Use memmove instead of memcpy for overlapping areas
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 607c357f5ad8ff07c69dec7c185d667db4e43397..fb8243261cb3cc9165dbe990586d3865fad4ee61 100644 (file)
--- a/diff.c
+++ b/diff.c
 #include "diffcore.h"
 #include "delta.h"
 #include "xdiff-interface.h"
+#include "color.h"
 
 static int use_size_cache;
 
-static int diff_detect_rename_default = 0;
+static int diff_detect_rename_default;
 static int diff_rename_limit_default = -1;
-static int diff_use_color_default = 0;
+static int diff_use_color_default;
 
-/* "\033[1;38;5;2xx;48;5;2xxm\0" is 23 bytes */
-static char diff_colors[][24] = {
+static char diff_colors[][COLOR_MAXLEN] = {
        "\033[m",       /* reset */
-       "",             /* normal */
-       "\033[1m",      /* bold */
-       "\033[36m",     /* cyan */
-       "\033[31m",     /* red */
-       "\033[32m",     /* green */
-       "\033[33m"      /* yellow */
+       "",             /* 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) */
 };
 
 static int parse_diff_color_slot(const char *var, int ofs)
@@ -42,122 +43,11 @@ static int parse_diff_color_slot(const char *var, int ofs)
                return DIFF_FILE_NEW;
        if (!strcasecmp(var+ofs, "commit"))
                return DIFF_COMMIT;
+       if (!strcasecmp(var+ofs, "whitespace"))
+               return DIFF_WHITESPACE;
        die("bad config variable '%s'", var);
 }
 
-static int parse_color(const char *name, int len)
-{
-       static const char * const color_names[] = {
-               "normal", "black", "red", "green", "yellow",
-               "blue", "magenta", "cyan", "white"
-       };
-       char *end;
-       int i;
-       for (i = 0; i < ARRAY_SIZE(color_names); i++) {
-               const char *str = color_names[i];
-               if (!strncasecmp(name, str, len) && !str[len])
-                       return i - 1;
-       }
-       i = strtol(name, &end, 10);
-       if (*name && !*end && i >= -1 && i <= 255)
-               return i;
-       return -2;
-}
-
-static int parse_attr(const char *name, int len)
-{
-       static const int attr_values[] = { 1, 2, 4, 5, 7 };
-       static const char * const attr_names[] = {
-               "bold", "dim", "ul", "blink", "reverse"
-       };
-       int i;
-       for (i = 0; i < ARRAY_SIZE(attr_names); i++) {
-               const char *str = attr_names[i];
-               if (!strncasecmp(name, str, len) && !str[len])
-                       return attr_values[i];
-       }
-       return -1;
-}
-
-static void parse_diff_color_value(const char *value, const char *var, char *dst)
-{
-       const char *ptr = value;
-       int attr = -1;
-       int fg = -2;
-       int bg = -2;
-
-       if (!strcasecmp(value, "reset")) {
-               strcpy(dst, "\033[m");
-               return;
-       }
-
-       /* [fg [bg]] [attr] */
-       while (*ptr) {
-               const char *word = ptr;
-               int val, len = 0;
-
-               while (word[len] && !isspace(word[len]))
-                       len++;
-
-               ptr = word + len;
-               while (*ptr && isspace(*ptr))
-                       ptr++;
-
-               val = parse_color(word, len);
-               if (val >= -1) {
-                       if (fg == -2) {
-                               fg = val;
-                               continue;
-                       }
-                       if (bg == -2) {
-                               bg = val;
-                               continue;
-                       }
-                       goto bad;
-               }
-               val = parse_attr(word, len);
-               if (val < 0 || attr != -1)
-                       goto bad;
-               attr = val;
-       }
-
-       if (attr >= 0 || fg >= 0 || bg >= 0) {
-               int sep = 0;
-
-               *dst++ = '\033';
-               *dst++ = '[';
-               if (attr >= 0) {
-                       *dst++ = '0' + attr;
-                       sep++;
-               }
-               if (fg >= 0) {
-                       if (sep++)
-                               *dst++ = ';';
-                       if (fg < 8) {
-                               *dst++ = '3';
-                               *dst++ = '0' + fg;
-                       } else {
-                               dst += sprintf(dst, "38;5;%d", fg);
-                       }
-               }
-               if (bg >= 0) {
-                       if (sep++)
-                               *dst++ = ';';
-                       if (bg < 8) {
-                               *dst++ = '4';
-                               *dst++ = '0' + bg;
-                       } else {
-                               dst += sprintf(dst, "48;5;%d", bg);
-                       }
-               }
-               *dst++ = 'm';
-       }
-       *dst = 0;
-       return;
-bad:
-       die("bad config value '%s' for variable '%s'", value, var);
-}
-
 /*
  * These are to give UI layer defaults.
  * The core-level commands such as git-diff-files should
@@ -171,22 +61,7 @@ int git_diff_ui_config(const char *var, const char *value)
                return 0;
        }
        if (!strcmp(var, "diff.color")) {
-               if (!value)
-                       diff_use_color_default = 1; /* bool */
-               else if (!strcasecmp(value, "auto")) {
-                       diff_use_color_default = 0;
-                       if (isatty(1) || (pager_in_use && pager_use_color)) {
-                               char *term = getenv("TERM");
-                               if (term && strcmp(term, "dumb"))
-                                       diff_use_color_default = 1;
-                       }
-               }
-               else if (!strcasecmp(value, "never"))
-                       diff_use_color_default = 0;
-               else if (!strcasecmp(value, "always"))
-                       diff_use_color_default = 1;
-               else
-                       diff_use_color_default = git_config_bool(var, value);
+               diff_use_color_default = git_config_colorbool(var, value);
                return 0;
        }
        if (!strcmp(var, "diff.renames")) {
@@ -201,7 +76,7 @@ int git_diff_ui_config(const char *var, const char *value)
        }
        if (!strncmp(var, "diff.color.", 11)) {
                int slot = parse_diff_color_slot(var, 11);
-               parse_diff_color_value(value, var, diff_colors[slot]);
+               color_parse(value, var, diff_colors[slot]);
                return 0;
        }
        return git_default_config(var, value);
@@ -216,7 +91,7 @@ static char *quote_one(const char *str)
                return NULL;
        needlen = quote_c_style(str, NULL, NULL, 0);
        if (!needlen)
-               return strdup(str);
+               return xstrdup(str);
        xp = xmalloc(needlen + 1);
        quote_c_style(str, xp, NULL, 0);
        return xp;
@@ -333,7 +208,7 @@ static void emit_rewrite_diff(const char *name_a,
        diff_populate_filespec(two, 0);
        lc_a = count_lines(one->data, one->size);
        lc_b = count_lines(two->data, two->size);
-       printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
+       printf("--- a/%s\n+++ b/%s\n@@ -", name_a, name_b);
        print_line_count(lc_a);
        printf(" +");
        print_line_count(lc_b);
@@ -358,12 +233,152 @@ static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
        return 0;
 }
 
+struct diff_words_buffer {
+       mmfile_t text;
+       long alloc;
+       long current; /* output pointer */
+       int suppressed_newline;
+};
+
+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);
+       }
+       line++;
+       len--;
+       memcpy(buffer->text.ptr + buffer->text.size, line, len);
+       buffer->text.size += len;
+}
+
+struct diff_words_data {
+       struct xdiff_emit_state xm;
+       struct diff_words_buffer minus, plus;
+};
+
+static void print_word(struct diff_words_buffer *buffer, int len, int color,
+               int suppress_newline)
+{
+       const char *ptr;
+       int eol = 0;
+
+       if (len == 0)
+               return;
+
+       ptr  = buffer->text.ptr + buffer->current;
+       buffer->current += len;
+
+       if (ptr[len - 1] == '\n') {
+               eol = 1;
+               len--;
+       }
+
+       fputs(diff_get_color(1, color), stdout);
+       fwrite(ptr, len, 1, stdout);
+       fputs(diff_get_color(1, DIFF_RESET), stdout);
+
+       if (eol) {
+               if (suppress_newline)
+                       buffer->suppressed_newline = 1;
+               else
+                       putchar('\n');
+       }
+}
+
+static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
+{
+       struct diff_words_data *diff_words = priv;
+
+       if (diff_words->minus.suppressed_newline) {
+               if (line[0] != '+')
+                       putchar('\n');
+               diff_words->minus.suppressed_newline = 0;
+       }
+
+       len--;
+       switch (line[0]) {
+               case '-':
+                       print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
+                       break;
+               case '+':
+                       print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
+                       break;
+               case ' ':
+                       print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
+                       diff_words->minus.current += len;
+                       break;
+       }
+}
+
+/* this executes the word diff on the accumulated buffers */
+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;
+
+       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;
+       xecfg.flags = 0;
+       ecb.outf = xdiff_outf;
+       ecb.priv = diff_words;
+       diff_words->xm.consume = fn_out_diff_words_aux;
+       xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
+
+       free(minus.ptr);
+       free(plus.ptr);
+       diff_words->minus.text.size = diff_words->plus.text.size = 0;
+
+       if (diff_words->minus.suppressed_newline) {
+               putchar('\n');
+               diff_words->minus.suppressed_newline = 0;
+       }
+}
+
 struct emit_callback {
        struct xdiff_emit_state xm;
        int nparents, color_diff;
        const char **label_path;
+       struct diff_words_data *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);
+
+               if (ecbdata->diff_words->minus.text.ptr)
+                       free (ecbdata->diff_words->minus.text.ptr);
+               if (ecbdata->diff_words->plus.text.ptr)
+                       free (ecbdata->diff_words->plus.text.ptr);
+               free(ecbdata->diff_words);
+               ecbdata->diff_words = NULL;
+       }
+}
+
 const char *diff_get_color(int diff_use_color, enum color_diff ix)
 {
        if (diff_use_color)
@@ -371,9 +386,89 @@ const char *diff_get_color(int diff_use_color, enum color_diff ix)
        return "";
 }
 
+static void emit_line(const char *set, const char *reset, const char *line, int len)
+{
+       if (len > 0 && line[len-1] == '\n')
+               len--;
+       fputs(set, stdout);
+       fwrite(line, len, 1, stdout);
+       puts(reset);
+}
+
+static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
+{
+       int col0 = ecbdata->nparents;
+       int last_tab_in_indent = -1;
+       int last_space_in_indent = -1;
+       int i;
+       int tail = len;
+       int need_highlight_leading_space = 0;
+       const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
+       const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
+
+       if (!*ws) {
+               emit_line(set, reset, line, len);
+               return;
+       }
+
+       /* The line is a newly added line.  Does it have funny leading
+        * whitespaces?  In indent, SP should never precede a TAB.
+        */
+       for (i = col0; i < len; i++) {
+               if (line[i] == '\t') {
+                       last_tab_in_indent = i;
+                       if (0 <= last_space_in_indent)
+                               need_highlight_leading_space = 1;
+               }
+               else if (line[i] == ' ')
+                       last_space_in_indent = i;
+               else
+                       break;
+       }
+       fputs(set, stdout);
+       fwrite(line, col0, 1, stdout);
+       fputs(reset, stdout);
+       if (((i == len) || line[i] == '\n') && i != col0) {
+               /* The whole line was indent */
+               emit_line(ws, reset, line + col0, len - col0);
+               return;
+       }
+       i = col0;
+       if (need_highlight_leading_space) {
+               while (i < last_tab_in_indent) {
+                       if (line[i] == ' ') {
+                               fputs(ws, stdout);
+                               putchar(' ');
+                               fputs(reset, stdout);
+                       }
+                       else
+                               putchar(line[i]);
+                       i++;
+               }
+       }
+       tail = len - 1;
+       if (line[tail] == '\n' && i < tail)
+               tail--;
+       while (i < tail) {
+               if (!isspace(line[tail]))
+                       break;
+               tail--;
+       }
+       if ((i < tail && line[tail + 1] != '\n')) {
+               /* This has whitespace between tail+1..len */
+               fputs(set, stdout);
+               fwrite(line + i, tail - i + 1, 1, stdout);
+               fputs(reset, stdout);
+               emit_line(ws, reset, line + tail + 1, len - tail - 1);
+       }
+       else
+               emit_line(set, reset, line + i, len - i);
+}
+
 static void fn_out_consume(void *priv, char *line, unsigned long len)
 {
        int i;
+       int color;
        struct emit_callback *ecbdata = priv;
        const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
@@ -391,26 +486,52 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
                ;
        if (2 <= i && i < len && line[i] == ' ') {
                ecbdata->nparents = i - 1;
-               set = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
+               emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
+                         reset, line, len);
+               return;
        }
-       else if (len < ecbdata->nparents)
+
+       if (len < ecbdata->nparents) {
                set = reset;
-       else {
-               int nparents = ecbdata->nparents;
-               int color = DIFF_PLAIN;
-               for (i = 0; i < nparents && len; i++) {
-                       if (line[i] == '-')
-                               color = DIFF_FILE_OLD;
-                       else if (line[i] == '+')
-                               color = DIFF_FILE_NEW;
-               }
-               set = diff_get_color(ecbdata->color_diff, color);
+               emit_line(reset, reset, line, len);
+               return;
        }
-       if (len > 0 && line[len-1] == '\n')
+
+       color = DIFF_PLAIN;
+       if (ecbdata->diff_words && ecbdata->nparents != 1)
+               /* fall back to normal diff */
+               free_diff_words_data(ecbdata);
+       if (ecbdata->diff_words) {
+               if (line[0] == '-') {
+                       diff_words_append(line, len,
+                                         &ecbdata->diff_words->minus);
+                       return;
+               } else if (line[0] == '+') {
+                       diff_words_append(line, 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--;
-       fputs (set, stdout);
-       fwrite (line, len, 1, stdout);
-       puts (reset);
+               emit_line(set, reset, line, len);
+               return;
+       }
+       for (i = 0; i < ecbdata->nparents && len; i++) {
+               if (line[i] == '-')
+                       color = DIFF_FILE_OLD;
+               else if (line[i] == '+')
+                       color = DIFF_FILE_NEW;
+       }
+
+       if (color != DIFF_FILE_NEW) {
+               emit_line(diff_get_color(ecbdata->color_diff, color),
+                         reset, line, len);
+               return;
+       }
+       emit_add_line(reset, ecbdata, line, len);
 }
 
 static char *pprint_rename(const char *a, const char *b)
@@ -499,7 +620,7 @@ static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
                x->is_renamed = 1;
        }
        else
-               x->name = strdup(name_a);
+               x->name = xstrdup(name_a);
        return x;
 }
 
@@ -514,21 +635,76 @@ static void diffstat_consume(void *priv, char *line, unsigned long len)
                x->deleted++;
 }
 
-static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
-static const char minuses[]= "----------------------------------------------------------------------";
 const char mime_boundary_leader[] = "------------";
 
-static void show_stats(struct diffstat_t* data)
+static int scale_linear(int it, int width, int max_change)
+{
+       /*
+        * make sure that at least one '-' is printed if there were deletions,
+        * and likewise for '+'.
+        */
+       if (max_change < 2)
+               return it;
+       return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
+}
+
+static void show_name(const char *prefix, const char *name, int len,
+                     const char *reset, const char *set)
+{
+       printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
+}
+
+static void show_graph(char ch, int cnt, const char *set, const char *reset)
+{
+       if (cnt <= 0)
+               return;
+       printf("%s", set);
+       while (cnt--)
+               putchar(ch);
+       printf("%s", reset);
+}
+
+static void show_stats(struct diffstat_t* data, struct diff_options *options)
 {
        int i, len, add, del, total, adds = 0, dels = 0;
-       int max, max_change = 0, max_len = 0;
+       int max_change = 0, max_len = 0;
        int total_files = data->nr;
+       int width, name_width;
+       const char *reset, *set, *add_c, *del_c;
 
        if (data->nr == 0)
                return;
 
+       width = options->stat_width ? options->stat_width : 80;
+       name_width = options->stat_name_width ? options->stat_name_width : 50;
+
+       /* 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;
+       }
+
+       /* Find the longest filename and max number of changes */
+       reset = diff_get_color(options->color_diff, DIFF_RESET);
+       set = diff_get_color(options->color_diff, DIFF_PLAIN);
+       add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
+       del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
+
        for (i = 0; i < data->nr; i++) {
                struct diffstat_file *file = data->files[i];
+               int change = file->added + file->deleted;
+
+               len = quote_c_style(file->name, NULL, NULL, 0);
+               if (len) {
+                       char *qname = xmalloc(len + 1);
+                       quote_c_style(file->name, qname, NULL, 0);
+                       free(file->name);
+                       file->name = qname;
+               }
 
                len = strlen(file->name);
                if (max_len < len)
@@ -536,54 +712,53 @@ static void show_stats(struct diffstat_t* data)
 
                if (file->is_binary || file->is_unmerged)
                        continue;
-               if (max_change < file->added + file->deleted)
-                       max_change = file->added + file->deleted;
+               if (max_change < change)
+                       max_change = change;
        }
 
+       /* Compute the width of the graph part;
+        * 10 is for one blank at the beginning of the line plus
+        * " | count " between the name and the graph.
+        *
+        * From here on, name_width is the width of the name area,
+        * and width is the width of the graph area.
+        */
+       name_width = (name_width < max_len) ? name_width : max_len;
+       if (width < (name_width + 10) + max_change)
+               width = width - (name_width + 10);
+       else
+               width = max_change;
+
        for (i = 0; i < data->nr; i++) {
                const char *prefix = "";
                char *name = data->files[i]->name;
                int added = data->files[i]->added;
                int deleted = data->files[i]->deleted;
-
-               if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
-                       char *qname = xmalloc(len + 1);
-                       quote_c_style(name, qname, NULL, 0);
-                       free(name);
-                       data->files[i]->name = name = qname;
-               }
+               int name_len;
 
                /*
                 * "scale" the filename
                 */
-               len = strlen(name);
-               max = max_len;
-               if (max > 50)
-                       max = 50;
-               if (len > max) {
+               len = name_width;
+               name_len = strlen(name);
+               if (name_width < name_len) {
                        char *slash;
                        prefix = "...";
-                       max -= 3;
-                       name += len - max;
+                       len -= 3;
+                       name += name_len - len;
                        slash = strchr(name, '/');
                        if (slash)
                                name = slash;
                }
-               len = max;
-
-               /*
-                * scale the add/delete
-                */
-               max = max_change;
-               if (max + len > 70)
-                       max = 70 - len;
 
                if (data->files[i]->is_binary) {
-                       printf(" %s%-*s |  Bin\n", prefix, len, name);
+                       show_name(prefix, name, len, reset, set);
+                       printf("  Bin\n");
                        goto free_diffstat_file;
                }
                else if (data->files[i]->is_unmerged) {
-                       printf(" %s%-*s |  Unmerged\n", prefix, len, name);
+                       show_name(prefix, name, len, reset, set);
+                       printf("  Unmerged\n");
                        goto free_diffstat_file;
                }
                else if (!data->files[i]->is_renamed &&
@@ -592,27 +767,32 @@ static void show_stats(struct diffstat_t* data)
                        goto free_diffstat_file;
                }
 
+               /*
+                * scale the add/delete
+                */
                add = added;
                del = deleted;
                total = add + del;
                adds += add;
                dels += del;
 
-               if (max_change > 0) {
-                       total = (total * max + max_change / 2) / max_change;
-                       add = (add * max + max_change / 2) / max_change;
-                       del = total - add;
+               if (width <= max_change) {
+                       add = scale_linear(add, width, max_change);
+                       del = scale_linear(del, width, max_change);
+                       total = add + del;
                }
-               printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
-                               len, name, added + deleted,
-                               add, pluses, del, minuses);
+               show_name(prefix, name, len, reset, set);
+               printf("%5d ", added + deleted);
+               show_graph('+', add, add_c, reset);
+               show_graph('-', del, del_c, reset);
+               putchar('\n');
        free_diffstat_file:
                free(data->files[i]->name);
                free(data->files[i]);
        }
        free(data->files);
-       printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
-                       total_files, adds, dels);
+       printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
+              set, total_files, adds, dels, reset);
 }
 
 struct checkdiff_t {
@@ -679,7 +859,7 @@ static unsigned char *deflate_it(char *data,
        return deflated;
 }
 
-static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
+static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
 {
        void *cp;
        void *delta;
@@ -690,7 +870,6 @@ static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
        unsigned long deflate_size;
        unsigned long data_size;
 
-       printf("GIT binary patch\n");
        /* We could do deflated delta, or we could do just deflated two,
         * whichever is smaller.
         */
@@ -739,15 +918,20 @@ static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
        free(data);
 }
 
+static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
+{
+       printf("GIT binary patch\n");
+       emit_binary_diff_body(one, two);
+       emit_binary_diff_body(two, one);
+}
+
 #define FIRST_FEW_BYTES 8000
 static int mmfile_is_binary(mmfile_t *mf)
 {
        long sz = mf->size;
        if (FIRST_FEW_BYTES < sz)
                sz = FIRST_FEW_BYTES;
-       if (memchr(mf->ptr, 0, sz))
-               return 1;
-       return 0;
+       return !!memchr(mf->ptr, 0, sz);
 }
 
 static void builtin_diff(const char *name_a,
@@ -836,7 +1020,12 @@ static void builtin_diff(const char *name_a,
                ecb.outf = xdiff_outf;
                ecb.priv = &ecbdata;
                ecbdata.xm.consume = fn_out_consume;
+               if (o->color_diff_words)
+                       ecbdata.diff_words =
+                               xcalloc(1, sizeof(struct diff_words_data));
                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+               if (o->color_diff_words)
+                       free_diff_words_data(&ecbdata);
        }
 
  free_ab_and_return:
@@ -939,8 +1128,8 @@ void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
 {
        if (mode) {
                spec->mode = canon_mode(mode);
-               memcpy(spec->sha1, sha1, 20);
-               spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
+               hashcpy(spec->sha1, sha1);
+               spec->sha1_valid = !is_null_sha1(sha1);
        }
 }
 
@@ -978,7 +1167,7 @@ static int work_tree_matches(const char *name, const unsigned char *sha1)
        if ((lstat(name, &st) < 0) ||
            !S_ISREG(st.st_mode) || /* careful! */
            ce_match_stat(ce, &st, 0) ||
-           memcmp(sha1, ce->sha1, 20))
+           hashcmp(sha1, ce->sha1))
                return 0;
        /* we return 1 only when we can stat, it is a regular file,
         * stat information matches, and sha1 recorded in the cache
@@ -1006,7 +1195,7 @@ static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
        while (last > first) {
                int cmp, next = (last + first) >> 1;
                e = sha1_size_cache[next];
-               cmp = memcmp(e->sha1, sha1, 20);
+               cmp = hashcmp(e->sha1, sha1);
                if (!cmp)
                        return e;
                if (cmp < 0) {
@@ -1032,7 +1221,7 @@ static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
                        sizeof(*sha1_size_cache));
        e = xmalloc(sizeof(struct sha1_size_cache));
        sha1_size_cache[first] = e;
-       memcpy(e->sha1, sha1, 20);
+       hashcpy(e->sha1, sha1);
        e->size = size;
        return e;
 }
@@ -1354,7 +1543,7 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
                }
        }
        else
-               memset(one->sha1, 0, 20);
+               hashclr(one->sha1);
 }
 
 static void run_diff(struct diff_filepair *p, struct diff_options *o)
@@ -1417,9 +1606,15 @@ static void run_diff(struct diff_filepair *p, struct diff_options *o)
                ;
        }
 
-       if (memcmp(one->sha1, two->sha1, 20)) {
+       if (hashcmp(one->sha1, two->sha1)) {
                int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
 
+               if (o->binary) {
+                       mmfile_t mf;
+                       if ((!fill_mmfile(&mf, one) && mmfile_is_binary(&mf)) ||
+                           (!fill_mmfile(&mf, two) && mmfile_is_binary(&mf)))
+                               abbrev = 40;
+               }
                len += snprintf(msg + len, sizeof(msg) - len,
                                "index %.*s..%.*s",
                                abbrev, sha1_to_hex(one->sha1),
@@ -1515,10 +1710,21 @@ void diff_setup(struct diff_options *options)
 
 int diff_setup_done(struct diff_options *options)
 {
-       if ((options->find_copies_harder &&
-            options->detect_rename != DIFF_DETECT_COPY) ||
-           (0 <= options->rename_limit && !options->detect_rename))
-               return -1;
+       int count = 0;
+
+       if (options->output_format & DIFF_FORMAT_NAME)
+               count++;
+       if (options->output_format & DIFF_FORMAT_NAME_STATUS)
+               count++;
+       if (options->output_format & DIFF_FORMAT_CHECKDIFF)
+               count++;
+       if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
+               count++;
+       if (count > 1)
+               die("--name-only, --name-status, --check and -s are mutually exclusive");
+
+       if (options->find_copies_harder)
+               options->detect_rename = DIFF_DETECT_COPY;
 
        if (options->output_format & (DIFF_FORMAT_NAME |
                                      DIFF_FORMAT_NAME_STATUS |
@@ -1622,8 +1828,33 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        else if (!strcmp(arg, "--patch-with-raw")) {
                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
        }
-       else if (!strcmp(arg, "--stat"))
+       else if (!strncmp(arg, "--stat", 6)) {
+               char *end;
+               int width = options->stat_width;
+               int name_width = options->stat_name_width;
+               arg += 6;
+               end = (char *)arg;
+
+               switch (*arg) {
+               case '-':
+                       if (!strncmp(arg, "-width=", 7))
+                               width = strtoul(arg + 7, &end, 10);
+                       else if (!strncmp(arg, "-name-width=", 12))
+                               name_width = strtoul(arg + 12, &end, 10);
+                       break;
+               case '=':
+                       width = strtoul(arg+1, &end, 10);
+                       if (*end == ',')
+                               name_width = strtoul(end+1, &end, 10);
+               }
+
+               /* Important! This checks all the error cases! */
+               if (*end)
+                       return 0;
                options->output_format |= DIFF_FORMAT_DIFFSTAT;
+               options->stat_name_width = name_width;
+               options->stat_width = width;
+       }
        else if (!strcmp(arg, "--check"))
                options->output_format |= DIFF_FORMAT_CHECKDIFF;
        else if (!strcmp(arg, "--summary"))
@@ -1639,7 +1870,7 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->full_index = 1;
        else if (!strcmp(arg, "--binary")) {
                options->output_format |= DIFF_FORMAT_PATCH;
-               options->full_index = options->binary = 1;
+               options->binary = 1;
        }
        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
                options->text = 1;
@@ -1699,6 +1930,8 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->xdl_opts |= XDF_IGNORE_WHITESPACE;
        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
                options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
+       else if (!strcmp(arg, "--color-words"))
+               options->color_diff = options->color_diff_words = 1;
        else if (!strcmp(arg, "--no-renames"))
                options->detect_rename = 0;
        else
@@ -1786,13 +2019,9 @@ struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
                                 struct diff_filespec *one,
                                 struct diff_filespec *two)
 {
-       struct diff_filepair *dp = xmalloc(sizeof(*dp));
+       struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
        dp->one = one;
        dp->two = two;
-       dp->score = 0;
-       dp->status = 0;
-       dp->source_stays = 0;
-       dp->broken_pair = 0;
        if (queue)
                diff_q(queue, dp);
        return dp;
@@ -1927,7 +2156,7 @@ int diff_unmodified_pair(struct diff_filepair *p)
         * dealing with a change.
         */
        if (one->sha1_valid && two->sha1_valid &&
-           !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
+           !hashcmp(one->sha1, two->sha1))
                return 1; /* no change */
        if (!one->sha1_valid && !two->sha1_valid)
                return 1; /* both look at the same file on the filesystem. */
@@ -2066,7 +2295,7 @@ static void diff_resolve_rename_copy(void)
                        if (!p->status)
                                p->status = DIFF_STATUS_RENAMED;
                }
-               else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
+               else if (hashcmp(p->one->sha1, p->two->sha1) ||
                         p->one->mode != p->two->mode)
                        p->status = DIFF_STATUS_MODIFIED;
                else {
@@ -2383,7 +2612,7 @@ void diff_flush(struct diff_options *options)
                        if (check_pair_status(p))
                                diff_flush_stat(p, options, &diffstat);
                }
-               show_stats(&diffstat);
+               show_stats(&diffstat, options);
                separator++;
        }
 
@@ -2410,6 +2639,9 @@ void diff_flush(struct diff_options *options)
                }
        }
 
+       if (output_format & DIFF_FORMAT_CALLBACK)
+               options->format_callback(q, options, options->format_callback_data);
+
        for (i = 0; i < q->nr; i++)
                diff_free_filepair(q->queue[i]);
 free_queue: