t5302: Use prerequisite tags to skip 64-bit offset tests
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 20135cbe7c8209835dca8d52d1d48e7aa871c1b3..75d9fab8f87033ba86f687fbc92c5d4ee4c68d7b 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -11,6 +11,8 @@
 #include "attr.h"
 #include "run-command.h"
 #include "utf8.h"
+#include "userdiff.h"
+#include "sigchain.h"
 
 #ifdef NO_FAST_WORKING_DIRECTORY
 #define FAST_WORKING_DIRECTORY 0
 #endif
 
 static int diff_detect_rename_default;
-static int diff_rename_limit_default = 100;
+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 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 */
 };
 
+static void diff_filespec_load_driver(struct diff_filespec *one);
+static char *run_textconv(const char *, struct diff_filespec *, size_t *);
+
 static int parse_diff_color_slot(const char *var, int ofs)
 {
        if (!strcasecmp(var+ofs, "plain"))
@@ -54,87 +62,14 @@ static int parse_diff_color_slot(const char *var, int ofs)
        die("bad config variable '%s'", var);
 }
 
-static struct ll_diff_driver {
-       const char *name;
-       struct ll_diff_driver *next;
-       const char *cmd;
-} *user_diff, **user_diff_tail;
-
-/*
- * Currently there is only "diff.<drivername>.command" variable;
- * because there are "diff.color.<slot>" variables, we are parsing
- * this in a bit convoluted way to allow low level diff driver
- * called "color".
- */
-static int parse_lldiff_command(const char *var, const char *ep, const char *value)
-{
-       const char *name;
-       int namelen;
-       struct ll_diff_driver *drv;
-
-       name = var + 5;
-       namelen = ep - name;
-       for (drv = user_diff; drv; drv = drv->next)
-               if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
-                       break;
-       if (!drv) {
-               drv = xcalloc(1, sizeof(struct ll_diff_driver));
-               drv->name = xmemdupz(name, namelen);
-               if (!user_diff_tail)
-                       user_diff_tail = &user_diff;
-               *user_diff_tail = drv;
-               user_diff_tail = &(drv->next);
-       }
-
-       return git_config_string(&(drv->cmd), var, value);
-}
-
-/*
- * 'diff.<what>.funcname' attribute can be specified in the configuration
- * to define a customized regexp to find the beginning of a function to
- * be used for hunk header lines of "diff -p" style output.
- */
-static struct funcname_pattern {
-       char *name;
-       char *pattern;
-       struct funcname_pattern *next;
-} *funcname_pattern_list;
-
-static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
-{
-       const char *name;
-       int namelen;
-       struct funcname_pattern *pp;
-
-       name = var + 5; /* "diff." */
-       namelen = ep - name;
-
-       for (pp = funcname_pattern_list; pp; pp = pp->next)
-               if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
-                       break;
-       if (!pp) {
-               pp = xcalloc(1, sizeof(*pp));
-               pp->name = xmemdupz(name, namelen);
-               pp->next = funcname_pattern_list;
-               funcname_pattern_list = pp;
-       }
-       free(pp->pattern);
-       pp->pattern = xstrdup(value);
-       return 0;
-}
-
 /*
  * These are to give UI layer defaults.
  * The core-level commands such as git-diff-files should
  * never be affected by the setting of diff.renames
  * the user happens to have in the configuration file.
  */
-int git_diff_ui_config(const char *var, const char *value)
+int git_diff_ui_config(const char *var, const char *value, void *cb)
 {
-       if (!strcmp(var, "diff.renamelimit")) {
-               diff_rename_limit_default = git_config_int(var, value);
-               return 0;
-       }
        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
                diff_use_color_default = git_config_colorbool(var, value, -1);
                return 0;
@@ -153,24 +88,31 @@ int git_diff_ui_config(const char *var, const char *value)
                diff_auto_refresh_index = git_config_bool(var, value);
                return 0;
        }
-       if (!strcmp(var, "diff.external")) {
-               if (!value)
-                       return config_error_nonbool(var);
-               external_diff_cmd_cfg = xstrdup(value);
+       if (!strcmp(var, "diff.mnemonicprefix")) {
+               diff_mnemonic_prefix = git_config_bool(var, value);
                return 0;
        }
-       if (!prefixcmp(var, "diff.")) {
-               const char *ep = strrchr(var, '.');
+       if (!strcmp(var, "diff.external"))
+               return git_config_string(&external_diff_cmd_cfg, var, value);
+       if (!strcmp(var, "diff.wordregex"))
+               return git_config_string(&diff_word_regex_cfg, var, value);
 
-               if (ep != var + 4 && !strcmp(ep, ".command"))
-                       return parse_lldiff_command(var, ep, value);
-       }
-
-       return git_diff_basic_config(var, value);
+       return git_diff_basic_config(var, value, cb);
 }
 
-int git_diff_basic_config(const char *var, const char *value)
+int git_diff_basic_config(const char *var, const char *value, void *cb)
 {
+       if (!strcmp(var, "diff.renamelimit")) {
+               diff_rename_limit_default = git_config_int(var, value);
+               return 0;
+       }
+
+       switch (userdiff_config(var, value)) {
+               case 0: break;
+               case -1: return -1;
+               default: return 0;
+       }
+
        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
                int slot = parse_diff_color_slot(var, 11);
                if (!value)
@@ -179,27 +121,23 @@ int git_diff_basic_config(const char *var, const char *value)
                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);
-                       }
-               }
+       /* 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);
+       return git_color_default_config(var, value, cb);
 }
 
 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);
@@ -233,6 +171,33 @@ static struct diff_tempfile {
        char tmp_path[PATH_MAX];
 } diff_temp[2];
 
+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(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 int count_lines(const char *data, int size)
 {
        int count, ch, completely_empty = 1, nl_just_seen = 0;
@@ -297,6 +262,8 @@ static void emit_rewrite_diff(const char *name_a,
                              const char *name_b,
                              struct diff_filespec *one,
                              struct diff_filespec *two,
+                             const char *textconv_one,
+                             const char *textconv_two,
                              struct diff_options *o)
 {
        int lc_a, lc_b;
@@ -308,6 +275,17 @@ static void emit_rewrite_diff(const char *name_a,
        const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
        const char *reset = diff_get_color(color_diff, DIFF_RESET);
        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
+       const char *a_prefix, *b_prefix;
+       const char *data_one, *data_two;
+       size_t size_one, size_two;
+
+       if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
+               a_prefix = o->b_prefix;
+               b_prefix = o->a_prefix;
+       } else {
+               a_prefix = o->a_prefix;
+               b_prefix = o->b_prefix;
+       }
 
        name_a += (*name_a == '/');
        name_b += (*name_b == '/');
@@ -316,13 +294,32 @@ static void emit_rewrite_diff(const char *name_a,
 
        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);
+       quote_two_c_style(&a_name, a_prefix, name_a, 0);
+       quote_two_c_style(&b_name, 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);
+       if (textconv_one) {
+               data_one = run_textconv(textconv_one, one, &size_one);
+               if (!data_one)
+                       die("unable to read files to diff");
+       }
+       else {
+               data_one = one->data;
+               size_one = one->size;
+       }
+       if (textconv_two) {
+               data_two = run_textconv(textconv_two, two, &size_two);
+               if (!data_two)
+                       die("unable to read files to diff");
+       }
+       else {
+               data_two = two->data;
+               size_two = two->size;
+       }
+
+       lc_a = count_lines(data_one, size_one);
+       lc_b = count_lines(data_two, size_two);
        fprintf(o->file,
                "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
                metainfo, a_name.buf, name_a_tab, reset,
@@ -332,9 +329,9 @@ static void emit_rewrite_diff(const char *name_a,
        print_line_count(o->file, lc_b);
        fprintf(o->file, " @@%s\n", reset);
        if (lc_a)
-               copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
+               copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
        if (lc_b)
-               copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
+               copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
 }
 
 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
@@ -346,6 +343,7 @@ 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;
@@ -354,83 +352,138 @@ static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 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_data {
-       struct xdiff_emit_state xm;
        struct diff_words_buffer minus, plus;
+       const char *current_plus;
        FILE *file;
+       regex_t *word_regex;
 };
 
-static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
-               int suppress_newline)
+static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 {
-       const char *ptr;
-       int eol = 0;
+       struct diff_words_data *diff_words = priv;
+       int minus_first, minus_len, plus_first, plus_len;
+       const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 
-       if (len == 0)
+       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;
+       /* 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 (ptr[len - 1] == '\n') {
-               eol = 1;
-               len--;
+       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 (diff_words->current_plus != plus_begin)
+               fwrite(diff_words->current_plus,
+                               plus_begin - diff_words->current_plus, 1,
+                               diff_words->file);
+       if (minus_begin != minus_end)
+               color_fwrite_lines(diff_words->file,
+                               diff_get_color(1, DIFF_FILE_OLD),
+                               minus_end - minus_begin, minus_begin);
+       if (plus_begin != plus_end)
+               color_fwrite_lines(diff_words->file,
+                               diff_get_color(1, DIFF_FILE_NEW),
+                               plus_end - plus_begin, plus_begin);
+
+       diff_words->current_plus = plus_end;
+}
+
+/* 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;
        }
 
-       fputs(diff_get_color(1, color), file);
-       fwrite(ptr, len, 1, file);
-       fputs(diff_get_color(1, DIFF_RESET), file);
+       /* find the next word */
+       while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
+               (*begin)++;
+       if (*begin >= buffer->size)
+               return -1;
 
-       if (eol) {
-               if (suppress_newline)
-                       buffer->suppressed_newline = 1;
-               else
-                       putc('\n', file);
-       }
+       /* 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;
        }
 }
 
@@ -441,46 +494,41 @@ static void diff_words_show(struct diff_words_data *diff_words)
        xdemitconf_t xecfg;
        xdemitcb_t ecb;
        mmfile_t minus, plus;
-       int i;
 
-       memset(&xecfg, 0, sizeof(xecfg));
-       minus.size = diff_words->minus.text.size;
-       minus.ptr = xmalloc(minus.size);
-       memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
-       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;
+       /* special case: only removal */
+       if (!diff_words->plus.text.size) {
+               color_fwrite_lines(diff_words->file,
+                       diff_get_color(1, DIFF_FILE_OLD),
+                       diff_words->minus.text.size, diff_words->minus.text.ptr);
+               diff_words->minus.text.size = 0;
+               return;
+       }
 
-       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);
+       diff_words->current_plus = diff_words->plus.text.ptr;
 
+       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 = XDF_NEED_MINIMAL;
+       /* 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, &ecb);
        free(minus.ptr);
        free(plus.ptr);
+       if (diff_words->current_plus != diff_words->plus.text.ptr +
+                       diff_words->plus.text.size)
+               fwrite(diff_words->current_plus,
+                       diff_words->plus.text.ptr + diff_words->plus.text.size
+                       - diff_words->current_plus, 1,
+                       diff_words->file);
        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;
-       }
 }
 
 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
 
 struct emit_callback {
-       struct xdiff_emit_state xm;
        int nparents, color_diff;
        unsigned ws_rule;
        sane_truncate_fn truncate;
@@ -499,7 +547,10 @@ static void free_diff_words_data(struct emit_callback *ecbdata)
                        diff_words_show(ecbdata->diff_words);
 
                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;
        }
@@ -514,13 +565,20 @@ const char *diff_get_color(int diff_use_color, enum color_diff ix)
 
 static void emit_line(FILE *file, const char *set, const char *reset, const char *line, int len)
 {
-       int has_trailing_newline = (len > 0 && line[len-1] == '\n');
+       int has_trailing_newline, has_trailing_carriage_return;
+
+       has_trailing_newline = (len > 0 && line[len-1] == '\n');
        if (has_trailing_newline)
                len--;
+       has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
+       if (has_trailing_carriage_return)
+               len--;
 
        fputs(set, file);
        fwrite(line, len, 1, file);
        fputs(reset, file);
+       if (has_trailing_carriage_return)
+               fputc('\r', file);
        if (has_trailing_newline)
                fputc('\n', file);
 }
@@ -535,9 +593,9 @@ static void emit_add_line(const char *reset, struct emit_callback *ecbdata, cons
        else {
                /* Emit just the prefix, then the rest. */
                emit_line(ecbdata->file, set, reset, line, ecbdata->nparents);
-               (void)check_and_emit_line(line + ecbdata->nparents,
-                   len - ecbdata->nparents, ecbdata->ws_rule,
-                   ecbdata->file, set, reset, ws);
+               ws_check_emit(line + ecbdata->nparents,
+                             len - ecbdata->nparents, ecbdata->ws_rule,
+                             ecbdata->file, set, reset, ws);
        }
 }
 
@@ -583,6 +641,12 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
                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;
+       }
+
        /* This is not really necessary for now because
         * this codepath only deals with two-way diffs.
         */
@@ -646,7 +710,7 @@ 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);
@@ -654,7 +718,6 @@ 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, " => ");
@@ -711,8 +774,6 @@ 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 {
@@ -799,8 +860,7 @@ 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 {
@@ -815,7 +875,7 @@ static void fill_print_name(struct diffstat_file *file)
 
 static void show_stats(struct diffstat_t* data, struct diff_options *options)
 {
-       int i, len, add, del, total, adds = 0, dels = 0;
+       int i, len, add, del, adds = 0, dels = 0;
        int max_change = 0, max_len = 0;
        int total_files = data->nr;
        int width, name_width;
@@ -830,12 +890,12 @@ static void show_stats(struct diffstat_t* data, struct diff_options *options)
        /* Sanity: give at least 5 columns to the graph,
         * but leave at least 10 columns for the name.
         */
-       if (width < name_width + 15) {
-               if (name_width <= 25)
-                       width = name_width + 15;
-               else
-                       name_width = width - 15;
-       }
+       if (width < 25)
+               width = 25;
+       if (name_width < 10)
+               name_width = 10;
+       else if (width < name_width + 15)
+               name_width = width - 15;
 
        /* Find the longest filename and max number of changes */
        reset = diff_get_color_opt(options, DIFF_RESET);
@@ -918,17 +978,16 @@ 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 ", added + deleted);
+               fprintf(options->file, "%5d%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");
@@ -997,18 +1056,23 @@ static void show_numstat(struct diffstat_t* data, struct diff_options *options)
        }
 }
 
-struct diffstat_dir {
-       struct diffstat_file **files;
-       int nr, percent, cumulative;
+struct dirstat_file {
+       const char *name;
+       unsigned long changed;
+};
+
+struct dirstat_dir {
+       struct dirstat_file *files;
+       int alloc, nr, percent, cumulative;
 };
 
-static long gather_dirstat(FILE *file, struct diffstat_dir *dir, unsigned long changed, const char *base, int baselen)
+static long gather_dirstat(FILE *file, struct dirstat_dir *dir, unsigned long changed, const char *base, int baselen)
 {
        unsigned long this_dir = 0;
        unsigned int sources = 0;
 
        while (dir->nr) {
-               struct diffstat_file *f = *dir->files;
+               struct dirstat_file *f = dir->files;
                int namelen = strlen(f->name);
                unsigned long this;
                char *slash;
@@ -1023,10 +1087,7 @@ static long gather_dirstat(FILE *file, struct diffstat_dir *dir, unsigned long c
                        this = gather_dirstat(file, dir, changed, f->name, newbaselen);
                        sources++;
                } else {
-                       if (f->is_unmerged || f->is_binary)
-                               this = 0;
-                       else
-                               this = f->added + f->deleted;
+                       this = f->changed;
                        dir->files++;
                        dir->nr--;
                        sources += 2;
@@ -1054,19 +1115,69 @@ static long gather_dirstat(FILE *file, struct diffstat_dir *dir, unsigned long c
        return this_dir;
 }
 
-static void show_dirstat(struct diffstat_t *data, struct diff_options *options)
+static int dirstat_compare(const void *_a, const void *_b)
+{
+       const struct dirstat_file *a = _a;
+       const struct dirstat_file *b = _b;
+       return strcmp(a->name, b->name);
+}
+
+static void show_dirstat(struct diff_options *options)
 {
        int i;
        unsigned long changed;
-       struct diffstat_dir dir;
+       struct dirstat_dir dir;
+       struct diff_queue_struct *q = &diff_queued_diff;
+
+       dir.files = NULL;
+       dir.alloc = 0;
+       dir.nr = 0;
+       dir.percent = options->dirstat_percent;
+       dir.cumulative = DIFF_OPT_TST(options, DIRSTAT_CUMULATIVE);
 
-       /* Calculate total changes */
        changed = 0;
-       for (i = 0; i < data->nr; i++) {
-               if (data->files[i]->is_binary || data->files[i]->is_unmerged)
+       for (i = 0; i < q->nr; i++) {
+               struct diff_filepair *p = q->queue[i];
+               const char *name;
+               unsigned long copied, added, damage;
+
+               name = p->one->path ? p->one->path : p->two->path;
+
+               if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
+                       diff_populate_filespec(p->one, 0);
+                       diff_populate_filespec(p->two, 0);
+                       diffcore_count_changes(p->one, p->two, NULL, NULL, 0,
+                                              &copied, &added);
+                       diff_free_filespec_data(p->one);
+                       diff_free_filespec_data(p->two);
+               } else if (DIFF_FILE_VALID(p->one)) {
+                       diff_populate_filespec(p->one, 1);
+                       copied = added = 0;
+                       diff_free_filespec_data(p->one);
+               } else if (DIFF_FILE_VALID(p->two)) {
+                       diff_populate_filespec(p->two, 1);
+                       copied = 0;
+                       added = p->two->size;
+                       diff_free_filespec_data(p->two);
+               } else
                        continue;
-               changed += data->files[i]->added;
-               changed += data->files[i]->deleted;
+
+               /*
+                * Original minus copied is the removed material,
+                * added is the new material.  They are both damages
+                * 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;
+               dir.files[dir.nr].changed = damage;
+               changed += damage;
+               dir.nr++;
        }
 
        /* This can happen even with many files, if everything was renames */
@@ -1074,10 +1185,7 @@ static void show_dirstat(struct diffstat_t *data, struct diff_options *options)
                return;
 
        /* Show all directories with more than x% of the changes */
-       dir.files = data->files;
-       dir.nr = data->nr;
-       dir.percent = options->dirstat_percent;
-       dir.cumulative = options->output_format & DIFF_FORMAT_CUMULATIVE;
+       qsort(dir.files, dir.nr, sizeof(dir.files[0]), dirstat_compare);
        gather_dirstat(options->file, &dir, changed, "", 0);
 }
 
@@ -1096,42 +1204,86 @@ static void free_diffstat_info(struct diffstat_t *diffstat)
 }
 
 struct checkdiff_t {
-       struct xdiff_emit_state xm;
        const char *filename;
-       int lineno, color_diff;
+       int lineno;
+       struct diff_options *o;
        unsigned ws_rule;
        unsigned status;
-       FILE *file;
+       int trailing_blanks_start;
 };
 
+static int is_conflict_marker(const char *line, unsigned long len)
+{
+       char firstchar;
+       int cnt;
+
+       if (len < 8)
+               return 0;
+       firstchar = line[0];
+       switch (firstchar) {
+       case '=': case '>': case '<':
+               break;
+       default:
+               return 0;
+       }
+       for (cnt = 1; cnt < 7; 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 */
+               return 0;
+       }
+       return 1;
+}
+
 static void checkdiff_consume(void *priv, char *line, unsigned long len)
 {
        struct checkdiff_t *data = priv;
-       const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
-       const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
-       const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
+       int color_diff = DIFF_OPT_TST(data->o, COLOR_DIFF);
+       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;
 
        if (line[0] == '+') {
+               unsigned bad;
                data->lineno++;
-               data->status = check_and_emit_line(line + 1, len - 1,
-                   data->ws_rule, NULL, NULL, NULL, NULL);
-               if (!data->status)
+               if (!ws_blank_line(line + 1, len - 1, data->ws_rule))
+                       data->trailing_blanks_start = 0;
+               else if (!data->trailing_blanks_start)
+                       data->trailing_blanks_start = data->lineno;
+               if (is_conflict_marker(line + 1, len - 1)) {
+                       data->status |= 1;
+                       fprintf(data->o->file,
+                               "%s:%d: leftover conflict marker\n",
+                               data->filename, data->lineno);
+               }
+               bad = ws_check(line + 1, len - 1, data->ws_rule);
+               if (!bad)
                        return;
-               err = whitespace_error_string(data->status);
-               fprintf(data->file, "%s:%d: %s.\n", data->filename, data->lineno, err);
+               data->status |= bad;
+               err = whitespace_error_string(bad);
+               fprintf(data->o->file, "%s:%d: %s.\n",
+                       data->filename, data->lineno, err);
                free(err);
-               emit_line(data->file, set, reset, line, 1);
-               (void)check_and_emit_line(line + 1, len - 1, data->ws_rule,
-                   data->file, set, reset, ws);
-       } else if (line[0] == ' ')
+               emit_line(data->o->file, set, reset, line, 1);
+               ws_check_emit(line + 1, len - 1, data->ws_rule,
+                             data->o->file, set, reset, ws);
+       } else if (line[0] == ' ') {
                data->lineno++;
-       else if (line[0] == '@') {
+               data->trailing_blanks_start = 0;
+       } else if (line[0] == '@') {
                char *plus = strchr(line, '+');
                if (plus)
                        data->lineno = strtol(plus, NULL, 10) - 1;
                else
                        die("invalid diff");
+               data->trailing_blanks_start = 0;
        }
 }
 
@@ -1226,114 +1378,61 @@ static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two)
        emit_binary_diff_body(file, two, one);
 }
 
-static void setup_diff_attr_check(struct git_attr_check *check)
+static 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 char *funcname_pattern(const char *ident)
+static const char *userdiff_word_regex(struct diff_filespec *one)
 {
-       struct funcname_pattern *pp;
-
-       for (pp = funcname_pattern_list; pp; pp = pp->next)
-               if (!strcmp(ident, pp->name))
-                       return pp->pattern;
-       return NULL;
+       diff_filespec_load_driver(one);
+       return one->driver->word_regex;
 }
 
-static struct builtin_funcname_pattern {
-       const char *name;
-       const char *pattern;
-} builtin_funcname_pattern[] = {
-       { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
-                       "new\\|return\\|switch\\|throw\\|while\\)\n"
-                       "^[     ]*\\(\\([       ]*"
-                       "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
-                       "[      ]*([^;]*\\)$" },
-       { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
-};
-
-static const char *diff_funcname_pattern(struct diff_filespec *one)
+void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
 {
-       const char *ident, *pattern;
-       int i;
-
-       diff_filespec_check_attr(one);
-       ident = one->funcname_pattern_ident;
-
-       if (!ident)
-               /*
-                * If the config file has "funcname.default" defined, that
-                * regexp is used; otherwise NULL is returned and xemit uses
-                * the built-in default.
-                */
-               return funcname_pattern("default");
-
-       /* Look up custom "funcname.$ident" regexp from config. */
-       pattern = funcname_pattern(ident);
-       if (pattern)
-               return pattern;
-
-       /*
-        * And define built-in fallback patterns here.  Note that
-        * these can be overridden by the user's config settings.
-        */
-       for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
-               if (!strcmp(ident, builtin_funcname_pattern[i].name))
-                       return builtin_funcname_pattern[i].pattern;
+       if (!options->a_prefix)
+               options->a_prefix = a;
+       if (!options->b_prefix)
+               options->b_prefix = b;
+}
 
-       return NULL;
+static const char *get_textconv(struct diff_filespec *one)
+{
+       if (!DIFF_FILE_VALID(one))
+               return NULL;
+       if (!S_ISREG(one->mode))
+               return NULL;
+       diff_filespec_load_driver(one);
+       return one->driver->textconv;
 }
 
 static void builtin_diff(const char *name_a,
@@ -1349,9 +1448,29 @@ 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;
+       const char *textconv_one = NULL, *textconv_two = NULL;
 
-       a_one = quote_two(o->a_prefix, name_a + (*name_a == '/'));
-       b_two = quote_two(o->b_prefix, name_b + (*name_b == '/'));
+       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(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);
@@ -1379,8 +1498,11 @@ static void builtin_diff(const char *name_a,
                 */
                if ((one->mode ^ two->mode) & S_IFMT)
                        goto free_ab_and_return;
-               if (complete_rewrite) {
-                       emit_rewrite_diff(name_a, name_b, one, two, o);
+               if (complete_rewrite &&
+                   (textconv_one || !diff_filespec_is_binary(one)) &&
+                   (textconv_two || !diff_filespec_is_binary(two))) {
+                       emit_rewrite_diff(name_a, name_b, one, two,
+                                               textconv_one, textconv_two, o);
                        o->found_changes = 1;
                        goto free_ab_and_return;
                }
@@ -1390,7 +1512,8 @@ static void builtin_diff(const char *name_a,
                die("unable to read files to diff");
 
        if (!DIFF_OPT_TST(o, TEXT) &&
-           (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
+           ( (diff_filespec_is_binary(one) && !textconv_one) ||
+             (diff_filespec_is_binary(two) && !textconv_two) )) {
                /* Quite common confusing case */
                if (mf1.size == mf2.size &&
                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
@@ -1409,12 +1532,28 @@ static void builtin_diff(const char *name_a,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
                struct emit_callback ecbdata;
-               const char *funcname_pattern;
+               const struct userdiff_funcname *pe;
+
+               if (textconv_one) {
+                       size_t size;
+                       mf1.ptr = run_textconv(textconv_one, one, &size);
+                       if (!mf1.ptr)
+                               die("unable to read files to diff");
+                       mf1.size = size;
+               }
+               if (textconv_two) {
+                       size_t size;
+                       mf2.ptr = run_textconv(textconv_two, two, &size);
+                       if (!mf2.ptr)
+                               die("unable to read files to diff");
+                       mf2.size = size;
+               }
 
-               funcname_pattern = diff_funcname_pattern(one);
-               if (!funcname_pattern)
-                       funcname_pattern = diff_funcname_pattern(two);
+               pe = diff_funcname_pattern(one);
+               if (!pe)
+                       pe = diff_funcname_pattern(two);
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                memset(&ecbdata, 0, sizeof(ecbdata));
                ecbdata.label_path = lbl;
@@ -1424,26 +1563,44 @@ static void builtin_diff(const char *name_a,
                ecbdata.file = o->file;
                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
                xecfg.ctxlen = o->context;
+               xecfg.interhunkctxlen = o->interhunkcontext;
                xecfg.flags = XDL_EMIT_FUNCNAMES;
-               if (funcname_pattern)
-                       xdiff_set_find_func(&xecfg, funcname_pattern);
+               if (pe)
+                       xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
                if (!diffopts)
                        ;
                else if (!prefixcmp(diffopts, "--unified="))
                        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)) {
                        ecbdata.diff_words =
                                xcalloc(1, sizeof(struct diff_words_data));
                        ecbdata.diff_words->file = o->file;
+                       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);
+                       }
                }
-               xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+               xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
+                             &xpp, &xecfg, &ecb);
                if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
                        free_diff_words_data(&ecbdata);
+               if (textconv_one)
+                       free(mf1.ptr);
+               if (textconv_two)
+                       free(mf2.ptr);
        }
 
  free_ab_and_return:
@@ -1490,11 +1647,11 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
-               ecb.outf = xdiff_outf;
-               ecb.priv = diffstat;
-               xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+               xdi_diff_outf(&mf1, &mf2, diffstat_consume, diffstat,
+                             &xpp, &xecfg, &ecb);
        }
 
  free_and_return:
@@ -1504,8 +1661,9 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
 
 static void builtin_checkdiff(const char *name_a, const char *name_b,
                              const char *attr_path,
-                            struct diff_filespec *one,
-                            struct diff_filespec *two, struct diff_options *o)
+                             struct diff_filespec *one,
+                             struct diff_filespec *two,
+                             struct diff_options *o)
 {
        mmfile_t mf1, mf2;
        struct checkdiff_t data;
@@ -1514,16 +1672,20 @@ 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.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
+       data.o = o;
        data.ws_rule = whitespace_rule(attr_path);
-       data.file = o->file;
 
        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
                die("unable to read files to diff");
 
+       /*
+        * All the other codepaths check both sides, but not checking
+        * the "old" side here is deliberate.  We are checking the newly
+        * introduced changes, and as long as the "new" side is text, we
+        * can and should check what it introduces.
+        */
        if (diff_filespec_is_binary(two))
                goto free_and_return;
        else {
@@ -1532,11 +1694,19 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
                xdemitconf_t xecfg;
                xdemitcb_t ecb;
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
+               xecfg.ctxlen = 1; /* at least one context line */
                xpp.flags = XDF_NEED_MINIMAL;
-               ecb.outf = xdiff_outf;
-               ecb.priv = &data;
-               xdi_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+               xdi_diff_outf(&mf1, &mf2, checkdiff_consume, &data,
+                             &xpp, &xecfg, &ecb);
+
+               if ((data.ws_rule & WS_TRAILING_SPACE) &&
+                   data.trailing_blanks_start) {
+                       fprintf(o->file, "%s:%d: ends with blank lines.\n",
+                               data.filename, data.trailing_blanks_start);
+                       data.status = 1; /* report errors */
+               }
        }
  free_and_return:
        diff_free_filespec_data(one);
@@ -1554,6 +1724,7 @@ 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;
 }
 
@@ -1610,7 +1781,7 @@ 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);
@@ -1638,10 +1809,9 @@ 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));
@@ -1693,7 +1863,7 @@ 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;
 
@@ -1713,19 +1883,18 @@ 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;
@@ -1736,7 +1905,6 @@ 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);
@@ -1798,10 +1966,11 @@ static void prep_temp_blob(struct diff_tempfile *temp,
        sprintf(temp->mode, "%06o", mode);
 }
 
-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
@@ -1810,7 +1979,13 @@ 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 ||
@@ -1824,13 +1999,12 @@ static void prepare_temp_file(const char *name,
                if (S_ISLNK(st.st_mode)) {
                        int ret;
                        char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
-                       size_t sz = xsize_t(st.st_size);
-                       if (sizeof(buf) <= st.st_size)
-                               die("symlink too long: %s", name);
-                       ret = readlink(name, buf, sz);
+                       ret = readlink(name, buf, sizeof(buf));
                        if (ret < 0)
                                die("readlink(%s)", name);
-                       prep_temp_blob(temp, buf, sz,
+                       if (ret == sizeof(buf))
+                               die("symlink too long: %s", name);
+                       prep_temp_blob(temp, buf, ret,
                                       (one->sha1_valid ?
                                        one->sha1 : null_sha1),
                                       (one->sha1_valid ?
@@ -1851,7 +2025,7 @@ static void prepare_temp_file(const char *name,
                         */
                        sprintf(temp->mode, "%06o", one->mode);
                }
-               return;
+               return temp;
        }
        else {
                if (diff_populate_filespec(one, 0))
@@ -1859,24 +2033,7 @@ static void prepare_temp_file(const char *name,
                prep_temp_blob(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:
@@ -1894,34 +2051,22 @@ 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;
@@ -1940,27 +2085,66 @@ static void run_external_diff(const char *pgm,
        }
 }
 
-static const char *external_diff_attr(const char *name)
+static int similarity_index(struct diff_filepair *p)
 {
-       struct git_attr_check attr_diff_check;
+       return p->score * 100 / MAX_SCORE;
+}
 
-       if (!name)
-               return NULL;
+static void fill_metainfo(struct strbuf *msg,
+                         const char *name,
+                         const char *other,
+                         struct diff_filespec *one,
+                         struct diff_filespec *two,
+                         struct diff_options *o,
+                         struct diff_filepair *p)
+{
+       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 ");
+               quote_c_style(name, msg, NULL, 0);
+               strbuf_addstr(msg, "\ncopy to ");
+               quote_c_style(other, msg, NULL, 0);
+               strbuf_addch(msg, '\n');
+               break;
+       case DIFF_STATUS_RENAMED:
+               strbuf_addf(msg, "similarity index %d%%", similarity_index(p));
+               strbuf_addstr(msg, "\nrename from ");
+               quote_c_style(name, msg, NULL, 0);
+               strbuf_addstr(msg, "\nrename to ");
+               quote_c_style(other, msg, NULL, 0);
+               strbuf_addch(msg, '\n');
+               break;
+       case DIFF_STATUS_MODIFIED:
+               if (p->score) {
+                       strbuf_addf(msg, "dissimilarity index %d%%\n",
+                                   similarity_index(p));
+                       break;
+               }
+               /* fallthru */
+       default:
+               /* nothing */
+               ;
+       }
+       if (one && two && hashcmp(one->sha1, two->sha1)) {
+               int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
 
-       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;
+               if (DIFF_OPT_TST(o, BINARY)) {
+                       mmfile_t mf;
+                       if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
+                           (!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));
+               if (one->mode == two->mode)
+                       strbuf_addf(msg, " %06o", one->mode);
+               strbuf_addch(msg, '\n');
        }
-       return NULL;
+       if (msg->len)
+               strbuf_setlen(msg, msg->len - 1);
 }
 
 static void run_diff_cmd(const char *pgm,
@@ -1969,16 +2153,24 @@ static void run_diff_cmd(const char *pgm,
                         const char *attr_path,
                         struct diff_filespec *one,
                         struct diff_filespec *two,
-                        const char *xfrm_msg,
+                        struct strbuf *msg,
                         struct diff_options *o,
-                        int complete_rewrite)
+                        struct diff_filepair *p)
 {
+       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;
+       }
+
        if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
                pgm = NULL;
        else {
-               const char *cmd = external_diff_attr(attr_path);
-               if (cmd)
-                       pgm = cmd;
+               struct userdiff_driver *drv = userdiff_find_by_path(attr_path);
+               if (drv && drv->external)
+                       pgm = drv->external;
        }
 
        if (pgm) {
@@ -2005,18 +2197,13 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
                        if (lstat(one->path, &st) < 0)
                                die("stat %s", one->path);
                        if (index_path(one->sha1, one->path, &st, 0))
-                               die("cannot hash %s\n", one->path);
+                               die("cannot hash %s", one->path);
                }
        }
        else
                hashclr(one->sha1);
 }
 
-static int similarity_index(struct diff_filepair *p)
-{
-       return p->score * 100 / MAX_SCORE;
-}
-
 static void strip_prefix(int prefix_length, const char **namep, const char **otherp)
 {
        /* Strip the prefix but do not molest /dev/null and absolute paths */
@@ -2030,13 +2217,11 @@ static void run_diff(struct diff_filepair *p, struct diff_options *o)
 {
        const char *pgm = external_diff();
        struct strbuf msg;
-       char *xfrm_msg;
        struct diff_filespec *one = p->one;
        struct diff_filespec *two = p->two;
        const char *name;
        const char *other;
        const char *attr_path;
-       int complete_rewrite = 0;
 
        name  = p->one->path;
        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
@@ -2046,83 +2231,34 @@ static void run_diff(struct diff_filepair *p, struct diff_options *o)
 
        if (DIFF_PAIR_UNMERGED(p)) {
                run_diff_cmd(pgm, name, NULL, attr_path,
-                            NULL, NULL, NULL, o, 0);
+                            NULL, NULL, NULL, o, p);
                return;
        }
 
        diff_fill_sha1_info(one);
        diff_fill_sha1_info(two);
 
-       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 ");
-               quote_c_style(name, &msg, NULL, 0);
-               strbuf_addstr(&msg, "\ncopy to ");
-               quote_c_style(other, &msg, NULL, 0);
-               strbuf_addch(&msg, '\n');
-               break;
-       case DIFF_STATUS_RENAMED:
-               strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
-               strbuf_addstr(&msg, "\nrename from ");
-               quote_c_style(name, &msg, NULL, 0);
-               strbuf_addstr(&msg, "\nrename to ");
-               quote_c_style(other, &msg, NULL, 0);
-               strbuf_addch(&msg, '\n');
-               break;
-       case DIFF_STATUS_MODIFIED:
-               if (p->score) {
-                       strbuf_addf(&msg, "dissimilarity index %d%%\n",
-                                       similarity_index(p));
-                       complete_rewrite = 1;
-                       break;
-               }
-               /* fallthru */
-       default:
-               /* nothing */
-               ;
-       }
-
-       if (hashcmp(one->sha1, two->sha1)) {
-               int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
-
-               if (DIFF_OPT_TST(o, BINARY)) {
-                       mmfile_t mf;
-                       if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
-                           (!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));
-               if (one->mode == two->mode)
-                       strbuf_addf(&msg, " %06o", one->mode);
-               strbuf_addch(&msg, '\n');
-       }
-
-       if (msg.len)
-               strbuf_setlen(&msg, msg.len - 1);
-       xfrm_msg = msg.len ? msg.buf : NULL;
-
        if (!pgm &&
            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
-               /* a filepair that changes between file and symlink
+               /*
+                * a filepair that changes between file and symlink
                 * needs to be split into deletion and creation.
                 */
                struct diff_filespec *null = alloc_filespec(two->path);
                run_diff_cmd(NULL, name, other, attr_path,
-                            one, null, xfrm_msg, o, 0);
+                            one, null, &msg, o, p);
                free(null);
+               strbuf_release(&msg);
+
                null = alloc_filespec(one->path);
                run_diff_cmd(NULL, name, other, attr_path,
-                            null, two, xfrm_msg, o, 0);
+                            null, two, &msg, o, p);
                free(null);
        }
        else
                run_diff_cmd(pgm, name, other, attr_path,
-                            one, two, xfrm_msg, o, complete_rewrite);
+                            one, two, &msg, o, p);
 
        strbuf_release(&msg);
 }
@@ -2189,18 +2325,17 @@ void diff_setup(struct diff_options *options)
        options->rename_limit = -1;
        options->dirstat_percent = 3;
        options->context = 3;
-       options->msg_sep = "";
 
        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_mnemonic_prefix) {
+               options->a_prefix = "a/";
+               options->b_prefix = "b/";
+       }
 }
 
 int diff_setup_done(struct diff_options *options)
@@ -2283,13 +2418,6 @@ int diff_setup_done(struct diff_options *options)
                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
        }
 
-       /*
-        * If we postprocess in diffcore, we cannot simply return
-        * upon the first hit.  We need to run diff as usual.
-        */
-       if (options->pickaxe || options->filter)
-               DIFF_OPT_CLR(options, QUIET);
-
        return 0;
 }
 
@@ -2361,8 +2489,14 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->output_format |= DIFF_FORMAT_SHORTSTAT;
        else if (opt_arg(arg, 'X', "dirstat", &options->dirstat_percent))
                options->output_format |= DIFF_FORMAT_DIRSTAT;
-       else if (!strcmp(arg, "--cumulative"))
-               options->output_format |= DIFF_FORMAT_CUMULATIVE;
+       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;
        else if (!strcmp(arg, "--summary"))
@@ -2431,11 +2565,13 @@ 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")) {
@@ -2456,8 +2592,15 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                DIFF_OPT_SET(options, COLOR_DIFF);
        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);
+               DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
+       }
+       else if (!prefixcmp(arg, "--color-words=")) {
+               DIFF_OPT_SET(options, COLOR_DIFF);
+               DIFF_OPT_SET(options, COLOR_DIFF_WORDS);
+               options->word_regex = arg + 14;
+       }
        else if (!strcmp(arg, "--exit-code"))
                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
        else if (!strcmp(arg, "--quiet"))
@@ -2466,6 +2609,12 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
        else if (!strcmp(arg, "--no-ext-diff"))
                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
+       else if (!strcmp(arg, "--textconv"))
+               DIFF_OPT_SET(options, ALLOW_TEXTCONV);
+       else if (!strcmp(arg, "--no-textconv"))
+               DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
+       else if (!strcmp(arg, "--ignore-submodules"))
+               DIFF_OPT_SET(options, IGNORE_SUBMODULES);
 
        /* misc options */
        else if (!strcmp(arg, "-z"))
@@ -2497,6 +2646,9 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->b_prefix = arg + 13;
        else if (!strcmp(arg, "--no-prefix"))
                options->a_prefix = options->b_prefix = "";
+       else if (opt_arg(arg, '\0', "inter-hunk-context",
+                        &options->interhunkcontext))
+               ;
        else if (!prefixcmp(arg, "--output=")) {
                options->file = fopen(arg + strlen("--output="), "w");
                options->close_file = 1;
@@ -2916,8 +3068,7 @@ static void diff_summary(FILE *file, struct diff_filepair *p)
 }
 
 struct patch_id_t {
-       struct xdiff_emit_state xm;
-       SHA_CTX *ctx;
+       git_SHA_CTX *ctx;
        int patchlen;
 };
 
@@ -2945,7 +3096,7 @@ 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;
 }
 
@@ -2954,14 +3105,13 @@ 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;
@@ -2971,6 +3121,7 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                struct diff_filepair *p = q->queue[i];
                int len1, len2;
 
+               memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                if (p->status == 0)
                        return error("internal diff status error");
@@ -3021,17 +3172,16 @@ 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;
                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, &ecb);
        }
 
-       SHA1_Final(sha1, &ctx);
+       git_SHA1_Final(sha1, &ctx);
        return 0;
 }
 
@@ -3101,18 +3251,15 @@ void diff_flush(struct diff_options *options)
                separator++;
        }
 
-       if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT|DIFF_FORMAT_DIRSTAT)) {
+       if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
                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))
                                diff_flush_stat(p, options, &diffstat);
                }
-               if (output_format & DIFF_FORMAT_DIRSTAT)
-                       show_dirstat(&diffstat, options);
                if (output_format & DIFF_FORMAT_NUMSTAT)
                        show_numstat(&diffstat, options);
                if (output_format & DIFF_FORMAT_DIFFSTAT)
@@ -3122,6 +3269,8 @@ void diff_flush(struct diff_options *options)
                free_diffstat_info(&diffstat);
                separator++;
        }
+       if (output_format & DIFF_FORMAT_DIRSTAT)
+               show_dirstat(options);
 
        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
                for (i = 0; i < q->nr; i++)
@@ -3131,11 +3280,10 @@ void diff_flush(struct diff_options *options)
 
        if (output_format & DIFF_FORMAT_PATCH) {
                if (separator) {
+                       putc(options->line_termination, options->file);
                        if (options->stat_sep) {
                                /* attach patch instead of inline */
                                fputs(options->stat_sep, options->file);
-                       } else {
-                               putc(options->line_termination, options->file);
                        }
                }
 
@@ -3278,10 +3426,7 @@ static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
 
 void diffcore_std(struct diff_options *options)
 {
-       if (DIFF_OPT_TST(options, QUIET))
-               return;
-
-       if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
+       if (options->skip_stat_unmatch)
                diffcore_skip_stat_unmatch(options);
        if (options->break_opt != -1)
                diffcore_break(options->break_opt);
@@ -3320,11 +3465,13 @@ 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 *base, const char *path)
+                   const char *concatpath)
 {
-       char concatpath[PATH_MAX];
        struct diff_filespec *one, *two;
 
+       if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
+               return;
+
        /* This may look odd, but it is a preparation for
         * feeding "there are unchanged files which should
         * not produce diffs, but when you are doing copy
@@ -3341,9 +3488,6 @@ void diff_addremove(struct diff_options *options,
                addremove = (addremove == '+' ? '-' :
                             addremove == '-' ? '+' : addremove);
 
-       if (!path) path = "";
-       sprintf(concatpath, "%s%s", base, path);
-
        if (options->prefix &&
            strncmp(concatpath, options->prefix, options->prefix_length))
                return;
@@ -3364,19 +3508,20 @@ void diff_change(struct diff_options *options,
                 unsigned old_mode, unsigned new_mode,
                 const unsigned char *old_sha1,
                 const unsigned char *new_sha1,
-                const char *base, const char *path)
+                const char *concatpath)
 {
-       char concatpath[PATH_MAX];
        struct diff_filespec *one, *two;
 
+       if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
+                       && S_ISGITLINK(new_mode))
+               return;
+
        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
                unsigned tmp;
                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;
        }
-       if (!path) path = "";
-       sprintf(concatpath, "%s%s", base, path);
 
        if (options->prefix &&
            strncmp(concatpath, options->prefix, options->prefix_length))
@@ -3406,3 +3551,32 @@ 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;
+
+       temp = prepare_temp_file(spec->path, spec);
+       *arg++ = pgm;
+       *arg++ = temp->name;
+       *arg = NULL;
+
+       memset(&child, 0, sizeof(child));
+       child.argv = argv;
+       child.out = -1;
+       if (start_command(&child) != 0 ||
+           strbuf_read(&buf, child.out, 0) < 0 ||
+           finish_command(&child) != 0) {
+               remove_tempfile();
+               error("error running textconv command '%s'", pgm);
+               return NULL;
+       }
+       remove_tempfile();
+
+       return strbuf_detach(&buf, outsize);
+}