diff.con commit Use gitattributes to define per-path whitespace rule (cf1b786)
   1/*
   2 * Copyright (C) 2005 Junio C Hamano
   3 */
   4#include "cache.h"
   5#include "quote.h"
   6#include "diff.h"
   7#include "diffcore.h"
   8#include "delta.h"
   9#include "xdiff-interface.h"
  10#include "color.h"
  11#include "attr.h"
  12#include "run-command.h"
  13
  14#ifdef NO_FAST_WORKING_DIRECTORY
  15#define FAST_WORKING_DIRECTORY 0
  16#else
  17#define FAST_WORKING_DIRECTORY 1
  18#endif
  19
  20static int diff_detect_rename_default;
  21static int diff_rename_limit_default = 100;
  22static int diff_use_color_default;
  23int diff_auto_refresh_index = 1;
  24
  25static char diff_colors[][COLOR_MAXLEN] = {
  26        "\033[m",       /* reset */
  27        "",             /* PLAIN (normal) */
  28        "\033[1m",      /* METAINFO (bold) */
  29        "\033[36m",     /* FRAGINFO (cyan) */
  30        "\033[31m",     /* OLD (red) */
  31        "\033[32m",     /* NEW (green) */
  32        "\033[33m",     /* COMMIT (yellow) */
  33        "\033[41m",     /* WHITESPACE (red background) */
  34};
  35
  36static int parse_diff_color_slot(const char *var, int ofs)
  37{
  38        if (!strcasecmp(var+ofs, "plain"))
  39                return DIFF_PLAIN;
  40        if (!strcasecmp(var+ofs, "meta"))
  41                return DIFF_METAINFO;
  42        if (!strcasecmp(var+ofs, "frag"))
  43                return DIFF_FRAGINFO;
  44        if (!strcasecmp(var+ofs, "old"))
  45                return DIFF_FILE_OLD;
  46        if (!strcasecmp(var+ofs, "new"))
  47                return DIFF_FILE_NEW;
  48        if (!strcasecmp(var+ofs, "commit"))
  49                return DIFF_COMMIT;
  50        if (!strcasecmp(var+ofs, "whitespace"))
  51                return DIFF_WHITESPACE;
  52        die("bad config variable '%s'", var);
  53}
  54
  55static struct ll_diff_driver {
  56        const char *name;
  57        struct ll_diff_driver *next;
  58        char *cmd;
  59} *user_diff, **user_diff_tail;
  60
  61static void read_config_if_needed(void)
  62{
  63        if (!user_diff_tail) {
  64                user_diff_tail = &user_diff;
  65                git_config(git_diff_ui_config);
  66        }
  67}
  68
  69/*
  70 * Currently there is only "diff.<drivername>.command" variable;
  71 * because there are "diff.color.<slot>" variables, we are parsing
  72 * this in a bit convoluted way to allow low level diff driver
  73 * called "color".
  74 */
  75static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  76{
  77        const char *name;
  78        int namelen;
  79        struct ll_diff_driver *drv;
  80
  81        name = var + 5;
  82        namelen = ep - name;
  83        for (drv = user_diff; drv; drv = drv->next)
  84                if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  85                        break;
  86        if (!drv) {
  87                drv = xcalloc(1, sizeof(struct ll_diff_driver));
  88                drv->name = xmemdupz(name, namelen);
  89                if (!user_diff_tail)
  90                        user_diff_tail = &user_diff;
  91                *user_diff_tail = drv;
  92                user_diff_tail = &(drv->next);
  93        }
  94
  95        if (!value)
  96                return error("%s: lacks value", var);
  97        drv->cmd = strdup(value);
  98        return 0;
  99}
 100
 101/*
 102 * 'diff.<what>.funcname' attribute can be specified in the configuration
 103 * to define a customized regexp to find the beginning of a function to
 104 * be used for hunk header lines of "diff -p" style output.
 105 */
 106static struct funcname_pattern {
 107        char *name;
 108        char *pattern;
 109        struct funcname_pattern *next;
 110} *funcname_pattern_list;
 111
 112static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 113{
 114        const char *name;
 115        int namelen;
 116        struct funcname_pattern *pp;
 117
 118        name = var + 5; /* "diff." */
 119        namelen = ep - name;
 120
 121        for (pp = funcname_pattern_list; pp; pp = pp->next)
 122                if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 123                        break;
 124        if (!pp) {
 125                pp = xcalloc(1, sizeof(*pp));
 126                pp->name = xmemdupz(name, namelen);
 127                pp->next = funcname_pattern_list;
 128                funcname_pattern_list = pp;
 129        }
 130        if (pp->pattern)
 131                free(pp->pattern);
 132        pp->pattern = xstrdup(value);
 133        return 0;
 134}
 135
 136/*
 137 * These are to give UI layer defaults.
 138 * The core-level commands such as git-diff-files should
 139 * never be affected by the setting of diff.renames
 140 * the user happens to have in the configuration file.
 141 */
 142int git_diff_ui_config(const char *var, const char *value)
 143{
 144        if (!strcmp(var, "diff.renamelimit")) {
 145                diff_rename_limit_default = git_config_int(var, value);
 146                return 0;
 147        }
 148        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 149                diff_use_color_default = git_config_colorbool(var, value);
 150                return 0;
 151        }
 152        if (!strcmp(var, "diff.renames")) {
 153                if (!value)
 154                        diff_detect_rename_default = DIFF_DETECT_RENAME;
 155                else if (!strcasecmp(value, "copies") ||
 156                         !strcasecmp(value, "copy"))
 157                        diff_detect_rename_default = DIFF_DETECT_COPY;
 158                else if (git_config_bool(var,value))
 159                        diff_detect_rename_default = DIFF_DETECT_RENAME;
 160                return 0;
 161        }
 162        if (!strcmp(var, "diff.autorefreshindex")) {
 163                diff_auto_refresh_index = git_config_bool(var, value);
 164                return 0;
 165        }
 166        if (!prefixcmp(var, "diff.")) {
 167                const char *ep = strrchr(var, '.');
 168
 169                if (ep != var + 4) {
 170                        if (!strcmp(ep, ".command"))
 171                                return parse_lldiff_command(var, ep, value);
 172                        if (!strcmp(ep, ".funcname"))
 173                                return parse_funcname_pattern(var, ep, value);
 174                }
 175        }
 176        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 177                int slot = parse_diff_color_slot(var, 11);
 178                color_parse(value, var, diff_colors[slot]);
 179                return 0;
 180        }
 181
 182        return git_default_config(var, value);
 183}
 184
 185static char *quote_two(const char *one, const char *two)
 186{
 187        int need_one = quote_c_style(one, NULL, NULL, 1);
 188        int need_two = quote_c_style(two, NULL, NULL, 1);
 189        struct strbuf res;
 190
 191        strbuf_init(&res, 0);
 192        if (need_one + need_two) {
 193                strbuf_addch(&res, '"');
 194                quote_c_style(one, &res, NULL, 1);
 195                quote_c_style(two, &res, NULL, 1);
 196                strbuf_addch(&res, '"');
 197        } else {
 198                strbuf_addstr(&res, one);
 199                strbuf_addstr(&res, two);
 200        }
 201        return strbuf_detach(&res, NULL);
 202}
 203
 204static const char *external_diff(void)
 205{
 206        static const char *external_diff_cmd = NULL;
 207        static int done_preparing = 0;
 208
 209        if (done_preparing)
 210                return external_diff_cmd;
 211        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 212        done_preparing = 1;
 213        return external_diff_cmd;
 214}
 215
 216static struct diff_tempfile {
 217        const char *name; /* filename external diff should read from */
 218        char hex[41];
 219        char mode[10];
 220        char tmp_path[PATH_MAX];
 221} diff_temp[2];
 222
 223static int count_lines(const char *data, int size)
 224{
 225        int count, ch, completely_empty = 1, nl_just_seen = 0;
 226        count = 0;
 227        while (0 < size--) {
 228                ch = *data++;
 229                if (ch == '\n') {
 230                        count++;
 231                        nl_just_seen = 1;
 232                        completely_empty = 0;
 233                }
 234                else {
 235                        nl_just_seen = 0;
 236                        completely_empty = 0;
 237                }
 238        }
 239        if (completely_empty)
 240                return 0;
 241        if (!nl_just_seen)
 242                count++; /* no trailing newline */
 243        return count;
 244}
 245
 246static void print_line_count(int count)
 247{
 248        switch (count) {
 249        case 0:
 250                printf("0,0");
 251                break;
 252        case 1:
 253                printf("1");
 254                break;
 255        default:
 256                printf("1,%d", count);
 257                break;
 258        }
 259}
 260
 261static void copy_file(int prefix, const char *data, int size,
 262                const char *set, const char *reset)
 263{
 264        int ch, nl_just_seen = 1;
 265        while (0 < size--) {
 266                ch = *data++;
 267                if (nl_just_seen) {
 268                        fputs(set, stdout);
 269                        putchar(prefix);
 270                }
 271                if (ch == '\n') {
 272                        nl_just_seen = 1;
 273                        fputs(reset, stdout);
 274                } else
 275                        nl_just_seen = 0;
 276                putchar(ch);
 277        }
 278        if (!nl_just_seen)
 279                printf("%s\n\\ No newline at end of file\n", reset);
 280}
 281
 282static void emit_rewrite_diff(const char *name_a,
 283                              const char *name_b,
 284                              struct diff_filespec *one,
 285                              struct diff_filespec *two,
 286                              int color_diff)
 287{
 288        int lc_a, lc_b;
 289        const char *name_a_tab, *name_b_tab;
 290        const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 291        const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 292        const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 293        const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 294        const char *reset = diff_get_color(color_diff, DIFF_RESET);
 295
 296        name_a += (*name_a == '/');
 297        name_b += (*name_b == '/');
 298        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 299        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 300
 301        diff_populate_filespec(one, 0);
 302        diff_populate_filespec(two, 0);
 303        lc_a = count_lines(one->data, one->size);
 304        lc_b = count_lines(two->data, two->size);
 305        printf("%s--- a/%s%s%s\n%s+++ b/%s%s%s\n%s@@ -",
 306               metainfo, name_a, name_a_tab, reset,
 307               metainfo, name_b, name_b_tab, reset, fraginfo);
 308        print_line_count(lc_a);
 309        printf(" +");
 310        print_line_count(lc_b);
 311        printf(" @@%s\n", reset);
 312        if (lc_a)
 313                copy_file('-', one->data, one->size, old, reset);
 314        if (lc_b)
 315                copy_file('+', two->data, two->size, new, reset);
 316}
 317
 318static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 319{
 320        if (!DIFF_FILE_VALID(one)) {
 321                mf->ptr = (char *)""; /* does not matter */
 322                mf->size = 0;
 323                return 0;
 324        }
 325        else if (diff_populate_filespec(one, 0))
 326                return -1;
 327        mf->ptr = one->data;
 328        mf->size = one->size;
 329        return 0;
 330}
 331
 332struct diff_words_buffer {
 333        mmfile_t text;
 334        long alloc;
 335        long current; /* output pointer */
 336        int suppressed_newline;
 337};
 338
 339static void diff_words_append(char *line, unsigned long len,
 340                struct diff_words_buffer *buffer)
 341{
 342        if (buffer->text.size + len > buffer->alloc) {
 343                buffer->alloc = (buffer->text.size + len) * 3 / 2;
 344                buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 345        }
 346        line++;
 347        len--;
 348        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 349        buffer->text.size += len;
 350}
 351
 352struct diff_words_data {
 353        struct xdiff_emit_state xm;
 354        struct diff_words_buffer minus, plus;
 355};
 356
 357static void print_word(struct diff_words_buffer *buffer, int len, int color,
 358                int suppress_newline)
 359{
 360        const char *ptr;
 361        int eol = 0;
 362
 363        if (len == 0)
 364                return;
 365
 366        ptr  = buffer->text.ptr + buffer->current;
 367        buffer->current += len;
 368
 369        if (ptr[len - 1] == '\n') {
 370                eol = 1;
 371                len--;
 372        }
 373
 374        fputs(diff_get_color(1, color), stdout);
 375        fwrite(ptr, len, 1, stdout);
 376        fputs(diff_get_color(1, DIFF_RESET), stdout);
 377
 378        if (eol) {
 379                if (suppress_newline)
 380                        buffer->suppressed_newline = 1;
 381                else
 382                        putchar('\n');
 383        }
 384}
 385
 386static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 387{
 388        struct diff_words_data *diff_words = priv;
 389
 390        if (diff_words->minus.suppressed_newline) {
 391                if (line[0] != '+')
 392                        putchar('\n');
 393                diff_words->minus.suppressed_newline = 0;
 394        }
 395
 396        len--;
 397        switch (line[0]) {
 398                case '-':
 399                        print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
 400                        break;
 401                case '+':
 402                        print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
 403                        break;
 404                case ' ':
 405                        print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
 406                        diff_words->minus.current += len;
 407                        break;
 408        }
 409}
 410
 411/* this executes the word diff on the accumulated buffers */
 412static void diff_words_show(struct diff_words_data *diff_words)
 413{
 414        xpparam_t xpp;
 415        xdemitconf_t xecfg;
 416        xdemitcb_t ecb;
 417        mmfile_t minus, plus;
 418        int i;
 419
 420        memset(&xecfg, 0, sizeof(xecfg));
 421        minus.size = diff_words->minus.text.size;
 422        minus.ptr = xmalloc(minus.size);
 423        memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 424        for (i = 0; i < minus.size; i++)
 425                if (isspace(minus.ptr[i]))
 426                        minus.ptr[i] = '\n';
 427        diff_words->minus.current = 0;
 428
 429        plus.size = diff_words->plus.text.size;
 430        plus.ptr = xmalloc(plus.size);
 431        memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 432        for (i = 0; i < plus.size; i++)
 433                if (isspace(plus.ptr[i]))
 434                        plus.ptr[i] = '\n';
 435        diff_words->plus.current = 0;
 436
 437        xpp.flags = XDF_NEED_MINIMAL;
 438        xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 439        ecb.outf = xdiff_outf;
 440        ecb.priv = diff_words;
 441        diff_words->xm.consume = fn_out_diff_words_aux;
 442        xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 443
 444        free(minus.ptr);
 445        free(plus.ptr);
 446        diff_words->minus.text.size = diff_words->plus.text.size = 0;
 447
 448        if (diff_words->minus.suppressed_newline) {
 449                putchar('\n');
 450                diff_words->minus.suppressed_newline = 0;
 451        }
 452}
 453
 454struct emit_callback {
 455        struct xdiff_emit_state xm;
 456        int nparents, color_diff;
 457        unsigned ws_rule;
 458        const char **label_path;
 459        struct diff_words_data *diff_words;
 460        int *found_changesp;
 461};
 462
 463static void free_diff_words_data(struct emit_callback *ecbdata)
 464{
 465        if (ecbdata->diff_words) {
 466                /* flush buffers */
 467                if (ecbdata->diff_words->minus.text.size ||
 468                                ecbdata->diff_words->plus.text.size)
 469                        diff_words_show(ecbdata->diff_words);
 470
 471                if (ecbdata->diff_words->minus.text.ptr)
 472                        free (ecbdata->diff_words->minus.text.ptr);
 473                if (ecbdata->diff_words->plus.text.ptr)
 474                        free (ecbdata->diff_words->plus.text.ptr);
 475                free(ecbdata->diff_words);
 476                ecbdata->diff_words = NULL;
 477        }
 478}
 479
 480const char *diff_get_color(int diff_use_color, enum color_diff ix)
 481{
 482        if (diff_use_color)
 483                return diff_colors[ix];
 484        return "";
 485}
 486
 487static void emit_line(const char *set, const char *reset, const char *line, int len)
 488{
 489        if (len > 0 && line[len-1] == '\n')
 490                len--;
 491        fputs(set, stdout);
 492        fwrite(line, len, 1, stdout);
 493        puts(reset);
 494}
 495
 496static void emit_line_with_ws(int nparents,
 497                              const char *set, const char *reset, const char *ws,
 498                              const char *line, int len, unsigned ws_rule)
 499{
 500        int col0 = nparents;
 501        int last_tab_in_indent = -1;
 502        int last_space_in_indent = -1;
 503        int i;
 504        int tail = len;
 505        int need_highlight_leading_space = 0;
 506        /*
 507         * The line is a newly added line.  Does it have funny leading
 508         * whitespaces?  In indent, SP should never precede a TAB.  In
 509         * addition, under "indent with non tab" rule, there should not
 510         * be more than 8 consecutive spaces.
 511         */
 512        for (i = col0; i < len; i++) {
 513                if (line[i] == '\t') {
 514                        last_tab_in_indent = i;
 515                        if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 516                            0 <= last_space_in_indent)
 517                                need_highlight_leading_space = 1;
 518                }
 519                else if (line[i] == ' ')
 520                        last_space_in_indent = i;
 521                else
 522                        break;
 523        }
 524        if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 525            0 <= last_space_in_indent &&
 526            last_tab_in_indent < 0 &&
 527            8 <= (i - col0)) {
 528                last_tab_in_indent = i;
 529                need_highlight_leading_space = 1;
 530        }
 531        fputs(set, stdout);
 532        fwrite(line, col0, 1, stdout);
 533        fputs(reset, stdout);
 534        if (((i == len) || line[i] == '\n') && i != col0) {
 535                /* The whole line was indent */
 536                emit_line(ws, reset, line + col0, len - col0);
 537                return;
 538        }
 539        i = col0;
 540        if (need_highlight_leading_space) {
 541                while (i < last_tab_in_indent) {
 542                        if (line[i] == ' ') {
 543                                fputs(ws, stdout);
 544                                putchar(' ');
 545                                fputs(reset, stdout);
 546                        }
 547                        else
 548                                putchar(line[i]);
 549                        i++;
 550                }
 551        }
 552        tail = len - 1;
 553        if (line[tail] == '\n' && i < tail)
 554                tail--;
 555        if (ws_rule & WS_TRAILING_SPACE) {
 556                while (i < tail) {
 557                        if (!isspace(line[tail]))
 558                                break;
 559                        tail--;
 560                }
 561        }
 562        if ((i < tail && line[tail + 1] != '\n')) {
 563                /* This has whitespace between tail+1..len */
 564                fputs(set, stdout);
 565                fwrite(line + i, tail - i + 1, 1, stdout);
 566                fputs(reset, stdout);
 567                emit_line(ws, reset, line + tail + 1, len - tail - 1);
 568        }
 569        else
 570                emit_line(set, reset, line + i, len - i);
 571}
 572
 573static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 574{
 575        const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 576        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 577
 578        if (!*ws)
 579                emit_line(set, reset, line, len);
 580        else
 581                emit_line_with_ws(ecbdata->nparents, set, reset, ws,
 582                                  line, len, ecbdata->ws_rule);
 583}
 584
 585static void fn_out_consume(void *priv, char *line, unsigned long len)
 586{
 587        int i;
 588        int color;
 589        struct emit_callback *ecbdata = priv;
 590        const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 591        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 592
 593        *(ecbdata->found_changesp) = 1;
 594
 595        if (ecbdata->label_path[0]) {
 596                const char *name_a_tab, *name_b_tab;
 597
 598                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 599                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 600
 601                printf("%s--- %s%s%s\n",
 602                       set, ecbdata->label_path[0], reset, name_a_tab);
 603                printf("%s+++ %s%s%s\n",
 604                       set, ecbdata->label_path[1], reset, name_b_tab);
 605                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 606        }
 607
 608        /* This is not really necessary for now because
 609         * this codepath only deals with two-way diffs.
 610         */
 611        for (i = 0; i < len && line[i] == '@'; i++)
 612                ;
 613        if (2 <= i && i < len && line[i] == ' ') {
 614                ecbdata->nparents = i - 1;
 615                emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 616                          reset, line, len);
 617                return;
 618        }
 619
 620        if (len < ecbdata->nparents) {
 621                set = reset;
 622                emit_line(reset, reset, line, len);
 623                return;
 624        }
 625
 626        color = DIFF_PLAIN;
 627        if (ecbdata->diff_words && ecbdata->nparents != 1)
 628                /* fall back to normal diff */
 629                free_diff_words_data(ecbdata);
 630        if (ecbdata->diff_words) {
 631                if (line[0] == '-') {
 632                        diff_words_append(line, len,
 633                                          &ecbdata->diff_words->minus);
 634                        return;
 635                } else if (line[0] == '+') {
 636                        diff_words_append(line, len,
 637                                          &ecbdata->diff_words->plus);
 638                        return;
 639                }
 640                if (ecbdata->diff_words->minus.text.size ||
 641                    ecbdata->diff_words->plus.text.size)
 642                        diff_words_show(ecbdata->diff_words);
 643                line++;
 644                len--;
 645                emit_line(set, reset, line, len);
 646                return;
 647        }
 648        for (i = 0; i < ecbdata->nparents && len; i++) {
 649                if (line[i] == '-')
 650                        color = DIFF_FILE_OLD;
 651                else if (line[i] == '+')
 652                        color = DIFF_FILE_NEW;
 653        }
 654
 655        if (color != DIFF_FILE_NEW) {
 656                emit_line(diff_get_color(ecbdata->color_diff, color),
 657                          reset, line, len);
 658                return;
 659        }
 660        emit_add_line(reset, ecbdata, line, len);
 661}
 662
 663static char *pprint_rename(const char *a, const char *b)
 664{
 665        const char *old = a;
 666        const char *new = b;
 667        struct strbuf name;
 668        int pfx_length, sfx_length;
 669        int len_a = strlen(a);
 670        int len_b = strlen(b);
 671        int a_midlen, b_midlen;
 672        int qlen_a = quote_c_style(a, NULL, NULL, 0);
 673        int qlen_b = quote_c_style(b, NULL, NULL, 0);
 674
 675        strbuf_init(&name, 0);
 676        if (qlen_a || qlen_b) {
 677                quote_c_style(a, &name, NULL, 0);
 678                strbuf_addstr(&name, " => ");
 679                quote_c_style(b, &name, NULL, 0);
 680                return strbuf_detach(&name, NULL);
 681        }
 682
 683        /* Find common prefix */
 684        pfx_length = 0;
 685        while (*old && *new && *old == *new) {
 686                if (*old == '/')
 687                        pfx_length = old - a + 1;
 688                old++;
 689                new++;
 690        }
 691
 692        /* Find common suffix */
 693        old = a + len_a;
 694        new = b + len_b;
 695        sfx_length = 0;
 696        while (a <= old && b <= new && *old == *new) {
 697                if (*old == '/')
 698                        sfx_length = len_a - (old - a);
 699                old--;
 700                new--;
 701        }
 702
 703        /*
 704         * pfx{mid-a => mid-b}sfx
 705         * {pfx-a => pfx-b}sfx
 706         * pfx{sfx-a => sfx-b}
 707         * name-a => name-b
 708         */
 709        a_midlen = len_a - pfx_length - sfx_length;
 710        b_midlen = len_b - pfx_length - sfx_length;
 711        if (a_midlen < 0)
 712                a_midlen = 0;
 713        if (b_midlen < 0)
 714                b_midlen = 0;
 715
 716        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 717        if (pfx_length + sfx_length) {
 718                strbuf_add(&name, a, pfx_length);
 719                strbuf_addch(&name, '{');
 720        }
 721        strbuf_add(&name, a + pfx_length, a_midlen);
 722        strbuf_addstr(&name, " => ");
 723        strbuf_add(&name, b + pfx_length, b_midlen);
 724        if (pfx_length + sfx_length) {
 725                strbuf_addch(&name, '}');
 726                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 727        }
 728        return strbuf_detach(&name, NULL);
 729}
 730
 731struct diffstat_t {
 732        struct xdiff_emit_state xm;
 733
 734        int nr;
 735        int alloc;
 736        struct diffstat_file {
 737                char *name;
 738                unsigned is_unmerged:1;
 739                unsigned is_binary:1;
 740                unsigned is_renamed:1;
 741                unsigned int added, deleted;
 742        } **files;
 743};
 744
 745static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 746                                          const char *name_a,
 747                                          const char *name_b)
 748{
 749        struct diffstat_file *x;
 750        x = xcalloc(sizeof (*x), 1);
 751        if (diffstat->nr == diffstat->alloc) {
 752                diffstat->alloc = alloc_nr(diffstat->alloc);
 753                diffstat->files = xrealloc(diffstat->files,
 754                                diffstat->alloc * sizeof(x));
 755        }
 756        diffstat->files[diffstat->nr++] = x;
 757        if (name_b) {
 758                x->name = pprint_rename(name_a, name_b);
 759                x->is_renamed = 1;
 760        }
 761        else
 762                x->name = xstrdup(name_a);
 763        return x;
 764}
 765
 766static void diffstat_consume(void *priv, char *line, unsigned long len)
 767{
 768        struct diffstat_t *diffstat = priv;
 769        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 770
 771        if (line[0] == '+')
 772                x->added++;
 773        else if (line[0] == '-')
 774                x->deleted++;
 775}
 776
 777const char mime_boundary_leader[] = "------------";
 778
 779static int scale_linear(int it, int width, int max_change)
 780{
 781        /*
 782         * make sure that at least one '-' is printed if there were deletions,
 783         * and likewise for '+'.
 784         */
 785        if (max_change < 2)
 786                return it;
 787        return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 788}
 789
 790static void show_name(const char *prefix, const char *name, int len,
 791                      const char *reset, const char *set)
 792{
 793        printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
 794}
 795
 796static void show_graph(char ch, int cnt, const char *set, const char *reset)
 797{
 798        if (cnt <= 0)
 799                return;
 800        printf("%s", set);
 801        while (cnt--)
 802                putchar(ch);
 803        printf("%s", reset);
 804}
 805
 806static void show_stats(struct diffstat_t* data, struct diff_options *options)
 807{
 808        int i, len, add, del, total, adds = 0, dels = 0;
 809        int max_change = 0, max_len = 0;
 810        int total_files = data->nr;
 811        int width, name_width;
 812        const char *reset, *set, *add_c, *del_c;
 813
 814        if (data->nr == 0)
 815                return;
 816
 817        width = options->stat_width ? options->stat_width : 80;
 818        name_width = options->stat_name_width ? options->stat_name_width : 50;
 819
 820        /* Sanity: give at least 5 columns to the graph,
 821         * but leave at least 10 columns for the name.
 822         */
 823        if (width < name_width + 15) {
 824                if (name_width <= 25)
 825                        width = name_width + 15;
 826                else
 827                        name_width = width - 15;
 828        }
 829
 830        /* Find the longest filename and max number of changes */
 831        reset = diff_get_color(options->color_diff, DIFF_RESET);
 832        set = diff_get_color(options->color_diff, DIFF_PLAIN);
 833        add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
 834        del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
 835
 836        for (i = 0; i < data->nr; i++) {
 837                struct diffstat_file *file = data->files[i];
 838                int change = file->added + file->deleted;
 839
 840                if (!file->is_renamed) {  /* renames are already quoted by pprint_rename */
 841                        struct strbuf buf;
 842                        strbuf_init(&buf, 0);
 843                        if (quote_c_style(file->name, &buf, NULL, 0)) {
 844                                free(file->name);
 845                                file->name = strbuf_detach(&buf, NULL);
 846                        } else {
 847                                strbuf_release(&buf);
 848                        }
 849                }
 850
 851                len = strlen(file->name);
 852                if (max_len < len)
 853                        max_len = len;
 854
 855                if (file->is_binary || file->is_unmerged)
 856                        continue;
 857                if (max_change < change)
 858                        max_change = change;
 859        }
 860
 861        /* Compute the width of the graph part;
 862         * 10 is for one blank at the beginning of the line plus
 863         * " | count " between the name and the graph.
 864         *
 865         * From here on, name_width is the width of the name area,
 866         * and width is the width of the graph area.
 867         */
 868        name_width = (name_width < max_len) ? name_width : max_len;
 869        if (width < (name_width + 10) + max_change)
 870                width = width - (name_width + 10);
 871        else
 872                width = max_change;
 873
 874        for (i = 0; i < data->nr; i++) {
 875                const char *prefix = "";
 876                char *name = data->files[i]->name;
 877                int added = data->files[i]->added;
 878                int deleted = data->files[i]->deleted;
 879                int name_len;
 880
 881                /*
 882                 * "scale" the filename
 883                 */
 884                len = name_width;
 885                name_len = strlen(name);
 886                if (name_width < name_len) {
 887                        char *slash;
 888                        prefix = "...";
 889                        len -= 3;
 890                        name += name_len - len;
 891                        slash = strchr(name, '/');
 892                        if (slash)
 893                                name = slash;
 894                }
 895
 896                if (data->files[i]->is_binary) {
 897                        show_name(prefix, name, len, reset, set);
 898                        printf("  Bin ");
 899                        printf("%s%d%s", del_c, deleted, reset);
 900                        printf(" -> ");
 901                        printf("%s%d%s", add_c, added, reset);
 902                        printf(" bytes");
 903                        printf("\n");
 904                        goto free_diffstat_file;
 905                }
 906                else if (data->files[i]->is_unmerged) {
 907                        show_name(prefix, name, len, reset, set);
 908                        printf("  Unmerged\n");
 909                        goto free_diffstat_file;
 910                }
 911                else if (!data->files[i]->is_renamed &&
 912                         (added + deleted == 0)) {
 913                        total_files--;
 914                        goto free_diffstat_file;
 915                }
 916
 917                /*
 918                 * scale the add/delete
 919                 */
 920                add = added;
 921                del = deleted;
 922                total = add + del;
 923                adds += add;
 924                dels += del;
 925
 926                if (width <= max_change) {
 927                        add = scale_linear(add, width, max_change);
 928                        del = scale_linear(del, width, max_change);
 929                        total = add + del;
 930                }
 931                show_name(prefix, name, len, reset, set);
 932                printf("%5d ", added + deleted);
 933                show_graph('+', add, add_c, reset);
 934                show_graph('-', del, del_c, reset);
 935                putchar('\n');
 936        free_diffstat_file:
 937                free(data->files[i]->name);
 938                free(data->files[i]);
 939        }
 940        free(data->files);
 941        printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 942               set, total_files, adds, dels, reset);
 943}
 944
 945static void show_shortstats(struct diffstat_t* data)
 946{
 947        int i, adds = 0, dels = 0, total_files = data->nr;
 948
 949        if (data->nr == 0)
 950                return;
 951
 952        for (i = 0; i < data->nr; i++) {
 953                if (!data->files[i]->is_binary &&
 954                    !data->files[i]->is_unmerged) {
 955                        int added = data->files[i]->added;
 956                        int deleted= data->files[i]->deleted;
 957                        if (!data->files[i]->is_renamed &&
 958                            (added + deleted == 0)) {
 959                                total_files--;
 960                        } else {
 961                                adds += added;
 962                                dels += deleted;
 963                        }
 964                }
 965                free(data->files[i]->name);
 966                free(data->files[i]);
 967        }
 968        free(data->files);
 969
 970        printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 971               total_files, adds, dels);
 972}
 973
 974static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 975{
 976        int i;
 977
 978        for (i = 0; i < data->nr; i++) {
 979                struct diffstat_file *file = data->files[i];
 980
 981                if (file->is_binary)
 982                        printf("-\t-\t");
 983                else
 984                        printf("%d\t%d\t", file->added, file->deleted);
 985                if (!file->is_renamed) {
 986                        write_name_quoted(file->name, stdout, options->line_termination);
 987                } else {
 988                        fputs(file->name, stdout);
 989                        putchar(options->line_termination);
 990                }
 991        }
 992}
 993
 994struct checkdiff_t {
 995        struct xdiff_emit_state xm;
 996        const char *filename;
 997        int lineno, color_diff;
 998        unsigned ws_rule;
 999};
1000
1001static void checkdiff_consume(void *priv, char *line, unsigned long len)
1002{
1003        struct checkdiff_t *data = priv;
1004        const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1005        const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1006        const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1007
1008        if (line[0] == '+') {
1009                int i, spaces = 0, space_before_tab = 0, white_space_at_end = 0;
1010
1011                /* check space before tab */
1012                for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
1013                        if (line[i] == ' ')
1014                                spaces++;
1015                if (line[i - 1] == '\t' && spaces)
1016                        space_before_tab = 1;
1017
1018                /* check white space at line end */
1019                if (line[len - 1] == '\n')
1020                        len--;
1021                if (isspace(line[len - 1]))
1022                        white_space_at_end = 1;
1023
1024                if (space_before_tab || white_space_at_end) {
1025                        printf("%s:%d: %s", data->filename, data->lineno, ws);
1026                        if (space_before_tab) {
1027                                printf("space before tab");
1028                                if (white_space_at_end)
1029                                        putchar(',');
1030                        }
1031                        if (white_space_at_end)
1032                                printf("white space at end");
1033                        printf(":%s ", reset);
1034                        emit_line_with_ws(1, set, reset, ws, line, len,
1035                                          data->ws_rule);
1036                }
1037
1038                data->lineno++;
1039        } else if (line[0] == ' ')
1040                data->lineno++;
1041        else if (line[0] == '@') {
1042                char *plus = strchr(line, '+');
1043                if (plus)
1044                        data->lineno = strtol(plus, NULL, 10);
1045                else
1046                        die("invalid diff");
1047        }
1048}
1049
1050static unsigned char *deflate_it(char *data,
1051                                 unsigned long size,
1052                                 unsigned long *result_size)
1053{
1054        int bound;
1055        unsigned char *deflated;
1056        z_stream stream;
1057
1058        memset(&stream, 0, sizeof(stream));
1059        deflateInit(&stream, zlib_compression_level);
1060        bound = deflateBound(&stream, size);
1061        deflated = xmalloc(bound);
1062        stream.next_out = deflated;
1063        stream.avail_out = bound;
1064
1065        stream.next_in = (unsigned char *)data;
1066        stream.avail_in = size;
1067        while (deflate(&stream, Z_FINISH) == Z_OK)
1068                ; /* nothing */
1069        deflateEnd(&stream);
1070        *result_size = stream.total_out;
1071        return deflated;
1072}
1073
1074static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1075{
1076        void *cp;
1077        void *delta;
1078        void *deflated;
1079        void *data;
1080        unsigned long orig_size;
1081        unsigned long delta_size;
1082        unsigned long deflate_size;
1083        unsigned long data_size;
1084
1085        /* We could do deflated delta, or we could do just deflated two,
1086         * whichever is smaller.
1087         */
1088        delta = NULL;
1089        deflated = deflate_it(two->ptr, two->size, &deflate_size);
1090        if (one->size && two->size) {
1091                delta = diff_delta(one->ptr, one->size,
1092                                   two->ptr, two->size,
1093                                   &delta_size, deflate_size);
1094                if (delta) {
1095                        void *to_free = delta;
1096                        orig_size = delta_size;
1097                        delta = deflate_it(delta, delta_size, &delta_size);
1098                        free(to_free);
1099                }
1100        }
1101
1102        if (delta && delta_size < deflate_size) {
1103                printf("delta %lu\n", orig_size);
1104                free(deflated);
1105                data = delta;
1106                data_size = delta_size;
1107        }
1108        else {
1109                printf("literal %lu\n", two->size);
1110                free(delta);
1111                data = deflated;
1112                data_size = deflate_size;
1113        }
1114
1115        /* emit data encoded in base85 */
1116        cp = data;
1117        while (data_size) {
1118                int bytes = (52 < data_size) ? 52 : data_size;
1119                char line[70];
1120                data_size -= bytes;
1121                if (bytes <= 26)
1122                        line[0] = bytes + 'A' - 1;
1123                else
1124                        line[0] = bytes - 26 + 'a' - 1;
1125                encode_85(line + 1, cp, bytes);
1126                cp = (char *) cp + bytes;
1127                puts(line);
1128        }
1129        printf("\n");
1130        free(data);
1131}
1132
1133static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1134{
1135        printf("GIT binary patch\n");
1136        emit_binary_diff_body(one, two);
1137        emit_binary_diff_body(two, one);
1138}
1139
1140static void setup_diff_attr_check(struct git_attr_check *check)
1141{
1142        static struct git_attr *attr_diff;
1143
1144        if (!attr_diff) {
1145                attr_diff = git_attr("diff", 4);
1146        }
1147        check[0].attr = attr_diff;
1148}
1149
1150static void diff_filespec_check_attr(struct diff_filespec *one)
1151{
1152        struct git_attr_check attr_diff_check;
1153        int check_from_data = 0;
1154
1155        if (one->checked_attr)
1156                return;
1157
1158        setup_diff_attr_check(&attr_diff_check);
1159        one->is_binary = 0;
1160        one->funcname_pattern_ident = NULL;
1161
1162        if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1163                const char *value;
1164
1165                /* binaryness */
1166                value = attr_diff_check.value;
1167                if (ATTR_TRUE(value))
1168                        ;
1169                else if (ATTR_FALSE(value))
1170                        one->is_binary = 1;
1171                else
1172                        check_from_data = 1;
1173
1174                /* funcname pattern ident */
1175                if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1176                        ;
1177                else
1178                        one->funcname_pattern_ident = value;
1179        }
1180
1181        if (check_from_data) {
1182                if (!one->data && DIFF_FILE_VALID(one))
1183                        diff_populate_filespec(one, 0);
1184
1185                if (one->data)
1186                        one->is_binary = buffer_is_binary(one->data, one->size);
1187        }
1188}
1189
1190int diff_filespec_is_binary(struct diff_filespec *one)
1191{
1192        diff_filespec_check_attr(one);
1193        return one->is_binary;
1194}
1195
1196static const char *funcname_pattern(const char *ident)
1197{
1198        struct funcname_pattern *pp;
1199
1200        read_config_if_needed();
1201        for (pp = funcname_pattern_list; pp; pp = pp->next)
1202                if (!strcmp(ident, pp->name))
1203                        return pp->pattern;
1204        return NULL;
1205}
1206
1207static struct builtin_funcname_pattern {
1208        const char *name;
1209        const char *pattern;
1210} builtin_funcname_pattern[] = {
1211        { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1212                        "new\\|return\\|switch\\|throw\\|while\\)\n"
1213                        "^[     ]*\\(\\([       ]*"
1214                        "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1215                        "[      ]*([^;]*$\\)" },
1216        { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1217};
1218
1219static const char *diff_funcname_pattern(struct diff_filespec *one)
1220{
1221        const char *ident, *pattern;
1222        int i;
1223
1224        diff_filespec_check_attr(one);
1225        ident = one->funcname_pattern_ident;
1226
1227        if (!ident)
1228                /*
1229                 * If the config file has "funcname.default" defined, that
1230                 * regexp is used; otherwise NULL is returned and xemit uses
1231                 * the built-in default.
1232                 */
1233                return funcname_pattern("default");
1234
1235        /* Look up custom "funcname.$ident" regexp from config. */
1236        pattern = funcname_pattern(ident);
1237        if (pattern)
1238                return pattern;
1239
1240        /*
1241         * And define built-in fallback patterns here.  Note that
1242         * these can be overriden by the user's config settings.
1243         */
1244        for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1245                if (!strcmp(ident, builtin_funcname_pattern[i].name))
1246                        return builtin_funcname_pattern[i].pattern;
1247
1248        return NULL;
1249}
1250
1251static void builtin_diff(const char *name_a,
1252                         const char *name_b,
1253                         struct diff_filespec *one,
1254                         struct diff_filespec *two,
1255                         const char *xfrm_msg,
1256                         struct diff_options *o,
1257                         int complete_rewrite)
1258{
1259        mmfile_t mf1, mf2;
1260        const char *lbl[2];
1261        char *a_one, *b_two;
1262        const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
1263        const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
1264
1265        a_one = quote_two("a/", name_a + (*name_a == '/'));
1266        b_two = quote_two("b/", name_b + (*name_b == '/'));
1267        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1268        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1269        printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1270        if (lbl[0][0] == '/') {
1271                /* /dev/null */
1272                printf("%snew file mode %06o%s\n", set, two->mode, reset);
1273                if (xfrm_msg && xfrm_msg[0])
1274                        printf("%s%s%s\n", set, xfrm_msg, reset);
1275        }
1276        else if (lbl[1][0] == '/') {
1277                printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1278                if (xfrm_msg && xfrm_msg[0])
1279                        printf("%s%s%s\n", set, xfrm_msg, reset);
1280        }
1281        else {
1282                if (one->mode != two->mode) {
1283                        printf("%sold mode %06o%s\n", set, one->mode, reset);
1284                        printf("%snew mode %06o%s\n", set, two->mode, reset);
1285                }
1286                if (xfrm_msg && xfrm_msg[0])
1287                        printf("%s%s%s\n", set, xfrm_msg, reset);
1288                /*
1289                 * we do not run diff between different kind
1290                 * of objects.
1291                 */
1292                if ((one->mode ^ two->mode) & S_IFMT)
1293                        goto free_ab_and_return;
1294                if (complete_rewrite) {
1295                        emit_rewrite_diff(name_a, name_b, one, two,
1296                                        o->color_diff);
1297                        o->found_changes = 1;
1298                        goto free_ab_and_return;
1299                }
1300        }
1301
1302        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1303                die("unable to read files to diff");
1304
1305        if (!o->text &&
1306            (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1307                /* Quite common confusing case */
1308                if (mf1.size == mf2.size &&
1309                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1310                        goto free_ab_and_return;
1311                if (o->binary)
1312                        emit_binary_diff(&mf1, &mf2);
1313                else
1314                        printf("Binary files %s and %s differ\n",
1315                               lbl[0], lbl[1]);
1316                o->found_changes = 1;
1317        }
1318        else {
1319                /* Crazy xdl interfaces.. */
1320                const char *diffopts = getenv("GIT_DIFF_OPTS");
1321                xpparam_t xpp;
1322                xdemitconf_t xecfg;
1323                xdemitcb_t ecb;
1324                struct emit_callback ecbdata;
1325                const char *funcname_pattern;
1326
1327                funcname_pattern = diff_funcname_pattern(one);
1328                if (!funcname_pattern)
1329                        funcname_pattern = diff_funcname_pattern(two);
1330
1331                memset(&xecfg, 0, sizeof(xecfg));
1332                memset(&ecbdata, 0, sizeof(ecbdata));
1333                ecbdata.label_path = lbl;
1334                ecbdata.color_diff = o->color_diff;
1335                ecbdata.found_changesp = &o->found_changes;
1336                ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1337                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1338                xecfg.ctxlen = o->context;
1339                xecfg.flags = XDL_EMIT_FUNCNAMES;
1340                if (funcname_pattern)
1341                        xdiff_set_find_func(&xecfg, funcname_pattern);
1342                if (!diffopts)
1343                        ;
1344                else if (!prefixcmp(diffopts, "--unified="))
1345                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1346                else if (!prefixcmp(diffopts, "-u"))
1347                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1348                ecb.outf = xdiff_outf;
1349                ecb.priv = &ecbdata;
1350                ecbdata.xm.consume = fn_out_consume;
1351                if (o->color_diff_words)
1352                        ecbdata.diff_words =
1353                                xcalloc(1, sizeof(struct diff_words_data));
1354                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1355                if (o->color_diff_words)
1356                        free_diff_words_data(&ecbdata);
1357        }
1358
1359 free_ab_and_return:
1360        diff_free_filespec_data(one);
1361        diff_free_filespec_data(two);
1362        free(a_one);
1363        free(b_two);
1364        return;
1365}
1366
1367static void builtin_diffstat(const char *name_a, const char *name_b,
1368                             struct diff_filespec *one,
1369                             struct diff_filespec *two,
1370                             struct diffstat_t *diffstat,
1371                             struct diff_options *o,
1372                             int complete_rewrite)
1373{
1374        mmfile_t mf1, mf2;
1375        struct diffstat_file *data;
1376
1377        data = diffstat_add(diffstat, name_a, name_b);
1378
1379        if (!one || !two) {
1380                data->is_unmerged = 1;
1381                return;
1382        }
1383        if (complete_rewrite) {
1384                diff_populate_filespec(one, 0);
1385                diff_populate_filespec(two, 0);
1386                data->deleted = count_lines(one->data, one->size);
1387                data->added = count_lines(two->data, two->size);
1388                goto free_and_return;
1389        }
1390        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1391                die("unable to read files to diff");
1392
1393        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1394                data->is_binary = 1;
1395                data->added = mf2.size;
1396                data->deleted = mf1.size;
1397        } else {
1398                /* Crazy xdl interfaces.. */
1399                xpparam_t xpp;
1400                xdemitconf_t xecfg;
1401                xdemitcb_t ecb;
1402
1403                memset(&xecfg, 0, sizeof(xecfg));
1404                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1405                ecb.outf = xdiff_outf;
1406                ecb.priv = diffstat;
1407                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1408        }
1409
1410 free_and_return:
1411        diff_free_filespec_data(one);
1412        diff_free_filespec_data(two);
1413}
1414
1415static void builtin_checkdiff(const char *name_a, const char *name_b,
1416                             struct diff_filespec *one,
1417                             struct diff_filespec *two, struct diff_options *o)
1418{
1419        mmfile_t mf1, mf2;
1420        struct checkdiff_t data;
1421
1422        if (!two)
1423                return;
1424
1425        memset(&data, 0, sizeof(data));
1426        data.xm.consume = checkdiff_consume;
1427        data.filename = name_b ? name_b : name_a;
1428        data.lineno = 0;
1429        data.color_diff = o->color_diff;
1430        data.ws_rule = whitespace_rule(data.filename);
1431
1432        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1433                die("unable to read files to diff");
1434
1435        if (diff_filespec_is_binary(two))
1436                goto free_and_return;
1437        else {
1438                /* Crazy xdl interfaces.. */
1439                xpparam_t xpp;
1440                xdemitconf_t xecfg;
1441                xdemitcb_t ecb;
1442
1443                memset(&xecfg, 0, sizeof(xecfg));
1444                xpp.flags = XDF_NEED_MINIMAL;
1445                ecb.outf = xdiff_outf;
1446                ecb.priv = &data;
1447                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1448        }
1449 free_and_return:
1450        diff_free_filespec_data(one);
1451        diff_free_filespec_data(two);
1452}
1453
1454struct diff_filespec *alloc_filespec(const char *path)
1455{
1456        int namelen = strlen(path);
1457        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1458
1459        memset(spec, 0, sizeof(*spec));
1460        spec->path = (char *)(spec + 1);
1461        memcpy(spec->path, path, namelen+1);
1462        spec->count = 1;
1463        return spec;
1464}
1465
1466void free_filespec(struct diff_filespec *spec)
1467{
1468        if (!--spec->count) {
1469                diff_free_filespec_data(spec);
1470                free(spec);
1471        }
1472}
1473
1474void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1475                   unsigned short mode)
1476{
1477        if (mode) {
1478                spec->mode = canon_mode(mode);
1479                hashcpy(spec->sha1, sha1);
1480                spec->sha1_valid = !is_null_sha1(sha1);
1481        }
1482}
1483
1484/*
1485 * Given a name and sha1 pair, if the index tells us the file in
1486 * the work tree has that object contents, return true, so that
1487 * prepare_temp_file() does not have to inflate and extract.
1488 */
1489static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1490{
1491        struct cache_entry *ce;
1492        struct stat st;
1493        int pos, len;
1494
1495        /* We do not read the cache ourselves here, because the
1496         * benchmark with my previous version that always reads cache
1497         * shows that it makes things worse for diff-tree comparing
1498         * two linux-2.6 kernel trees in an already checked out work
1499         * tree.  This is because most diff-tree comparisons deal with
1500         * only a small number of files, while reading the cache is
1501         * expensive for a large project, and its cost outweighs the
1502         * savings we get by not inflating the object to a temporary
1503         * file.  Practically, this code only helps when we are used
1504         * by diff-cache --cached, which does read the cache before
1505         * calling us.
1506         */
1507        if (!active_cache)
1508                return 0;
1509
1510        /* We want to avoid the working directory if our caller
1511         * doesn't need the data in a normal file, this system
1512         * is rather slow with its stat/open/mmap/close syscalls,
1513         * and the object is contained in a pack file.  The pack
1514         * is probably already open and will be faster to obtain
1515         * the data through than the working directory.  Loose
1516         * objects however would tend to be slower as they need
1517         * to be individually opened and inflated.
1518         */
1519        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1520                return 0;
1521
1522        len = strlen(name);
1523        pos = cache_name_pos(name, len);
1524        if (pos < 0)
1525                return 0;
1526        ce = active_cache[pos];
1527        if ((lstat(name, &st) < 0) ||
1528            !S_ISREG(st.st_mode) || /* careful! */
1529            ce_match_stat(ce, &st, 0) ||
1530            hashcmp(sha1, ce->sha1))
1531                return 0;
1532        /* we return 1 only when we can stat, it is a regular file,
1533         * stat information matches, and sha1 recorded in the cache
1534         * matches.  I.e. we know the file in the work tree really is
1535         * the same as the <name, sha1> pair.
1536         */
1537        return 1;
1538}
1539
1540static int populate_from_stdin(struct diff_filespec *s)
1541{
1542        struct strbuf buf;
1543        size_t size = 0;
1544
1545        strbuf_init(&buf, 0);
1546        if (strbuf_read(&buf, 0, 0) < 0)
1547                return error("error while reading from stdin %s",
1548                                     strerror(errno));
1549
1550        s->should_munmap = 0;
1551        s->data = strbuf_detach(&buf, &size);
1552        s->size = size;
1553        s->should_free = 1;
1554        return 0;
1555}
1556
1557static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1558{
1559        int len;
1560        char *data = xmalloc(100);
1561        len = snprintf(data, 100,
1562                "Subproject commit %s\n", sha1_to_hex(s->sha1));
1563        s->data = data;
1564        s->size = len;
1565        s->should_free = 1;
1566        if (size_only) {
1567                s->data = NULL;
1568                free(data);
1569        }
1570        return 0;
1571}
1572
1573/*
1574 * While doing rename detection and pickaxe operation, we may need to
1575 * grab the data for the blob (or file) for our own in-core comparison.
1576 * diff_filespec has data and size fields for this purpose.
1577 */
1578int diff_populate_filespec(struct diff_filespec *s, int size_only)
1579{
1580        int err = 0;
1581        if (!DIFF_FILE_VALID(s))
1582                die("internal error: asking to populate invalid file.");
1583        if (S_ISDIR(s->mode))
1584                return -1;
1585
1586        if (s->data)
1587                return 0;
1588
1589        if (size_only && 0 < s->size)
1590                return 0;
1591
1592        if (S_ISGITLINK(s->mode))
1593                return diff_populate_gitlink(s, size_only);
1594
1595        if (!s->sha1_valid ||
1596            reuse_worktree_file(s->path, s->sha1, 0)) {
1597                struct strbuf buf;
1598                struct stat st;
1599                int fd;
1600
1601                if (!strcmp(s->path, "-"))
1602                        return populate_from_stdin(s);
1603
1604                if (lstat(s->path, &st) < 0) {
1605                        if (errno == ENOENT) {
1606                        err_empty:
1607                                err = -1;
1608                        empty:
1609                                s->data = (char *)"";
1610                                s->size = 0;
1611                                return err;
1612                        }
1613                }
1614                s->size = xsize_t(st.st_size);
1615                if (!s->size)
1616                        goto empty;
1617                if (size_only)
1618                        return 0;
1619                if (S_ISLNK(st.st_mode)) {
1620                        int ret;
1621                        s->data = xmalloc(s->size);
1622                        s->should_free = 1;
1623                        ret = readlink(s->path, s->data, s->size);
1624                        if (ret < 0) {
1625                                free(s->data);
1626                                goto err_empty;
1627                        }
1628                        return 0;
1629                }
1630                fd = open(s->path, O_RDONLY);
1631                if (fd < 0)
1632                        goto err_empty;
1633                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1634                close(fd);
1635                s->should_munmap = 1;
1636
1637                /*
1638                 * Convert from working tree format to canonical git format
1639                 */
1640                strbuf_init(&buf, 0);
1641                if (convert_to_git(s->path, s->data, s->size, &buf)) {
1642                        size_t size = 0;
1643                        munmap(s->data, s->size);
1644                        s->should_munmap = 0;
1645                        s->data = strbuf_detach(&buf, &size);
1646                        s->size = size;
1647                        s->should_free = 1;
1648                }
1649        }
1650        else {
1651                enum object_type type;
1652                if (size_only)
1653                        type = sha1_object_info(s->sha1, &s->size);
1654                else {
1655                        s->data = read_sha1_file(s->sha1, &type, &s->size);
1656                        s->should_free = 1;
1657                }
1658        }
1659        return 0;
1660}
1661
1662void diff_free_filespec_blob(struct diff_filespec *s)
1663{
1664        if (s->should_free)
1665                free(s->data);
1666        else if (s->should_munmap)
1667                munmap(s->data, s->size);
1668
1669        if (s->should_free || s->should_munmap) {
1670                s->should_free = s->should_munmap = 0;
1671                s->data = NULL;
1672        }
1673}
1674
1675void diff_free_filespec_data(struct diff_filespec *s)
1676{
1677        diff_free_filespec_blob(s);
1678        free(s->cnt_data);
1679        s->cnt_data = NULL;
1680}
1681
1682static void prep_temp_blob(struct diff_tempfile *temp,
1683                           void *blob,
1684                           unsigned long size,
1685                           const unsigned char *sha1,
1686                           int mode)
1687{
1688        int fd;
1689
1690        fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1691        if (fd < 0)
1692                die("unable to create temp-file: %s", strerror(errno));
1693        if (write_in_full(fd, blob, size) != size)
1694                die("unable to write temp-file");
1695        close(fd);
1696        temp->name = temp->tmp_path;
1697        strcpy(temp->hex, sha1_to_hex(sha1));
1698        temp->hex[40] = 0;
1699        sprintf(temp->mode, "%06o", mode);
1700}
1701
1702static void prepare_temp_file(const char *name,
1703                              struct diff_tempfile *temp,
1704                              struct diff_filespec *one)
1705{
1706        if (!DIFF_FILE_VALID(one)) {
1707        not_a_valid_file:
1708                /* A '-' entry produces this for file-2, and
1709                 * a '+' entry produces this for file-1.
1710                 */
1711                temp->name = "/dev/null";
1712                strcpy(temp->hex, ".");
1713                strcpy(temp->mode, ".");
1714                return;
1715        }
1716
1717        if (!one->sha1_valid ||
1718            reuse_worktree_file(name, one->sha1, 1)) {
1719                struct stat st;
1720                if (lstat(name, &st) < 0) {
1721                        if (errno == ENOENT)
1722                                goto not_a_valid_file;
1723                        die("stat(%s): %s", name, strerror(errno));
1724                }
1725                if (S_ISLNK(st.st_mode)) {
1726                        int ret;
1727                        char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1728                        size_t sz = xsize_t(st.st_size);
1729                        if (sizeof(buf) <= st.st_size)
1730                                die("symlink too long: %s", name);
1731                        ret = readlink(name, buf, sz);
1732                        if (ret < 0)
1733                                die("readlink(%s)", name);
1734                        prep_temp_blob(temp, buf, sz,
1735                                       (one->sha1_valid ?
1736                                        one->sha1 : null_sha1),
1737                                       (one->sha1_valid ?
1738                                        one->mode : S_IFLNK));
1739                }
1740                else {
1741                        /* we can borrow from the file in the work tree */
1742                        temp->name = name;
1743                        if (!one->sha1_valid)
1744                                strcpy(temp->hex, sha1_to_hex(null_sha1));
1745                        else
1746                                strcpy(temp->hex, sha1_to_hex(one->sha1));
1747                        /* Even though we may sometimes borrow the
1748                         * contents from the work tree, we always want
1749                         * one->mode.  mode is trustworthy even when
1750                         * !(one->sha1_valid), as long as
1751                         * DIFF_FILE_VALID(one).
1752                         */
1753                        sprintf(temp->mode, "%06o", one->mode);
1754                }
1755                return;
1756        }
1757        else {
1758                if (diff_populate_filespec(one, 0))
1759                        die("cannot read data blob for %s", one->path);
1760                prep_temp_blob(temp, one->data, one->size,
1761                               one->sha1, one->mode);
1762        }
1763}
1764
1765static void remove_tempfile(void)
1766{
1767        int i;
1768
1769        for (i = 0; i < 2; i++)
1770                if (diff_temp[i].name == diff_temp[i].tmp_path) {
1771                        unlink(diff_temp[i].name);
1772                        diff_temp[i].name = NULL;
1773                }
1774}
1775
1776static void remove_tempfile_on_signal(int signo)
1777{
1778        remove_tempfile();
1779        signal(SIGINT, SIG_DFL);
1780        raise(signo);
1781}
1782
1783/* An external diff command takes:
1784 *
1785 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1786 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1787 *
1788 */
1789static void run_external_diff(const char *pgm,
1790                              const char *name,
1791                              const char *other,
1792                              struct diff_filespec *one,
1793                              struct diff_filespec *two,
1794                              const char *xfrm_msg,
1795                              int complete_rewrite)
1796{
1797        const char *spawn_arg[10];
1798        struct diff_tempfile *temp = diff_temp;
1799        int retval;
1800        static int atexit_asked = 0;
1801        const char *othername;
1802        const char **arg = &spawn_arg[0];
1803
1804        othername = (other? other : name);
1805        if (one && two) {
1806                prepare_temp_file(name, &temp[0], one);
1807                prepare_temp_file(othername, &temp[1], two);
1808                if (! atexit_asked &&
1809                    (temp[0].name == temp[0].tmp_path ||
1810                     temp[1].name == temp[1].tmp_path)) {
1811                        atexit_asked = 1;
1812                        atexit(remove_tempfile);
1813                }
1814                signal(SIGINT, remove_tempfile_on_signal);
1815        }
1816
1817        if (one && two) {
1818                *arg++ = pgm;
1819                *arg++ = name;
1820                *arg++ = temp[0].name;
1821                *arg++ = temp[0].hex;
1822                *arg++ = temp[0].mode;
1823                *arg++ = temp[1].name;
1824                *arg++ = temp[1].hex;
1825                *arg++ = temp[1].mode;
1826                if (other) {
1827                        *arg++ = other;
1828                        *arg++ = xfrm_msg;
1829                }
1830        } else {
1831                *arg++ = pgm;
1832                *arg++ = name;
1833        }
1834        *arg = NULL;
1835        fflush(NULL);
1836        retval = run_command_v_opt(spawn_arg, 0);
1837        remove_tempfile();
1838        if (retval) {
1839                fprintf(stderr, "external diff died, stopping at %s.\n", name);
1840                exit(1);
1841        }
1842}
1843
1844static const char *external_diff_attr(const char *name)
1845{
1846        struct git_attr_check attr_diff_check;
1847
1848        setup_diff_attr_check(&attr_diff_check);
1849        if (!git_checkattr(name, 1, &attr_diff_check)) {
1850                const char *value = attr_diff_check.value;
1851                if (!ATTR_TRUE(value) &&
1852                    !ATTR_FALSE(value) &&
1853                    !ATTR_UNSET(value)) {
1854                        struct ll_diff_driver *drv;
1855
1856                        read_config_if_needed();
1857                        for (drv = user_diff; drv; drv = drv->next)
1858                                if (!strcmp(drv->name, value))
1859                                        return drv->cmd;
1860                }
1861        }
1862        return NULL;
1863}
1864
1865static void run_diff_cmd(const char *pgm,
1866                         const char *name,
1867                         const char *other,
1868                         struct diff_filespec *one,
1869                         struct diff_filespec *two,
1870                         const char *xfrm_msg,
1871                         struct diff_options *o,
1872                         int complete_rewrite)
1873{
1874        if (!o->allow_external)
1875                pgm = NULL;
1876        else {
1877                const char *cmd = external_diff_attr(name);
1878                if (cmd)
1879                        pgm = cmd;
1880        }
1881
1882        if (pgm) {
1883                run_external_diff(pgm, name, other, one, two, xfrm_msg,
1884                                  complete_rewrite);
1885                return;
1886        }
1887        if (one && two)
1888                builtin_diff(name, other ? other : name,
1889                             one, two, xfrm_msg, o, complete_rewrite);
1890        else
1891                printf("* Unmerged path %s\n", name);
1892}
1893
1894static void diff_fill_sha1_info(struct diff_filespec *one)
1895{
1896        if (DIFF_FILE_VALID(one)) {
1897                if (!one->sha1_valid) {
1898                        struct stat st;
1899                        if (!strcmp(one->path, "-")) {
1900                                hashcpy(one->sha1, null_sha1);
1901                                return;
1902                        }
1903                        if (lstat(one->path, &st) < 0)
1904                                die("stat %s", one->path);
1905                        if (index_path(one->sha1, one->path, &st, 0))
1906                                die("cannot hash %s\n", one->path);
1907                }
1908        }
1909        else
1910                hashclr(one->sha1);
1911}
1912
1913static int similarity_index(struct diff_filepair *p)
1914{
1915        return p->score * 100 / MAX_SCORE;
1916}
1917
1918static void run_diff(struct diff_filepair *p, struct diff_options *o)
1919{
1920        const char *pgm = external_diff();
1921        struct strbuf msg;
1922        char *xfrm_msg;
1923        struct diff_filespec *one = p->one;
1924        struct diff_filespec *two = p->two;
1925        const char *name;
1926        const char *other;
1927        int complete_rewrite = 0;
1928
1929
1930        if (DIFF_PAIR_UNMERGED(p)) {
1931                run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1932                return;
1933        }
1934
1935        name  = p->one->path;
1936        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1937        diff_fill_sha1_info(one);
1938        diff_fill_sha1_info(two);
1939
1940        strbuf_init(&msg, PATH_MAX * 2 + 300);
1941        switch (p->status) {
1942        case DIFF_STATUS_COPIED:
1943                strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1944                strbuf_addstr(&msg, "\ncopy from ");
1945                quote_c_style(name, &msg, NULL, 0);
1946                strbuf_addstr(&msg, "\ncopy to ");
1947                quote_c_style(other, &msg, NULL, 0);
1948                strbuf_addch(&msg, '\n');
1949                break;
1950        case DIFF_STATUS_RENAMED:
1951                strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1952                strbuf_addstr(&msg, "\nrename from ");
1953                quote_c_style(name, &msg, NULL, 0);
1954                strbuf_addstr(&msg, "\nrename to ");
1955                quote_c_style(other, &msg, NULL, 0);
1956                strbuf_addch(&msg, '\n');
1957                break;
1958        case DIFF_STATUS_MODIFIED:
1959                if (p->score) {
1960                        strbuf_addf(&msg, "dissimilarity index %d%%\n",
1961                                        similarity_index(p));
1962                        complete_rewrite = 1;
1963                        break;
1964                }
1965                /* fallthru */
1966        default:
1967                /* nothing */
1968                ;
1969        }
1970
1971        if (hashcmp(one->sha1, two->sha1)) {
1972                int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1973
1974                if (o->binary) {
1975                        mmfile_t mf;
1976                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
1977                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
1978                                abbrev = 40;
1979                }
1980                strbuf_addf(&msg, "index %.*s..%.*s",
1981                                abbrev, sha1_to_hex(one->sha1),
1982                                abbrev, sha1_to_hex(two->sha1));
1983                if (one->mode == two->mode)
1984                        strbuf_addf(&msg, " %06o", one->mode);
1985                strbuf_addch(&msg, '\n');
1986        }
1987
1988        if (msg.len)
1989                strbuf_setlen(&msg, msg.len - 1);
1990        xfrm_msg = msg.len ? msg.buf : NULL;
1991
1992        if (!pgm &&
1993            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1994            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1995                /* a filepair that changes between file and symlink
1996                 * needs to be split into deletion and creation.
1997                 */
1998                struct diff_filespec *null = alloc_filespec(two->path);
1999                run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
2000                free(null);
2001                null = alloc_filespec(one->path);
2002                run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
2003                free(null);
2004        }
2005        else
2006                run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
2007                             complete_rewrite);
2008
2009        strbuf_release(&msg);
2010}
2011
2012static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2013                         struct diffstat_t *diffstat)
2014{
2015        const char *name;
2016        const char *other;
2017        int complete_rewrite = 0;
2018
2019        if (DIFF_PAIR_UNMERGED(p)) {
2020                /* unmerged */
2021                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2022                return;
2023        }
2024
2025        name = p->one->path;
2026        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2027
2028        diff_fill_sha1_info(p->one);
2029        diff_fill_sha1_info(p->two);
2030
2031        if (p->status == DIFF_STATUS_MODIFIED && p->score)
2032                complete_rewrite = 1;
2033        builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2034}
2035
2036static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2037{
2038        const char *name;
2039        const char *other;
2040
2041        if (DIFF_PAIR_UNMERGED(p)) {
2042                /* unmerged */
2043                return;
2044        }
2045
2046        name = p->one->path;
2047        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2048
2049        diff_fill_sha1_info(p->one);
2050        diff_fill_sha1_info(p->two);
2051
2052        builtin_checkdiff(name, other, p->one, p->two, o);
2053}
2054
2055void diff_setup(struct diff_options *options)
2056{
2057        memset(options, 0, sizeof(*options));
2058        options->line_termination = '\n';
2059        options->break_opt = -1;
2060        options->rename_limit = -1;
2061        options->context = 3;
2062        options->msg_sep = "";
2063
2064        options->change = diff_change;
2065        options->add_remove = diff_addremove;
2066        options->color_diff = diff_use_color_default;
2067        options->detect_rename = diff_detect_rename_default;
2068}
2069
2070int diff_setup_done(struct diff_options *options)
2071{
2072        int count = 0;
2073
2074        if (options->output_format & DIFF_FORMAT_NAME)
2075                count++;
2076        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2077                count++;
2078        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2079                count++;
2080        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2081                count++;
2082        if (count > 1)
2083                die("--name-only, --name-status, --check and -s are mutually exclusive");
2084
2085        if (options->find_copies_harder)
2086                options->detect_rename = DIFF_DETECT_COPY;
2087
2088        if (options->output_format & (DIFF_FORMAT_NAME |
2089                                      DIFF_FORMAT_NAME_STATUS |
2090                                      DIFF_FORMAT_CHECKDIFF |
2091                                      DIFF_FORMAT_NO_OUTPUT))
2092                options->output_format &= ~(DIFF_FORMAT_RAW |
2093                                            DIFF_FORMAT_NUMSTAT |
2094                                            DIFF_FORMAT_DIFFSTAT |
2095                                            DIFF_FORMAT_SHORTSTAT |
2096                                            DIFF_FORMAT_SUMMARY |
2097                                            DIFF_FORMAT_PATCH);
2098
2099        /*
2100         * These cases always need recursive; we do not drop caller-supplied
2101         * recursive bits for other formats here.
2102         */
2103        if (options->output_format & (DIFF_FORMAT_PATCH |
2104                                      DIFF_FORMAT_NUMSTAT |
2105                                      DIFF_FORMAT_DIFFSTAT |
2106                                      DIFF_FORMAT_SHORTSTAT |
2107                                      DIFF_FORMAT_SUMMARY |
2108                                      DIFF_FORMAT_CHECKDIFF))
2109                options->recursive = 1;
2110        /*
2111         * Also pickaxe would not work very well if you do not say recursive
2112         */
2113        if (options->pickaxe)
2114                options->recursive = 1;
2115
2116        if (options->detect_rename && options->rename_limit < 0)
2117                options->rename_limit = diff_rename_limit_default;
2118        if (options->setup & DIFF_SETUP_USE_CACHE) {
2119                if (!active_cache)
2120                        /* read-cache does not die even when it fails
2121                         * so it is safe for us to do this here.  Also
2122                         * it does not smudge active_cache or active_nr
2123                         * when it fails, so we do not have to worry about
2124                         * cleaning it up ourselves either.
2125                         */
2126                        read_cache();
2127        }
2128        if (options->abbrev <= 0 || 40 < options->abbrev)
2129                options->abbrev = 40; /* full */
2130
2131        /*
2132         * It does not make sense to show the first hit we happened
2133         * to have found.  It does not make sense not to return with
2134         * exit code in such a case either.
2135         */
2136        if (options->quiet) {
2137                options->output_format = DIFF_FORMAT_NO_OUTPUT;
2138                options->exit_with_status = 1;
2139        }
2140
2141        /*
2142         * If we postprocess in diffcore, we cannot simply return
2143         * upon the first hit.  We need to run diff as usual.
2144         */
2145        if (options->pickaxe || options->filter)
2146                options->quiet = 0;
2147
2148        return 0;
2149}
2150
2151static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2152{
2153        char c, *eq;
2154        int len;
2155
2156        if (*arg != '-')
2157                return 0;
2158        c = *++arg;
2159        if (!c)
2160                return 0;
2161        if (c == arg_short) {
2162                c = *++arg;
2163                if (!c)
2164                        return 1;
2165                if (val && isdigit(c)) {
2166                        char *end;
2167                        int n = strtoul(arg, &end, 10);
2168                        if (*end)
2169                                return 0;
2170                        *val = n;
2171                        return 1;
2172                }
2173                return 0;
2174        }
2175        if (c != '-')
2176                return 0;
2177        arg++;
2178        eq = strchr(arg, '=');
2179        if (eq)
2180                len = eq - arg;
2181        else
2182                len = strlen(arg);
2183        if (!len || strncmp(arg, arg_long, len))
2184                return 0;
2185        if (eq) {
2186                int n;
2187                char *end;
2188                if (!isdigit(*++eq))
2189                        return 0;
2190                n = strtoul(eq, &end, 10);
2191                if (*end)
2192                        return 0;
2193                *val = n;
2194        }
2195        return 1;
2196}
2197
2198static int diff_scoreopt_parse(const char *opt);
2199
2200int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2201{
2202        const char *arg = av[0];
2203        if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2204                options->output_format |= DIFF_FORMAT_PATCH;
2205        else if (opt_arg(arg, 'U', "unified", &options->context))
2206                options->output_format |= DIFF_FORMAT_PATCH;
2207        else if (!strcmp(arg, "--raw"))
2208                options->output_format |= DIFF_FORMAT_RAW;
2209        else if (!strcmp(arg, "--patch-with-raw")) {
2210                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2211        }
2212        else if (!strcmp(arg, "--numstat")) {
2213                options->output_format |= DIFF_FORMAT_NUMSTAT;
2214        }
2215        else if (!strcmp(arg, "--shortstat")) {
2216                options->output_format |= DIFF_FORMAT_SHORTSTAT;
2217        }
2218        else if (!prefixcmp(arg, "--stat")) {
2219                char *end;
2220                int width = options->stat_width;
2221                int name_width = options->stat_name_width;
2222                arg += 6;
2223                end = (char *)arg;
2224
2225                switch (*arg) {
2226                case '-':
2227                        if (!prefixcmp(arg, "-width="))
2228                                width = strtoul(arg + 7, &end, 10);
2229                        else if (!prefixcmp(arg, "-name-width="))
2230                                name_width = strtoul(arg + 12, &end, 10);
2231                        break;
2232                case '=':
2233                        width = strtoul(arg+1, &end, 10);
2234                        if (*end == ',')
2235                                name_width = strtoul(end+1, &end, 10);
2236                }
2237
2238                /* Important! This checks all the error cases! */
2239                if (*end)
2240                        return 0;
2241                options->output_format |= DIFF_FORMAT_DIFFSTAT;
2242                options->stat_name_width = name_width;
2243                options->stat_width = width;
2244        }
2245        else if (!strcmp(arg, "--check"))
2246                options->output_format |= DIFF_FORMAT_CHECKDIFF;
2247        else if (!strcmp(arg, "--summary"))
2248                options->output_format |= DIFF_FORMAT_SUMMARY;
2249        else if (!strcmp(arg, "--patch-with-stat")) {
2250                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2251        }
2252        else if (!strcmp(arg, "-z"))
2253                options->line_termination = 0;
2254        else if (!prefixcmp(arg, "-l"))
2255                options->rename_limit = strtoul(arg+2, NULL, 10);
2256        else if (!strcmp(arg, "--full-index"))
2257                options->full_index = 1;
2258        else if (!strcmp(arg, "--binary")) {
2259                options->output_format |= DIFF_FORMAT_PATCH;
2260                options->binary = 1;
2261        }
2262        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
2263                options->text = 1;
2264        }
2265        else if (!strcmp(arg, "--name-only"))
2266                options->output_format |= DIFF_FORMAT_NAME;
2267        else if (!strcmp(arg, "--name-status"))
2268                options->output_format |= DIFF_FORMAT_NAME_STATUS;
2269        else if (!strcmp(arg, "-R"))
2270                options->reverse_diff = 1;
2271        else if (!prefixcmp(arg, "-S"))
2272                options->pickaxe = arg + 2;
2273        else if (!strcmp(arg, "-s")) {
2274                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2275        }
2276        else if (!prefixcmp(arg, "-O"))
2277                options->orderfile = arg + 2;
2278        else if (!prefixcmp(arg, "--diff-filter="))
2279                options->filter = arg + 14;
2280        else if (!strcmp(arg, "--pickaxe-all"))
2281                options->pickaxe_opts = DIFF_PICKAXE_ALL;
2282        else if (!strcmp(arg, "--pickaxe-regex"))
2283                options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2284        else if (!prefixcmp(arg, "-B")) {
2285                if ((options->break_opt =
2286                     diff_scoreopt_parse(arg)) == -1)
2287                        return -1;
2288        }
2289        else if (!prefixcmp(arg, "-M")) {
2290                if ((options->rename_score =
2291                     diff_scoreopt_parse(arg)) == -1)
2292                        return -1;
2293                options->detect_rename = DIFF_DETECT_RENAME;
2294        }
2295        else if (!prefixcmp(arg, "-C")) {
2296                if (options->detect_rename == DIFF_DETECT_COPY)
2297                        options->find_copies_harder = 1;
2298                if ((options->rename_score =
2299                     diff_scoreopt_parse(arg)) == -1)
2300                        return -1;
2301                options->detect_rename = DIFF_DETECT_COPY;
2302        }
2303        else if (!strcmp(arg, "--find-copies-harder"))
2304                options->find_copies_harder = 1;
2305        else if (!strcmp(arg, "--follow"))
2306                options->follow_renames = 1;
2307        else if (!strcmp(arg, "--abbrev"))
2308                options->abbrev = DEFAULT_ABBREV;
2309        else if (!prefixcmp(arg, "--abbrev=")) {
2310                options->abbrev = strtoul(arg + 9, NULL, 10);
2311                if (options->abbrev < MINIMUM_ABBREV)
2312                        options->abbrev = MINIMUM_ABBREV;
2313                else if (40 < options->abbrev)
2314                        options->abbrev = 40;
2315        }
2316        else if (!strcmp(arg, "--color"))
2317                options->color_diff = 1;
2318        else if (!strcmp(arg, "--no-color"))
2319                options->color_diff = 0;
2320        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2321                options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2322        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2323                options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2324        else if (!strcmp(arg, "--ignore-space-at-eol"))
2325                options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2326        else if (!strcmp(arg, "--color-words"))
2327                options->color_diff = options->color_diff_words = 1;
2328        else if (!strcmp(arg, "--no-renames"))
2329                options->detect_rename = 0;
2330        else if (!strcmp(arg, "--exit-code"))
2331                options->exit_with_status = 1;
2332        else if (!strcmp(arg, "--quiet"))
2333                options->quiet = 1;
2334        else if (!strcmp(arg, "--ext-diff"))
2335                options->allow_external = 1;
2336        else if (!strcmp(arg, "--no-ext-diff"))
2337                options->allow_external = 0;
2338        else
2339                return 0;
2340        return 1;
2341}
2342
2343static int parse_num(const char **cp_p)
2344{
2345        unsigned long num, scale;
2346        int ch, dot;
2347        const char *cp = *cp_p;
2348
2349        num = 0;
2350        scale = 1;
2351        dot = 0;
2352        for(;;) {
2353                ch = *cp;
2354                if ( !dot && ch == '.' ) {
2355                        scale = 1;
2356                        dot = 1;
2357                } else if ( ch == '%' ) {
2358                        scale = dot ? scale*100 : 100;
2359                        cp++;   /* % is always at the end */
2360                        break;
2361                } else if ( ch >= '0' && ch <= '9' ) {
2362                        if ( scale < 100000 ) {
2363                                scale *= 10;
2364                                num = (num*10) + (ch-'0');
2365                        }
2366                } else {
2367                        break;
2368                }
2369                cp++;
2370        }
2371        *cp_p = cp;
2372
2373        /* user says num divided by scale and we say internally that
2374         * is MAX_SCORE * num / scale.
2375         */
2376        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2377}
2378
2379static int diff_scoreopt_parse(const char *opt)
2380{
2381        int opt1, opt2, cmd;
2382
2383        if (*opt++ != '-')
2384                return -1;
2385        cmd = *opt++;
2386        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2387                return -1; /* that is not a -M, -C nor -B option */
2388
2389        opt1 = parse_num(&opt);
2390        if (cmd != 'B')
2391                opt2 = 0;
2392        else {
2393                if (*opt == 0)
2394                        opt2 = 0;
2395                else if (*opt != '/')
2396                        return -1; /* we expect -B80/99 or -B80 */
2397                else {
2398                        opt++;
2399                        opt2 = parse_num(&opt);
2400                }
2401        }
2402        if (*opt != 0)
2403                return -1;
2404        return opt1 | (opt2 << 16);
2405}
2406
2407struct diff_queue_struct diff_queued_diff;
2408
2409void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2410{
2411        if (queue->alloc <= queue->nr) {
2412                queue->alloc = alloc_nr(queue->alloc);
2413                queue->queue = xrealloc(queue->queue,
2414                                        sizeof(dp) * queue->alloc);
2415        }
2416        queue->queue[queue->nr++] = dp;
2417}
2418
2419struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2420                                 struct diff_filespec *one,
2421                                 struct diff_filespec *two)
2422{
2423        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2424        dp->one = one;
2425        dp->two = two;
2426        if (queue)
2427                diff_q(queue, dp);
2428        return dp;
2429}
2430
2431void diff_free_filepair(struct diff_filepair *p)
2432{
2433        free_filespec(p->one);
2434        free_filespec(p->two);
2435        free(p);
2436}
2437
2438/* This is different from find_unique_abbrev() in that
2439 * it stuffs the result with dots for alignment.
2440 */
2441const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2442{
2443        int abblen;
2444        const char *abbrev;
2445        if (len == 40)
2446                return sha1_to_hex(sha1);
2447
2448        abbrev = find_unique_abbrev(sha1, len);
2449        if (!abbrev)
2450                return sha1_to_hex(sha1);
2451        abblen = strlen(abbrev);
2452        if (abblen < 37) {
2453                static char hex[41];
2454                if (len < abblen && abblen <= len + 2)
2455                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2456                else
2457                        sprintf(hex, "%s...", abbrev);
2458                return hex;
2459        }
2460        return sha1_to_hex(sha1);
2461}
2462
2463static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2464{
2465        int line_termination = opt->line_termination;
2466        int inter_name_termination = line_termination ? '\t' : '\0';
2467
2468        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2469                printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2470                       diff_unique_abbrev(p->one->sha1, opt->abbrev));
2471                printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2472        }
2473        if (p->score) {
2474                printf("%c%03d%c", p->status, similarity_index(p),
2475                           inter_name_termination);
2476        } else {
2477                printf("%c%c", p->status, inter_name_termination);
2478        }
2479
2480        if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2481                write_name_quoted(p->one->path, stdout, inter_name_termination);
2482                write_name_quoted(p->two->path, stdout, line_termination);
2483        } else {
2484                const char *path = p->one->mode ? p->one->path : p->two->path;
2485                write_name_quoted(path, stdout, line_termination);
2486        }
2487}
2488
2489int diff_unmodified_pair(struct diff_filepair *p)
2490{
2491        /* This function is written stricter than necessary to support
2492         * the currently implemented transformers, but the idea is to
2493         * let transformers to produce diff_filepairs any way they want,
2494         * and filter and clean them up here before producing the output.
2495         */
2496        struct diff_filespec *one = p->one, *two = p->two;
2497
2498        if (DIFF_PAIR_UNMERGED(p))
2499                return 0; /* unmerged is interesting */
2500
2501        /* deletion, addition, mode or type change
2502         * and rename are all interesting.
2503         */
2504        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2505            DIFF_PAIR_MODE_CHANGED(p) ||
2506            strcmp(one->path, two->path))
2507                return 0;
2508
2509        /* both are valid and point at the same path.  that is, we are
2510         * dealing with a change.
2511         */
2512        if (one->sha1_valid && two->sha1_valid &&
2513            !hashcmp(one->sha1, two->sha1))
2514                return 1; /* no change */
2515        if (!one->sha1_valid && !two->sha1_valid)
2516                return 1; /* both look at the same file on the filesystem. */
2517        return 0;
2518}
2519
2520static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2521{
2522        if (diff_unmodified_pair(p))
2523                return;
2524
2525        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2526            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2527                return; /* no tree diffs in patch format */
2528
2529        run_diff(p, o);
2530}
2531
2532static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2533                            struct diffstat_t *diffstat)
2534{
2535        if (diff_unmodified_pair(p))
2536                return;
2537
2538        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2539            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2540                return; /* no tree diffs in patch format */
2541
2542        run_diffstat(p, o, diffstat);
2543}
2544
2545static void diff_flush_checkdiff(struct diff_filepair *p,
2546                struct diff_options *o)
2547{
2548        if (diff_unmodified_pair(p))
2549                return;
2550
2551        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2552            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2553                return; /* no tree diffs in patch format */
2554
2555        run_checkdiff(p, o);
2556}
2557
2558int diff_queue_is_empty(void)
2559{
2560        struct diff_queue_struct *q = &diff_queued_diff;
2561        int i;
2562        for (i = 0; i < q->nr; i++)
2563                if (!diff_unmodified_pair(q->queue[i]))
2564                        return 0;
2565        return 1;
2566}
2567
2568#if DIFF_DEBUG
2569void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2570{
2571        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2572                x, one ? one : "",
2573                s->path,
2574                DIFF_FILE_VALID(s) ? "valid" : "invalid",
2575                s->mode,
2576                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2577        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2578                x, one ? one : "",
2579                s->size, s->xfrm_flags);
2580}
2581
2582void diff_debug_filepair(const struct diff_filepair *p, int i)
2583{
2584        diff_debug_filespec(p->one, i, "one");
2585        diff_debug_filespec(p->two, i, "two");
2586        fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2587                p->score, p->status ? p->status : '?',
2588                p->one->rename_used, p->broken_pair);
2589}
2590
2591void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2592{
2593        int i;
2594        if (msg)
2595                fprintf(stderr, "%s\n", msg);
2596        fprintf(stderr, "q->nr = %d\n", q->nr);
2597        for (i = 0; i < q->nr; i++) {
2598                struct diff_filepair *p = q->queue[i];
2599                diff_debug_filepair(p, i);
2600        }
2601}
2602#endif
2603
2604static void diff_resolve_rename_copy(void)
2605{
2606        int i;
2607        struct diff_filepair *p;
2608        struct diff_queue_struct *q = &diff_queued_diff;
2609
2610        diff_debug_queue("resolve-rename-copy", q);
2611
2612        for (i = 0; i < q->nr; i++) {
2613                p = q->queue[i];
2614                p->status = 0; /* undecided */
2615                if (DIFF_PAIR_UNMERGED(p))
2616                        p->status = DIFF_STATUS_UNMERGED;
2617                else if (!DIFF_FILE_VALID(p->one))
2618                        p->status = DIFF_STATUS_ADDED;
2619                else if (!DIFF_FILE_VALID(p->two))
2620                        p->status = DIFF_STATUS_DELETED;
2621                else if (DIFF_PAIR_TYPE_CHANGED(p))
2622                        p->status = DIFF_STATUS_TYPE_CHANGED;
2623
2624                /* from this point on, we are dealing with a pair
2625                 * whose both sides are valid and of the same type, i.e.
2626                 * either in-place edit or rename/copy edit.
2627                 */
2628                else if (DIFF_PAIR_RENAME(p)) {
2629                        /*
2630                         * A rename might have re-connected a broken
2631                         * pair up, causing the pathnames to be the
2632                         * same again. If so, that's not a rename at
2633                         * all, just a modification..
2634                         *
2635                         * Otherwise, see if this source was used for
2636                         * multiple renames, in which case we decrement
2637                         * the count, and call it a copy.
2638                         */
2639                        if (!strcmp(p->one->path, p->two->path))
2640                                p->status = DIFF_STATUS_MODIFIED;
2641                        else if (--p->one->rename_used > 0)
2642                                p->status = DIFF_STATUS_COPIED;
2643                        else
2644                                p->status = DIFF_STATUS_RENAMED;
2645                }
2646                else if (hashcmp(p->one->sha1, p->two->sha1) ||
2647                         p->one->mode != p->two->mode ||
2648                         is_null_sha1(p->one->sha1))
2649                        p->status = DIFF_STATUS_MODIFIED;
2650                else {
2651                        /* This is a "no-change" entry and should not
2652                         * happen anymore, but prepare for broken callers.
2653                         */
2654                        error("feeding unmodified %s to diffcore",
2655                              p->one->path);
2656                        p->status = DIFF_STATUS_UNKNOWN;
2657                }
2658        }
2659        diff_debug_queue("resolve-rename-copy done", q);
2660}
2661
2662static int check_pair_status(struct diff_filepair *p)
2663{
2664        switch (p->status) {
2665        case DIFF_STATUS_UNKNOWN:
2666                return 0;
2667        case 0:
2668                die("internal error in diff-resolve-rename-copy");
2669        default:
2670                return 1;
2671        }
2672}
2673
2674static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2675{
2676        int fmt = opt->output_format;
2677
2678        if (fmt & DIFF_FORMAT_CHECKDIFF)
2679                diff_flush_checkdiff(p, opt);
2680        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2681                diff_flush_raw(p, opt);
2682        else if (fmt & DIFF_FORMAT_NAME)
2683                write_name_quoted(p->two->path, stdout, opt->line_termination);
2684}
2685
2686static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2687{
2688        if (fs->mode)
2689                printf(" %s mode %06o ", newdelete, fs->mode);
2690        else
2691                printf(" %s ", newdelete);
2692        write_name_quoted(fs->path, stdout, '\n');
2693}
2694
2695
2696static void show_mode_change(struct diff_filepair *p, int show_name)
2697{
2698        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2699                printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2700                        show_name ? ' ' : '\n');
2701                if (show_name) {
2702                        write_name_quoted(p->two->path, stdout, '\n');
2703                }
2704        }
2705}
2706
2707static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2708{
2709        char *names = pprint_rename(p->one->path, p->two->path);
2710
2711        printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2712        free(names);
2713        show_mode_change(p, 0);
2714}
2715
2716static void diff_summary(struct diff_filepair *p)
2717{
2718        switch(p->status) {
2719        case DIFF_STATUS_DELETED:
2720                show_file_mode_name("delete", p->one);
2721                break;
2722        case DIFF_STATUS_ADDED:
2723                show_file_mode_name("create", p->two);
2724                break;
2725        case DIFF_STATUS_COPIED:
2726                show_rename_copy("copy", p);
2727                break;
2728        case DIFF_STATUS_RENAMED:
2729                show_rename_copy("rename", p);
2730                break;
2731        default:
2732                if (p->score) {
2733                        puts(" rewrite ");
2734                        write_name_quoted(p->two->path, stdout, ' ');
2735                        printf("(%d%%)\n", similarity_index(p));
2736                }
2737                show_mode_change(p, !p->score);
2738                break;
2739        }
2740}
2741
2742struct patch_id_t {
2743        struct xdiff_emit_state xm;
2744        SHA_CTX *ctx;
2745        int patchlen;
2746};
2747
2748static int remove_space(char *line, int len)
2749{
2750        int i;
2751        char *dst = line;
2752        unsigned char c;
2753
2754        for (i = 0; i < len; i++)
2755                if (!isspace((c = line[i])))
2756                        *dst++ = c;
2757
2758        return dst - line;
2759}
2760
2761static void patch_id_consume(void *priv, char *line, unsigned long len)
2762{
2763        struct patch_id_t *data = priv;
2764        int new_len;
2765
2766        /* Ignore line numbers when computing the SHA1 of the patch */
2767        if (!prefixcmp(line, "@@ -"))
2768                return;
2769
2770        new_len = remove_space(line, len);
2771
2772        SHA1_Update(data->ctx, line, new_len);
2773        data->patchlen += new_len;
2774}
2775
2776/* returns 0 upon success, and writes result into sha1 */
2777static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2778{
2779        struct diff_queue_struct *q = &diff_queued_diff;
2780        int i;
2781        SHA_CTX ctx;
2782        struct patch_id_t data;
2783        char buffer[PATH_MAX * 4 + 20];
2784
2785        SHA1_Init(&ctx);
2786        memset(&data, 0, sizeof(struct patch_id_t));
2787        data.ctx = &ctx;
2788        data.xm.consume = patch_id_consume;
2789
2790        for (i = 0; i < q->nr; i++) {
2791                xpparam_t xpp;
2792                xdemitconf_t xecfg;
2793                xdemitcb_t ecb;
2794                mmfile_t mf1, mf2;
2795                struct diff_filepair *p = q->queue[i];
2796                int len1, len2;
2797
2798                memset(&xecfg, 0, sizeof(xecfg));
2799                if (p->status == 0)
2800                        return error("internal diff status error");
2801                if (p->status == DIFF_STATUS_UNKNOWN)
2802                        continue;
2803                if (diff_unmodified_pair(p))
2804                        continue;
2805                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2806                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2807                        continue;
2808                if (DIFF_PAIR_UNMERGED(p))
2809                        continue;
2810
2811                diff_fill_sha1_info(p->one);
2812                diff_fill_sha1_info(p->two);
2813                if (fill_mmfile(&mf1, p->one) < 0 ||
2814                                fill_mmfile(&mf2, p->two) < 0)
2815                        return error("unable to read files to diff");
2816
2817                len1 = remove_space(p->one->path, strlen(p->one->path));
2818                len2 = remove_space(p->two->path, strlen(p->two->path));
2819                if (p->one->mode == 0)
2820                        len1 = snprintf(buffer, sizeof(buffer),
2821                                        "diff--gita/%.*sb/%.*s"
2822                                        "newfilemode%06o"
2823                                        "---/dev/null"
2824                                        "+++b/%.*s",
2825                                        len1, p->one->path,
2826                                        len2, p->two->path,
2827                                        p->two->mode,
2828                                        len2, p->two->path);
2829                else if (p->two->mode == 0)
2830                        len1 = snprintf(buffer, sizeof(buffer),
2831                                        "diff--gita/%.*sb/%.*s"
2832                                        "deletedfilemode%06o"
2833                                        "---a/%.*s"
2834                                        "+++/dev/null",
2835                                        len1, p->one->path,
2836                                        len2, p->two->path,
2837                                        p->one->mode,
2838                                        len1, p->one->path);
2839                else
2840                        len1 = snprintf(buffer, sizeof(buffer),
2841                                        "diff--gita/%.*sb/%.*s"
2842                                        "---a/%.*s"
2843                                        "+++b/%.*s",
2844                                        len1, p->one->path,
2845                                        len2, p->two->path,
2846                                        len1, p->one->path,
2847                                        len2, p->two->path);
2848                SHA1_Update(&ctx, buffer, len1);
2849
2850                xpp.flags = XDF_NEED_MINIMAL;
2851                xecfg.ctxlen = 3;
2852                xecfg.flags = XDL_EMIT_FUNCNAMES;
2853                ecb.outf = xdiff_outf;
2854                ecb.priv = &data;
2855                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2856        }
2857
2858        SHA1_Final(sha1, &ctx);
2859        return 0;
2860}
2861
2862int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2863{
2864        struct diff_queue_struct *q = &diff_queued_diff;
2865        int i;
2866        int result = diff_get_patch_id(options, sha1);
2867
2868        for (i = 0; i < q->nr; i++)
2869                diff_free_filepair(q->queue[i]);
2870
2871        free(q->queue);
2872        q->queue = NULL;
2873        q->nr = q->alloc = 0;
2874
2875        return result;
2876}
2877
2878static int is_summary_empty(const struct diff_queue_struct *q)
2879{
2880        int i;
2881
2882        for (i = 0; i < q->nr; i++) {
2883                const struct diff_filepair *p = q->queue[i];
2884
2885                switch (p->status) {
2886                case DIFF_STATUS_DELETED:
2887                case DIFF_STATUS_ADDED:
2888                case DIFF_STATUS_COPIED:
2889                case DIFF_STATUS_RENAMED:
2890                        return 0;
2891                default:
2892                        if (p->score)
2893                                return 0;
2894                        if (p->one->mode && p->two->mode &&
2895                            p->one->mode != p->two->mode)
2896                                return 0;
2897                        break;
2898                }
2899        }
2900        return 1;
2901}
2902
2903void diff_flush(struct diff_options *options)
2904{
2905        struct diff_queue_struct *q = &diff_queued_diff;
2906        int i, output_format = options->output_format;
2907        int separator = 0;
2908
2909        /*
2910         * Order: raw, stat, summary, patch
2911         * or:    name/name-status/checkdiff (other bits clear)
2912         */
2913        if (!q->nr)
2914                goto free_queue;
2915
2916        if (output_format & (DIFF_FORMAT_RAW |
2917                             DIFF_FORMAT_NAME |
2918                             DIFF_FORMAT_NAME_STATUS |
2919                             DIFF_FORMAT_CHECKDIFF)) {
2920                for (i = 0; i < q->nr; i++) {
2921                        struct diff_filepair *p = q->queue[i];
2922                        if (check_pair_status(p))
2923                                flush_one_pair(p, options);
2924                }
2925                separator++;
2926        }
2927
2928        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2929                struct diffstat_t diffstat;
2930
2931                memset(&diffstat, 0, sizeof(struct diffstat_t));
2932                diffstat.xm.consume = diffstat_consume;
2933                for (i = 0; i < q->nr; i++) {
2934                        struct diff_filepair *p = q->queue[i];
2935                        if (check_pair_status(p))
2936                                diff_flush_stat(p, options, &diffstat);
2937                }
2938                if (output_format & DIFF_FORMAT_NUMSTAT)
2939                        show_numstat(&diffstat, options);
2940                if (output_format & DIFF_FORMAT_DIFFSTAT)
2941                        show_stats(&diffstat, options);
2942                else if (output_format & DIFF_FORMAT_SHORTSTAT)
2943                        show_shortstats(&diffstat);
2944                separator++;
2945        }
2946
2947        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2948                for (i = 0; i < q->nr; i++)
2949                        diff_summary(q->queue[i]);
2950                separator++;
2951        }
2952
2953        if (output_format & DIFF_FORMAT_PATCH) {
2954                if (separator) {
2955                        if (options->stat_sep) {
2956                                /* attach patch instead of inline */
2957                                fputs(options->stat_sep, stdout);
2958                        } else {
2959                                putchar(options->line_termination);
2960                        }
2961                }
2962
2963                for (i = 0; i < q->nr; i++) {
2964                        struct diff_filepair *p = q->queue[i];
2965                        if (check_pair_status(p))
2966                                diff_flush_patch(p, options);
2967                }
2968        }
2969
2970        if (output_format & DIFF_FORMAT_CALLBACK)
2971                options->format_callback(q, options, options->format_callback_data);
2972
2973        for (i = 0; i < q->nr; i++)
2974                diff_free_filepair(q->queue[i]);
2975free_queue:
2976        free(q->queue);
2977        q->queue = NULL;
2978        q->nr = q->alloc = 0;
2979}
2980
2981static void diffcore_apply_filter(const char *filter)
2982{
2983        int i;
2984        struct diff_queue_struct *q = &diff_queued_diff;
2985        struct diff_queue_struct outq;
2986        outq.queue = NULL;
2987        outq.nr = outq.alloc = 0;
2988
2989        if (!filter)
2990                return;
2991
2992        if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2993                int found;
2994                for (i = found = 0; !found && i < q->nr; i++) {
2995                        struct diff_filepair *p = q->queue[i];
2996                        if (((p->status == DIFF_STATUS_MODIFIED) &&
2997                             ((p->score &&
2998                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2999                              (!p->score &&
3000                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3001                            ((p->status != DIFF_STATUS_MODIFIED) &&
3002                             strchr(filter, p->status)))
3003                                found++;
3004                }
3005                if (found)
3006                        return;
3007
3008                /* otherwise we will clear the whole queue
3009                 * by copying the empty outq at the end of this
3010                 * function, but first clear the current entries
3011                 * in the queue.
3012                 */
3013                for (i = 0; i < q->nr; i++)
3014                        diff_free_filepair(q->queue[i]);
3015        }
3016        else {
3017                /* Only the matching ones */
3018                for (i = 0; i < q->nr; i++) {
3019                        struct diff_filepair *p = q->queue[i];
3020
3021                        if (((p->status == DIFF_STATUS_MODIFIED) &&
3022                             ((p->score &&
3023                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3024                              (!p->score &&
3025                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3026                            ((p->status != DIFF_STATUS_MODIFIED) &&
3027                             strchr(filter, p->status)))
3028                                diff_q(&outq, p);
3029                        else
3030                                diff_free_filepair(p);
3031                }
3032        }
3033        free(q->queue);
3034        *q = outq;
3035}
3036
3037/* Check whether two filespecs with the same mode and size are identical */
3038static int diff_filespec_is_identical(struct diff_filespec *one,
3039                                      struct diff_filespec *two)
3040{
3041        if (S_ISGITLINK(one->mode)) {
3042                diff_fill_sha1_info(one);
3043                diff_fill_sha1_info(two);
3044                return !hashcmp(one->sha1, two->sha1);
3045        }
3046        if (diff_populate_filespec(one, 0))
3047                return 0;
3048        if (diff_populate_filespec(two, 0))
3049                return 0;
3050        return !memcmp(one->data, two->data, one->size);
3051}
3052
3053static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3054{
3055        int i;
3056        struct diff_queue_struct *q = &diff_queued_diff;
3057        struct diff_queue_struct outq;
3058        outq.queue = NULL;
3059        outq.nr = outq.alloc = 0;
3060
3061        for (i = 0; i < q->nr; i++) {
3062                struct diff_filepair *p = q->queue[i];
3063
3064                /*
3065                 * 1. Entries that come from stat info dirtyness
3066                 *    always have both sides (iow, not create/delete),
3067                 *    one side of the object name is unknown, with
3068                 *    the same mode and size.  Keep the ones that
3069                 *    do not match these criteria.  They have real
3070                 *    differences.
3071                 *
3072                 * 2. At this point, the file is known to be modified,
3073                 *    with the same mode and size, and the object
3074                 *    name of one side is unknown.  Need to inspect
3075                 *    the identical contents.
3076                 */
3077                if (!DIFF_FILE_VALID(p->one) || /* (1) */
3078                    !DIFF_FILE_VALID(p->two) ||
3079                    (p->one->sha1_valid && p->two->sha1_valid) ||
3080                    (p->one->mode != p->two->mode) ||
3081                    diff_populate_filespec(p->one, 1) ||
3082                    diff_populate_filespec(p->two, 1) ||
3083                    (p->one->size != p->two->size) ||
3084                    !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3085                        diff_q(&outq, p);
3086                else {
3087                        /*
3088                         * The caller can subtract 1 from skip_stat_unmatch
3089                         * to determine how many paths were dirty only
3090                         * due to stat info mismatch.
3091                         */
3092                        if (!diffopt->no_index)
3093                                diffopt->skip_stat_unmatch++;
3094                        diff_free_filepair(p);
3095                }
3096        }
3097        free(q->queue);
3098        *q = outq;
3099}
3100
3101void diffcore_std(struct diff_options *options)
3102{
3103        if (options->quiet)
3104                return;
3105
3106        if (options->skip_stat_unmatch && !options->find_copies_harder)
3107                diffcore_skip_stat_unmatch(options);
3108        if (options->break_opt != -1)
3109                diffcore_break(options->break_opt);
3110        if (options->detect_rename)
3111                diffcore_rename(options);
3112        if (options->break_opt != -1)
3113                diffcore_merge_broken();
3114        if (options->pickaxe)
3115                diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3116        if (options->orderfile)
3117                diffcore_order(options->orderfile);
3118        diff_resolve_rename_copy();
3119        diffcore_apply_filter(options->filter);
3120
3121        options->has_changes = !!diff_queued_diff.nr;
3122}
3123
3124
3125void diff_addremove(struct diff_options *options,
3126                    int addremove, unsigned mode,
3127                    const unsigned char *sha1,
3128                    const char *base, const char *path)
3129{
3130        char concatpath[PATH_MAX];
3131        struct diff_filespec *one, *two;
3132
3133        /* This may look odd, but it is a preparation for
3134         * feeding "there are unchanged files which should
3135         * not produce diffs, but when you are doing copy
3136         * detection you would need them, so here they are"
3137         * entries to the diff-core.  They will be prefixed
3138         * with something like '=' or '*' (I haven't decided
3139         * which but should not make any difference).
3140         * Feeding the same new and old to diff_change()
3141         * also has the same effect.
3142         * Before the final output happens, they are pruned after
3143         * merged into rename/copy pairs as appropriate.
3144         */
3145        if (options->reverse_diff)
3146                addremove = (addremove == '+' ? '-' :
3147                             addremove == '-' ? '+' : addremove);
3148
3149        if (!path) path = "";
3150        sprintf(concatpath, "%s%s", base, path);
3151        one = alloc_filespec(concatpath);
3152        two = alloc_filespec(concatpath);
3153
3154        if (addremove != '+')
3155                fill_filespec(one, sha1, mode);
3156        if (addremove != '-')
3157                fill_filespec(two, sha1, mode);
3158
3159        diff_queue(&diff_queued_diff, one, two);
3160        options->has_changes = 1;
3161}
3162
3163void diff_change(struct diff_options *options,
3164                 unsigned old_mode, unsigned new_mode,
3165                 const unsigned char *old_sha1,
3166                 const unsigned char *new_sha1,
3167                 const char *base, const char *path)
3168{
3169        char concatpath[PATH_MAX];
3170        struct diff_filespec *one, *two;
3171
3172        if (options->reverse_diff) {
3173                unsigned tmp;
3174                const unsigned char *tmp_c;
3175                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3176                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3177        }
3178        if (!path) path = "";
3179        sprintf(concatpath, "%s%s", base, path);
3180        one = alloc_filespec(concatpath);
3181        two = alloc_filespec(concatpath);
3182        fill_filespec(one, old_sha1, old_mode);
3183        fill_filespec(two, new_sha1, new_mode);
3184
3185        diff_queue(&diff_queued_diff, one, two);
3186        options->has_changes = 1;
3187}
3188
3189void diff_unmerge(struct diff_options *options,
3190                  const char *path,
3191                  unsigned mode, const unsigned char *sha1)
3192{
3193        struct diff_filespec *one, *two;
3194        one = alloc_filespec(path);
3195        two = alloc_filespec(path);
3196        fill_filespec(one, sha1, mode);
3197        diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3198}