+                        &options->interhunkcontext))
+               ;
        else if (!prefixcmp(arg, "--output=")) {
                options->file = fopen(arg + strlen("--output="), "w");
                options->close_file = 1;
@@ -3067,7 +2961,7 @@ static void diff_summary(FILE *file, struct diff_filepair *p)
 }
 
 struct patch_id_t {
-       SHA_CTX *ctx;
+       git_SHA_CTX *ctx;
        int patchlen;
 };
 
@@ -3095,7 +2989,7 @@ static void patch_id_consume(void *priv, char *line, unsigned long len)
 
        new_len = remove_space(line, len);
 
-       SHA1_Update(data->ctx, line, new_len);
+       git_SHA1_Update(data->ctx, line, new_len);
        data->patchlen += new_len;
 }
 
@@ -3104,11 +2998,11 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
 {
        struct diff_queue_struct *q = &diff_queued_diff;
        int i;
-       SHA_CTX ctx;
+       git_SHA_CTX ctx;
        struct patch_id_t data;
        char buffer[PATH_MAX * 4 + 20];
 
-       SHA1_Init(&ctx);
+       git_SHA1_Init(&ctx);
        memset(&data, 0, sizeof(struct patch_id_t));
        data.ctx = &ctx;
 
@@ -3120,6 +3014,7 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                struct diff_filepair *p = q->queue[i];
                int len1, len2;
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                if (p->status == 0)
                        return error("internal diff status error");
@@ -3170,7 +3065,7 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                                        len2, p->two->path,
                                        len1, p->one->path,
                                        len2, p->two->path);
-               SHA1_Update(&ctx, buffer, len1);
+               git_SHA1_Update(&ctx, buffer, len1);
 
                xpp.flags = XDF_NEED_MINIMAL;
                xecfg.ctxlen = 3;
@@ -3179,7 +3074,7 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                              &xpp, &xecfg, &ecb);
        }
 
-       SHA1_Final(sha1, &ctx);
+       git_SHA1_Final(sha1, &ctx);
        return 0;
 }
 
@@ -3549,3 +3444,34 @@ void diff_unmerge(struct diff_options *options,
        fill_filespec(one, sha1, mode);
        diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
 }
+
+static char *run_textconv(const char *pgm, struct diff_filespec *spec,
+               size_t *outsize)
+{
+       struct diff_tempfile temp;
+       const char *argv[3];
+       const char **arg = argv;
+       struct child_process child;
+       struct strbuf buf = STRBUF_INIT;
+
+       prepare_temp_file(spec->path, &temp, spec);
+       *arg++ = pgm;
+       *arg++ = temp.name;
+       *arg = NULL;
+
+       memset(&child, 0, sizeof(child));
+       child.argv = argv;
+       child.out = -1;
+       if (start_command(&child) != 0 ||
+           strbuf_read(&buf, child.out, 0) < 0 ||
+           finish_command(&child) != 0) {
+               if (temp.name == temp.tmp_path)
+                       unlink(temp.name);
+               error("error running textconv command '%s'", pgm);
+               return NULL;
+       }
+       if (temp.name == temp.tmp_path)
+               unlink(temp.name);
+
+       return strbuf_detach(&buf, outsize);
+}