git-pickaxe: optimize by avoiding repeated read_sha1_file().
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 443e24861ba1857822dbe8e3c831f8037e4e832e..33153787b8117396cf906e69e656849ac04f3257 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -20,12 +20,13 @@ static int diff_use_color_default;
 
 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)
@@ -42,6 +43,8 @@ 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);
 }
 
@@ -205,7 +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);
@@ -383,9 +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);
@@ -403,45 +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;
-               if (ecbdata->diff_words && 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--;
-               } else
-                       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)
@@ -545,21 +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)
@@ -567,54 +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 &&
@@ -623,27 +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 {
@@ -1582,6 +1748,7 @@ int diff_setup_done(struct diff_options *options)
                                      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);
@@ -1591,7 +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;
        /*
@@ -1679,8 +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"))
@@ -2428,7 +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));
@@ -2438,7 +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++;
        }