Try normalizing files to avoid delete/modify conflicts when merging
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 8414478032632c2b9b877c65ddcf955451083934..2332fa6798d9e40e83a07a318675e70d9b3169aa 100644 (file)
--- a/diff.c
+++ b/diff.c
 #include "attr.h"
 #include "run-command.h"
 #include "utf8.h"
+#include "userdiff.h"
+#include "sigchain.h"
+#include "submodule.h"
+#include "ll-merge.h"
 
 #ifdef NO_FAST_WORKING_DIRECTORY
 #define FAST_WORKING_DIRECTORY 0
 
 static int diff_detect_rename_default;
 static int diff_rename_limit_default = 200;
+static int diff_suppress_blank_empty;
 int diff_use_color_default = -1;
+static const char *diff_word_regex_cfg;
 static const char *external_diff_cmd_cfg;
 int diff_auto_refresh_index = 1;
+static int diff_mnemonic_prefix;
+static int diff_no_prefix;
 
 static char diff_colors[][COLOR_MAXLEN] = {
-       "\033[m",       /* reset */
-       "",             /* PLAIN (normal) */
-       "\033[1m",      /* METAINFO (bold) */
-       "\033[36m",     /* FRAGINFO (cyan) */
-       "\033[31m",     /* OLD (red) */
-       "\033[32m",     /* NEW (green) */
-       "\033[33m",     /* COMMIT (yellow) */
-       "\033[41m",     /* WHITESPACE (red background) */
+       GIT_COLOR_RESET,
+       GIT_COLOR_NORMAL,       /* PLAIN */
+       GIT_COLOR_BOLD,         /* METAINFO */
+       GIT_COLOR_CYAN,         /* FRAGINFO */
+       GIT_COLOR_RED,          /* OLD */
+       GIT_COLOR_GREEN,        /* NEW */
+       GIT_COLOR_YELLOW,       /* COMMIT */
+       GIT_COLOR_BG_RED,       /* WHITESPACE */
+       GIT_COLOR_NORMAL,       /* FUNCINFO */
 };
 
 static int parse_diff_color_slot(const char *var, int ofs)
@@ -51,81 +60,18 @@ static int parse_diff_color_slot(const char *var, int ofs)
                return DIFF_COMMIT;
        if (!strcasecmp(var+ofs, "whitespace"))
                return DIFF_WHITESPACE;
-       die("bad config variable '%s'", var);
+       if (!strcasecmp(var+ofs, "func"))
+               return DIFF_FUNCINFO;
+       return -1;
 }
 
-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.
- */
-struct funcname_pattern_entry {
-       char *name;
-       char *pattern;
-       int cflags;
-};
-static struct funcname_pattern_list {
-       struct funcname_pattern_list *next;
-       struct funcname_pattern_entry e;
-} *funcname_pattern_list;
-
-static int parse_funcname_pattern(const char *var, const char *ep, const char *value, int cflags)
+static int git_config_rename(const char *var, const char *value)
 {
-       const char *name;
-       int namelen;
-       struct funcname_pattern_list *pp;
-
-       name = var + 5; /* "diff." */
-       namelen = ep - name;
-
-       for (pp = funcname_pattern_list; pp; pp = pp->next)
-               if (!strncmp(pp->e.name, name, namelen) && !pp->e.name[namelen])
-                       break;
-       if (!pp) {
-               pp = xcalloc(1, sizeof(*pp));
-               pp->e.name = xmemdupz(name, namelen);
-               pp->next = funcname_pattern_list;
-               funcname_pattern_list = pp;
-       }
-       free(pp->e.pattern);
-       pp->e.pattern = xstrdup(value);
-       pp->e.cflags = cflags;
-       return 0;
+       if (!value)
+               return DIFF_DETECT_RENAME;
+       if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
+               return  DIFF_DETECT_COPY;
+       return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
 }
 
 /*
@@ -141,27 +87,25 @@ int git_diff_ui_config(const char *var, const char *value, void *cb)
                return 0;
        }
        if (!strcmp(var, "diff.renames")) {
-               if (!value)
-                       diff_detect_rename_default = DIFF_DETECT_RENAME;
-               else if (!strcasecmp(value, "copies") ||
-                        !strcasecmp(value, "copy"))
-                       diff_detect_rename_default = DIFF_DETECT_COPY;
-               else if (git_config_bool(var,value))
-                       diff_detect_rename_default = DIFF_DETECT_RENAME;
+               diff_detect_rename_default = git_config_rename(var, value);
                return 0;
        }
        if (!strcmp(var, "diff.autorefreshindex")) {
                diff_auto_refresh_index = git_config_bool(var, value);
                return 0;
        }
+       if (!strcmp(var, "diff.mnemonicprefix")) {
+               diff_mnemonic_prefix = git_config_bool(var, value);
+               return 0;
+       }
+       if (!strcmp(var, "diff.noprefix")) {
+               diff_no_prefix = git_config_bool(var, value);
+               return 0;
+       }
        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);
-       }
+       if (!strcmp(var, "diff.wordregex"))
+               return git_config_string(&diff_word_regex_cfg, var, value);
 
        return git_diff_basic_config(var, value, cb);
 }
@@ -173,29 +117,28 @@ 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 (slot < 0)
+                       return 0;
                if (!value)
                        return config_error_nonbool(var);
                color_parse(value, var, diff_colors[slot]);
                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,
-                                       0);
-                       } else if (!strcmp(ep, ".xfuncname")) {
-                               if (!value)
-                                       return config_error_nonbool(var);
-                               return parse_funcname_pattern(var, ep, value,
-                                       REG_EXTENDED);
-                       }
-               }
+       /* like GNU diff's --suppress-blank-empty option  */
+       if (!strcmp(var, "diff.suppressblankempty") ||
+                       /* for backwards compatibility */
+                       !strcmp(var, "diff.suppress-blank-empty")) {
+               diff_suppress_blank_empty = git_config_bool(var, value);
+               return 0;
        }
 
        return git_color_default_config(var, value, cb);
@@ -205,9 +148,8 @@ 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);
@@ -244,7 +186,6 @@ static struct diff_tempfile {
 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 
 struct emit_callback {
-       struct xdiff_emit_state xm;
        int color_diff;
        unsigned ws_rule;
        int blank_at_eof_in_preimage;
@@ -254,8 +195,9 @@ struct emit_callback {
        sane_truncate_fn truncate;
        const char **label_path;
        struct diff_words_data *diff_words;
+       struct diff_options *opt;
        int *found_changesp;
-       FILE *file;
+       struct strbuf *header;
 };
 
 static int count_lines(const char *data, int size)
@@ -281,21 +223,6 @@ static int count_lines(const char *data, int size)
        return count;
 }
 
-static void print_line_count(FILE *file, int count)
-{
-       switch (count) {
-       case 0:
-               fprintf(file, "0,0");
-               break;
-       case 1:
-               fprintf(file, "1");
-               break;
-       default:
-               fprintf(file, "1,%d", count);
-               break;
-       }
-}
-
 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 {
        if (!DIFF_FILE_VALID(one)) {
@@ -305,6 +232,7 @@ 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;
@@ -355,11 +283,19 @@ static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
        ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
 }
 
-static void emit_line_0(FILE *file, const char *set, const char *reset,
+static void emit_line_0(struct diff_options *o, const char *set, const char *reset,
                        int first, const char *line, int len)
 {
        int has_trailing_newline, has_trailing_carriage_return;
        int nofirst;
+       FILE *file = o->file;
+
+       if (o->output_prefix) {
+               struct strbuf *msg = NULL;
+               msg = o->output_prefix(o, o->output_prefix_data);
+               assert(msg);
+               fwrite(msg->buf, msg->len, 1, file);
+       }
 
        if (len == 0) {
                has_trailing_newline = (first == '\n');
@@ -376,22 +312,23 @@ static void emit_line_0(FILE *file, const char *set, const char *reset,
                nofirst = 0;
        }
 
-       fputs(set, file);
-
-       if (!nofirst)
-               fputc(first, file);
-       fwrite(line, len, 1, file);
-       fputs(reset, file);
+       if (len || !nofirst) {
+               fputs(set, file);
+               if (!nofirst)
+                       fputc(first, file);
+               fwrite(line, len, 1, file);
+               fputs(reset, file);
+       }
        if (has_trailing_carriage_return)
                fputc('\r', file);
        if (has_trailing_newline)
                fputc('\n', file);
 }
 
-static void emit_line(FILE *file, const char *set, const char *reset,
+static void emit_line(struct diff_options *o, const char *set, const char *reset,
                      const char *line, int len)
 {
-       emit_line_0(file, set, reset, line[0], line+1, len-1);
+       emit_line_0(o, set, reset, line[0], line+1, len-1);
 }
 
 static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
@@ -413,15 +350,115 @@ static void emit_add_line(const char *reset,
        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 
        if (!*ws)
-               emit_line_0(ecbdata->file, set, reset, '+', line, len);
+               emit_line_0(ecbdata->opt, set, reset, '+', line, len);
        else if (new_blank_line_at_eof(ecbdata, line, len))
                /* Blank line at EOF - paint '+' as well */
-               emit_line_0(ecbdata->file, ws, reset, '+', line, len);
+               emit_line_0(ecbdata->opt, ws, reset, '+', line, len);
        else {
                /* Emit just the prefix, then the rest. */
-               emit_line_0(ecbdata->file, set, reset, '+', "", 0);
+               emit_line_0(ecbdata->opt, set, reset, '+', "", 0);
                ws_check_emit(line, len, ecbdata->ws_rule,
-                             ecbdata->file, set, reset, ws);
+                             ecbdata->opt->file, set, reset, ws);
+       }
+}
+
+static void emit_hunk_header(struct emit_callback *ecbdata,
+                            const char *line, int len)
+{
+       const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
+       const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
+       const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
+       const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
+       static const char atat[2] = { '@', '@' };
+       const char *cp, *ep;
+       struct strbuf msgbuf = STRBUF_INIT;
+       int org_len = len;
+       int i = 1;
+
+       /*
+        * As a hunk header must begin with "@@ -<old>, +<new> @@",
+        * it always is at least 10 bytes long.
+        */
+       if (len < 10 ||
+           memcmp(line, atat, 2) ||
+           !(ep = memmem(line + 2, len - 2, atat, 2))) {
+               emit_line(ecbdata->opt, plain, reset, line, len);
+               return;
+       }
+       ep += 2; /* skip over @@ */
+
+       /* The hunk header in fraginfo color */
+       strbuf_add(&msgbuf, frag, strlen(frag));
+       strbuf_add(&msgbuf, line, ep - line);
+       strbuf_add(&msgbuf, reset, strlen(reset));
+
+       /*
+        * trailing "\r\n"
+        */
+       for ( ; i < 3; i++)
+               if (line[len - i] == '\r' || line[len - i] == '\n')
+                       len--;
+
+       /* blank before the func header */
+       for (cp = ep; ep - line < len; ep++)
+               if (*ep != ' ' && *ep != '\t')
+                       break;
+       if (ep != cp) {
+               strbuf_add(&msgbuf, plain, strlen(plain));
+               strbuf_add(&msgbuf, cp, ep - cp);
+               strbuf_add(&msgbuf, reset, strlen(reset));
+       }
+
+       if (ep < line + len) {
+               strbuf_add(&msgbuf, func, strlen(func));
+               strbuf_add(&msgbuf, ep, line + len - ep);
+               strbuf_add(&msgbuf, reset, strlen(reset));
+       }
+
+       strbuf_add(&msgbuf, line + len, org_len - len);
+       emit_line(ecbdata->opt, "", "", msgbuf.buf, msgbuf.len);
+       strbuf_release(&msgbuf);
+}
+
+static struct diff_tempfile *claim_diff_tempfile(void) {
+       int i;
+       for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
+               if (!diff_temp[i].name)
+                       return diff_temp + i;
+       die("BUG: diff is failing to clean up its tempfiles");
+}
+
+static int remove_tempfile_installed;
+
+static void remove_tempfile(void)
+{
+       int i;
+       for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
+               if (diff_temp[i].name == diff_temp[i].tmp_path)
+                       unlink_or_warn(diff_temp[i].name);
+               diff_temp[i].name = NULL;
+       }
+}
+
+static void remove_tempfile_on_signal(int signo)
+{
+       remove_tempfile();
+       sigchain_pop(signo);
+       raise(signo);
+}
+
+static void print_line_count(FILE *file, int count)
+{
+       switch (count) {
+       case 0:
+               fprintf(file, "0,0");
+               break;
+       case 1:
+               fprintf(file, "1");
+               break;
+       default:
+               fprintf(file, "1,%d", count);
+               break;
        }
 }
 
@@ -440,7 +477,7 @@ static void emit_rewrite_lines(struct emit_callback *ecb,
                len = endp ? (endp - data + 1) : size;
                if (prefix != '+') {
                        ecb->lno_in_preimage++;
-                       emit_line_0(ecb->file, old, reset, '-',
+                       emit_line_0(ecb->opt, old, reset, '-',
                                    data, len);
                } else {
                        ecb->lno_in_postimage++;
@@ -452,7 +489,7 @@ static void emit_rewrite_lines(struct emit_callback *ecb,
        if (!endp) {
                const char *plain = diff_get_color(ecb->color_diff,
                                                   DIFF_PLAIN);
-               emit_line_0(ecb->file, plain, reset, '\\',
+               emit_line_0(ecb->opt, plain, reset, '\\',
                            nneof, strlen(nneof));
        }
 }
@@ -461,6 +498,8 @@ static void emit_rewrite_diff(const char *name_a,
                              const char *name_b,
                              struct diff_filespec *one,
                              struct diff_filespec *two,
+                             struct userdiff_driver *textconv_one,
+                             struct userdiff_driver *textconv_two,
                              struct diff_options *o)
 {
        int lc_a, lc_b;
@@ -470,130 +509,334 @@ static void emit_rewrite_diff(const char *name_a,
        const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
        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;
+       char *data_one, *data_two;
+       size_t size_one, size_two;
        struct emit_callback ecbdata;
+       char *line_prefix = "";
+       struct strbuf *msgbuf;
+
+       if (o && o->output_prefix) {
+               msgbuf = o->output_prefix(o, o->output_prefix_data);
+               line_prefix = msgbuf->buf;
+       }
+
+       if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
+               a_prefix = o->b_prefix;
+               b_prefix = o->a_prefix;
+       } else {
+               a_prefix = o->a_prefix;
+               b_prefix = o->b_prefix;
+       }
+
+       name_a += (*name_a == '/');
+       name_b += (*name_b == '/');
+       name_a_tab = strchr(name_a, ' ') ? "\t" : "";
+       name_b_tab = strchr(name_b, ' ') ? "\t" : "";
+
+       strbuf_reset(&a_name);
+       strbuf_reset(&b_name);
+       quote_two_c_style(&a_name, a_prefix, name_a, 0);
+       quote_two_c_style(&b_name, b_prefix, name_b, 0);
+
+       size_one = fill_textconv(textconv_one, one, &data_one);
+       size_two = fill_textconv(textconv_two, two, &data_two);
 
        memset(&ecbdata, 0, sizeof(ecbdata));
        ecbdata.color_diff = color_diff;
        ecbdata.found_changesp = &o->found_changes;
        ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
-       ecbdata.file = o->file;
+       ecbdata.opt = o;
        if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
                mmfile_t mf1, mf2;
-               fill_mmfile(&mf1, one);
-               fill_mmfile(&mf2, two);
+               mf1.ptr = (char *)data_one;
+               mf2.ptr = (char *)data_two;
+               mf1.size = size_one;
+               mf2.size = size_two;
                check_blank_at_eof(&mf1, &mf2, &ecbdata);
        }
        ecbdata.lno_in_preimage = 1;
        ecbdata.lno_in_postimage = 1;
 
-       name_a += (*name_a == '/');
-       name_b += (*name_b == '/');
-       name_a_tab = strchr(name_a, ' ') ? "\t" : "";
-       name_b_tab = strchr(name_b, ' ') ? "\t" : "";
-
-       strbuf_reset(&a_name);
-       strbuf_reset(&b_name);
-       quote_two_c_style(&a_name, o->a_prefix, name_a, 0);
-       quote_two_c_style(&b_name, o->b_prefix, name_b, 0);
-
-       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);
+       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,
-               metainfo, b_name.buf, name_b_tab, reset, fraginfo);
+               "%s%s--- %s%s%s\n%s%s+++ %s%s%s\n%s%s@@ -",
+               line_prefix, metainfo, a_name.buf, name_a_tab, reset,
+               line_prefix, metainfo, b_name.buf, name_b_tab, reset,
+               line_prefix, fraginfo);
        print_line_count(o->file, lc_a);
        fprintf(o->file, " +");
        print_line_count(o->file, lc_b);
        fprintf(o->file, " @@%s\n", reset);
        if (lc_a)
-               emit_rewrite_lines(&ecbdata, '-', one->data, one->size);
+               emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
        if (lc_b)
-               emit_rewrite_lines(&ecbdata, '+', two->data, two->size);
+               emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
+       if (textconv_one)
+               free((char *)data_one);
+       if (textconv_two)
+               free((char *)data_two);
 }
 
 struct diff_words_buffer {
        mmfile_t text;
        long alloc;
-       long current; /* output pointer */
-       int suppressed_newline;
+       struct diff_words_orig {
+               const char *begin, *end;
+       } *orig;
+       int orig_nr, orig_alloc;
 };
 
 static void diff_words_append(char *line, unsigned long len,
                struct diff_words_buffer *buffer)
 {
-       if (buffer->text.size + len > buffer->alloc) {
-               buffer->alloc = (buffer->text.size + len) * 3 / 2;
-               buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
-       }
+       ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
        line++;
        len--;
        memcpy(buffer->text.ptr + buffer->text.size, line, len);
        buffer->text.size += len;
+       buffer->text.ptr[buffer->text.size] = '\0';
 }
 
+struct diff_words_style_elem
+{
+       const char *prefix;
+       const char *suffix;
+       const char *color; /* NULL; filled in by the setup code if
+                           * color is enabled */
+};
+
+struct diff_words_style
+{
+       enum diff_words_type type;
+       struct diff_words_style_elem new, old, ctx;
+       const char *newline;
+};
+
+struct diff_words_style diff_words_styles[] = {
+       { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
+       { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
+       { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
+};
+
 struct diff_words_data {
-       struct xdiff_emit_state xm;
        struct diff_words_buffer minus, plus;
-       FILE *file;
+       const char *current_plus;
+       int last_minus;
+       struct diff_options *opt;
+       regex_t *word_regex;
+       enum diff_words_type type;
+       struct diff_words_style *style;
 };
 
-static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
-               int suppress_newline)
+static int fn_out_diff_words_write_helper(FILE *fp,
+                                         struct diff_words_style_elem *st_el,
+                                         const char *newline,
+                                         size_t count, const char *buf,
+                                         const char *line_prefix)
 {
-       const char *ptr;
-       int eol = 0;
+       int print = 0;
+
+       while (count) {
+               char *p = memchr(buf, '\n', count);
+               if (print)
+                       fputs(line_prefix, fp);
+               if (p != buf) {
+                       if (st_el->color && fputs(st_el->color, fp) < 0)
+                               return -1;
+                       if (fputs(st_el->prefix, fp) < 0 ||
+                           fwrite(buf, p ? p - buf : count, 1, fp) != 1 ||
+                           fputs(st_el->suffix, fp) < 0)
+                               return -1;
+                       if (st_el->color && *st_el->color
+                           && fputs(GIT_COLOR_RESET, fp) < 0)
+                               return -1;
+               }
+               if (!p)
+                       return 0;
+               if (fputs(newline, fp) < 0)
+                       return -1;
+               count -= p + 1 - buf;
+               buf = p + 1;
+               print = 1;
+       }
+       return 0;
+}
 
-       if (len == 0)
+/*
+ * '--color-words' algorithm can be described as:
+ *
+ *   1. collect a the minus/plus lines of a diff hunk, divided into
+ *      minus-lines and plus-lines;
+ *
+ *   2. break both minus-lines and plus-lines into words and
+ *      place them into two mmfile_t with one word for each line;
+ *
+ *   3. use xdiff to run diff on the two mmfile_t to get the words level diff;
+ *
+ * And for the common parts of the both file, we output the plus side text.
+ * diff_words->current_plus is used to trace the current position of the plus file
+ * which printed. diff_words->last_minus is used to trace the last minus word
+ * printed.
+ *
+ * For '--graph' to work with '--color-words', we need to output the graph prefix
+ * on each line of color words output. Generally, there are two conditions on
+ * which we should output the prefix.
+ *
+ *   1. diff_words->last_minus == 0 &&
+ *      diff_words->current_plus == diff_words->plus.text.ptr
+ *
+ *      that is: the plus text must start as a new line, and if there is no minus
+ *      word printed, a graph prefix must be printed.
+ *
+ *   2. diff_words->current_plus > diff_words->plus.text.ptr &&
+ *      *(diff_words->current_plus - 1) == '\n'
+ *
+ *      that is: a graph prefix must be printed following a '\n'
+ */
+static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
+{
+       if ((diff_words->last_minus == 0 &&
+               diff_words->current_plus == diff_words->plus.text.ptr) ||
+               (diff_words->current_plus > diff_words->plus.text.ptr &&
+               *(diff_words->current_plus - 1) == '\n')) {
+               return 1;
+       } else {
+               return 0;
+       }
+}
+
+static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
+{
+       struct diff_words_data *diff_words = priv;
+       struct diff_words_style *style = diff_words->style;
+       int minus_first, minus_len, plus_first, plus_len;
+       const char *minus_begin, *minus_end, *plus_begin, *plus_end;
+       struct diff_options *opt = diff_words->opt;
+       struct strbuf *msgbuf;
+       char *line_prefix = "";
+
+       if (line[0] != '@' || parse_hunk_header(line, len,
+                       &minus_first, &minus_len, &plus_first, &plus_len))
                return;
 
-       ptr  = buffer->text.ptr + buffer->current;
-       buffer->current += len;
+       assert(opt);
+       if (opt->output_prefix) {
+               msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
+               line_prefix = msgbuf->buf;
+       }
+
+       /* POSIX requires that first be decremented by one if len == 0... */
+       if (minus_len) {
+               minus_begin = diff_words->minus.orig[minus_first].begin;
+               minus_end =
+                       diff_words->minus.orig[minus_first + minus_len - 1].end;
+       } else
+               minus_begin = minus_end =
+                       diff_words->minus.orig[minus_first].end;
+
+       if (plus_len) {
+               plus_begin = diff_words->plus.orig[plus_first].begin;
+               plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
+       } else
+               plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
 
-       if (ptr[len - 1] == '\n') {
-               eol = 1;
-               len--;
+       if (color_words_output_graph_prefix(diff_words)) {
+               fputs(line_prefix, diff_words->opt->file);
+       }
+       if (diff_words->current_plus != plus_begin) {
+               fn_out_diff_words_write_helper(diff_words->opt->file,
+                               &style->ctx, style->newline,
+                               plus_begin - diff_words->current_plus,
+                               diff_words->current_plus, line_prefix);
+               if (*(plus_begin - 1) == '\n')
+                       fputs(line_prefix, diff_words->opt->file);
+       }
+       if (minus_begin != minus_end) {
+               fn_out_diff_words_write_helper(diff_words->opt->file,
+                               &style->old, style->newline,
+                               minus_end - minus_begin, minus_begin,
+                               line_prefix);
+       }
+       if (plus_begin != plus_end) {
+               fn_out_diff_words_write_helper(diff_words->opt->file,
+                               &style->new, style->newline,
+                               plus_end - plus_begin, plus_begin,
+                               line_prefix);
        }
 
-       fputs(diff_get_color(1, color), file);
-       fwrite(ptr, len, 1, file);
-       fputs(diff_get_color(1, DIFF_RESET), file);
+       diff_words->current_plus = plus_end;
+       diff_words->last_minus = minus_first;
+}
 
-       if (eol) {
-               if (suppress_newline)
-                       buffer->suppressed_newline = 1;
-               else
-                       putc('\n', file);
+/* This function starts looking at *begin, and returns 0 iff a word was found. */
+static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
+               int *begin, int *end)
+{
+       if (word_regex && *begin < buffer->size) {
+               regmatch_t match[1];
+               if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
+                       char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
+                                       '\n', match[0].rm_eo - match[0].rm_so);
+                       *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
+                       *begin += match[0].rm_so;
+                       return *begin >= *end;
+               }
+               return -1;
        }
+
+       /* find the next word */
+       while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
+               (*begin)++;
+       if (*begin >= buffer->size)
+               return -1;
+
+       /* find the end of the word */
+       *end = *begin + 1;
+       while (*end < buffer->size && !isspace(buffer->ptr[*end]))
+               (*end)++;
+
+       return 0;
 }
 
-static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
+/*
+ * This function splits the words in buffer->text, stores the list with
+ * newline separator into out, and saves the offsets of the original words
+ * in buffer->orig.
+ */
+static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
+               regex_t *word_regex)
 {
-       struct diff_words_data *diff_words = priv;
+       int i, j;
+       long alloc = 0;
 
-       if (diff_words->minus.suppressed_newline) {
-               if (line[0] != '+')
-                       putc('\n', diff_words->file);
-               diff_words->minus.suppressed_newline = 0;
-       }
+       out->size = 0;
+       out->ptr = NULL;
 
-       len--;
-       switch (line[0]) {
-               case '-':
-                       print_word(diff_words->file,
-                                  &diff_words->minus, len, DIFF_FILE_OLD, 1);
-                       break;
-               case '+':
-                       print_word(diff_words->file,
-                                  &diff_words->plus, len, DIFF_FILE_NEW, 0);
-                       break;
-               case ' ':
-                       print_word(diff_words->file,
-                                  &diff_words->plus, len, DIFF_PLAIN, 0);
-                       diff_words->minus.current += len;
-                       break;
+       /* fake an empty "0th" word */
+       ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
+       buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
+       buffer->orig_nr = 1;
+
+       for (i = 0; i < buffer->text.size; i++) {
+               if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
+                       return;
+
+               /* store original boundaries */
+               ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
+                               buffer->orig_alloc);
+               buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
+               buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
+               buffer->orig_nr++;
+
+               /* store one word */
+               ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
+               memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
+               out->ptr[out->size + j - i] = '\n';
+               out->size += j - i + 1;
+
+               i = j - 1;
        }
 }
 
@@ -602,54 +845,74 @@ static void diff_words_show(struct diff_words_data *diff_words)
 {
        xpparam_t xpp;
        xdemitconf_t xecfg;
-       xdemitcb_t ecb;
        mmfile_t minus, plus;
-       int i;
+       struct diff_words_style *style = diff_words->style;
 
-       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);
-       for (i = 0; i < minus.size; i++)
-               if (isspace(minus.ptr[i]))
-                       minus.ptr[i] = '\n';
-       diff_words->minus.current = 0;
-
-       plus.size = diff_words->plus.text.size;
-       plus.ptr = xmalloc(plus.size);
-       memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
-       for (i = 0; i < plus.size; i++)
-               if (isspace(plus.ptr[i]))
-                       plus.ptr[i] = '\n';
-       diff_words->plus.current = 0;
-
-       xpp.flags = XDF_NEED_MINIMAL;
-       xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
-       ecb.outf = xdiff_outf;
-       ecb.priv = diff_words;
-       diff_words->xm.consume = fn_out_diff_words_aux;
-       xdi_diff(&minus, &plus, &xpp, &xecfg, &ecb);
+       struct diff_options *opt = diff_words->opt;
+       struct strbuf *msgbuf;
+       char *line_prefix = "";
+
+       assert(opt);
+       if (opt->output_prefix) {
+               msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
+               line_prefix = msgbuf->buf;
+       }
+
+       /* special case: only removal */
+       if (!diff_words->plus.text.size) {
+               fputs(line_prefix, diff_words->opt->file);
+               fn_out_diff_words_write_helper(diff_words->opt->file,
+                       &style->old, style->newline,
+                       diff_words->minus.text.size,
+                       diff_words->minus.text.ptr, line_prefix);
+               diff_words->minus.text.size = 0;
+               return;
+       }
 
+       diff_words->current_plus = diff_words->plus.text.ptr;
+       diff_words->last_minus = 0;
+
+       memset(&xpp, 0, sizeof(xpp));
+       memset(&xecfg, 0, sizeof(xecfg));
+       diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
+       diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
+       xpp.flags = 0;
+       /* as only the hunk header will be parsed, we need a 0-context */
+       xecfg.ctxlen = 0;
+       xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
+                     &xpp, &xecfg);
        free(minus.ptr);
        free(plus.ptr);
+       if (diff_words->current_plus != diff_words->plus.text.ptr +
+                       diff_words->plus.text.size) {
+               if (color_words_output_graph_prefix(diff_words))
+                       fputs(line_prefix, diff_words->opt->file);
+               fn_out_diff_words_write_helper(diff_words->opt->file,
+                       &style->ctx, style->newline,
+                       diff_words->plus.text.ptr + diff_words->plus.text.size
+                       - diff_words->current_plus, diff_words->current_plus,
+                       line_prefix);
+       }
        diff_words->minus.text.size = diff_words->plus.text.size = 0;
+}
 
-       if (diff_words->minus.suppressed_newline) {
-               putc('\n', diff_words->file);
-               diff_words->minus.suppressed_newline = 0;
-       }
+/* In "color-words" mode, show word-diff of words accumulated in the buffer */
+static void diff_words_flush(struct emit_callback *ecbdata)
+{
+       if (ecbdata->diff_words->minus.text.size ||
+           ecbdata->diff_words->plus.text.size)
+               diff_words_show(ecbdata->diff_words);
 }
 
 static void free_diff_words_data(struct emit_callback *ecbdata)
 {
        if (ecbdata->diff_words) {
-               /* flush buffers */
-               if (ecbdata->diff_words->minus.text.size ||
-                               ecbdata->diff_words->plus.text.size)
-                       diff_words_show(ecbdata->diff_words);
-
+               diff_words_flush(ecbdata);
                free (ecbdata->diff_words->minus.text.ptr);
+               free (ecbdata->diff_words->minus.orig);
                free (ecbdata->diff_words->plus.text.ptr);
+               free (ecbdata->diff_words->plus.orig);
+               free(ecbdata->diff_words->word_regex);
                free(ecbdata->diff_words);
                ecbdata->diff_words = NULL;
        }
@@ -701,7 +964,20 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
        const char *meta = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
        const char *plain = diff_get_color(ecbdata->color_diff, DIFF_PLAIN);
        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
+       struct diff_options *o = ecbdata->opt;
+       char *line_prefix = "";
+       struct strbuf *msgbuf;
 
+       if (o && o->output_prefix) {
+               msgbuf = o->output_prefix(o, o->output_prefix_data);
+               line_prefix = msgbuf->buf;
+       }
+
+       if (ecbdata->header) {
+               fprintf(ecbdata->opt->file, "%s", ecbdata->header->buf);
+               strbuf_reset(ecbdata->header);
+               ecbdata->header = NULL;
+       }
        *(ecbdata->found_changesp) = 1;
 
        if (ecbdata->label_path[0]) {
@@ -710,26 +986,35 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 
-               fprintf(ecbdata->file, "%s--- %s%s%s\n",
-                       meta, ecbdata->label_path[0], reset, name_a_tab);
-               fprintf(ecbdata->file, "%s+++ %s%s%s\n",
-                       meta, ecbdata->label_path[1], reset, name_b_tab);
+               fprintf(ecbdata->opt->file, "%s%s--- %s%s%s\n",
+                       line_prefix, meta, ecbdata->label_path[0], reset, name_a_tab);
+               fprintf(ecbdata->opt->file, "%s%s+++ %s%s%s\n",
+                       line_prefix, meta, ecbdata->label_path[1], reset, name_b_tab);
                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
        }
 
+       if (diff_suppress_blank_empty
+           && len == 2 && line[0] == ' ' && line[1] == '\n') {
+               line[0] = '\n';
+               len = 1;
+       }
+
        if (line[0] == '@') {
+               if (ecbdata->diff_words)
+                       diff_words_flush(ecbdata);
                len = sane_truncate_line(ecbdata, line, len);
                find_lno(line, ecbdata);
-               emit_line(ecbdata->file,
-                         diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
-                         reset, line, len);
+               emit_hunk_header(ecbdata, line, len);
                if (line[len-1] != '\n')
-                       putc('\n', ecbdata->file);
+                       putc('\n', ecbdata->opt->file);
                return;
        }
 
        if (len < 1) {
-               emit_line(ecbdata->file, reset, reset, line, len);
+               emit_line(ecbdata->opt, reset, reset, line, len);
+               if (ecbdata->diff_words
+                   && ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN)
+                       fputs("~\n", ecbdata->opt->file);
                return;
        }
 
@@ -743,12 +1028,14 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
                                          &ecbdata->diff_words->plus);
                        return;
                }
-               if (ecbdata->diff_words->minus.text.size ||
-                   ecbdata->diff_words->plus.text.size)
-                       diff_words_show(ecbdata->diff_words);
-               line++;
-               len--;
-               emit_line(ecbdata->file, plain, reset, line, len);
+               diff_words_flush(ecbdata);
+               if (ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN) {
+                       emit_line(ecbdata->opt, plain, reset, line, len);
+                       fputs("~\n", ecbdata->opt->file);
+               } else {
+                       /* don't print the prefix character */
+                       emit_line(ecbdata->opt, plain, reset, line+1, len-1);
+               }
                return;
        }
 
@@ -759,7 +1046,7 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
                ecbdata->lno_in_preimage++;
                if (line[0] == ' ')
                        ecbdata->lno_in_postimage++;
-               emit_line(ecbdata->file, color, reset, line, len);
+               emit_line(ecbdata->opt, color, reset, line, len);
        } else {
                ecbdata->lno_in_postimage++;
                emit_add_line(reset, ecbdata, line + 1, len - 1);
@@ -770,7 +1057,7 @@ 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);
@@ -778,7 +1065,6 @@ 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, " => ");
@@ -835,8 +1121,6 @@ static char *pprint_rename(const char *a, const char *b)
 }
 
 struct diffstat_t {
-       struct xdiff_emit_state xm;
-
        int nr;
        int alloc;
        struct diffstat_file {
@@ -846,7 +1130,7 @@ struct diffstat_t {
                unsigned is_unmerged:1;
                unsigned is_binary:1;
                unsigned is_renamed:1;
-               unsigned int added, deleted;
+               uintmax_t added, deleted;
        } **files;
 };
 
@@ -899,10 +1183,9 @@ static int scale_linear(int it, int width, int max_change)
 }
 
 static void show_name(FILE *file,
-                     const char *prefix, const char *name, int len,
-                     const char *reset, const char *set)
+                     const char *prefix, const char *name, int len)
 {
-       fprintf(file, " %s%s%-*s%s |", set, prefix, len, name, reset);
+       fprintf(file, " %s%-*s |", prefix, len, name);
 }
 
 static void show_graph(FILE *file, char ch, int cnt, const char *set, const char *reset)
@@ -923,8 +1206,7 @@ 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 {
@@ -937,17 +1219,24 @@ static void fill_print_name(struct diffstat_file *file)
        file->print_name = pname;
 }
 
-static void show_stats(struct diffstat_tdata, struct diff_options *options)
+static void show_stats(struct diffstat_t *data, struct diff_options *options)
 {
-       int i, len, add, del, total, adds = 0, dels = 0;
-       int max_change = 0, max_len = 0;
+       int i, len, add, del, adds = 0, dels = 0;
+       uintmax_t max_change = 0, max_len = 0;
        int total_files = data->nr;
        int width, name_width;
        const char *reset, *set, *add_c, *del_c;
+       const char *line_prefix = "";
+       struct strbuf *msg = NULL;
 
        if (data->nr == 0)
                return;
 
+       if (options->output_prefix) {
+               msg = options->output_prefix(options, options->output_prefix_data);
+               line_prefix = msg->buf;
+       }
+
        width = options->stat_width ? options->stat_width : 80;
        name_width = options->stat_name_width ? options->stat_name_width : 50;
 
@@ -969,7 +1258,7 @@ static void show_stats(struct diffstat_t* data, struct diff_options *options)
 
        for (i = 0; i < data->nr; i++) {
                struct diffstat_file *file = data->files[i];
-               int change = file->added + file->deleted;
+               uintmax_t change = file->added + file->deleted;
                fill_print_name(file);
                len = strlen(file->print_name);
                if (max_len < len)
@@ -997,8 +1286,8 @@ static void show_stats(struct diffstat_t* data, struct diff_options *options)
        for (i = 0; i < data->nr; i++) {
                const char *prefix = "";
                char *name = data->files[i]->print_name;
-               int added = data->files[i]->added;
-               int deleted = data->files[i]->deleted;
+               uintmax_t added = data->files[i]->added;
+               uintmax_t deleted = data->files[i]->deleted;
                int name_len;
 
                /*
@@ -1017,17 +1306,21 @@ static void show_stats(struct diffstat_t* data, struct diff_options *options)
                }
 
                if (data->files[i]->is_binary) {
-                       show_name(options->file, prefix, name, len, reset, set);
+                       fprintf(options->file, "%s", line_prefix);
+                       show_name(options->file, prefix, name, len);
                        fprintf(options->file, "  Bin ");
-                       fprintf(options->file, "%s%d%s", del_c, deleted, reset);
+                       fprintf(options->file, "%s%"PRIuMAX"%s",
+                               del_c, deleted, reset);
                        fprintf(options->file, " -> ");
-                       fprintf(options->file, "%s%d%s", add_c, added, reset);
+                       fprintf(options->file, "%s%"PRIuMAX"%s",
+                               add_c, added, reset);
                        fprintf(options->file, " bytes");
                        fprintf(options->file, "\n");
                        continue;
                }
                else if (data->files[i]->is_unmerged) {
-                       show_name(options->file, prefix, name, len, reset, set);
+                       fprintf(options->file, "%s", line_prefix);
+                       show_name(options->file, prefix, name, len);
                        fprintf(options->file, "  Unmerged\n");
                        continue;
                }
@@ -1042,28 +1335,28 @@ static void show_stats(struct diffstat_t* data, struct diff_options *options)
                 */
                add = added;
                del = deleted;
-               total = add + del;
                adds += add;
                dels += del;
 
                if (width <= max_change) {
                        add = scale_linear(add, width, max_change);
                        del = scale_linear(del, width, max_change);
-                       total = add + del;
                }
-               show_name(options->file, prefix, name, len, reset, set);
-               fprintf(options->file, "%5d%s", added + deleted,
+               fprintf(options->file, "%s", line_prefix);
+               show_name(options->file, prefix, name, len);
+               fprintf(options->file, "%5"PRIuMAX"%s", added + deleted,
                                added + deleted ? " " : "");
                show_graph(options->file, '+', add, add_c, reset);
                show_graph(options->file, '-', del, del_c, reset);
                fprintf(options->file, "\n");
        }
+       fprintf(options->file, "%s", line_prefix);
        fprintf(options->file,
-              "%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
-              set, total_files, adds, dels, reset);
+              " %d files changed, %d insertions(+), %d deletions(-)\n",
+              total_files, adds, dels);
 }
 
-static void show_shortstats(struct diffstat_tdata, struct diff_options *options)
+static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
 {
        int i, adds = 0, dels = 0, total_files = data->nr;
 
@@ -1084,11 +1377,17 @@ static void show_shortstats(struct diffstat_t* data, struct diff_options *option
                        }
                }
        }
+       if (options->output_prefix) {
+               struct strbuf *msg = NULL;
+               msg = options->output_prefix(options,
+                               options->output_prefix_data);
+               fprintf(options->file, "%s", msg->buf);
+       }
        fprintf(options->file, " %d files changed, %d insertions(+), %d deletions(-)\n",
               total_files, adds, dels);
 }
 
-static void show_numstat(struct diffstat_tdata, struct diff_options *options)
+static void show_numstat(struct diffstat_t *data, struct diff_options *options)
 {
        int i;
 
@@ -1098,11 +1397,19 @@ static void show_numstat(struct diffstat_t* data, struct diff_options *options)
        for (i = 0; i < data->nr; i++) {
                struct diffstat_file *file = data->files[i];
 
+               if (options->output_prefix) {
+                       struct strbuf *msg = NULL;
+                       msg = options->output_prefix(options,
+                                       options->output_prefix_data);
+                       fprintf(options->file, "%s", msg->buf);
+               }
+
                if (file->is_binary)
                        fprintf(options->file, "-\t-\t");
                else
                        fprintf(options->file,
-                               "%d\t%d\t", file->added, file->deleted);
+                               "%"PRIuMAX"\t%"PRIuMAX"\t",
+                               file->added, file->deleted);
                if (options->line_termination) {
                        fill_print_name(file);
                        if (!file->is_renamed)
@@ -1132,10 +1439,18 @@ struct dirstat_dir {
        int alloc, nr, percent, cumulative;
 };
 
-static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
+static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
+               unsigned long changed, const char *base, int baselen)
 {
        unsigned long this_dir = 0;
        unsigned int sources = 0;
+       const char *line_prefix = "";
+       struct strbuf *msg = NULL;
+
+       if (opt->output_prefix) {
+               msg = opt->output_prefix(opt, opt->output_prefix_data);
+               line_prefix = msg->buf;
+       }
 
        while (dir->nr) {
                struct dirstat_file *f = dir->files;
@@ -1150,7 +1465,7 @@ static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long ch
                slash = strchr(f->name + baselen, '/');
                if (slash) {
                        int newbaselen = slash + 1 - f->name;
-                       this = gather_dirstat(file, dir, changed, f->name, newbaselen);
+                       this = gather_dirstat(opt, dir, changed, f->name, newbaselen);
                        sources++;
                } else {
                        this = f->changed;
@@ -1172,7 +1487,8 @@ static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long ch
                if (permille) {
                        int percent = permille / 10;
                        if (percent >= dir->percent) {
-                               fprintf(file, "%4d.%01d%% %.*s\n", percent, permille % 10, baselen, base);
+                               fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
+                                       percent, permille % 10, baselen, base);
                                if (!dir->cumulative)
                                        return 0;
                        }
@@ -1231,9 +1547,13 @@ 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;
@@ -1248,7 +1568,7 @@ static void show_dirstat(struct diff_options *options)
 
        /* Show all directories with more than x% of the changes */
        qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
-       gather_dirstat(options->file, &dir, changed, "", 0);
+       gather_dirstat(options, &dir, changed, "", 0);
 }
 
 static void free_diffstat_info(struct diffstat_t *diffstat)
@@ -1266,40 +1586,34 @@ static void free_diffstat_info(struct diffstat_t *diffstat)
 }
 
 struct checkdiff_t {
-       struct xdiff_emit_state xm;
        const char *filename;
        int lineno;
+       int conflict_marker_size;
        struct diff_options *o;
        unsigned ws_rule;
        unsigned status;
 };
 
-static int is_conflict_marker(const char *line, unsigned long len)
+static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
 {
        char firstchar;
        int cnt;
 
-       if (len < 8)
+       if (len < marker_size + 1)
                return 0;
        firstchar = line[0];
        switch (firstchar) {
-       case '=': case '>': case '<':
+       case '=': case '>': case '<': case '|':
                break;
        default:
                return 0;
        }
-       for (cnt = 1; cnt < 7; cnt++)
+       for (cnt = 1; cnt < marker_size; cnt++)
                if (line[cnt] != firstchar)
                        return 0;
-       /* line[0] thru line[6] are same as firstchar */
-       if (firstchar == '=') {
-               /* divider between ours and theirs? */
-               if (len != 8 || line[7] != '\n')
-                       return 0;
-       } else if (len < 8 || !isspace(line[7])) {
-               /* not divider before ours nor after theirs */
+       /* line[1] thru line[marker_size-1] are same as firstchar */
+       if (len < marker_size + 1 || !isspace(line[marker_size]))
                return 0;
-       }
        return 1;
 }
 
@@ -1307,29 +1621,39 @@ static void checkdiff_consume(void *priv, char *line, unsigned long len)
 {
        struct checkdiff_t *data = priv;
        int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
+       int marker_size = data->conflict_marker_size;
        const char *ws = diff_get_color(color_diff, DIFF_WHITESPACE);
        const char *reset = diff_get_color(color_diff, DIFF_RESET);
        const char *set = diff_get_color(color_diff, DIFF_FILE_NEW);
        char *err;
+       char *line_prefix = "";
+       struct strbuf *msgbuf;
+
+       assert(data->o);
+       if (data->o->output_prefix) {
+               msgbuf = data->o->output_prefix(data->o,
+                       data->o->output_prefix_data);
+               line_prefix = msgbuf->buf;
+       }
 
        if (line[0] == '+') {
                unsigned bad;
                data->lineno++;
-               if (is_conflict_marker(line + 1, len - 1)) {
+               if (is_conflict_marker(line + 1, marker_size, len - 1)) {
                        data->status |= 1;
                        fprintf(data->o->file,
-                               "%s:%d: leftover conflict marker\n",
-                               data->filename, data->lineno);
+                               "%s%s:%d: leftover conflict marker\n",
+                               line_prefix, data->filename, data->lineno);
                }
                bad = ws_check(line + 1, len - 1, data->ws_rule);
                if (!bad)
                        return;
                data->status |= bad;
                err = whitespace_error_string(bad);
-               fprintf(data->o->file, "%s:%d: %s.\n",
-                       data->filename, data->lineno, err);
+               fprintf(data->o->file, "%s%s:%d: %s.\n",
+                       line_prefix, data->filename, data->lineno, err);
                free(err);
-               emit_line(data->o->file, set, reset, line, 1);
+               emit_line(data->o, set, reset, line, 1);
                ws_check_emit(line + 1, len - 1, data->ws_rule,
                              data->o->file, set, reset, ws);
        } else if (line[0] == ' ') {
@@ -1367,7 +1691,7 @@ static unsigned char *deflate_it(char *data,
        return deflated;
 }
 
-static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
+static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
 {
        void *cp;
        void *delta;
@@ -1396,13 +1720,13 @@ static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
        }
 
        if (delta && delta_size < deflate_size) {
-               fprintf(file, "delta %lu\n", orig_size);
+               fprintf(file, "%sdelta %lu\n", prefix, orig_size);
                free(deflated);
                data = delta;
                data_size = delta_size;
        }
        else {
-               fprintf(file, "literal %lu\n", two->size);
+               fprintf(file, "%sliteral %lu\n", prefix, two->size);
                free(delta);
                data = deflated;
                data_size = deflate_size;
@@ -1420,137 +1744,88 @@ static void emit_binary_diff_body(FILE *file, mmfile_t *one, mmfile_t *two)
                        line[0] = bytes - 26 + 'a' - 1;
                encode_85(line + 1, cp, bytes);
                cp = (char *) cp + bytes;
+               fprintf(file, "%s", prefix);
                fputs(line, file);
                fputc('\n', file);
        }
-       fprintf(file, "\n");
+       fprintf(file, "%s\n", prefix);
        free(data);
 }
 
-static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
+static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *prefix)
 {
-       fprintf(file, "GIT binary patch\n");
-       emit_binary_diff_body(file, one, two);
-       emit_binary_diff_body(file, two, one);
+       fprintf(file, "%sGIT binary patch\n", prefix);
+       emit_binary_diff_body(file, one, two, prefix);
+       emit_binary_diff_body(file, two, one, prefix);
 }
 
-static void setup_diff_attr_check(struct git_attr_check *check)
+static void diff_filespec_load_driver(struct diff_filespec *one)
 {
-       static struct git_attr *attr_diff;
-
-       if (!attr_diff) {
-               attr_diff = git_attr("diff", 4);
-       }
-       check[0].attr = attr_diff;
+       if (!one->driver)
+               one->driver = userdiff_find_by_path(one->path);
+       if (!one->driver)
+               one->driver = userdiff_find_by_name("default");
 }
 
-static void diff_filespec_check_attr(struct diff_filespec *one)
+int diff_filespec_is_binary(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->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;
 }
 
-int diff_filespec_is_binary(struct diff_filespec *one)
+static const struct userdiff_funcname *diff_funcname_pattern(struct diff_filespec *one)
 {
-       diff_filespec_check_attr(one);
-       return one->is_binary;
+       diff_filespec_load_driver(one);
+       return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
 }
 
-static const struct funcname_pattern_entry *funcname_pattern(const char *ident)
-{
-       struct funcname_pattern_list *pp;
-
-       for (pp = funcname_pattern_list; pp; pp = pp->next)
-               if (!strcmp(ident, pp->e.name))
-                       return &pp->e;
-       return NULL;
-}
-
-static const struct funcname_pattern_entry builtin_funcname_pattern[] = {
-       { "java",
-         "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
-         "^[ \t]*(([ \t]*[A-Za-z_][A-Za-z_0-9]*){2,}[ \t]*\\([^;]*)$",
-         REG_EXTENDED },
-       { "pascal",
-         "^((procedure|function|constructor|destructor|interface|"
-               "implementation|initialization|finalization)[ \t]*.*)$"
-         "|"
-         "^(.*=[ \t]*(class|record).*)$",
-         REG_EXTENDED },
-       { "bibtex", "(@[a-zA-Z]{1,}[ \t]*\\{{0,1}[ \t]*[^ \t\"@',\\#}{~%]*).*$",
-         REG_EXTENDED },
-       { "tex",
-         "^(\\\\((sub)*section|chapter|part)\\*{0,1}\\{.*)$",
-         REG_EXTENDED },
-       { "ruby", "^[ \t]*((class|module|def)[ \t].*)$",
-         REG_EXTENDED },
-};
-
-static const struct funcname_pattern_entry *diff_funcname_pattern(struct diff_filespec *one)
+static const char *userdiff_word_regex(struct diff_filespec *one)
 {
-       const char *ident;
-       const struct funcname_pattern_entry *pe;
-       int i;
+       diff_filespec_load_driver(one);
+       return one->driver->word_regex;
+}
 
-       diff_filespec_check_attr(one);
-       ident = one->funcname_pattern_ident;
+void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
+{
+       if (!options->a_prefix)
+               options->a_prefix = a;
+       if (!options->b_prefix)
+               options->b_prefix = b;
+}
 
-       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");
+struct userdiff_driver *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);
+       if (!one->driver->textconv)
+               return NULL;
 
-       /* Look up custom "funcname.$ident" regexp from config. */
-       pe = funcname_pattern(ident);
-       if (pe)
-               return pe;
+       if (one->driver->textconv_want_cache && !one->driver->textconv_cache) {
+               struct notes_cache *c = xmalloc(sizeof(*c));
+               struct strbuf name = STRBUF_INIT;
 
-       /*
-        * 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];
+               strbuf_addf(&name, "textconv/%s", one->driver->name);
+               notes_cache_init(c, name.buf, one->driver->textconv);
+               one->driver->textconv_cache = c;
+       }
 
-       return NULL;
+       return one->driver;
 }
 
 static void builtin_diff(const char *name_a,
@@ -1558,6 +1833,7 @@ static void builtin_diff(const char *name_a,
                         struct diff_filespec *one,
                         struct diff_filespec *two,
                         const char *xfrm_msg,
+                        int must_show_header,
                         struct diff_options *o,
                         int complete_rewrite)
 {
@@ -1566,61 +1842,111 @@ static void builtin_diff(const char *name_a,
        char *a_one, *b_two;
        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;
+       struct userdiff_driver *textconv_one = NULL;
+       struct userdiff_driver *textconv_two = NULL;
+       struct strbuf header = STRBUF_INIT;
+       struct strbuf *msgbuf;
+       char *line_prefix = "";
+
+       if (o->output_prefix) {
+               msgbuf = o->output_prefix(o, o->output_prefix_data);
+               line_prefix = msgbuf->buf;
+       }
+
+       if (DIFF_OPT_TST(o, SUBMODULE_LOG) &&
+                       (!one->mode || S_ISGITLINK(one->mode)) &&
+                       (!two->mode || S_ISGITLINK(two->mode))) {
+               const char *del = diff_get_color_opt(o, DIFF_FILE_OLD);
+               const char *add = diff_get_color_opt(o, DIFF_FILE_NEW);
+               show_submodule_summary(o->file, one ? one->path : two->path,
+                               one->sha1, two->sha1, two->dirty_submodule,
+                               del, add, reset);
+               return;
+       }
+
+       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)) {
+               a_prefix = o->b_prefix;
+               b_prefix = o->a_prefix;
+       } else {
+               a_prefix = o->a_prefix;
+               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(o->a_prefix, name_a + (*name_a == '/'));
-       b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
+       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";
        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
-       fprintf(o->file, "%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
+       strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, set, a_one, b_two, reset);
        if (lbl[0][0] == '/') {
                /* /dev/null */
-               fprintf(o->file, "%snew file mode %06o%s\n", set, two->mode, reset);
-               if (xfrm_msg && xfrm_msg[0])
-                       fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
+               strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, set, two->mode, reset);
+               if (xfrm_msg)
+                       strbuf_addstr(&header, xfrm_msg);
+               must_show_header = 1;
        }
        else if (lbl[1][0] == '/') {
-               fprintf(o->file, "%sdeleted file mode %06o%s\n", set, one->mode, reset);
-               if (xfrm_msg && xfrm_msg[0])
-                       fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
+               strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, set, one->mode, reset);
+               if (xfrm_msg)
+                       strbuf_addstr(&header, xfrm_msg);
+               must_show_header = 1;
        }
        else {
                if (one->mode != two->mode) {
-                       fprintf(o->file, "%sold mode %06o%s\n", set, one->mode, reset);
-                       fprintf(o->file, "%snew mode %06o%s\n", set, two->mode, reset);
+                       strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, set, one->mode, reset);
+                       strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, set, two->mode, reset);
+                       must_show_header = 1;
                }
-               if (xfrm_msg && xfrm_msg[0])
-                       fprintf(o->file, "%s%s%s\n", set, xfrm_msg, reset);
+               if (xfrm_msg)
+                       strbuf_addstr(&header, xfrm_msg);
+
                /*
                 * we do not run diff between different kind
                 * of objects.
                 */
                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))) {
+                       fprintf(o->file, "%s", header.buf);
+                       strbuf_reset(&header);
+                       emit_rewrite_diff(name_a, name_b, one, two,
+                                               textconv_one, textconv_two, o);
                        o->found_changes = 1;
                        goto free_ab_and_return;
                }
        }
 
-       if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
-               die("unable to read files to diff");
-
        if (!DIFF_OPT_TST(o, TEXT) &&
-           (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
+           ( (!textconv_one && diff_filespec_is_binary(one)) ||
+             (!textconv_two && diff_filespec_is_binary(two)) )) {
+               if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
+                       die("unable to read files to diff");
                /* Quite common confusing case */
                if (mf1.size == mf2.size &&
-                   !memcmp(mf1.ptr, mf2.ptr, mf1.size))
+                   !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
+                       if (must_show_header)
+                               fprintf(o->file, "%s", header.buf);
                        goto free_ab_and_return;
+               }
+               fprintf(o->file, "%s", header.buf);
+               strbuf_reset(&header);
                if (DIFF_OPT_TST(o, BINARY))
-                       emit_binary_diff(o->file, &mf1, &mf2);
+                       emit_binary_diff(o->file, &mf1, &mf2, line_prefix);
                else
-                       fprintf(o->file, "Binary files %s and %s differ\n",
-                               lbl[0], lbl[1]);
+                       fprintf(o->file, "%sBinary files %s and %s differ\n",
+                               line_prefix, lbl[0], lbl[1]);
                o->found_changes = 1;
        }
        else {
@@ -1628,14 +1954,22 @@ static void builtin_diff(const char *name_a,
                const char *diffopts = getenv("GIT_DIFF_OPTS");
                xpparam_t xpp;
                xdemitconf_t xecfg;
-               xdemitcb_t ecb;
                struct emit_callback ecbdata;
-               const struct funcname_pattern_entry *pe;
+               const struct userdiff_funcname *pe;
+
+               if (!DIFF_XDL_TST(o, WHITESPACE_FLAGS) || must_show_header) {
+                       fprintf(o->file, "%s", header.buf);
+                       strbuf_reset(&header);
+               }
+
+               mf1.size = fill_textconv(textconv_one, one, &mf1.ptr);
+               mf2.size = fill_textconv(textconv_two, two, &mf2.ptr);
 
                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;
@@ -1644,9 +1978,11 @@ static void builtin_diff(const char *name_a,
                ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
                if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
-               ecbdata.file = o->file;
-               xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
+               ecbdata.opt = o;
+               ecbdata.header = header.len ? &header : NULL;
+               xpp.flags = o->xdl_opts;
                xecfg.ctxlen = o->context;
+               xecfg.interhunkctxlen = o->interhunkcontext;
                xecfg.flags = XDL_EMIT_FUNCNAMES;
                if (pe)
                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
@@ -1656,20 +1992,55 @@ static void builtin_diff(const char *name_a,
                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
                else if (!prefixcmp(diffopts, "-u"))
                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
-               ecb.outf = xdiff_outf;
-               ecb.priv = &ecbdata;
-               ecbdata.xm.consume = fn_out_consume;
-               if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS)) {
+               if (o->word_diff) {
+                       int i;
+
                        ecbdata.diff_words =
                                xcalloc(1, sizeof(struct diff_words_data));
-                       ecbdata.diff_words->file = o->file;
+                       ecbdata.diff_words->type = o->word_diff;
+                       ecbdata.diff_words->opt = o;
+                       if (!o->word_regex)
+                               o->word_regex = userdiff_word_regex(one);
+                       if (!o->word_regex)
+                               o->word_regex = userdiff_word_regex(two);
+                       if (!o->word_regex)
+                               o->word_regex = diff_word_regex_cfg;
+                       if (o->word_regex) {
+                               ecbdata.diff_words->word_regex = (regex_t *)
+                                       xmalloc(sizeof(regex_t));
+                               if (regcomp(ecbdata.diff_words->word_regex,
+                                               o->word_regex,
+                                               REG_EXTENDED | REG_NEWLINE))
+                                       die ("Invalid regular expression: %s",
+                                                       o->word_regex);
+                       }
+                       for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
+                               if (o->word_diff == diff_words_styles[i].type) {
+                                       ecbdata.diff_words->style =
+                                               &diff_words_styles[i];
+                                       break;
+                               }
+                       }
+                       if (DIFF_OPT_TST(o, COLOR_DIFF)) {
+                               struct diff_words_style *st = ecbdata.diff_words->style;
+                               st->old.color = diff_get_color_opt(o, DIFF_FILE_OLD);
+                               st->new.color = diff_get_color_opt(o, DIFF_FILE_NEW);
+                               st->ctx.color = diff_get_color_opt(o, DIFF_PLAIN);
+                       }
                }
-               xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
-               if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
+               xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
+                             &xpp, &xecfg);
+               if (o->word_diff)
                        free_diff_words_data(&ecbdata);
+               if (textconv_one)
+                       free(mf1.ptr);
+               if (textconv_two)
+                       free(mf2.ptr);
+               xdiff_clear_find_func(&xecfg);
        }
 
  free_ab_and_return:
+       strbuf_release(&header);
        diff_free_filespec_data(one);
        diff_free_filespec_data(two);
        free(a_one);
@@ -1711,13 +2082,12 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
                /* Crazy xdl interfaces.. */
                xpparam_t xpp;
                xdemitconf_t xecfg;
-               xdemitcb_t ecb;
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
-               xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
-               ecb.outf = xdiff_outf;
-               ecb.priv = diffstat;
-               xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+               xpp.flags = o->xdl_opts;
+               xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
+                             &xpp, &xecfg);
        }
 
  free_and_return:
@@ -1738,11 +2108,11 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
                return;
 
        memset(&data, 0, sizeof(data));
-       data.xm.consume = checkdiff_consume;
        data.filename = name_b ? name_b : name_a;
        data.lineno = 0;
        data.o = o;
        data.ws_rule = whitespace_rule(attr_path);
+       data.conflict_marker_size = ll_merge_marker_size(attr_path);
 
        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
                die("unable to read files to diff");
@@ -1759,14 +2129,13 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
                /* Crazy xdl interfaces.. */
                xpparam_t xpp;
                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;
-               ecb.outf = xdiff_outf;
-               ecb.priv = &data;
-               xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+               xpp.flags = 0;
+               xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
+                             &xpp, &xecfg);
 
                if (data.ws_rule & WS_BLANK_AT_EOF) {
                        struct emit_callback ecbdata;
@@ -1802,6 +2171,7 @@ 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;
 }
 
@@ -1834,7 +2204,8 @@ static int reuse_worktree_file(const char *name, const unsigned char *sha1, int
        struct stat st;
        int pos, len;
 
-       /* We do not read the cache ourselves here, because the
+       /*
+        * We do not read the cache ourselves here, because the
         * benchmark with my previous version that always reads cache
         * shows that it makes things worse for diff-tree comparing
         * two linux-2.6 kernel trees in an already checked out work
@@ -1858,7 +2229,7 @@ static int reuse_worktree_file(const char *name, const unsigned char *sha1, int
         * objects however would tend to be slower as they need
         * to be individually opened and inflated.
         */
-       if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
+       if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1))
                return 0;
 
        len = strlen(name);
@@ -1874,6 +2245,13 @@ static int reuse_worktree_file(const char *name, const unsigned char *sha1, int
        if (hashcmp(sha1, ce->sha1) || !S_ISREG(ce->ce_mode))
                return 0;
 
+       /*
+        * If ce is marked as "assume unchanged", there is no
+        * guarantee that work tree matches what we are looking for.
+        */
+       if ((ce->ce_flags & CE_VALID) || ce_skip_worktree(ce))
+               return 0;
+
        /*
         * If ce matches the file in the work tree, we can reuse it.
         */
@@ -1886,10 +2264,9 @@ 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));
@@ -1904,9 +2281,14 @@ static int populate_from_stdin(struct diff_filespec *s)
 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
 {
        int len;
-       char *data = xmalloc(100);
+       char *data = xmalloc(100), *dirty = "";
+
+       /* Are we looking at the work tree? */
+       if (s->dirty_submodule)
+               dirty = "-dirty";
+
        len = snprintf(data, 100,
-               "Subproject commit %s\n", sha1_to_hex(s->sha1));
+                      "Subproject commit %s%s\n", sha1_to_hex(s->sha1), dirty);
        s->data = data;
        s->size = len;
        s->should_free = 1;
@@ -1941,7 +2323,7 @@ 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;
 
@@ -1961,19 +2343,18 @@ 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;
@@ -1984,7 +2365,6 @@ 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);
@@ -2026,30 +2406,48 @@ void diff_free_filespec_data(struct diff_filespec *s)
        s->cnt_data = NULL;
 }
 
-static void prep_temp_blob(struct diff_tempfile *temp,
+static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
                           void *blob,
                           unsigned long size,
                           const unsigned char *sha1,
                           int mode)
 {
        int fd;
+       struct strbuf buf = STRBUF_INIT;
+       struct strbuf template = STRBUF_INIT;
+       char *path_dup = xstrdup(path);
+       const char *base = basename(path_dup);
+
+       /* Generate "XXXXXX_basename.ext" */
+       strbuf_addstr(&template, "XXXXXX_");
+       strbuf_addstr(&template, base);
 
-       fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
+       fd = git_mkstemps(temp->tmp_path, PATH_MAX, template.buf,
+                       strlen(base) + 1);
        if (fd < 0)
-               die("unable to create temp-file: %s", strerror(errno));
+               die_errno("unable to create temp-file");
+       if (convert_to_working_tree(path,
+                       (const char *)blob, (size_t)size, &buf)) {
+               blob = buf.buf;
+               size = buf.len;
+       }
        if (write_in_full(fd, blob, size) != size)
-               die("unable to write temp-file");
+               die_errno("unable to write temp-file");
        close(fd);
        temp->name = temp->tmp_path;
        strcpy(temp->hex, sha1_to_hex(sha1));
        temp->hex[40] = 0;
        sprintf(temp->mode, "%06o", mode);
+       strbuf_release(&buf);
+       strbuf_release(&template);
+       free(path_dup);
 }
 
-static void prepare_temp_file(const char *name,
-                             struct diff_tempfile *temp,
-                             struct diff_filespec *one)
+static struct diff_tempfile *prepare_temp_file(const char *name,
+               struct diff_filespec *one)
 {
+       struct diff_tempfile *temp = claim_diff_tempfile();
+
        if (!DIFF_FILE_VALID(one)) {
        not_a_valid_file:
                /* A '-' entry produces this for file-2, and
@@ -2058,7 +2456,13 @@ static void prepare_temp_file(const char *name,
                temp->name = "/dev/null";
                strcpy(temp->hex, ".");
                strcpy(temp->mode, ".");
-               return;
+               return temp;
+       }
+
+       if (!remove_tempfile_installed) {
+               atexit(remove_tempfile);
+               sigchain_push_common(remove_tempfile_on_signal);
+               remove_tempfile_installed = 1;
        }
 
        if (!one->sha1_valid ||
@@ -2067,22 +2471,18 @@ static void prepare_temp_file(const char *name,
                if (lstat(name, &st) < 0) {
                        if (errno == ENOENT)
                                goto not_a_valid_file;
-                       die("stat(%s): %s", name, strerror(errno));
+                       die_errno("stat(%s)", 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);
-                       if (ret < 0)
-                               die("readlink(%s)", name);
-                       prep_temp_blob(temp, buf, sz,
+                       struct strbuf sb = STRBUF_INIT;
+                       if (strbuf_readlink(&sb, name, st.st_size) < 0)
+                               die_errno("readlink(%s)", name);
+                       prep_temp_blob(name, temp, sb.buf, sb.len,
                                       (one->sha1_valid ?
                                        one->sha1 : null_sha1),
                                       (one->sha1_valid ?
                                        one->mode : S_IFLNK));
+                       strbuf_release(&sb);
                }
                else {
                        /* we can borrow from the file in the work tree */
@@ -2099,32 +2499,15 @@ static void prepare_temp_file(const char *name,
                         */
                        sprintf(temp->mode, "%06o", one->mode);
                }
-               return;
+               return temp;
        }
        else {
                if (diff_populate_filespec(one, 0))
                        die("cannot read data blob for %s", one->path);
-               prep_temp_blob(temp, one->data, one->size,
+               prep_temp_blob(name, temp, one->data, one->size,
                               one->sha1, one->mode);
        }
-}
-
-static void remove_tempfile(void)
-{
-       int i;
-
-       for (i = 0; i < 2; i++)
-               if (diff_temp[i].name == diff_temp[i].tmp_path) {
-                       unlink(diff_temp[i].name);
-                       diff_temp[i].name = NULL;
-               }
-}
-
-static void remove_tempfile_on_signal(int signo)
-{
-       remove_tempfile();
-       signal(SIGINT, SIG_DFL);
-       raise(signo);
+       return temp;
 }
 
 /* An external diff command takes:
@@ -2142,34 +2525,22 @@ static void run_external_diff(const char *pgm,
                              int complete_rewrite)
 {
        const char *spawn_arg[10];
-       struct diff_tempfile *temp = diff_temp;
        int retval;
-       static int atexit_asked = 0;
-       const char *othername;
        const char **arg = &spawn_arg[0];
 
-       othername = (other? other : name);
-       if (one && two) {
-               prepare_temp_file(name, &temp[0], one);
-               prepare_temp_file(othername, &temp[1], two);
-               if (! atexit_asked &&
-                   (temp[0].name == temp[0].tmp_path ||
-                    temp[1].name == temp[1].tmp_path)) {
-                       atexit_asked = 1;
-                       atexit(remove_tempfile);
-               }
-               signal(SIGINT, remove_tempfile_on_signal);
-       }
-
        if (one && two) {
+               struct diff_tempfile *temp_one, *temp_two;
+               const char *othername = (other ? other : name);
+               temp_one = prepare_temp_file(name, one);
+               temp_two = prepare_temp_file(othername, two);
                *arg++ = pgm;
                *arg++ = name;
-               *arg++ = temp[0].name;
-               *arg++ = temp[0].hex;
-               *arg++ = temp[0].mode;
-               *arg++ = temp[1].name;
-               *arg++ = temp[1].hex;
-               *arg++ = temp[1].mode;
+               *arg++ = temp_one->name;
+               *arg++ = temp_one->hex;
+               *arg++ = temp_one->mode;
+               *arg++ = temp_two->name;
+               *arg++ = temp_two->hex;
+               *arg++ = temp_two->mode;
                if (other) {
                        *arg++ = other;
                        *arg++ = xfrm_msg;
@@ -2180,7 +2551,7 @@ static void run_external_diff(const char *pgm,
        }
        *arg = NULL;
        fflush(NULL);
-       retval = run_command_v_opt(spawn_arg, 0);
+       retval = run_command_v_opt(spawn_arg, RUN_USING_SHELL);
        remove_tempfile();
        if (retval) {
                fprintf(stderr, "external diff died, stopping at %s.\n", name);
@@ -2188,29 +2559,6 @@ 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 int similarity_index(struct diff_filepair *p)
 {
        return p->score * 100 / MAX_SCORE;
@@ -2222,36 +2570,53 @@ static void fill_metainfo(struct strbuf *msg,
                          struct diff_filespec *one,
                          struct diff_filespec *two,
                          struct diff_options *o,
-                         struct diff_filepair *p)
+                         struct diff_filepair *p,
+                         int *must_show_header,
+                         int use_color)
 {
+       const char *set = diff_get_color(use_color, DIFF_METAINFO);
+       const char *reset = diff_get_color(use_color, DIFF_RESET);
+       struct strbuf *msgbuf;
+       char *line_prefix = "";
+
+       *must_show_header = 1;
+       if (o->output_prefix) {
+               msgbuf = o->output_prefix(o, o->output_prefix_data);
+               line_prefix = msgbuf->buf;
+       }
        strbuf_init(msg, PATH_MAX * 2 + 300);
        switch (p->status) {
        case DIFF_STATUS_COPIED:
-               strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
-               strbuf_addstr(msg, "\ncopy from ");
+               strbuf_addf(msg, "%s%ssimilarity index %d%%",
+                           line_prefix, set, similarity_index(p));
+               strbuf_addf(msg, "%s\n%s%scopy from ",
+                           reset,  line_prefix, set);
                quote_c_style(name, msg, NULL, 0);
-               strbuf_addstr(msg, "\ncopy to ");
+               strbuf_addf(msg, "%s\n%s%scopy to ", reset, line_prefix, set);
                quote_c_style(other, msg, NULL, 0);
-               strbuf_addch(msg, '\n');
+               strbuf_addf(msg, "%s\n", reset);
                break;
        case DIFF_STATUS_RENAMED:
-               strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
-               strbuf_addstr(msg, "\nrename from ");
+               strbuf_addf(msg, "%s%ssimilarity index %d%%",
+                           line_prefix, set, similarity_index(p));
+               strbuf_addf(msg, "%s\n%s%srename from ",
+                           reset, line_prefix, set);
                quote_c_style(name, msg, NULL, 0);
-               strbuf_addstr(msg, "\nrename to ");
+               strbuf_addf(msg, "%s\n%s%srename to ",
+                           reset, line_prefix, set);
                quote_c_style(other, msg, NULL, 0);
-               strbuf_addch(msg, '\n');
+               strbuf_addf(msg, "%s\n", reset);
                break;
        case DIFF_STATUS_MODIFIED:
                if (p->score) {
-                       strbuf_addf(msg, "dissimilarity index %d%%\n",
-                                   similarity_index(p));
+                       strbuf_addf(msg, "%s%sdissimilarity index %d%%%s\n",
+                                   line_prefix,
+                                   set, similarity_index(p), reset);
                        break;
                }
                /* fallthru */
        default:
-               /* nothing */
-               ;
+               *must_show_header = 0;
        }
        if (one && two && hashcmp(one->sha1, two->sha1)) {
                int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
@@ -2262,15 +2627,14 @@ static void fill_metainfo(struct strbuf *msg,
                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
                                abbrev = 40;
                }
-               strbuf_addf(msg, "index %.*s..%.*s",
-                           abbrev, sha1_to_hex(one->sha1),
-                           abbrev, sha1_to_hex(two->sha1));
+               strbuf_addf(msg, "%s%sindex %s..", set,
+                           line_prefix,
+                           find_unique_abbrev(one->sha1, abbrev));
+               strbuf_addstr(msg, find_unique_abbrev(two->sha1, abbrev));
                if (one->mode == two->mode)
                        strbuf_addf(msg, " %06o", one->mode);
-               strbuf_addch(msg, '\n');
+               strbuf_addf(msg, "%s\n", reset);
        }
-       if (msg->len)
-               strbuf_setlen(msg, msg->len - 1);
 }
 
 static void run_diff_cmd(const char *pgm,
@@ -2285,18 +2649,25 @@ static void run_diff_cmd(const char *pgm,
 {
        const char *xfrm_msg = NULL;
        int complete_rewrite = (p->status == DIFF_STATUS_MODIFIED) && p->score;
-
-       if (msg) {
-               fill_metainfo(msg, name, other, one, two, o, p);
-               xfrm_msg = msg->len ? msg->buf : NULL;
-       }
+       int must_show_header = 0;
 
        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 (msg) {
+               /*
+                * don't use colors when the header is intended for an
+                * external diff driver
+                */
+               fill_metainfo(msg, name, other, one, two, o, p,
+                             &must_show_header,
+                             DIFF_OPT_TST(o, COLOR_DIFF) && !pgm);
+               xfrm_msg = msg->len ? msg->buf : NULL;
        }
 
        if (pgm) {
@@ -2306,7 +2677,8 @@ static void run_diff_cmd(const char *pgm,
        }
        if (one && two)
                builtin_diff(name, other ? other : name,
-                            one, two, xfrm_msg, o, complete_rewrite);
+                            one, two, xfrm_msg, must_show_header,
+                            o, complete_rewrite);
        else
                fprintf(o->file, "* Unmerged path %s\n", name);
 }
@@ -2321,9 +2693,9 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
                                return;
                        }
                        if (lstat(one->path, &st) < 0)
-                               die("stat %s", one->path);
+                               die_errno("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
@@ -2443,6 +2815,7 @@ static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
 void diff_setup(struct diff_options *options)
 {
        memset(options, 0, sizeof(*options));
+       memset(&diff_queued_diff, 0, sizeof(diff_queued_diff));
 
        options->file = stdout;
 
@@ -2450,19 +2823,20 @@ void diff_setup(struct diff_options *options)
        options->break_opt = -1;
        options->rename_limit = -1;
        options->dirstat_percent = 3;
-       DIFF_OPT_CLR(options, DIRSTAT_CUMULATIVE);
        options->context = 3;
 
        options->change = diff_change;
        options->add_remove = diff_addremove;
        if (diff_use_color_default > 0)
                DIFF_OPT_SET(options, COLOR_DIFF);
-       else
-               DIFF_OPT_CLR(options, COLOR_DIFF);
        options->detect_rename = diff_detect_rename_default;
 
-       options->a_prefix = "a/";
-       options->b_prefix = "b/";
+       if (diff_no_prefix) {
+               options->a_prefix = options->b_prefix = "";
+       } else if (!diff_mnemonic_prefix) {
+               options->a_prefix = "a/";
+               options->b_prefix = "b/";
+       }
 }
 
 int diff_setup_done(struct diff_options *options)
@@ -2480,6 +2854,20 @@ int diff_setup_done(struct diff_options *options)
        if (count > 1)
                die("--name-only, --name-status, --check and -s are mutually exclusive");
 
+       /*
+        * Most of the time we can say "there are changes"
+        * only by checking if there are changed paths, but
+        * --ignore-whitespace* options force us to look
+        * inside contents.
+        */
+
+       if (DIFF_XDL_TST(options, IGNORE_WHITESPACE) ||
+           DIFF_XDL_TST(options, IGNORE_WHITESPACE_CHANGE) ||
+           DIFF_XDL_TST(options, IGNORE_WHITESPACE_AT_EOL))
+               DIFF_OPT_SET(options, DIFF_FROM_CONTENTS);
+       else
+               DIFF_OPT_CLR(options, DIFF_FROM_CONTENTS);
+
        if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
                options->detect_rename = DIFF_DETECT_COPY;
 
@@ -2519,6 +2907,12 @@ int diff_setup_done(struct diff_options *options)
         */
        if (options->pickaxe)
                DIFF_OPT_SET(options, RECURSIVE);
+       /*
+        * When patches are generated, submodules diffed against the work tree
+        * must be checked for dirtiness too so it can be shown in the output
+        */
+       if (options->output_format & DIFF_FORMAT_PATCH)
+               DIFF_OPT_SET(options, DIRTY_SUBMODULES);
 
        if (options->detect_rename && options->rename_limit < 0)
                options->rename_limit = diff_rename_limit_default;
@@ -2540,7 +2934,7 @@ int diff_setup_done(struct diff_options *options)
         * to have found.  It does not make sense not to return with
         * exit code in such a case either.
         */
-       if (DIFF_OPT_TST(options, QUIET)) {
+       if (DIFF_OPT_TST(options, QUICK)) {
                options->output_format = DIFF_FORMAT_NO_OUTPUT;
                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
        }
@@ -2602,7 +2996,7 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        const char *arg = av[0];
 
        /* Output format options */
-       if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
+       if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
                options->output_format |= DIFF_FORMAT_PATCH;
        else if (opt_arg(arg, 'U', "unified", &options->context))
                options->output_format |= DIFF_FORMAT_PATCH;
@@ -2619,6 +3013,10 @@ 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;
@@ -2688,11 +3086,13 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 
        /* xdiff options */
        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
-               options->xdl_opts |= XDF_IGNORE_WHITESPACE;
+               DIFF_XDL_SET(options, IGNORE_WHITESPACE);
        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
-               options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
+               DIFF_XDL_SET(options, IGNORE_WHITESPACE_CHANGE);
        else if (!strcmp(arg, "--ignore-space-at-eol"))
-               options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
+               DIFF_XDL_SET(options, IGNORE_WHITESPACE_AT_EOL);
+       else if (!strcmp(arg, "--patience"))
+               DIFF_XDL_SET(options, PATIENCE_DIFF);
 
        /* flags options */
        else if (!strcmp(arg, "--binary")) {
@@ -2711,20 +3111,70 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                DIFF_OPT_SET(options, FOLLOW_RENAMES);
        else if (!strcmp(arg, "--color"))
                DIFF_OPT_SET(options, COLOR_DIFF);
+       else if (!prefixcmp(arg, "--color=")) {
+               int value = git_config_colorbool(NULL, arg+8, -1);
+               if (value == 0)
+                       DIFF_OPT_CLR(options, COLOR_DIFF);
+               else if (value > 0)
+                       DIFF_OPT_SET(options, COLOR_DIFF);
+               else
+                       return error("option `color' expects \"always\", \"auto\", or \"never\"");
+       }
        else if (!strcmp(arg, "--no-color"))
                DIFF_OPT_CLR(options, COLOR_DIFF);
-       else if (!strcmp(arg, "--color-words"))
-               options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
+       else if (!strcmp(arg, "--color-words")) {
+               DIFF_OPT_SET(options, COLOR_DIFF);
+               options->word_diff = DIFF_WORDS_COLOR;
+       }
+       else if (!prefixcmp(arg, "--color-words=")) {
+               DIFF_OPT_SET(options, COLOR_DIFF);
+               options->word_diff = DIFF_WORDS_COLOR;
+               options->word_regex = arg + 14;
+       }
+       else if (!strcmp(arg, "--word-diff")) {
+               if (options->word_diff == DIFF_WORDS_NONE)
+                       options->word_diff = DIFF_WORDS_PLAIN;
+       }
+       else if (!prefixcmp(arg, "--word-diff=")) {
+               const char *type = arg + 12;
+               if (!strcmp(type, "plain"))
+                       options->word_diff = DIFF_WORDS_PLAIN;
+               else if (!strcmp(type, "color")) {
+                       DIFF_OPT_SET(options, COLOR_DIFF);
+                       options->word_diff = DIFF_WORDS_COLOR;
+               }
+               else if (!strcmp(type, "porcelain"))
+                       options->word_diff = DIFF_WORDS_PORCELAIN;
+               else if (!strcmp(type, "none"))
+                       options->word_diff = DIFF_WORDS_NONE;
+               else
+                       die("bad --word-diff argument: %s", type);
+       }
+       else if (!prefixcmp(arg, "--word-diff-regex=")) {
+               if (options->word_diff == DIFF_WORDS_NONE)
+                       options->word_diff = DIFF_WORDS_PLAIN;
+               options->word_regex = arg + 18;
+       }
        else if (!strcmp(arg, "--exit-code"))
                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
        else if (!strcmp(arg, "--quiet"))
-               DIFF_OPT_SET(options, QUIET);
+               DIFF_OPT_SET(options, QUICK);
        else if (!strcmp(arg, "--ext-diff"))
                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);
+       else if (!strcmp(arg, "--submodule"))
+               DIFF_OPT_SET(options, SUBMODULE_LOG);
+       else if (!prefixcmp(arg, "--submodule=")) {
+               if (!strcmp(arg + 12, "log"))
+                       DIFF_OPT_SET(options, SUBMODULE_LOG);
+       }
 
        /* misc options */
        else if (!strcmp(arg, "-z"))
@@ -2756,8 +3206,13 @@ 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");
+               if (!options->file)
+                       die_errno("Could not open '%s'", arg + strlen("--output="));
                options->close_file = 1;
        } else
                return 0;
@@ -2773,7 +3228,7 @@ static int parse_num(const char **cp_p)
        num = 0;
        scale = 1;
        dot = 0;
-       for(;;) {
+       for (;;) {
                ch = *cp;
                if ( !dot && ch == '.' ) {
                        scale = 1;
@@ -2886,6 +3341,11 @@ static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
 {
        int line_termination = opt->line_termination;
        int inter_name_termination = line_termination ? '\t' : '\0';
+       if (opt->output_prefix) {
+               struct strbuf *msg = NULL;
+               msg = opt->output_prefix(opt, opt->output_prefix_data);
+               fprintf(opt->file, "%s", msg->buf);
+       }
 
        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
                fprintf(opt->file, ":%06o %06o %s ", p->one->mode, p->two->mode,
@@ -2940,7 +3400,8 @@ int diff_unmodified_pair(struct diff_filepair *p)
         * dealing with a change.
         */
        if (one->sha1_valid && two->sha1_valid &&
-           !hashcmp(one->sha1, two->sha1))
+           !hashcmp(one->sha1, two->sha1) &&
+           !one->dirty_submodule && !two->dirty_submodule)
                return 1; /* no change */
        if (!one->sha1_valid && !two->sha1_valid)
                return 1; /* both look at the same file on the filesystem. */
@@ -3075,6 +3536,8 @@ static void diff_resolve_rename_copy(void)
                }
                else if (hashcmp(p->one->sha1, p->two->sha1) ||
                         p->one->mode != p->two->mode ||
+                        p->one->dirty_submodule ||
+                        p->two->dirty_submodule ||
                         is_null_sha1(p->one->sha1))
                        p->status = DIFF_STATUS_MODIFIED;
                else {
@@ -3128,55 +3591,68 @@ static void show_file_mode_name(FILE *file, const char *newdelete, struct diff_f
 }
 
 
-static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name)
+static void show_mode_change(FILE *file, struct diff_filepair *p, int show_name,
+               const char *line_prefix)
 {
        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
-               fprintf(file, " mode change %06o => %06o%c", p->one->mode, p->two->mode,
-                       show_name ? ' ' : '\n');
+               fprintf(file, "%s mode change %06o => %06o%c", line_prefix, p->one->mode,
+                       p->two->mode, show_name ? ' ' : '\n');
                if (show_name) {
                        write_name_quoted(p->two->path, file, '\n');
                }
        }
 }
 
-static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p)
+static void show_rename_copy(FILE *file, const char *renamecopy, struct diff_filepair *p,
+                       const char *line_prefix)
 {
        char *names = pprint_rename(p->one->path, p->two->path);
 
        fprintf(file, " %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
        free(names);
-       show_mode_change(file, p, 0);
+       show_mode_change(file, p, 0, line_prefix);
 }
 
-static void diff_summary(FILE *file, struct diff_filepair *p)
+static void diff_summary(struct diff_options *opt, struct diff_filepair *p)
 {
+       FILE *file = opt->file;
+       char *line_prefix = "";
+
+       if (opt->output_prefix) {
+               struct strbuf *buf = opt->output_prefix(opt, opt->output_prefix_data);
+               line_prefix = buf->buf;
+       }
+
        switch(p->status) {
        case DIFF_STATUS_DELETED:
+               fputs(line_prefix, file);
                show_file_mode_name(file, "delete", p->one);
                break;
        case DIFF_STATUS_ADDED:
+               fputs(line_prefix, file);
                show_file_mode_name(file, "create", p->two);
                break;
        case DIFF_STATUS_COPIED:
-               show_rename_copy(file, "copy", p);
+               fputs(line_prefix, file);
+               show_rename_copy(file, "copy", p, line_prefix);
                break;
        case DIFF_STATUS_RENAMED:
-               show_rename_copy(file, "rename", p);
+               fputs(line_prefix, file);
+               show_rename_copy(file, "rename", p, line_prefix);
                break;
        default:
                if (p->score) {
-                       fputs(" rewrite ", file);
+                       fprintf(file, "%s rewrite ", line_prefix);
                        write_name_quoted(p->two->path, file, ' ');
                        fprintf(file, "(%d%%)\n", similarity_index(p));
                }
-               show_mode_change(file, p, !p->score);
+               show_mode_change(file, p, !p->score, line_prefix);
                break;
        }
 }
 
 struct patch_id_t {
-       struct xdiff_emit_state xm;
-       SHA_CTX *ctx;
+       git_SHA_CTX *ctx;
        int patchlen;
 };
 
@@ -3204,7 +3680,7 @@ 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;
 }
 
@@ -3213,23 +3689,22 @@ 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;
-       data.xm.consume = patch_id_consume;
 
        for (i = 0; i < q->nr; i++) {
                xpparam_t xpp;
                xdemitconf_t xecfg;
-               xdemitcb_t ecb;
                mmfile_t mf1, mf2;
                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");
@@ -3280,17 +3755,16 @@ 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;
+               xpp.flags = 0;
                xecfg.ctxlen = 3;
                xecfg.flags = XDL_EMIT_FUNCNAMES;
-               ecb.outf = xdiff_outf;
-               ecb.priv = &data;
-               xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+               xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
+                             &xpp, &xecfg);
        }
 
-       SHA1_Final(sha1, &ctx);
+       git_SHA1_Final(sha1, &ctx);
        return 0;
 }
 
@@ -3304,8 +3778,7 @@ int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
                diff_free_filepair(q->queue[i]);
 
        free(q->queue);
-       q->queue = NULL;
-       q->nr = q->alloc = 0;
+       DIFF_QUEUE_CLEAR(q);
 
        return result;
 }
@@ -3364,7 +3837,6 @@ void diff_flush(struct diff_options *options)
                struct diffstat_t diffstat;
 
                memset(&diffstat, 0, sizeof(struct diffstat_t));
-               diffstat.xm.consume = diffstat_consume;
                for (i = 0; i < q->nr; i++) {
                        struct diff_filepair *p = q->queue[i];
                        if (check_pair_status(p))
@@ -3383,11 +3855,35 @@ void diff_flush(struct diff_options *options)
                show_dirstat(options);
 
        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
-               for (i = 0; i < q->nr; i++)
-                       diff_summary(options->file, q->queue[i]);
+               for (i = 0; i < q->nr; i++) {
+                       diff_summary(options, q->queue[i]);
+               }
                separator++;
        }
 
+       if (output_format & DIFF_FORMAT_NO_OUTPUT &&
+           DIFF_OPT_TST(options, EXIT_WITH_STATUS) &&
+           DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
+               /*
+                * run diff_flush_patch for the exit status. setting
+                * options->file to /dev/null should be safe, becaue we
+                * aren't supposed to produce any output anyway.
+                */
+               if (options->close_file)
+                       fclose(options->file);
+               options->file = fopen("/dev/null", "w");
+               if (!options->file)
+                       die_errno("Could not open /dev/null");
+               options->close_file = 1;
+               for (i = 0; i < q->nr; i++) {
+                       struct diff_filepair *p = q->queue[i];
+                       if (check_pair_status(p))
+                               diff_flush_patch(p, options);
+                       if (options->found_changes)
+                               break;
+               }
+       }
+
        if (output_format & DIFF_FORMAT_PATCH) {
                if (separator) {
                        putc(options->line_termination, options->file);
@@ -3411,10 +3907,21 @@ void diff_flush(struct diff_options *options)
                diff_free_filepair(q->queue[i]);
 free_queue:
        free(q->queue);
-       q->queue = NULL;
-       q->nr = q->alloc = 0;
+       DIFF_QUEUE_CLEAR(q);
        if (options->close_file)
                fclose(options->file);
+
+       /*
+        * Report the content-level differences with HAS_CHANGES;
+        * diff_addremove/diff_change does not set the bit when
+        * DIFF_FROM_CONTENTS is in effect (e.g. with -w).
+        */
+       if (DIFF_OPT_TST(options, DIFF_FROM_CONTENTS)) {
+               if (options->found_changes)
+                       DIFF_OPT_SET(options, HAS_CHANGES);
+               else
+                       DIFF_OPT_CLR(options, HAS_CHANGES);
+       }
 }
 
 static void diffcore_apply_filter(const char *filter)
@@ -3422,8 +3929,7 @@ static void diffcore_apply_filter(const char *filter)
        int i;
        struct diff_queue_struct *q = &diff_queued_diff;
        struct diff_queue_struct outq;
-       outq.queue = NULL;
-       outq.nr = outq.alloc = 0;
+       DIFF_QUEUE_CLEAR(&outq);
 
        if (!filter)
                return;
@@ -3491,14 +3997,13 @@ 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;
+       DIFF_QUEUE_CLEAR(&outq);
 
        for (i = 0; i < q->nr; i++) {
                struct diff_filepair *p = q->queue[i];
 
                /*
-                * 1. Entries that come from stat info dirtyness
+                * 1. Entries that come from stat info dirtiness
                 *    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
@@ -3534,8 +4039,31 @@ static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
        *q = outq;
 }
 
+static int diffnamecmp(const void *a_, const void *b_)
+{
+       const struct diff_filepair *a = *((const struct diff_filepair **)a_);
+       const struct diff_filepair *b = *((const struct diff_filepair **)b_);
+       const char *name_a, *name_b;
+
+       name_a = a->one ? a->one->path : a->two->path;
+       name_b = b->one ? b->one->path : b->two->path;
+       return strcmp(name_a, name_b);
+}
+
+void diffcore_fix_diff_index(struct diff_options *options)
+{
+       struct diff_queue_struct *q = &diff_queued_diff;
+       qsort(q->queue, q->nr, sizeof(q->queue[0]), diffnamecmp);
+}
+
 void diffcore_std(struct diff_options *options)
 {
+       /* We never run this function more than one time, because the
+        * rename/copy detection logic can only run once.
+        */
+       if (diff_queued_diff.run)
+               return;
+
        if (options->skip_stat_unmatch)
                diffcore_skip_stat_unmatch(options);
        if (options->break_opt != -1)
@@ -3551,10 +4079,12 @@ void diffcore_std(struct diff_options *options)
        diff_resolve_rename_copy();
        diffcore_apply_filter(options->filter);
 
-       if (diff_queued_diff.nr)
+       if (diff_queued_diff.nr && !DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
                DIFF_OPT_SET(options, HAS_CHANGES);
        else
                DIFF_OPT_CLR(options, HAS_CHANGES);
+
+       diff_queued_diff.run = 1;
 }
 
 int diff_result_code(struct diff_options *opt, int status)
@@ -3575,7 +4105,7 @@ int diff_result_code(struct diff_options *opt, int status)
 void diff_addremove(struct diff_options *options,
                    int addremove, unsigned mode,
                    const unsigned char *sha1,
-                   const char *concatpath)
+                   const char *concatpath, unsigned dirty_submodule)
 {
        struct diff_filespec *one, *two;
 
@@ -3607,18 +4137,22 @@ void diff_addremove(struct diff_options *options,
 
        if (addremove != '+')
                fill_filespec(one, sha1, mode);
-       if (addremove != '-')
+       if (addremove != '-') {
                fill_filespec(two, sha1, mode);
+               two->dirty_submodule = dirty_submodule;
+       }
 
        diff_queue(&diff_queued_diff, one, two);
-       DIFF_OPT_SET(options, HAS_CHANGES);
+       if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
+               DIFF_OPT_SET(options, HAS_CHANGES);
 }
 
 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 *concatpath)
+                const char *concatpath,
+                unsigned old_dirty_submodule, unsigned new_dirty_submodule)
 {
        struct diff_filespec *one, *two;
 
@@ -3631,6 +4165,8 @@ void diff_change(struct diff_options *options,
                const unsigned char *tmp_c;
                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
+               tmp = old_dirty_submodule; old_dirty_submodule = new_dirty_submodule;
+                       new_dirty_submodule = tmp;
        }
 
        if (options->prefix &&
@@ -3641,9 +4177,12 @@ void diff_change(struct diff_options *options,
        two = alloc_filespec(concatpath);
        fill_filespec(one, old_sha1, old_mode);
        fill_filespec(two, new_sha1, new_mode);
+       one->dirty_submodule = old_dirty_submodule;
+       two->dirty_submodule = new_dirty_submodule;
 
        diff_queue(&diff_queued_diff, one, two);
-       DIFF_OPT_SET(options, HAS_CHANGES);
+       if (!DIFF_OPT_TST(options, DIFF_FROM_CONTENTS))
+               DIFF_OPT_SET(options, HAS_CHANGES);
 }
 
 void diff_unmerge(struct diff_options *options,
@@ -3661,3 +4200,85 @@ 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;
+       int err = 0;
+
+       temp = prepare_temp_file(spec->path, spec);
+       *arg++ = pgm;
+       *arg++ = temp->name;
+       *arg = NULL;
+
+       memset(&child, 0, sizeof(child));
+       child.use_shell = 1;
+       child.argv = argv;
+       child.out = -1;
+       if (start_command(&child)) {
+               remove_tempfile();
+               return NULL;
+       }
+
+       if (strbuf_read(&buf, child.out, 0) < 0)
+               err = error("error reading from textconv command '%s'", pgm);
+       close(child.out);
+
+       if (finish_command(&child) || err) {
+               strbuf_release(&buf);
+               remove_tempfile();
+               return NULL;
+       }
+       remove_tempfile();
+
+       return strbuf_detach(&buf, outsize);
+}
+
+size_t fill_textconv(struct userdiff_driver *driver,
+                    struct diff_filespec *df,
+                    char **outbuf)
+{
+       size_t size;
+
+       if (!driver || !driver->textconv) {
+               if (!DIFF_FILE_VALID(df)) {
+                       *outbuf = "";
+                       return 0;
+               }
+               if (diff_populate_filespec(df, 0))
+                       die("unable to read files to diff");
+               *outbuf = df->data;
+               return df->size;
+       }
+
+       if (driver->textconv_cache) {
+               *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
+                                         &size);
+               if (*outbuf)
+                       return size;
+       }
+
+       *outbuf = run_textconv(driver->textconv, df, &size);
+       if (!*outbuf)
+               die("unable to read files to diff");
+
+       if (driver->textconv_cache) {
+               /* ignore errors, as we might be in a readonly repository */
+               notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
+                               size);
+               /*
+                * we could save up changes and flush them all at the end,
+                * but we would need an extra call after all diffing is done.
+                * Since generating a cache entry is the slow path anyway,
+                * this extra overhead probably isn't a big deal.
+                */
+               notes_cache_write(driver->textconv_cache);
+       }
+
+       return size;
+}