Merge branch 'js/diff'
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 5358fd0c14a249c2d1a3fe417016264aba6fc745..33153787b8117396cf906e69e656849ac04f3257 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;
 static int diff_rename_limit_default = -1;
-static int diff_use_color_default = 0;
-
-enum color_diff {
-       DIFF_RESET = 0,
-       DIFF_PLAIN = 1,
-       DIFF_METAINFO = 2,
-       DIFF_FRAGINFO = 3,
-       DIFF_FILE_OLD = 4,
-       DIFF_FILE_NEW = 5,
-};
-
-#define COLOR_NORMAL  ""
-#define COLOR_BOLD    "\033[1m"
-#define COLOR_DIM     "\033[2m"
-#define COLOR_UL      "\033[4m"
-#define COLOR_BLINK   "\033[5m"
-#define COLOR_REVERSE "\033[7m"
-#define COLOR_RESET   "\033[m"
-
-#define COLOR_BLACK   "\033[30m"
-#define COLOR_RED     "\033[31m"
-#define COLOR_GREEN   "\033[32m"
-#define COLOR_YELLOW  "\033[33m"
-#define COLOR_BLUE    "\033[34m"
-#define COLOR_MAGENTA "\033[35m"
-#define COLOR_CYAN    "\033[36m"
-#define COLOR_WHITE   "\033[37m"
-
-static const char *diff_colors[] = {
-       [DIFF_RESET]    = COLOR_RESET,
-       [DIFF_PLAIN]    = COLOR_NORMAL,
-       [DIFF_METAINFO] = COLOR_BOLD,
-       [DIFF_FRAGINFO] = COLOR_CYAN,
-       [DIFF_FILE_OLD] = COLOR_RED,
-       [DIFF_FILE_NEW] = COLOR_GREEN,
+static int diff_use_color_default;
+
+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) */
 };
 
 static int parse_diff_color_slot(const char *var, int ofs)
@@ -63,66 +41,42 @@ static int parse_diff_color_slot(const char *var, int ofs)
                return DIFF_FILE_OLD;
        if (!strcasecmp(var+ofs, "new"))
                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 const char *parse_diff_color_value(const char *value, const char *var)
-{
-       if (!strcasecmp(value, "normal"))
-               return COLOR_NORMAL;
-       if (!strcasecmp(value, "bold"))
-               return COLOR_BOLD;
-       if (!strcasecmp(value, "dim"))
-               return COLOR_DIM;
-       if (!strcasecmp(value, "ul"))
-               return COLOR_UL;
-       if (!strcasecmp(value, "blink"))
-               return COLOR_BLINK;
-       if (!strcasecmp(value, "reverse"))
-               return COLOR_REVERSE;
-       if (!strcasecmp(value, "reset"))
-               return COLOR_RESET;
-       if (!strcasecmp(value, "black"))
-               return COLOR_BLACK;
-       if (!strcasecmp(value, "red"))
-               return COLOR_RED;
-       if (!strcasecmp(value, "green"))
-               return COLOR_GREEN;
-       if (!strcasecmp(value, "yellow"))
-               return COLOR_YELLOW;
-       if (!strcasecmp(value, "blue"))
-               return COLOR_BLUE;
-       if (!strcasecmp(value, "magenta"))
-               return COLOR_MAGENTA;
-       if (!strcasecmp(value, "cyan"))
-               return COLOR_CYAN;
-       if (!strcasecmp(value, "white"))
-               return COLOR_WHITE;
-       die("bad config value '%s' for variable '%s'", value, var);
-}
-
-int git_diff_config(const char *var, const char *value)
+/*
+ * 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)
 {
        if (!strcmp(var, "diff.renamelimit")) {
                diff_rename_limit_default = git_config_int(var, value);
                return 0;
        }
        if (!strcmp(var, "diff.color")) {
+               diff_use_color_default = git_config_colorbool(var, value);
+               return 0;
+       }
+       if (!strcmp(var, "diff.renames")) {
                if (!value)
-                       diff_use_color_default = 1; /* bool */
-               else if (!strcasecmp(value, "auto"))
-                       diff_use_color_default = isatty(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_detect_rename_default = DIFF_DETECT_RENAME;
+               else if (!strcasecmp(value, "copies") ||
+                        !strcasecmp(value, "copy"))
+                       diff_detect_rename_default = DIFF_DETECT_COPY;
+               else if (git_config_bool(var,value))
+                       diff_detect_rename_default = DIFF_DETECT_RENAME;
                return 0;
        }
        if (!strncmp(var, "diff.color.", 11)) {
                int slot = parse_diff_color_slot(var, 11);
-               diff_colors[slot] = parse_diff_color_value(value, var);
+               color_parse(value, var, diff_colors[slot]);
                return 0;
        }
        return git_default_config(var, value);
@@ -137,7 +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;
@@ -254,7 +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);
@@ -279,25 +233,245 @@ 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 inline const char *get_color(int diff_use_color, enum color_diff ix)
+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)
                return diff_colors[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 = get_color(ecbdata->color_diff, DIFF_METAINFO);
-       const char *reset = get_color(ecbdata->color_diff, DIFF_RESET);
+       const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
+       const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 
        if (ecbdata->label_path[0]) {
                printf("%s--- %s%s\n", set, ecbdata->label_path[0], reset);
@@ -312,24 +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 = 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 = 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--;
-       printf("%s%.*s%s\n", set, (int) len, line, 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)
@@ -418,7 +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;
 }
 
@@ -433,21 +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)
@@ -455,54 +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 &&
@@ -511,27 +767,49 @@ 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);
+}
+
+static void show_numstat(struct diffstat_t* data, struct diff_options *options)
+{
+       int i;
+
+       for (i = 0; i < data->nr; i++) {
+               struct diffstat_file *file = data->files[i];
+
+               printf("%d\t%d\t", file->added, file->deleted);
+               if (options->line_termination &&
+                   quote_c_style(file->name, NULL, NULL, 0))
+                       quote_c_style(file->name, NULL, stdout, 0);
+               else
+                       fputs(file->name, stdout);
+               putchar(options->line_termination);
+       }
 }
 
 struct checkdiff_t {
@@ -583,7 +861,7 @@ static unsigned char *deflate_it(char *data,
        z_stream stream;
 
        memset(&stream, 0, sizeof(stream));
-       deflateInit(&stream, Z_BEST_COMPRESSION);
+       deflateInit(&stream, zlib_compression_level);
        bound = deflateBound(&stream, size);
        deflated = xmalloc(bound);
        stream.next_out = deflated;
@@ -598,7 +876,7 @@ 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;
@@ -609,7 +887,6 @@ 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.
         */
@@ -658,15 +935,20 @@ 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,
@@ -680,8 +962,8 @@ static void builtin_diff(const char *name_a,
        mmfile_t mf1, mf2;
        const char *lbl[2];
        char *a_one, *b_two;
-       const char *set = get_color(o->color_diff, DIFF_METAINFO);
-       const char *reset = get_color(o->color_diff, DIFF_RESET);
+       const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
+       const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
 
        a_one = quote_two("a/", name_a);
        b_two = quote_two("b/", name_b);
@@ -721,7 +1003,7 @@ static void builtin_diff(const char *name_a,
        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
                die("unable to read files to diff");
 
-       if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2)) {
+       if (!o->text && (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))) {
                /* Quite common confusing case */
                if (mf1.size == mf2.size &&
                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
@@ -755,7 +1037,12 @@ 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:
@@ -858,8 +1145,8 @@ 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);
        }
 }
 
@@ -897,7 +1184,7 @@ 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
@@ -925,7 +1212,7 @@ 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) {
@@ -951,7 +1238,7 @@ 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;
 }
@@ -1273,7 +1560,7 @@ 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)
@@ -1336,9 +1623,15 @@ 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),
@@ -1429,6 +1722,7 @@ void diff_setup(struct diff_options *options)
        options->change = diff_change;
        options->add_remove = diff_addremove;
        options->color_diff = diff_use_color_default;
+       options->detect_rename = diff_detect_rename_default;
 }
 
 int diff_setup_done(struct diff_options *options)
@@ -1446,16 +1740,15 @@ int diff_setup_done(struct diff_options *options)
        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) ||
-           (0 <= options->rename_limit && !options->detect_rename))
-               return -1;
+       if (options->find_copies_harder)
+               options->detect_rename = DIFF_DETECT_COPY;
 
        if (options->output_format & (DIFF_FORMAT_NAME |
                                      DIFF_FORMAT_NAME_STATUS |
                                      DIFF_FORMAT_CHECKDIFF |
                                      DIFF_FORMAT_NO_OUTPUT))
                options->output_format &= ~(DIFF_FORMAT_RAW |
+                                           DIFF_FORMAT_NUMSTAT |
                                            DIFF_FORMAT_DIFFSTAT |
                                            DIFF_FORMAT_SUMMARY |
                                            DIFF_FORMAT_PATCH);
@@ -1465,7 +1758,9 @@ int diff_setup_done(struct diff_options *options)
         * recursive bits for other formats here.
         */
        if (options->output_format & (DIFF_FORMAT_PATCH |
+                                     DIFF_FORMAT_NUMSTAT |
                                      DIFF_FORMAT_DIFFSTAT |
+                                     DIFF_FORMAT_SUMMARY |
                                      DIFF_FORMAT_CHECKDIFF))
                options->recursive = 1;
        /*
@@ -1553,8 +1848,36 @@ 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 (!strcmp(arg, "--numstat")) {
+               options->output_format |= DIFF_FORMAT_NUMSTAT;
+       }
+       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"))
@@ -1570,7 +1893,10 @@ 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;
        }
        else if (!strcmp(arg, "--name-only"))
                options->output_format |= DIFF_FORMAT_NAME;
@@ -1621,10 +1947,16 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        }
        else if (!strcmp(arg, "--color"))
                options->color_diff = 1;
+       else if (!strcmp(arg, "--no-color"))
+               options->color_diff = 0;
        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
                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
                return 0;
        return 1;
@@ -1710,13 +2042,9 @@ 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;
@@ -1851,7 +2179,7 @@ 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. */
@@ -1990,7 +2318,7 @@ 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 {
@@ -2108,6 +2436,145 @@ static void diff_summary(struct diff_filepair *p)
        }
 }
 
+struct patch_id_t {
+       struct xdiff_emit_state xm;
+       SHA_CTX *ctx;
+       int patchlen;
+};
+
+static int remove_space(char *line, int len)
+{
+       int i;
+        char *dst = line;
+        unsigned char c;
+
+        for (i = 0; i < len; i++)
+                if (!isspace((c = line[i])))
+                        *dst++ = c;
+
+        return dst - line;
+}
+
+static void patch_id_consume(void *priv, char *line, unsigned long len)
+{
+       struct patch_id_t *data = priv;
+       int new_len;
+
+       /* Ignore line numbers when computing the SHA1 of the patch */
+       if (!strncmp(line, "@@ -", 4))
+               return;
+
+       new_len = remove_space(line, len);
+
+       SHA1_Update(data->ctx, line, new_len);
+       data->patchlen += new_len;
+}
+
+/* returns 0 upon success, and writes result into sha1 */
+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;
+       struct patch_id_t data;
+       char buffer[PATH_MAX * 4 + 20];
+
+       SHA1_Init(&ctx);
+       memset(&data, 0, sizeof(struct patch_id_t));
+       data.ctx = &ctx;
+       data.xm.consume = patch_id_consume;
+
+       for (i = 0; i < q->nr; i++) {
+               xpparam_t xpp;
+               xdemitconf_t xecfg;
+               xdemitcb_t ecb;
+               mmfile_t mf1, mf2;
+               struct diff_filepair *p = q->queue[i];
+               int len1, len2;
+
+               if (p->status == 0)
+                       return error("internal diff status error");
+               if (p->status == DIFF_STATUS_UNKNOWN)
+                       continue;
+               if (diff_unmodified_pair(p))
+                       continue;
+               if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
+                   (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
+                       continue;
+               if (DIFF_PAIR_UNMERGED(p))
+                       continue;
+
+               diff_fill_sha1_info(p->one);
+               diff_fill_sha1_info(p->two);
+               if (fill_mmfile(&mf1, p->one) < 0 ||
+                               fill_mmfile(&mf2, p->two) < 0)
+                       return error("unable to read files to diff");
+
+               /* Maybe hash p->two? into the patch id? */
+               if (mmfile_is_binary(&mf2))
+                       continue;
+
+               len1 = remove_space(p->one->path, strlen(p->one->path));
+               len2 = remove_space(p->two->path, strlen(p->two->path));
+               if (p->one->mode == 0)
+                       len1 = snprintf(buffer, sizeof(buffer),
+                                       "diff--gita/%.*sb/%.*s"
+                                       "newfilemode%06o"
+                                       "---/dev/null"
+                                       "+++b/%.*s",
+                                       len1, p->one->path,
+                                       len2, p->two->path,
+                                       p->two->mode,
+                                       len2, p->two->path);
+               else if (p->two->mode == 0)
+                       len1 = snprintf(buffer, sizeof(buffer),
+                                       "diff--gita/%.*sb/%.*s"
+                                       "deletedfilemode%06o"
+                                       "---a/%.*s"
+                                       "+++/dev/null",
+                                       len1, p->one->path,
+                                       len2, p->two->path,
+                                       p->one->mode,
+                                       len1, p->one->path);
+               else
+                       len1 = snprintf(buffer, sizeof(buffer),
+                                       "diff--gita/%.*sb/%.*s"
+                                       "---a/%.*s"
+                                       "+++b/%.*s",
+                                       len1, p->one->path,
+                                       len2, p->two->path,
+                                       len1, p->one->path,
+                                       len2, p->two->path);
+               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;
+               xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+       }
+
+       SHA1_Final(sha1, &ctx);
+       return 0;
+}
+
+int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
+{
+       struct diff_queue_struct *q = &diff_queued_diff;
+       int i;
+       int result = diff_get_patch_id(options, sha1);
+
+       for (i = 0; i < q->nr; i++)
+               diff_free_filepair(q->queue[i]);
+
+       free(q->queue);
+       q->queue = NULL;
+       q->nr = q->alloc = 0;
+
+       return result;
+}
+
 static int is_summary_empty(const struct diff_queue_struct *q)
 {
        int i;
@@ -2158,7 +2625,7 @@ void diff_flush(struct diff_options *options)
                separator++;
        }
 
-       if (output_format & DIFF_FORMAT_DIFFSTAT) {
+       if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_NUMSTAT)) {
                struct diffstat_t diffstat;
 
                memset(&diffstat, 0, sizeof(struct diffstat_t));
@@ -2168,7 +2635,10 @@ void diff_flush(struct diff_options *options)
                        if (check_pair_status(p))
                                diff_flush_stat(p, options, &diffstat);
                }
-               show_stats(&diffstat);
+               if (output_format & DIFF_FORMAT_NUMSTAT)
+                       show_numstat(&diffstat, options);
+               if (output_format & DIFF_FORMAT_DIFFSTAT)
+                       show_stats(&diffstat, options);
                separator++;
        }
 
@@ -2195,6 +2665,9 @@ 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: