diff.con commit Merge branch 'jn/web' into HEAD (6beb669)
   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
  13#ifdef NO_FAST_WORKING_DIRECTORY
  14#define FAST_WORKING_DIRECTORY 0
  15#else
  16#define FAST_WORKING_DIRECTORY 1
  17#endif
  18
  19static int diff_detect_rename_default;
  20static int diff_rename_limit_default = 100;
  21static int diff_use_color_default;
  22int diff_auto_refresh_index = 1;
  23
  24static char diff_colors[][COLOR_MAXLEN] = {
  25        "\033[m",       /* reset */
  26        "",             /* PLAIN (normal) */
  27        "\033[1m",      /* METAINFO (bold) */
  28        "\033[36m",     /* FRAGINFO (cyan) */
  29        "\033[31m",     /* OLD (red) */
  30        "\033[32m",     /* NEW (green) */
  31        "\033[33m",     /* COMMIT (yellow) */
  32        "\033[41m",     /* WHITESPACE (red background) */
  33};
  34
  35static int parse_diff_color_slot(const char *var, int ofs)
  36{
  37        if (!strcasecmp(var+ofs, "plain"))
  38                return DIFF_PLAIN;
  39        if (!strcasecmp(var+ofs, "meta"))
  40                return DIFF_METAINFO;
  41        if (!strcasecmp(var+ofs, "frag"))
  42                return DIFF_FRAGINFO;
  43        if (!strcasecmp(var+ofs, "old"))
  44                return DIFF_FILE_OLD;
  45        if (!strcasecmp(var+ofs, "new"))
  46                return DIFF_FILE_NEW;
  47        if (!strcasecmp(var+ofs, "commit"))
  48                return DIFF_COMMIT;
  49        if (!strcasecmp(var+ofs, "whitespace"))
  50                return DIFF_WHITESPACE;
  51        die("bad config variable '%s'", var);
  52}
  53
  54static struct ll_diff_driver {
  55        const char *name;
  56        struct ll_diff_driver *next;
  57        char *cmd;
  58} *user_diff, **user_diff_tail;
  59
  60static void read_config_if_needed(void)
  61{
  62        if (!user_diff_tail) {
  63                user_diff_tail = &user_diff;
  64                git_config(git_diff_ui_config);
  65        }
  66}
  67
  68/*
  69 * Currently there is only "diff.<drivername>.command" variable;
  70 * because there are "diff.color.<slot>" variables, we are parsing
  71 * this in a bit convoluted way to allow low level diff driver
  72 * called "color".
  73 */
  74static int parse_lldiff_command(const char *var, const char *ep, const char *value)
  75{
  76        const char *name;
  77        int namelen;
  78        struct ll_diff_driver *drv;
  79
  80        name = var + 5;
  81        namelen = ep - name;
  82        for (drv = user_diff; drv; drv = drv->next)
  83                if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
  84                        break;
  85        if (!drv) {
  86                drv = xcalloc(1, sizeof(struct ll_diff_driver));
  87                drv->name = xmemdupz(name, namelen);
  88                if (!user_diff_tail)
  89                        user_diff_tail = &user_diff;
  90                *user_diff_tail = drv;
  91                user_diff_tail = &(drv->next);
  92        }
  93
  94        if (!value)
  95                return error("%s: lacks value", var);
  96        drv->cmd = strdup(value);
  97        return 0;
  98}
  99
 100/*
 101 * 'diff.<what>.funcname' attribute can be specified in the configuration
 102 * to define a customized regexp to find the beginning of a function to
 103 * be used for hunk header lines of "diff -p" style output.
 104 */
 105static struct funcname_pattern {
 106        char *name;
 107        char *pattern;
 108        struct funcname_pattern *next;
 109} *funcname_pattern_list;
 110
 111static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
 112{
 113        const char *name;
 114        int namelen;
 115        struct funcname_pattern *pp;
 116
 117        name = var + 5; /* "diff." */
 118        namelen = ep - name;
 119
 120        for (pp = funcname_pattern_list; pp; pp = pp->next)
 121                if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
 122                        break;
 123        if (!pp) {
 124                pp = xcalloc(1, sizeof(*pp));
 125                pp->name = xmemdupz(name, namelen);
 126                pp->next = funcname_pattern_list;
 127                funcname_pattern_list = pp;
 128        }
 129        if (pp->pattern)
 130                free(pp->pattern);
 131        pp->pattern = xstrdup(value);
 132        return 0;
 133}
 134
 135/*
 136 * These are to give UI layer defaults.
 137 * The core-level commands such as git-diff-files should
 138 * never be affected by the setting of diff.renames
 139 * the user happens to have in the configuration file.
 140 */
 141int git_diff_ui_config(const char *var, const char *value)
 142{
 143        if (!strcmp(var, "diff.renamelimit")) {
 144                diff_rename_limit_default = git_config_int(var, value);
 145                return 0;
 146        }
 147        if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
 148                diff_use_color_default = git_config_colorbool(var, value);
 149                return 0;
 150        }
 151        if (!strcmp(var, "diff.renames")) {
 152                if (!value)
 153                        diff_detect_rename_default = DIFF_DETECT_RENAME;
 154                else if (!strcasecmp(value, "copies") ||
 155                         !strcasecmp(value, "copy"))
 156                        diff_detect_rename_default = DIFF_DETECT_COPY;
 157                else if (git_config_bool(var,value))
 158                        diff_detect_rename_default = DIFF_DETECT_RENAME;
 159                return 0;
 160        }
 161        if (!strcmp(var, "diff.autorefreshindex")) {
 162                diff_auto_refresh_index = git_config_bool(var, value);
 163                return 0;
 164        }
 165        if (!prefixcmp(var, "diff.")) {
 166                const char *ep = strrchr(var, '.');
 167
 168                if (ep != var + 4) {
 169                        if (!strcmp(ep, ".command"))
 170                                return parse_lldiff_command(var, ep, value);
 171                        if (!strcmp(ep, ".funcname"))
 172                                return parse_funcname_pattern(var, ep, value);
 173                }
 174        }
 175        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
 176                int slot = parse_diff_color_slot(var, 11);
 177                color_parse(value, var, diff_colors[slot]);
 178                return 0;
 179        }
 180
 181        return git_default_config(var, value);
 182}
 183
 184static char *quote_two(const char *one, const char *two)
 185{
 186        int need_one = quote_c_style(one, NULL, NULL, 1);
 187        int need_two = quote_c_style(two, NULL, NULL, 1);
 188        struct strbuf res;
 189
 190        strbuf_init(&res, 0);
 191        if (need_one + need_two) {
 192                strbuf_addch(&res, '"');
 193                quote_c_style(one, &res, NULL, 1);
 194                quote_c_style(two, &res, NULL, 1);
 195                strbuf_addch(&res, '"');
 196        } else {
 197                strbuf_addstr(&res, one);
 198                strbuf_addstr(&res, two);
 199        }
 200        return strbuf_detach(&res, NULL);
 201}
 202
 203static const char *external_diff(void)
 204{
 205        static const char *external_diff_cmd = NULL;
 206        static int done_preparing = 0;
 207
 208        if (done_preparing)
 209                return external_diff_cmd;
 210        external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
 211        done_preparing = 1;
 212        return external_diff_cmd;
 213}
 214
 215static struct diff_tempfile {
 216        const char *name; /* filename external diff should read from */
 217        char hex[41];
 218        char mode[10];
 219        char tmp_path[PATH_MAX];
 220} diff_temp[2];
 221
 222static int count_lines(const char *data, int size)
 223{
 224        int count, ch, completely_empty = 1, nl_just_seen = 0;
 225        count = 0;
 226        while (0 < size--) {
 227                ch = *data++;
 228                if (ch == '\n') {
 229                        count++;
 230                        nl_just_seen = 1;
 231                        completely_empty = 0;
 232                }
 233                else {
 234                        nl_just_seen = 0;
 235                        completely_empty = 0;
 236                }
 237        }
 238        if (completely_empty)
 239                return 0;
 240        if (!nl_just_seen)
 241                count++; /* no trailing newline */
 242        return count;
 243}
 244
 245static void print_line_count(int count)
 246{
 247        switch (count) {
 248        case 0:
 249                printf("0,0");
 250                break;
 251        case 1:
 252                printf("1");
 253                break;
 254        default:
 255                printf("1,%d", count);
 256                break;
 257        }
 258}
 259
 260static void copy_file(int prefix, const char *data, int size,
 261                const char *set, const char *reset)
 262{
 263        int ch, nl_just_seen = 1;
 264        while (0 < size--) {
 265                ch = *data++;
 266                if (nl_just_seen) {
 267                        fputs(set, stdout);
 268                        putchar(prefix);
 269                }
 270                if (ch == '\n') {
 271                        nl_just_seen = 1;
 272                        fputs(reset, stdout);
 273                } else
 274                        nl_just_seen = 0;
 275                putchar(ch);
 276        }
 277        if (!nl_just_seen)
 278                printf("%s\n\\ No newline at end of file\n", reset);
 279}
 280
 281static void emit_rewrite_diff(const char *name_a,
 282                              const char *name_b,
 283                              struct diff_filespec *one,
 284                              struct diff_filespec *two,
 285                              int color_diff)
 286{
 287        int lc_a, lc_b;
 288        const char *name_a_tab, *name_b_tab;
 289        const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
 290        const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
 291        const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
 292        const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
 293        const char *reset = diff_get_color(color_diff, DIFF_RESET);
 294
 295        name_a += (*name_a == '/');
 296        name_b += (*name_b == '/');
 297        name_a_tab = strchr(name_a, ' ') ? "\t" : "";
 298        name_b_tab = strchr(name_b, ' ') ? "\t" : "";
 299
 300        diff_populate_filespec(one, 0);
 301        diff_populate_filespec(two, 0);
 302        lc_a = count_lines(one->data, one->size);
 303        lc_b = count_lines(two->data, two->size);
 304        printf("%s--- a/%s%s%s\n%s+++ b/%s%s%s\n%s@@ -",
 305               metainfo, name_a, name_a_tab, reset,
 306               metainfo, name_b, name_b_tab, reset, fraginfo);
 307        print_line_count(lc_a);
 308        printf(" +");
 309        print_line_count(lc_b);
 310        printf(" @@%s\n", reset);
 311        if (lc_a)
 312                copy_file('-', one->data, one->size, old, reset);
 313        if (lc_b)
 314                copy_file('+', two->data, two->size, new, reset);
 315}
 316
 317static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 318{
 319        if (!DIFF_FILE_VALID(one)) {
 320                mf->ptr = (char *)""; /* does not matter */
 321                mf->size = 0;
 322                return 0;
 323        }
 324        else if (diff_populate_filespec(one, 0))
 325                return -1;
 326        mf->ptr = one->data;
 327        mf->size = one->size;
 328        return 0;
 329}
 330
 331struct diff_words_buffer {
 332        mmfile_t text;
 333        long alloc;
 334        long current; /* output pointer */
 335        int suppressed_newline;
 336};
 337
 338static void diff_words_append(char *line, unsigned long len,
 339                struct diff_words_buffer *buffer)
 340{
 341        if (buffer->text.size + len > buffer->alloc) {
 342                buffer->alloc = (buffer->text.size + len) * 3 / 2;
 343                buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
 344        }
 345        line++;
 346        len--;
 347        memcpy(buffer->text.ptr + buffer->text.size, line, len);
 348        buffer->text.size += len;
 349}
 350
 351struct diff_words_data {
 352        struct xdiff_emit_state xm;
 353        struct diff_words_buffer minus, plus;
 354};
 355
 356static void print_word(struct diff_words_buffer *buffer, int len, int color,
 357                int suppress_newline)
 358{
 359        const char *ptr;
 360        int eol = 0;
 361
 362        if (len == 0)
 363                return;
 364
 365        ptr  = buffer->text.ptr + buffer->current;
 366        buffer->current += len;
 367
 368        if (ptr[len - 1] == '\n') {
 369                eol = 1;
 370                len--;
 371        }
 372
 373        fputs(diff_get_color(1, color), stdout);
 374        fwrite(ptr, len, 1, stdout);
 375        fputs(diff_get_color(1, DIFF_RESET), stdout);
 376
 377        if (eol) {
 378                if (suppress_newline)
 379                        buffer->suppressed_newline = 1;
 380                else
 381                        putchar('\n');
 382        }
 383}
 384
 385static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 386{
 387        struct diff_words_data *diff_words = priv;
 388
 389        if (diff_words->minus.suppressed_newline) {
 390                if (line[0] != '+')
 391                        putchar('\n');
 392                diff_words->minus.suppressed_newline = 0;
 393        }
 394
 395        len--;
 396        switch (line[0]) {
 397                case '-':
 398                        print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
 399                        break;
 400                case '+':
 401                        print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
 402                        break;
 403                case ' ':
 404                        print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
 405                        diff_words->minus.current += len;
 406                        break;
 407        }
 408}
 409
 410/* this executes the word diff on the accumulated buffers */
 411static void diff_words_show(struct diff_words_data *diff_words)
 412{
 413        xpparam_t xpp;
 414        xdemitconf_t xecfg;
 415        xdemitcb_t ecb;
 416        mmfile_t minus, plus;
 417        int i;
 418
 419        memset(&xecfg, 0, sizeof(xecfg));
 420        minus.size = diff_words->minus.text.size;
 421        minus.ptr = xmalloc(minus.size);
 422        memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
 423        for (i = 0; i < minus.size; i++)
 424                if (isspace(minus.ptr[i]))
 425                        minus.ptr[i] = '\n';
 426        diff_words->minus.current = 0;
 427
 428        plus.size = diff_words->plus.text.size;
 429        plus.ptr = xmalloc(plus.size);
 430        memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
 431        for (i = 0; i < plus.size; i++)
 432                if (isspace(plus.ptr[i]))
 433                        plus.ptr[i] = '\n';
 434        diff_words->plus.current = 0;
 435
 436        xpp.flags = XDF_NEED_MINIMAL;
 437        xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
 438        ecb.outf = xdiff_outf;
 439        ecb.priv = diff_words;
 440        diff_words->xm.consume = fn_out_diff_words_aux;
 441        xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
 442
 443        free(minus.ptr);
 444        free(plus.ptr);
 445        diff_words->minus.text.size = diff_words->plus.text.size = 0;
 446
 447        if (diff_words->minus.suppressed_newline) {
 448                putchar('\n');
 449                diff_words->minus.suppressed_newline = 0;
 450        }
 451}
 452
 453struct emit_callback {
 454        struct xdiff_emit_state xm;
 455        int nparents, color_diff;
 456        const char **label_path;
 457        struct diff_words_data *diff_words;
 458        int *found_changesp;
 459};
 460
 461static void free_diff_words_data(struct emit_callback *ecbdata)
 462{
 463        if (ecbdata->diff_words) {
 464                /* flush buffers */
 465                if (ecbdata->diff_words->minus.text.size ||
 466                                ecbdata->diff_words->plus.text.size)
 467                        diff_words_show(ecbdata->diff_words);
 468
 469                if (ecbdata->diff_words->minus.text.ptr)
 470                        free (ecbdata->diff_words->minus.text.ptr);
 471                if (ecbdata->diff_words->plus.text.ptr)
 472                        free (ecbdata->diff_words->plus.text.ptr);
 473                free(ecbdata->diff_words);
 474                ecbdata->diff_words = NULL;
 475        }
 476}
 477
 478const char *diff_get_color(int diff_use_color, enum color_diff ix)
 479{
 480        if (diff_use_color)
 481                return diff_colors[ix];
 482        return "";
 483}
 484
 485static void emit_line(const char *set, const char *reset, const char *line, int len)
 486{
 487        if (len > 0 && line[len-1] == '\n')
 488                len--;
 489        fputs(set, stdout);
 490        fwrite(line, len, 1, stdout);
 491        puts(reset);
 492}
 493
 494static void emit_line_with_ws(int nparents,
 495                const char *set, const char *reset, const char *ws,
 496                const char *line, int len)
 497{
 498        int col0 = nparents;
 499        int last_tab_in_indent = -1;
 500        int last_space_in_indent = -1;
 501        int i;
 502        int tail = len;
 503        int need_highlight_leading_space = 0;
 504        /* The line is a newly added line.  Does it have funny leading
 505         * whitespaces?  In indent, SP should never precede a TAB.
 506         */
 507        for (i = col0; i < len; i++) {
 508                if (line[i] == '\t') {
 509                        last_tab_in_indent = i;
 510                        if (0 <= last_space_in_indent)
 511                                need_highlight_leading_space = 1;
 512                }
 513                else if (line[i] == ' ')
 514                        last_space_in_indent = i;
 515                else
 516                        break;
 517        }
 518        fputs(set, stdout);
 519        fwrite(line, col0, 1, stdout);
 520        fputs(reset, stdout);
 521        if (((i == len) || line[i] == '\n') && i != col0) {
 522                /* The whole line was indent */
 523                emit_line(ws, reset, line + col0, len - col0);
 524                return;
 525        }
 526        i = col0;
 527        if (need_highlight_leading_space) {
 528                while (i < last_tab_in_indent) {
 529                        if (line[i] == ' ') {
 530                                fputs(ws, stdout);
 531                                putchar(' ');
 532                                fputs(reset, stdout);
 533                        }
 534                        else
 535                                putchar(line[i]);
 536                        i++;
 537                }
 538        }
 539        tail = len - 1;
 540        if (line[tail] == '\n' && i < tail)
 541                tail--;
 542        while (i < tail) {
 543                if (!isspace(line[tail]))
 544                        break;
 545                tail--;
 546        }
 547        if ((i < tail && line[tail + 1] != '\n')) {
 548                /* This has whitespace between tail+1..len */
 549                fputs(set, stdout);
 550                fwrite(line + i, tail - i + 1, 1, stdout);
 551                fputs(reset, stdout);
 552                emit_line(ws, reset, line + tail + 1, len - tail - 1);
 553        }
 554        else
 555                emit_line(set, reset, line + i, len - i);
 556}
 557
 558static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
 559{
 560        const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
 561        const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
 562
 563        if (!*ws)
 564                emit_line(set, reset, line, len);
 565        else
 566                emit_line_with_ws(ecbdata->nparents, set, reset, ws,
 567                                line, len);
 568}
 569
 570static void fn_out_consume(void *priv, char *line, unsigned long len)
 571{
 572        int i;
 573        int color;
 574        struct emit_callback *ecbdata = priv;
 575        const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
 576        const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
 577
 578        *(ecbdata->found_changesp) = 1;
 579
 580        if (ecbdata->label_path[0]) {
 581                const char *name_a_tab, *name_b_tab;
 582
 583                name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
 584                name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
 585
 586                printf("%s--- %s%s%s\n",
 587                       set, ecbdata->label_path[0], reset, name_a_tab);
 588                printf("%s+++ %s%s%s\n",
 589                       set, ecbdata->label_path[1], reset, name_b_tab);
 590                ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
 591        }
 592
 593        /* This is not really necessary for now because
 594         * this codepath only deals with two-way diffs.
 595         */
 596        for (i = 0; i < len && line[i] == '@'; i++)
 597                ;
 598        if (2 <= i && i < len && line[i] == ' ') {
 599                ecbdata->nparents = i - 1;
 600                emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
 601                          reset, line, len);
 602                return;
 603        }
 604
 605        if (len < ecbdata->nparents) {
 606                set = reset;
 607                emit_line(reset, reset, line, len);
 608                return;
 609        }
 610
 611        color = DIFF_PLAIN;
 612        if (ecbdata->diff_words && ecbdata->nparents != 1)
 613                /* fall back to normal diff */
 614                free_diff_words_data(ecbdata);
 615        if (ecbdata->diff_words) {
 616                if (line[0] == '-') {
 617                        diff_words_append(line, len,
 618                                          &ecbdata->diff_words->minus);
 619                        return;
 620                } else if (line[0] == '+') {
 621                        diff_words_append(line, len,
 622                                          &ecbdata->diff_words->plus);
 623                        return;
 624                }
 625                if (ecbdata->diff_words->minus.text.size ||
 626                    ecbdata->diff_words->plus.text.size)
 627                        diff_words_show(ecbdata->diff_words);
 628                line++;
 629                len--;
 630                emit_line(set, reset, line, len);
 631                return;
 632        }
 633        for (i = 0; i < ecbdata->nparents && len; i++) {
 634                if (line[i] == '-')
 635                        color = DIFF_FILE_OLD;
 636                else if (line[i] == '+')
 637                        color = DIFF_FILE_NEW;
 638        }
 639
 640        if (color != DIFF_FILE_NEW) {
 641                emit_line(diff_get_color(ecbdata->color_diff, color),
 642                          reset, line, len);
 643                return;
 644        }
 645        emit_add_line(reset, ecbdata, line, len);
 646}
 647
 648static char *pprint_rename(const char *a, const char *b)
 649{
 650        const char *old = a;
 651        const char *new = b;
 652        struct strbuf name;
 653        int pfx_length, sfx_length;
 654        int len_a = strlen(a);
 655        int len_b = strlen(b);
 656        int a_midlen, b_midlen;
 657        int qlen_a = quote_c_style(a, NULL, NULL, 0);
 658        int qlen_b = quote_c_style(b, NULL, NULL, 0);
 659
 660        strbuf_init(&name, 0);
 661        if (qlen_a || qlen_b) {
 662                quote_c_style(a, &name, NULL, 0);
 663                strbuf_addstr(&name, " => ");
 664                quote_c_style(b, &name, NULL, 0);
 665                return strbuf_detach(&name, NULL);
 666        }
 667
 668        /* Find common prefix */
 669        pfx_length = 0;
 670        while (*old && *new && *old == *new) {
 671                if (*old == '/')
 672                        pfx_length = old - a + 1;
 673                old++;
 674                new++;
 675        }
 676
 677        /* Find common suffix */
 678        old = a + len_a;
 679        new = b + len_b;
 680        sfx_length = 0;
 681        while (a <= old && b <= new && *old == *new) {
 682                if (*old == '/')
 683                        sfx_length = len_a - (old - a);
 684                old--;
 685                new--;
 686        }
 687
 688        /*
 689         * pfx{mid-a => mid-b}sfx
 690         * {pfx-a => pfx-b}sfx
 691         * pfx{sfx-a => sfx-b}
 692         * name-a => name-b
 693         */
 694        a_midlen = len_a - pfx_length - sfx_length;
 695        b_midlen = len_b - pfx_length - sfx_length;
 696        if (a_midlen < 0)
 697                a_midlen = 0;
 698        if (b_midlen < 0)
 699                b_midlen = 0;
 700
 701        strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
 702        if (pfx_length + sfx_length) {
 703                strbuf_add(&name, a, pfx_length);
 704                strbuf_addch(&name, '{');
 705        }
 706        strbuf_add(&name, a + pfx_length, a_midlen);
 707        strbuf_addstr(&name, " => ");
 708        strbuf_add(&name, b + pfx_length, b_midlen);
 709        if (pfx_length + sfx_length) {
 710                strbuf_addch(&name, '}');
 711                strbuf_add(&name, a + len_a - sfx_length, sfx_length);
 712        }
 713        return strbuf_detach(&name, NULL);
 714}
 715
 716struct diffstat_t {
 717        struct xdiff_emit_state xm;
 718
 719        int nr;
 720        int alloc;
 721        struct diffstat_file {
 722                char *name;
 723                unsigned is_unmerged:1;
 724                unsigned is_binary:1;
 725                unsigned is_renamed:1;
 726                unsigned int added, deleted;
 727        } **files;
 728};
 729
 730static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
 731                                          const char *name_a,
 732                                          const char *name_b)
 733{
 734        struct diffstat_file *x;
 735        x = xcalloc(sizeof (*x), 1);
 736        if (diffstat->nr == diffstat->alloc) {
 737                diffstat->alloc = alloc_nr(diffstat->alloc);
 738                diffstat->files = xrealloc(diffstat->files,
 739                                diffstat->alloc * sizeof(x));
 740        }
 741        diffstat->files[diffstat->nr++] = x;
 742        if (name_b) {
 743                x->name = pprint_rename(name_a, name_b);
 744                x->is_renamed = 1;
 745        }
 746        else
 747                x->name = xstrdup(name_a);
 748        return x;
 749}
 750
 751static void diffstat_consume(void *priv, char *line, unsigned long len)
 752{
 753        struct diffstat_t *diffstat = priv;
 754        struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
 755
 756        if (line[0] == '+')
 757                x->added++;
 758        else if (line[0] == '-')
 759                x->deleted++;
 760}
 761
 762const char mime_boundary_leader[] = "------------";
 763
 764static int scale_linear(int it, int width, int max_change)
 765{
 766        /*
 767         * make sure that at least one '-' is printed if there were deletions,
 768         * and likewise for '+'.
 769         */
 770        if (max_change < 2)
 771                return it;
 772        return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
 773}
 774
 775static void show_name(const char *prefix, const char *name, int len,
 776                      const char *reset, const char *set)
 777{
 778        printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
 779}
 780
 781static void show_graph(char ch, int cnt, const char *set, const char *reset)
 782{
 783        if (cnt <= 0)
 784                return;
 785        printf("%s", set);
 786        while (cnt--)
 787                putchar(ch);
 788        printf("%s", reset);
 789}
 790
 791static void show_stats(struct diffstat_t* data, struct diff_options *options)
 792{
 793        int i, len, add, del, total, adds = 0, dels = 0;
 794        int max_change = 0, max_len = 0;
 795        int total_files = data->nr;
 796        int width, name_width;
 797        const char *reset, *set, *add_c, *del_c;
 798
 799        if (data->nr == 0)
 800                return;
 801
 802        width = options->stat_width ? options->stat_width : 80;
 803        name_width = options->stat_name_width ? options->stat_name_width : 50;
 804
 805        /* Sanity: give at least 5 columns to the graph,
 806         * but leave at least 10 columns for the name.
 807         */
 808        if (width < name_width + 15) {
 809                if (name_width <= 25)
 810                        width = name_width + 15;
 811                else
 812                        name_width = width - 15;
 813        }
 814
 815        /* Find the longest filename and max number of changes */
 816        reset = diff_get_color(options->color_diff, DIFF_RESET);
 817        set = diff_get_color(options->color_diff, DIFF_PLAIN);
 818        add_c = diff_get_color(options->color_diff, DIFF_FILE_NEW);
 819        del_c = diff_get_color(options->color_diff, DIFF_FILE_OLD);
 820
 821        for (i = 0; i < data->nr; i++) {
 822                struct diffstat_file *file = data->files[i];
 823                int change = file->added + file->deleted;
 824
 825                if (!file->is_renamed) {  /* renames are already quoted by pprint_rename */
 826                        struct strbuf buf;
 827                        strbuf_init(&buf, 0);
 828                        if (quote_c_style(file->name, &buf, NULL, 0)) {
 829                                free(file->name);
 830                                file->name = strbuf_detach(&buf, NULL);
 831                        } else {
 832                                strbuf_release(&buf);
 833                        }
 834                }
 835
 836                len = strlen(file->name);
 837                if (max_len < len)
 838                        max_len = len;
 839
 840                if (file->is_binary || file->is_unmerged)
 841                        continue;
 842                if (max_change < change)
 843                        max_change = change;
 844        }
 845
 846        /* Compute the width of the graph part;
 847         * 10 is for one blank at the beginning of the line plus
 848         * " | count " between the name and the graph.
 849         *
 850         * From here on, name_width is the width of the name area,
 851         * and width is the width of the graph area.
 852         */
 853        name_width = (name_width < max_len) ? name_width : max_len;
 854        if (width < (name_width + 10) + max_change)
 855                width = width - (name_width + 10);
 856        else
 857                width = max_change;
 858
 859        for (i = 0; i < data->nr; i++) {
 860                const char *prefix = "";
 861                char *name = data->files[i]->name;
 862                int added = data->files[i]->added;
 863                int deleted = data->files[i]->deleted;
 864                int name_len;
 865
 866                /*
 867                 * "scale" the filename
 868                 */
 869                len = name_width;
 870                name_len = strlen(name);
 871                if (name_width < name_len) {
 872                        char *slash;
 873                        prefix = "...";
 874                        len -= 3;
 875                        name += name_len - len;
 876                        slash = strchr(name, '/');
 877                        if (slash)
 878                                name = slash;
 879                }
 880
 881                if (data->files[i]->is_binary) {
 882                        show_name(prefix, name, len, reset, set);
 883                        printf("  Bin ");
 884                        printf("%s%d%s", del_c, deleted, reset);
 885                        printf(" -> ");
 886                        printf("%s%d%s", add_c, added, reset);
 887                        printf(" bytes");
 888                        printf("\n");
 889                        goto free_diffstat_file;
 890                }
 891                else if (data->files[i]->is_unmerged) {
 892                        show_name(prefix, name, len, reset, set);
 893                        printf("  Unmerged\n");
 894                        goto free_diffstat_file;
 895                }
 896                else if (!data->files[i]->is_renamed &&
 897                         (added + deleted == 0)) {
 898                        total_files--;
 899                        goto free_diffstat_file;
 900                }
 901
 902                /*
 903                 * scale the add/delete
 904                 */
 905                add = added;
 906                del = deleted;
 907                total = add + del;
 908                adds += add;
 909                dels += del;
 910
 911                if (width <= max_change) {
 912                        add = scale_linear(add, width, max_change);
 913                        del = scale_linear(del, width, max_change);
 914                        total = add + del;
 915                }
 916                show_name(prefix, name, len, reset, set);
 917                printf("%5d ", added + deleted);
 918                show_graph('+', add, add_c, reset);
 919                show_graph('-', del, del_c, reset);
 920                putchar('\n');
 921        free_diffstat_file:
 922                free(data->files[i]->name);
 923                free(data->files[i]);
 924        }
 925        free(data->files);
 926        printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
 927               set, total_files, adds, dels, reset);
 928}
 929
 930static void show_shortstats(struct diffstat_t* data)
 931{
 932        int i, adds = 0, dels = 0, total_files = data->nr;
 933
 934        if (data->nr == 0)
 935                return;
 936
 937        for (i = 0; i < data->nr; i++) {
 938                if (!data->files[i]->is_binary &&
 939                    !data->files[i]->is_unmerged) {
 940                        int added = data->files[i]->added;
 941                        int deleted= data->files[i]->deleted;
 942                        if (!data->files[i]->is_renamed &&
 943                            (added + deleted == 0)) {
 944                                total_files--;
 945                        } else {
 946                                adds += added;
 947                                dels += deleted;
 948                        }
 949                }
 950                free(data->files[i]->name);
 951                free(data->files[i]);
 952        }
 953        free(data->files);
 954
 955        printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
 956               total_files, adds, dels);
 957}
 958
 959static void show_numstat(struct diffstat_t* data, struct diff_options *options)
 960{
 961        int i;
 962
 963        for (i = 0; i < data->nr; i++) {
 964                struct diffstat_file *file = data->files[i];
 965
 966                if (file->is_binary)
 967                        printf("-\t-\t");
 968                else
 969                        printf("%d\t%d\t", file->added, file->deleted);
 970                if (!file->is_renamed) {
 971                        write_name_quoted(file->name, stdout, options->line_termination);
 972                } else {
 973                        fputs(file->name, stdout);
 974                        putchar(options->line_termination);
 975                }
 976        }
 977}
 978
 979struct checkdiff_t {
 980        struct xdiff_emit_state xm;
 981        const char *filename;
 982        int lineno, color_diff;
 983};
 984
 985static void checkdiff_consume(void *priv, char *line, unsigned long len)
 986{
 987        struct checkdiff_t *data = priv;
 988        const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
 989        const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
 990        const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
 991
 992        if (line[0] == '+') {
 993                int i, spaces = 0, space_before_tab = 0, white_space_at_end = 0;
 994
 995                /* check space before tab */
 996                for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
 997                        if (line[i] == ' ')
 998                                spaces++;
 999                if (line[i - 1] == '\t' && spaces)
1000                        space_before_tab = 1;
1001
1002                /* check white space at line end */
1003                if (line[len - 1] == '\n')
1004                        len--;
1005                if (isspace(line[len - 1]))
1006                        white_space_at_end = 1;
1007
1008                if (space_before_tab || white_space_at_end) {
1009                        printf("%s:%d: %s", data->filename, data->lineno, ws);
1010                        if (space_before_tab) {
1011                                printf("space before tab");
1012                                if (white_space_at_end)
1013                                        putchar(',');
1014                        }
1015                        if (white_space_at_end)
1016                                printf("white space at end");
1017                        printf(":%s ", reset);
1018                        emit_line_with_ws(1, set, reset, ws, line, len);
1019                }
1020
1021                data->lineno++;
1022        } else if (line[0] == ' ')
1023                data->lineno++;
1024        else if (line[0] == '@') {
1025                char *plus = strchr(line, '+');
1026                if (plus)
1027                        data->lineno = strtol(plus, NULL, 10);
1028                else
1029                        die("invalid diff");
1030        }
1031}
1032
1033static unsigned char *deflate_it(char *data,
1034                                 unsigned long size,
1035                                 unsigned long *result_size)
1036{
1037        int bound;
1038        unsigned char *deflated;
1039        z_stream stream;
1040
1041        memset(&stream, 0, sizeof(stream));
1042        deflateInit(&stream, zlib_compression_level);
1043        bound = deflateBound(&stream, size);
1044        deflated = xmalloc(bound);
1045        stream.next_out = deflated;
1046        stream.avail_out = bound;
1047
1048        stream.next_in = (unsigned char *)data;
1049        stream.avail_in = size;
1050        while (deflate(&stream, Z_FINISH) == Z_OK)
1051                ; /* nothing */
1052        deflateEnd(&stream);
1053        *result_size = stream.total_out;
1054        return deflated;
1055}
1056
1057static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1058{
1059        void *cp;
1060        void *delta;
1061        void *deflated;
1062        void *data;
1063        unsigned long orig_size;
1064        unsigned long delta_size;
1065        unsigned long deflate_size;
1066        unsigned long data_size;
1067
1068        /* We could do deflated delta, or we could do just deflated two,
1069         * whichever is smaller.
1070         */
1071        delta = NULL;
1072        deflated = deflate_it(two->ptr, two->size, &deflate_size);
1073        if (one->size && two->size) {
1074                delta = diff_delta(one->ptr, one->size,
1075                                   two->ptr, two->size,
1076                                   &delta_size, deflate_size);
1077                if (delta) {
1078                        void *to_free = delta;
1079                        orig_size = delta_size;
1080                        delta = deflate_it(delta, delta_size, &delta_size);
1081                        free(to_free);
1082                }
1083        }
1084
1085        if (delta && delta_size < deflate_size) {
1086                printf("delta %lu\n", orig_size);
1087                free(deflated);
1088                data = delta;
1089                data_size = delta_size;
1090        }
1091        else {
1092                printf("literal %lu\n", two->size);
1093                free(delta);
1094                data = deflated;
1095                data_size = deflate_size;
1096        }
1097
1098        /* emit data encoded in base85 */
1099        cp = data;
1100        while (data_size) {
1101                int bytes = (52 < data_size) ? 52 : data_size;
1102                char line[70];
1103                data_size -= bytes;
1104                if (bytes <= 26)
1105                        line[0] = bytes + 'A' - 1;
1106                else
1107                        line[0] = bytes - 26 + 'a' - 1;
1108                encode_85(line + 1, cp, bytes);
1109                cp = (char *) cp + bytes;
1110                puts(line);
1111        }
1112        printf("\n");
1113        free(data);
1114}
1115
1116static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1117{
1118        printf("GIT binary patch\n");
1119        emit_binary_diff_body(one, two);
1120        emit_binary_diff_body(two, one);
1121}
1122
1123static void setup_diff_attr_check(struct git_attr_check *check)
1124{
1125        static struct git_attr *attr_diff;
1126
1127        if (!attr_diff) {
1128                attr_diff = git_attr("diff", 4);
1129        }
1130        check[0].attr = attr_diff;
1131}
1132
1133static void diff_filespec_check_attr(struct diff_filespec *one)
1134{
1135        struct git_attr_check attr_diff_check;
1136        int check_from_data = 0;
1137
1138        if (one->checked_attr)
1139                return;
1140
1141        setup_diff_attr_check(&attr_diff_check);
1142        one->is_binary = 0;
1143        one->funcname_pattern_ident = NULL;
1144
1145        if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1146                const char *value;
1147
1148                /* binaryness */
1149                value = attr_diff_check.value;
1150                if (ATTR_TRUE(value))
1151                        ;
1152                else if (ATTR_FALSE(value))
1153                        one->is_binary = 1;
1154                else
1155                        check_from_data = 1;
1156
1157                /* funcname pattern ident */
1158                if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1159                        ;
1160                else
1161                        one->funcname_pattern_ident = value;
1162        }
1163
1164        if (check_from_data) {
1165                if (!one->data && DIFF_FILE_VALID(one))
1166                        diff_populate_filespec(one, 0);
1167
1168                if (one->data)
1169                        one->is_binary = buffer_is_binary(one->data, one->size);
1170        }
1171}
1172
1173int diff_filespec_is_binary(struct diff_filespec *one)
1174{
1175        diff_filespec_check_attr(one);
1176        return one->is_binary;
1177}
1178
1179static const char *funcname_pattern(const char *ident)
1180{
1181        struct funcname_pattern *pp;
1182
1183        read_config_if_needed();
1184        for (pp = funcname_pattern_list; pp; pp = pp->next)
1185                if (!strcmp(ident, pp->name))
1186                        return pp->pattern;
1187        return NULL;
1188}
1189
1190static struct builtin_funcname_pattern {
1191        const char *name;
1192        const char *pattern;
1193} builtin_funcname_pattern[] = {
1194        { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1195                        "new\\|return\\|switch\\|throw\\|while\\)\n"
1196                        "^[     ]*\\(\\([       ]*"
1197                        "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1198                        "[      ]*([^;]*$\\)" },
1199        { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1200};
1201
1202static const char *diff_funcname_pattern(struct diff_filespec *one)
1203{
1204        const char *ident, *pattern;
1205        int i;
1206
1207        diff_filespec_check_attr(one);
1208        ident = one->funcname_pattern_ident;
1209
1210        if (!ident)
1211                /*
1212                 * If the config file has "funcname.default" defined, that
1213                 * regexp is used; otherwise NULL is returned and xemit uses
1214                 * the built-in default.
1215                 */
1216                return funcname_pattern("default");
1217
1218        /* Look up custom "funcname.$ident" regexp from config. */
1219        pattern = funcname_pattern(ident);
1220        if (pattern)
1221                return pattern;
1222
1223        /*
1224         * And define built-in fallback patterns here.  Note that
1225         * these can be overriden by the user's config settings.
1226         */
1227        for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1228                if (!strcmp(ident, builtin_funcname_pattern[i].name))
1229                        return builtin_funcname_pattern[i].pattern;
1230
1231        return NULL;
1232}
1233
1234static void builtin_diff(const char *name_a,
1235                         const char *name_b,
1236                         struct diff_filespec *one,
1237                         struct diff_filespec *two,
1238                         const char *xfrm_msg,
1239                         struct diff_options *o,
1240                         int complete_rewrite)
1241{
1242        mmfile_t mf1, mf2;
1243        const char *lbl[2];
1244        char *a_one, *b_two;
1245        const char *set = diff_get_color(o->color_diff, DIFF_METAINFO);
1246        const char *reset = diff_get_color(o->color_diff, DIFF_RESET);
1247
1248        a_one = quote_two("a/", name_a + (*name_a == '/'));
1249        b_two = quote_two("b/", name_b + (*name_b == '/'));
1250        lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1251        lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1252        printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1253        if (lbl[0][0] == '/') {
1254                /* /dev/null */
1255                printf("%snew file mode %06o%s\n", set, two->mode, reset);
1256                if (xfrm_msg && xfrm_msg[0])
1257                        printf("%s%s%s\n", set, xfrm_msg, reset);
1258        }
1259        else if (lbl[1][0] == '/') {
1260                printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1261                if (xfrm_msg && xfrm_msg[0])
1262                        printf("%s%s%s\n", set, xfrm_msg, reset);
1263        }
1264        else {
1265                if (one->mode != two->mode) {
1266                        printf("%sold mode %06o%s\n", set, one->mode, reset);
1267                        printf("%snew mode %06o%s\n", set, two->mode, reset);
1268                }
1269                if (xfrm_msg && xfrm_msg[0])
1270                        printf("%s%s%s\n", set, xfrm_msg, reset);
1271                /*
1272                 * we do not run diff between different kind
1273                 * of objects.
1274                 */
1275                if ((one->mode ^ two->mode) & S_IFMT)
1276                        goto free_ab_and_return;
1277                if (complete_rewrite) {
1278                        emit_rewrite_diff(name_a, name_b, one, two,
1279                                        o->color_diff);
1280                        o->found_changes = 1;
1281                        goto free_ab_and_return;
1282                }
1283        }
1284
1285        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1286                die("unable to read files to diff");
1287
1288        if (!o->text &&
1289            (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1290                /* Quite common confusing case */
1291                if (mf1.size == mf2.size &&
1292                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1293                        goto free_ab_and_return;
1294                if (o->binary)
1295                        emit_binary_diff(&mf1, &mf2);
1296                else
1297                        printf("Binary files %s and %s differ\n",
1298                               lbl[0], lbl[1]);
1299                o->found_changes = 1;
1300        }
1301        else {
1302                /* Crazy xdl interfaces.. */
1303                const char *diffopts = getenv("GIT_DIFF_OPTS");
1304                xpparam_t xpp;
1305                xdemitconf_t xecfg;
1306                xdemitcb_t ecb;
1307                struct emit_callback ecbdata;
1308                const char *funcname_pattern;
1309
1310                funcname_pattern = diff_funcname_pattern(one);
1311                if (!funcname_pattern)
1312                        funcname_pattern = diff_funcname_pattern(two);
1313
1314                memset(&xecfg, 0, sizeof(xecfg));
1315                memset(&ecbdata, 0, sizeof(ecbdata));
1316                ecbdata.label_path = lbl;
1317                ecbdata.color_diff = o->color_diff;
1318                ecbdata.found_changesp = &o->found_changes;
1319                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1320                xecfg.ctxlen = o->context;
1321                xecfg.flags = XDL_EMIT_FUNCNAMES;
1322                if (funcname_pattern)
1323                        xdiff_set_find_func(&xecfg, funcname_pattern);
1324                if (!diffopts)
1325                        ;
1326                else if (!prefixcmp(diffopts, "--unified="))
1327                        xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1328                else if (!prefixcmp(diffopts, "-u"))
1329                        xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1330                ecb.outf = xdiff_outf;
1331                ecb.priv = &ecbdata;
1332                ecbdata.xm.consume = fn_out_consume;
1333                if (o->color_diff_words)
1334                        ecbdata.diff_words =
1335                                xcalloc(1, sizeof(struct diff_words_data));
1336                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1337                if (o->color_diff_words)
1338                        free_diff_words_data(&ecbdata);
1339        }
1340
1341 free_ab_and_return:
1342        diff_free_filespec_data(one);
1343        diff_free_filespec_data(two);
1344        free(a_one);
1345        free(b_two);
1346        return;
1347}
1348
1349static void builtin_diffstat(const char *name_a, const char *name_b,
1350                             struct diff_filespec *one,
1351                             struct diff_filespec *two,
1352                             struct diffstat_t *diffstat,
1353                             struct diff_options *o,
1354                             int complete_rewrite)
1355{
1356        mmfile_t mf1, mf2;
1357        struct diffstat_file *data;
1358
1359        data = diffstat_add(diffstat, name_a, name_b);
1360
1361        if (!one || !two) {
1362                data->is_unmerged = 1;
1363                return;
1364        }
1365        if (complete_rewrite) {
1366                diff_populate_filespec(one, 0);
1367                diff_populate_filespec(two, 0);
1368                data->deleted = count_lines(one->data, one->size);
1369                data->added = count_lines(two->data, two->size);
1370                goto free_and_return;
1371        }
1372        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1373                die("unable to read files to diff");
1374
1375        if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1376                data->is_binary = 1;
1377                data->added = mf2.size;
1378                data->deleted = mf1.size;
1379        } else {
1380                /* Crazy xdl interfaces.. */
1381                xpparam_t xpp;
1382                xdemitconf_t xecfg;
1383                xdemitcb_t ecb;
1384
1385                memset(&xecfg, 0, sizeof(xecfg));
1386                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1387                ecb.outf = xdiff_outf;
1388                ecb.priv = diffstat;
1389                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1390        }
1391
1392 free_and_return:
1393        diff_free_filespec_data(one);
1394        diff_free_filespec_data(two);
1395}
1396
1397static void builtin_checkdiff(const char *name_a, const char *name_b,
1398                             struct diff_filespec *one,
1399                             struct diff_filespec *two, struct diff_options *o)
1400{
1401        mmfile_t mf1, mf2;
1402        struct checkdiff_t data;
1403
1404        if (!two)
1405                return;
1406
1407        memset(&data, 0, sizeof(data));
1408        data.xm.consume = checkdiff_consume;
1409        data.filename = name_b ? name_b : name_a;
1410        data.lineno = 0;
1411        data.color_diff = o->color_diff;
1412
1413        if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1414                die("unable to read files to diff");
1415
1416        if (diff_filespec_is_binary(two))
1417                goto free_and_return;
1418        else {
1419                /* Crazy xdl interfaces.. */
1420                xpparam_t xpp;
1421                xdemitconf_t xecfg;
1422                xdemitcb_t ecb;
1423
1424                memset(&xecfg, 0, sizeof(xecfg));
1425                xpp.flags = XDF_NEED_MINIMAL;
1426                ecb.outf = xdiff_outf;
1427                ecb.priv = &data;
1428                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1429        }
1430 free_and_return:
1431        diff_free_filespec_data(one);
1432        diff_free_filespec_data(two);
1433}
1434
1435struct diff_filespec *alloc_filespec(const char *path)
1436{
1437        int namelen = strlen(path);
1438        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1439
1440        memset(spec, 0, sizeof(*spec));
1441        spec->path = (char *)(spec + 1);
1442        memcpy(spec->path, path, namelen+1);
1443        return spec;
1444}
1445
1446void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1447                   unsigned short mode)
1448{
1449        if (mode) {
1450                spec->mode = canon_mode(mode);
1451                hashcpy(spec->sha1, sha1);
1452                spec->sha1_valid = !is_null_sha1(sha1);
1453        }
1454}
1455
1456/*
1457 * Given a name and sha1 pair, if the index tells us the file in
1458 * the work tree has that object contents, return true, so that
1459 * prepare_temp_file() does not have to inflate and extract.
1460 */
1461static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1462{
1463        struct cache_entry *ce;
1464        struct stat st;
1465        int pos, len;
1466
1467        /* We do not read the cache ourselves here, because the
1468         * benchmark with my previous version that always reads cache
1469         * shows that it makes things worse for diff-tree comparing
1470         * two linux-2.6 kernel trees in an already checked out work
1471         * tree.  This is because most diff-tree comparisons deal with
1472         * only a small number of files, while reading the cache is
1473         * expensive for a large project, and its cost outweighs the
1474         * savings we get by not inflating the object to a temporary
1475         * file.  Practically, this code only helps when we are used
1476         * by diff-cache --cached, which does read the cache before
1477         * calling us.
1478         */
1479        if (!active_cache)
1480                return 0;
1481
1482        /* We want to avoid the working directory if our caller
1483         * doesn't need the data in a normal file, this system
1484         * is rather slow with its stat/open/mmap/close syscalls,
1485         * and the object is contained in a pack file.  The pack
1486         * is probably already open and will be faster to obtain
1487         * the data through than the working directory.  Loose
1488         * objects however would tend to be slower as they need
1489         * to be individually opened and inflated.
1490         */
1491        if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1492                return 0;
1493
1494        len = strlen(name);
1495        pos = cache_name_pos(name, len);
1496        if (pos < 0)
1497                return 0;
1498        ce = active_cache[pos];
1499        if ((lstat(name, &st) < 0) ||
1500            !S_ISREG(st.st_mode) || /* careful! */
1501            ce_match_stat(ce, &st, 0) ||
1502            hashcmp(sha1, ce->sha1))
1503                return 0;
1504        /* we return 1 only when we can stat, it is a regular file,
1505         * stat information matches, and sha1 recorded in the cache
1506         * matches.  I.e. we know the file in the work tree really is
1507         * the same as the <name, sha1> pair.
1508         */
1509        return 1;
1510}
1511
1512static int populate_from_stdin(struct diff_filespec *s)
1513{
1514        struct strbuf buf;
1515        size_t size = 0;
1516
1517        strbuf_init(&buf, 0);
1518        if (strbuf_read(&buf, 0, 0) < 0)
1519                return error("error while reading from stdin %s",
1520                                     strerror(errno));
1521
1522        s->should_munmap = 0;
1523        s->data = strbuf_detach(&buf, &size);
1524        s->size = size;
1525        s->should_free = 1;
1526        return 0;
1527}
1528
1529static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1530{
1531        int len;
1532        char *data = xmalloc(100);
1533        len = snprintf(data, 100,
1534                "Subproject commit %s\n", sha1_to_hex(s->sha1));
1535        s->data = data;
1536        s->size = len;
1537        s->should_free = 1;
1538        if (size_only) {
1539                s->data = NULL;
1540                free(data);
1541        }
1542        return 0;
1543}
1544
1545/*
1546 * While doing rename detection and pickaxe operation, we may need to
1547 * grab the data for the blob (or file) for our own in-core comparison.
1548 * diff_filespec has data and size fields for this purpose.
1549 */
1550int diff_populate_filespec(struct diff_filespec *s, int size_only)
1551{
1552        int err = 0;
1553        if (!DIFF_FILE_VALID(s))
1554                die("internal error: asking to populate invalid file.");
1555        if (S_ISDIR(s->mode))
1556                return -1;
1557
1558        if (s->data)
1559                return 0;
1560
1561        if (size_only && 0 < s->size)
1562                return 0;
1563
1564        if (S_ISGITLINK(s->mode))
1565                return diff_populate_gitlink(s, size_only);
1566
1567        if (!s->sha1_valid ||
1568            reuse_worktree_file(s->path, s->sha1, 0)) {
1569                struct strbuf buf;
1570                struct stat st;
1571                int fd;
1572
1573                if (!strcmp(s->path, "-"))
1574                        return populate_from_stdin(s);
1575
1576                if (lstat(s->path, &st) < 0) {
1577                        if (errno == ENOENT) {
1578                        err_empty:
1579                                err = -1;
1580                        empty:
1581                                s->data = (char *)"";
1582                                s->size = 0;
1583                                return err;
1584                        }
1585                }
1586                s->size = xsize_t(st.st_size);
1587                if (!s->size)
1588                        goto empty;
1589                if (size_only)
1590                        return 0;
1591                if (S_ISLNK(st.st_mode)) {
1592                        int ret;
1593                        s->data = xmalloc(s->size);
1594                        s->should_free = 1;
1595                        ret = readlink(s->path, s->data, s->size);
1596                        if (ret < 0) {
1597                                free(s->data);
1598                                goto err_empty;
1599                        }
1600                        return 0;
1601                }
1602                fd = open(s->path, O_RDONLY);
1603                if (fd < 0)
1604                        goto err_empty;
1605                s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1606                close(fd);
1607                s->should_munmap = 1;
1608
1609                /*
1610                 * Convert from working tree format to canonical git format
1611                 */
1612                strbuf_init(&buf, 0);
1613                if (convert_to_git(s->path, s->data, s->size, &buf)) {
1614                        size_t size = 0;
1615                        munmap(s->data, s->size);
1616                        s->should_munmap = 0;
1617                        s->data = strbuf_detach(&buf, &size);
1618                        s->size = size;
1619                        s->should_free = 1;
1620                }
1621        }
1622        else {
1623                enum object_type type;
1624                if (size_only)
1625                        type = sha1_object_info(s->sha1, &s->size);
1626                else {
1627                        s->data = read_sha1_file(s->sha1, &type, &s->size);
1628                        s->should_free = 1;
1629                }
1630        }
1631        return 0;
1632}
1633
1634void diff_free_filespec_blob(struct diff_filespec *s)
1635{
1636        if (s->should_free)
1637                free(s->data);
1638        else if (s->should_munmap)
1639                munmap(s->data, s->size);
1640
1641        if (s->should_free || s->should_munmap) {
1642                s->should_free = s->should_munmap = 0;
1643                s->data = NULL;
1644        }
1645}
1646
1647void diff_free_filespec_data(struct diff_filespec *s)
1648{
1649        diff_free_filespec_blob(s);
1650        free(s->cnt_data);
1651        s->cnt_data = NULL;
1652}
1653
1654static void prep_temp_blob(struct diff_tempfile *temp,
1655                           void *blob,
1656                           unsigned long size,
1657                           const unsigned char *sha1,
1658                           int mode)
1659{
1660        int fd;
1661
1662        fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1663        if (fd < 0)
1664                die("unable to create temp-file: %s", strerror(errno));
1665        if (write_in_full(fd, blob, size) != size)
1666                die("unable to write temp-file");
1667        close(fd);
1668        temp->name = temp->tmp_path;
1669        strcpy(temp->hex, sha1_to_hex(sha1));
1670        temp->hex[40] = 0;
1671        sprintf(temp->mode, "%06o", mode);
1672}
1673
1674static void prepare_temp_file(const char *name,
1675                              struct diff_tempfile *temp,
1676                              struct diff_filespec *one)
1677{
1678        if (!DIFF_FILE_VALID(one)) {
1679        not_a_valid_file:
1680                /* A '-' entry produces this for file-2, and
1681                 * a '+' entry produces this for file-1.
1682                 */
1683                temp->name = "/dev/null";
1684                strcpy(temp->hex, ".");
1685                strcpy(temp->mode, ".");
1686                return;
1687        }
1688
1689        if (!one->sha1_valid ||
1690            reuse_worktree_file(name, one->sha1, 1)) {
1691                struct stat st;
1692                if (lstat(name, &st) < 0) {
1693                        if (errno == ENOENT)
1694                                goto not_a_valid_file;
1695                        die("stat(%s): %s", name, strerror(errno));
1696                }
1697                if (S_ISLNK(st.st_mode)) {
1698                        int ret;
1699                        char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1700                        size_t sz = xsize_t(st.st_size);
1701                        if (sizeof(buf) <= st.st_size)
1702                                die("symlink too long: %s", name);
1703                        ret = readlink(name, buf, sz);
1704                        if (ret < 0)
1705                                die("readlink(%s)", name);
1706                        prep_temp_blob(temp, buf, sz,
1707                                       (one->sha1_valid ?
1708                                        one->sha1 : null_sha1),
1709                                       (one->sha1_valid ?
1710                                        one->mode : S_IFLNK));
1711                }
1712                else {
1713                        /* we can borrow from the file in the work tree */
1714                        temp->name = name;
1715                        if (!one->sha1_valid)
1716                                strcpy(temp->hex, sha1_to_hex(null_sha1));
1717                        else
1718                                strcpy(temp->hex, sha1_to_hex(one->sha1));
1719                        /* Even though we may sometimes borrow the
1720                         * contents from the work tree, we always want
1721                         * one->mode.  mode is trustworthy even when
1722                         * !(one->sha1_valid), as long as
1723                         * DIFF_FILE_VALID(one).
1724                         */
1725                        sprintf(temp->mode, "%06o", one->mode);
1726                }
1727                return;
1728        }
1729        else {
1730                if (diff_populate_filespec(one, 0))
1731                        die("cannot read data blob for %s", one->path);
1732                prep_temp_blob(temp, one->data, one->size,
1733                               one->sha1, one->mode);
1734        }
1735}
1736
1737static void remove_tempfile(void)
1738{
1739        int i;
1740
1741        for (i = 0; i < 2; i++)
1742                if (diff_temp[i].name == diff_temp[i].tmp_path) {
1743                        unlink(diff_temp[i].name);
1744                        diff_temp[i].name = NULL;
1745                }
1746}
1747
1748static void remove_tempfile_on_signal(int signo)
1749{
1750        remove_tempfile();
1751        signal(SIGINT, SIG_DFL);
1752        raise(signo);
1753}
1754
1755static int spawn_prog(const char *pgm, const char **arg)
1756{
1757        pid_t pid;
1758        int status;
1759
1760        fflush(NULL);
1761        pid = fork();
1762        if (pid < 0)
1763                die("unable to fork");
1764        if (!pid) {
1765                execvp(pgm, (char *const*) arg);
1766                exit(255);
1767        }
1768
1769        while (waitpid(pid, &status, 0) < 0) {
1770                if (errno == EINTR)
1771                        continue;
1772                return -1;
1773        }
1774
1775        /* Earlier we did not check the exit status because
1776         * diff exits non-zero if files are different, and
1777         * we are not interested in knowing that.  It was a
1778         * mistake which made it harder to quit a diff-*
1779         * session that uses the git-apply-patch-script as
1780         * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
1781         * should also exit non-zero only when it wants to
1782         * abort the entire diff-* session.
1783         */
1784        if (WIFEXITED(status) && !WEXITSTATUS(status))
1785                return 0;
1786        return -1;
1787}
1788
1789/* An external diff command takes:
1790 *
1791 * diff-cmd name infile1 infile1-sha1 infile1-mode \
1792 *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1793 *
1794 */
1795static void run_external_diff(const char *pgm,
1796                              const char *name,
1797                              const char *other,
1798                              struct diff_filespec *one,
1799                              struct diff_filespec *two,
1800                              const char *xfrm_msg,
1801                              int complete_rewrite)
1802{
1803        const char *spawn_arg[10];
1804        struct diff_tempfile *temp = diff_temp;
1805        int retval;
1806        static int atexit_asked = 0;
1807        const char *othername;
1808        const char **arg = &spawn_arg[0];
1809
1810        othername = (other? other : name);
1811        if (one && two) {
1812                prepare_temp_file(name, &temp[0], one);
1813                prepare_temp_file(othername, &temp[1], two);
1814                if (! atexit_asked &&
1815                    (temp[0].name == temp[0].tmp_path ||
1816                     temp[1].name == temp[1].tmp_path)) {
1817                        atexit_asked = 1;
1818                        atexit(remove_tempfile);
1819                }
1820                signal(SIGINT, remove_tempfile_on_signal);
1821        }
1822
1823        if (one && two) {
1824                *arg++ = pgm;
1825                *arg++ = name;
1826                *arg++ = temp[0].name;
1827                *arg++ = temp[0].hex;
1828                *arg++ = temp[0].mode;
1829                *arg++ = temp[1].name;
1830                *arg++ = temp[1].hex;
1831                *arg++ = temp[1].mode;
1832                if (other) {
1833                        *arg++ = other;
1834                        *arg++ = xfrm_msg;
1835                }
1836        } else {
1837                *arg++ = pgm;
1838                *arg++ = name;
1839        }
1840        *arg = NULL;
1841        retval = spawn_prog(pgm, spawn_arg);
1842        remove_tempfile();
1843        if (retval) {
1844                fprintf(stderr, "external diff died, stopping at %s.\n", name);
1845                exit(1);
1846        }
1847}
1848
1849static const char *external_diff_attr(const char *name)
1850{
1851        struct git_attr_check attr_diff_check;
1852
1853        setup_diff_attr_check(&attr_diff_check);
1854        if (!git_checkattr(name, 1, &attr_diff_check)) {
1855                const char *value = attr_diff_check.value;
1856                if (!ATTR_TRUE(value) &&
1857                    !ATTR_FALSE(value) &&
1858                    !ATTR_UNSET(value)) {
1859                        struct ll_diff_driver *drv;
1860
1861                        read_config_if_needed();
1862                        for (drv = user_diff; drv; drv = drv->next)
1863                                if (!strcmp(drv->name, value))
1864                                        return drv->cmd;
1865                }
1866        }
1867        return NULL;
1868}
1869
1870static void run_diff_cmd(const char *pgm,
1871                         const char *name,
1872                         const char *other,
1873                         struct diff_filespec *one,
1874                         struct diff_filespec *two,
1875                         const char *xfrm_msg,
1876                         struct diff_options *o,
1877                         int complete_rewrite)
1878{
1879        if (!o->allow_external)
1880                pgm = NULL;
1881        else {
1882                const char *cmd = external_diff_attr(name);
1883                if (cmd)
1884                        pgm = cmd;
1885        }
1886
1887        if (pgm) {
1888                run_external_diff(pgm, name, other, one, two, xfrm_msg,
1889                                  complete_rewrite);
1890                return;
1891        }
1892        if (one && two)
1893                builtin_diff(name, other ? other : name,
1894                             one, two, xfrm_msg, o, complete_rewrite);
1895        else
1896                printf("* Unmerged path %s\n", name);
1897}
1898
1899static void diff_fill_sha1_info(struct diff_filespec *one)
1900{
1901        if (DIFF_FILE_VALID(one)) {
1902                if (!one->sha1_valid) {
1903                        struct stat st;
1904                        if (!strcmp(one->path, "-")) {
1905                                hashcpy(one->sha1, null_sha1);
1906                                return;
1907                        }
1908                        if (lstat(one->path, &st) < 0)
1909                                die("stat %s", one->path);
1910                        if (index_path(one->sha1, one->path, &st, 0))
1911                                die("cannot hash %s\n", one->path);
1912                }
1913        }
1914        else
1915                hashclr(one->sha1);
1916}
1917
1918static int similarity_index(struct diff_filepair *p)
1919{
1920        return p->score * 100 / MAX_SCORE;
1921}
1922
1923static void run_diff(struct diff_filepair *p, struct diff_options *o)
1924{
1925        const char *pgm = external_diff();
1926        struct strbuf msg;
1927        char *xfrm_msg;
1928        struct diff_filespec *one = p->one;
1929        struct diff_filespec *two = p->two;
1930        const char *name;
1931        const char *other;
1932        int complete_rewrite = 0;
1933
1934
1935        if (DIFF_PAIR_UNMERGED(p)) {
1936                run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1937                return;
1938        }
1939
1940        name  = p->one->path;
1941        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1942        diff_fill_sha1_info(one);
1943        diff_fill_sha1_info(two);
1944
1945        strbuf_init(&msg, PATH_MAX * 2 + 300);
1946        switch (p->status) {
1947        case DIFF_STATUS_COPIED:
1948                strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1949                strbuf_addstr(&msg, "\ncopy from ");
1950                quote_c_style(name, &msg, NULL, 0);
1951                strbuf_addstr(&msg, "\ncopy to ");
1952                quote_c_style(other, &msg, NULL, 0);
1953                strbuf_addch(&msg, '\n');
1954                break;
1955        case DIFF_STATUS_RENAMED:
1956                strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1957                strbuf_addstr(&msg, "\nrename from ");
1958                quote_c_style(name, &msg, NULL, 0);
1959                strbuf_addstr(&msg, "\nrename to ");
1960                quote_c_style(other, &msg, NULL, 0);
1961                strbuf_addch(&msg, '\n');
1962                break;
1963        case DIFF_STATUS_MODIFIED:
1964                if (p->score) {
1965                        strbuf_addf(&msg, "dissimilarity index %d%%\n",
1966                                        similarity_index(p));
1967                        complete_rewrite = 1;
1968                        break;
1969                }
1970                /* fallthru */
1971        default:
1972                /* nothing */
1973                ;
1974        }
1975
1976        if (hashcmp(one->sha1, two->sha1)) {
1977                int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1978
1979                if (o->binary) {
1980                        mmfile_t mf;
1981                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
1982                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
1983                                abbrev = 40;
1984                }
1985                strbuf_addf(&msg, "index %.*s..%.*s",
1986                                abbrev, sha1_to_hex(one->sha1),
1987                                abbrev, sha1_to_hex(two->sha1));
1988                if (one->mode == two->mode)
1989                        strbuf_addf(&msg, " %06o", one->mode);
1990                strbuf_addch(&msg, '\n');
1991        }
1992
1993        if (msg.len)
1994                strbuf_setlen(&msg, msg.len - 1);
1995        xfrm_msg = msg.len ? msg.buf : NULL;
1996
1997        if (!pgm &&
1998            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1999            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2000                /* a filepair that changes between file and symlink
2001                 * needs to be split into deletion and creation.
2002                 */
2003                struct diff_filespec *null = alloc_filespec(two->path);
2004                run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
2005                free(null);
2006                null = alloc_filespec(one->path);
2007                run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
2008                free(null);
2009        }
2010        else
2011                run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
2012                             complete_rewrite);
2013
2014        strbuf_release(&msg);
2015}
2016
2017static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2018                         struct diffstat_t *diffstat)
2019{
2020        const char *name;
2021        const char *other;
2022        int complete_rewrite = 0;
2023
2024        if (DIFF_PAIR_UNMERGED(p)) {
2025                /* unmerged */
2026                builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2027                return;
2028        }
2029
2030        name = p->one->path;
2031        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2032
2033        diff_fill_sha1_info(p->one);
2034        diff_fill_sha1_info(p->two);
2035
2036        if (p->status == DIFF_STATUS_MODIFIED && p->score)
2037                complete_rewrite = 1;
2038        builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2039}
2040
2041static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2042{
2043        const char *name;
2044        const char *other;
2045
2046        if (DIFF_PAIR_UNMERGED(p)) {
2047                /* unmerged */
2048                return;
2049        }
2050
2051        name = p->one->path;
2052        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2053
2054        diff_fill_sha1_info(p->one);
2055        diff_fill_sha1_info(p->two);
2056
2057        builtin_checkdiff(name, other, p->one, p->two, o);
2058}
2059
2060void diff_setup(struct diff_options *options)
2061{
2062        memset(options, 0, sizeof(*options));
2063        options->line_termination = '\n';
2064        options->break_opt = -1;
2065        options->rename_limit = -1;
2066        options->context = 3;
2067        options->msg_sep = "";
2068
2069        options->change = diff_change;
2070        options->add_remove = diff_addremove;
2071        options->color_diff = diff_use_color_default;
2072        options->detect_rename = diff_detect_rename_default;
2073}
2074
2075int diff_setup_done(struct diff_options *options)
2076{
2077        int count = 0;
2078
2079        if (options->output_format & DIFF_FORMAT_NAME)
2080                count++;
2081        if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2082                count++;
2083        if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2084                count++;
2085        if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2086                count++;
2087        if (count > 1)
2088                die("--name-only, --name-status, --check and -s are mutually exclusive");
2089
2090        if (options->find_copies_harder)
2091                options->detect_rename = DIFF_DETECT_COPY;
2092
2093        if (options->output_format & (DIFF_FORMAT_NAME |
2094                                      DIFF_FORMAT_NAME_STATUS |
2095                                      DIFF_FORMAT_CHECKDIFF |
2096                                      DIFF_FORMAT_NO_OUTPUT))
2097                options->output_format &= ~(DIFF_FORMAT_RAW |
2098                                            DIFF_FORMAT_NUMSTAT |
2099                                            DIFF_FORMAT_DIFFSTAT |
2100                                            DIFF_FORMAT_SHORTSTAT |
2101                                            DIFF_FORMAT_SUMMARY |
2102                                            DIFF_FORMAT_PATCH);
2103
2104        /*
2105         * These cases always need recursive; we do not drop caller-supplied
2106         * recursive bits for other formats here.
2107         */
2108        if (options->output_format & (DIFF_FORMAT_PATCH |
2109                                      DIFF_FORMAT_NUMSTAT |
2110                                      DIFF_FORMAT_DIFFSTAT |
2111                                      DIFF_FORMAT_SHORTSTAT |
2112                                      DIFF_FORMAT_SUMMARY |
2113                                      DIFF_FORMAT_CHECKDIFF))
2114                options->recursive = 1;
2115        /*
2116         * Also pickaxe would not work very well if you do not say recursive
2117         */
2118        if (options->pickaxe)
2119                options->recursive = 1;
2120
2121        if (options->detect_rename && options->rename_limit < 0)
2122                options->rename_limit = diff_rename_limit_default;
2123        if (options->setup & DIFF_SETUP_USE_CACHE) {
2124                if (!active_cache)
2125                        /* read-cache does not die even when it fails
2126                         * so it is safe for us to do this here.  Also
2127                         * it does not smudge active_cache or active_nr
2128                         * when it fails, so we do not have to worry about
2129                         * cleaning it up ourselves either.
2130                         */
2131                        read_cache();
2132        }
2133        if (options->abbrev <= 0 || 40 < options->abbrev)
2134                options->abbrev = 40; /* full */
2135
2136        /*
2137         * It does not make sense to show the first hit we happened
2138         * to have found.  It does not make sense not to return with
2139         * exit code in such a case either.
2140         */
2141        if (options->quiet) {
2142                options->output_format = DIFF_FORMAT_NO_OUTPUT;
2143                options->exit_with_status = 1;
2144        }
2145
2146        /*
2147         * If we postprocess in diffcore, we cannot simply return
2148         * upon the first hit.  We need to run diff as usual.
2149         */
2150        if (options->pickaxe || options->filter)
2151                options->quiet = 0;
2152
2153        return 0;
2154}
2155
2156static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2157{
2158        char c, *eq;
2159        int len;
2160
2161        if (*arg != '-')
2162                return 0;
2163        c = *++arg;
2164        if (!c)
2165                return 0;
2166        if (c == arg_short) {
2167                c = *++arg;
2168                if (!c)
2169                        return 1;
2170                if (val && isdigit(c)) {
2171                        char *end;
2172                        int n = strtoul(arg, &end, 10);
2173                        if (*end)
2174                                return 0;
2175                        *val = n;
2176                        return 1;
2177                }
2178                return 0;
2179        }
2180        if (c != '-')
2181                return 0;
2182        arg++;
2183        eq = strchr(arg, '=');
2184        if (eq)
2185                len = eq - arg;
2186        else
2187                len = strlen(arg);
2188        if (!len || strncmp(arg, arg_long, len))
2189                return 0;
2190        if (eq) {
2191                int n;
2192                char *end;
2193                if (!isdigit(*++eq))
2194                        return 0;
2195                n = strtoul(eq, &end, 10);
2196                if (*end)
2197                        return 0;
2198                *val = n;
2199        }
2200        return 1;
2201}
2202
2203static int diff_scoreopt_parse(const char *opt);
2204
2205int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2206{
2207        const char *arg = av[0];
2208        if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2209                options->output_format |= DIFF_FORMAT_PATCH;
2210        else if (opt_arg(arg, 'U', "unified", &options->context))
2211                options->output_format |= DIFF_FORMAT_PATCH;
2212        else if (!strcmp(arg, "--raw"))
2213                options->output_format |= DIFF_FORMAT_RAW;
2214        else if (!strcmp(arg, "--patch-with-raw")) {
2215                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2216        }
2217        else if (!strcmp(arg, "--numstat")) {
2218                options->output_format |= DIFF_FORMAT_NUMSTAT;
2219        }
2220        else if (!strcmp(arg, "--shortstat")) {
2221                options->output_format |= DIFF_FORMAT_SHORTSTAT;
2222        }
2223        else if (!prefixcmp(arg, "--stat")) {
2224                char *end;
2225                int width = options->stat_width;
2226                int name_width = options->stat_name_width;
2227                arg += 6;
2228                end = (char *)arg;
2229
2230                switch (*arg) {
2231                case '-':
2232                        if (!prefixcmp(arg, "-width="))
2233                                width = strtoul(arg + 7, &end, 10);
2234                        else if (!prefixcmp(arg, "-name-width="))
2235                                name_width = strtoul(arg + 12, &end, 10);
2236                        break;
2237                case '=':
2238                        width = strtoul(arg+1, &end, 10);
2239                        if (*end == ',')
2240                                name_width = strtoul(end+1, &end, 10);
2241                }
2242
2243                /* Important! This checks all the error cases! */
2244                if (*end)
2245                        return 0;
2246                options->output_format |= DIFF_FORMAT_DIFFSTAT;
2247                options->stat_name_width = name_width;
2248                options->stat_width = width;
2249        }
2250        else if (!strcmp(arg, "--check"))
2251                options->output_format |= DIFF_FORMAT_CHECKDIFF;
2252        else if (!strcmp(arg, "--summary"))
2253                options->output_format |= DIFF_FORMAT_SUMMARY;
2254        else if (!strcmp(arg, "--patch-with-stat")) {
2255                options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2256        }
2257        else if (!strcmp(arg, "-z"))
2258                options->line_termination = 0;
2259        else if (!prefixcmp(arg, "-l"))
2260                options->rename_limit = strtoul(arg+2, NULL, 10);
2261        else if (!strcmp(arg, "--full-index"))
2262                options->full_index = 1;
2263        else if (!strcmp(arg, "--binary")) {
2264                options->output_format |= DIFF_FORMAT_PATCH;
2265                options->binary = 1;
2266        }
2267        else if (!strcmp(arg, "-a") || !strcmp(arg, "--text")) {
2268                options->text = 1;
2269        }
2270        else if (!strcmp(arg, "--name-only"))
2271                options->output_format |= DIFF_FORMAT_NAME;
2272        else if (!strcmp(arg, "--name-status"))
2273                options->output_format |= DIFF_FORMAT_NAME_STATUS;
2274        else if (!strcmp(arg, "-R"))
2275                options->reverse_diff = 1;
2276        else if (!prefixcmp(arg, "-S"))
2277                options->pickaxe = arg + 2;
2278        else if (!strcmp(arg, "-s")) {
2279                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2280        }
2281        else if (!prefixcmp(arg, "-O"))
2282                options->orderfile = arg + 2;
2283        else if (!prefixcmp(arg, "--diff-filter="))
2284                options->filter = arg + 14;
2285        else if (!strcmp(arg, "--pickaxe-all"))
2286                options->pickaxe_opts = DIFF_PICKAXE_ALL;
2287        else if (!strcmp(arg, "--pickaxe-regex"))
2288                options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2289        else if (!prefixcmp(arg, "-B")) {
2290                if ((options->break_opt =
2291                     diff_scoreopt_parse(arg)) == -1)
2292                        return -1;
2293        }
2294        else if (!prefixcmp(arg, "-M")) {
2295                if ((options->rename_score =
2296                     diff_scoreopt_parse(arg)) == -1)
2297                        return -1;
2298                options->detect_rename = DIFF_DETECT_RENAME;
2299        }
2300        else if (!prefixcmp(arg, "-C")) {
2301                if (options->detect_rename == DIFF_DETECT_COPY)
2302                        options->find_copies_harder = 1;
2303                if ((options->rename_score =
2304                     diff_scoreopt_parse(arg)) == -1)
2305                        return -1;
2306                options->detect_rename = DIFF_DETECT_COPY;
2307        }
2308        else if (!strcmp(arg, "--find-copies-harder"))
2309                options->find_copies_harder = 1;
2310        else if (!strcmp(arg, "--follow"))
2311                options->follow_renames = 1;
2312        else if (!strcmp(arg, "--abbrev"))
2313                options->abbrev = DEFAULT_ABBREV;
2314        else if (!prefixcmp(arg, "--abbrev=")) {
2315                options->abbrev = strtoul(arg + 9, NULL, 10);
2316                if (options->abbrev < MINIMUM_ABBREV)
2317                        options->abbrev = MINIMUM_ABBREV;
2318                else if (40 < options->abbrev)
2319                        options->abbrev = 40;
2320        }
2321        else if (!strcmp(arg, "--color"))
2322                options->color_diff = 1;
2323        else if (!strcmp(arg, "--no-color"))
2324                options->color_diff = 0;
2325        else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2326                options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2327        else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2328                options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2329        else if (!strcmp(arg, "--ignore-space-at-eol"))
2330                options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2331        else if (!strcmp(arg, "--color-words"))
2332                options->color_diff = options->color_diff_words = 1;
2333        else if (!strcmp(arg, "--no-renames"))
2334                options->detect_rename = 0;
2335        else if (!strcmp(arg, "--exit-code"))
2336                options->exit_with_status = 1;
2337        else if (!strcmp(arg, "--quiet"))
2338                options->quiet = 1;
2339        else if (!strcmp(arg, "--ext-diff"))
2340                options->allow_external = 1;
2341        else if (!strcmp(arg, "--no-ext-diff"))
2342                options->allow_external = 0;
2343        else
2344                return 0;
2345        return 1;
2346}
2347
2348static int parse_num(const char **cp_p)
2349{
2350        unsigned long num, scale;
2351        int ch, dot;
2352        const char *cp = *cp_p;
2353
2354        num = 0;
2355        scale = 1;
2356        dot = 0;
2357        for(;;) {
2358                ch = *cp;
2359                if ( !dot && ch == '.' ) {
2360                        scale = 1;
2361                        dot = 1;
2362                } else if ( ch == '%' ) {
2363                        scale = dot ? scale*100 : 100;
2364                        cp++;   /* % is always at the end */
2365                        break;
2366                } else if ( ch >= '0' && ch <= '9' ) {
2367                        if ( scale < 100000 ) {
2368                                scale *= 10;
2369                                num = (num*10) + (ch-'0');
2370                        }
2371                } else {
2372                        break;
2373                }
2374                cp++;
2375        }
2376        *cp_p = cp;
2377
2378        /* user says num divided by scale and we say internally that
2379         * is MAX_SCORE * num / scale.
2380         */
2381        return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2382}
2383
2384static int diff_scoreopt_parse(const char *opt)
2385{
2386        int opt1, opt2, cmd;
2387
2388        if (*opt++ != '-')
2389                return -1;
2390        cmd = *opt++;
2391        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2392                return -1; /* that is not a -M, -C nor -B option */
2393
2394        opt1 = parse_num(&opt);
2395        if (cmd != 'B')
2396                opt2 = 0;
2397        else {
2398                if (*opt == 0)
2399                        opt2 = 0;
2400                else if (*opt != '/')
2401                        return -1; /* we expect -B80/99 or -B80 */
2402                else {
2403                        opt++;
2404                        opt2 = parse_num(&opt);
2405                }
2406        }
2407        if (*opt != 0)
2408                return -1;
2409        return opt1 | (opt2 << 16);
2410}
2411
2412struct diff_queue_struct diff_queued_diff;
2413
2414void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2415{
2416        if (queue->alloc <= queue->nr) {
2417                queue->alloc = alloc_nr(queue->alloc);
2418                queue->queue = xrealloc(queue->queue,
2419                                        sizeof(dp) * queue->alloc);
2420        }
2421        queue->queue[queue->nr++] = dp;
2422}
2423
2424struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2425                                 struct diff_filespec *one,
2426                                 struct diff_filespec *two)
2427{
2428        struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2429        dp->one = one;
2430        dp->two = two;
2431        if (queue)
2432                diff_q(queue, dp);
2433        return dp;
2434}
2435
2436void diff_free_filepair(struct diff_filepair *p)
2437{
2438        diff_free_filespec_data(p->one);
2439        diff_free_filespec_data(p->two);
2440        free(p->one);
2441        free(p->two);
2442        free(p);
2443}
2444
2445/* This is different from find_unique_abbrev() in that
2446 * it stuffs the result with dots for alignment.
2447 */
2448const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2449{
2450        int abblen;
2451        const char *abbrev;
2452        if (len == 40)
2453                return sha1_to_hex(sha1);
2454
2455        abbrev = find_unique_abbrev(sha1, len);
2456        if (!abbrev)
2457                return sha1_to_hex(sha1);
2458        abblen = strlen(abbrev);
2459        if (abblen < 37) {
2460                static char hex[41];
2461                if (len < abblen && abblen <= len + 2)
2462                        sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2463                else
2464                        sprintf(hex, "%s...", abbrev);
2465                return hex;
2466        }
2467        return sha1_to_hex(sha1);
2468}
2469
2470static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2471{
2472        int line_termination = opt->line_termination;
2473        int inter_name_termination = line_termination ? '\t' : '\0';
2474
2475        if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2476                printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2477                       diff_unique_abbrev(p->one->sha1, opt->abbrev));
2478                printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2479        }
2480        if (p->score) {
2481                printf("%c%03d%c", p->status, similarity_index(p),
2482                           inter_name_termination);
2483        } else {
2484                printf("%c%c", p->status, inter_name_termination);
2485        }
2486
2487        if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2488                write_name_quoted(p->one->path, stdout, inter_name_termination);
2489                write_name_quoted(p->two->path, stdout, line_termination);
2490        } else {
2491                const char *path = p->one->mode ? p->one->path : p->two->path;
2492                write_name_quoted(path, stdout, line_termination);
2493        }
2494}
2495
2496int diff_unmodified_pair(struct diff_filepair *p)
2497{
2498        /* This function is written stricter than necessary to support
2499         * the currently implemented transformers, but the idea is to
2500         * let transformers to produce diff_filepairs any way they want,
2501         * and filter and clean them up here before producing the output.
2502         */
2503        struct diff_filespec *one = p->one, *two = p->two;
2504
2505        if (DIFF_PAIR_UNMERGED(p))
2506                return 0; /* unmerged is interesting */
2507
2508        /* deletion, addition, mode or type change
2509         * and rename are all interesting.
2510         */
2511        if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2512            DIFF_PAIR_MODE_CHANGED(p) ||
2513            strcmp(one->path, two->path))
2514                return 0;
2515
2516        /* both are valid and point at the same path.  that is, we are
2517         * dealing with a change.
2518         */
2519        if (one->sha1_valid && two->sha1_valid &&
2520            !hashcmp(one->sha1, two->sha1))
2521                return 1; /* no change */
2522        if (!one->sha1_valid && !two->sha1_valid)
2523                return 1; /* both look at the same file on the filesystem. */
2524        return 0;
2525}
2526
2527static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2528{
2529        if (diff_unmodified_pair(p))
2530                return;
2531
2532        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2533            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2534                return; /* no tree diffs in patch format */
2535
2536        run_diff(p, o);
2537}
2538
2539static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2540                            struct diffstat_t *diffstat)
2541{
2542        if (diff_unmodified_pair(p))
2543                return;
2544
2545        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2546            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2547                return; /* no tree diffs in patch format */
2548
2549        run_diffstat(p, o, diffstat);
2550}
2551
2552static void diff_flush_checkdiff(struct diff_filepair *p,
2553                struct diff_options *o)
2554{
2555        if (diff_unmodified_pair(p))
2556                return;
2557
2558        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2559            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2560                return; /* no tree diffs in patch format */
2561
2562        run_checkdiff(p, o);
2563}
2564
2565int diff_queue_is_empty(void)
2566{
2567        struct diff_queue_struct *q = &diff_queued_diff;
2568        int i;
2569        for (i = 0; i < q->nr; i++)
2570                if (!diff_unmodified_pair(q->queue[i]))
2571                        return 0;
2572        return 1;
2573}
2574
2575#if DIFF_DEBUG
2576void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2577{
2578        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2579                x, one ? one : "",
2580                s->path,
2581                DIFF_FILE_VALID(s) ? "valid" : "invalid",
2582                s->mode,
2583                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2584        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2585                x, one ? one : "",
2586                s->size, s->xfrm_flags);
2587}
2588
2589void diff_debug_filepair(const struct diff_filepair *p, int i)
2590{
2591        diff_debug_filespec(p->one, i, "one");
2592        diff_debug_filespec(p->two, i, "two");
2593        fprintf(stderr, "score %d, status %c stays %d broken %d\n",
2594                p->score, p->status ? p->status : '?',
2595                p->source_stays, p->broken_pair);
2596}
2597
2598void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2599{
2600        int i;
2601        if (msg)
2602                fprintf(stderr, "%s\n", msg);
2603        fprintf(stderr, "q->nr = %d\n", q->nr);
2604        for (i = 0; i < q->nr; i++) {
2605                struct diff_filepair *p = q->queue[i];
2606                diff_debug_filepair(p, i);
2607        }
2608}
2609#endif
2610
2611static void diff_resolve_rename_copy(void)
2612{
2613        int i, j;
2614        struct diff_filepair *p, *pp;
2615        struct diff_queue_struct *q = &diff_queued_diff;
2616
2617        diff_debug_queue("resolve-rename-copy", q);
2618
2619        for (i = 0; i < q->nr; i++) {
2620                p = q->queue[i];
2621                p->status = 0; /* undecided */
2622                if (DIFF_PAIR_UNMERGED(p))
2623                        p->status = DIFF_STATUS_UNMERGED;
2624                else if (!DIFF_FILE_VALID(p->one))
2625                        p->status = DIFF_STATUS_ADDED;
2626                else if (!DIFF_FILE_VALID(p->two))
2627                        p->status = DIFF_STATUS_DELETED;
2628                else if (DIFF_PAIR_TYPE_CHANGED(p))
2629                        p->status = DIFF_STATUS_TYPE_CHANGED;
2630
2631                /* from this point on, we are dealing with a pair
2632                 * whose both sides are valid and of the same type, i.e.
2633                 * either in-place edit or rename/copy edit.
2634                 */
2635                else if (DIFF_PAIR_RENAME(p)) {
2636                        if (p->source_stays) {
2637                                p->status = DIFF_STATUS_COPIED;
2638                                continue;
2639                        }
2640                        /* See if there is some other filepair that
2641                         * copies from the same source as us.  If so
2642                         * we are a copy.  Otherwise we are either a
2643                         * copy if the path stays, or a rename if it
2644                         * does not, but we already handled "stays" case.
2645                         */
2646                        for (j = i + 1; j < q->nr; j++) {
2647                                pp = q->queue[j];
2648                                if (strcmp(pp->one->path, p->one->path))
2649                                        continue; /* not us */
2650                                if (!DIFF_PAIR_RENAME(pp))
2651                                        continue; /* not a rename/copy */
2652                                /* pp is a rename/copy from the same source */
2653                                p->status = DIFF_STATUS_COPIED;
2654                                break;
2655                        }
2656                        if (!p->status)
2657                                p->status = DIFF_STATUS_RENAMED;
2658                }
2659                else if (hashcmp(p->one->sha1, p->two->sha1) ||
2660                         p->one->mode != p->two->mode ||
2661                         is_null_sha1(p->one->sha1))
2662                        p->status = DIFF_STATUS_MODIFIED;
2663                else {
2664                        /* This is a "no-change" entry and should not
2665                         * happen anymore, but prepare for broken callers.
2666                         */
2667                        error("feeding unmodified %s to diffcore",
2668                              p->one->path);
2669                        p->status = DIFF_STATUS_UNKNOWN;
2670                }
2671        }
2672        diff_debug_queue("resolve-rename-copy done", q);
2673}
2674
2675static int check_pair_status(struct diff_filepair *p)
2676{
2677        switch (p->status) {
2678        case DIFF_STATUS_UNKNOWN:
2679                return 0;
2680        case 0:
2681                die("internal error in diff-resolve-rename-copy");
2682        default:
2683                return 1;
2684        }
2685}
2686
2687static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2688{
2689        int fmt = opt->output_format;
2690
2691        if (fmt & DIFF_FORMAT_CHECKDIFF)
2692                diff_flush_checkdiff(p, opt);
2693        else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2694                diff_flush_raw(p, opt);
2695        else if (fmt & DIFF_FORMAT_NAME)
2696                write_name_quoted(p->two->path, stdout, opt->line_termination);
2697}
2698
2699static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2700{
2701        if (fs->mode)
2702                printf(" %s mode %06o ", newdelete, fs->mode);
2703        else
2704                printf(" %s ", newdelete);
2705        write_name_quoted(fs->path, stdout, '\n');
2706}
2707
2708
2709static void show_mode_change(struct diff_filepair *p, int show_name)
2710{
2711        if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2712                printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2713                        show_name ? ' ' : '\n');
2714                if (show_name) {
2715                        write_name_quoted(p->two->path, stdout, '\n');
2716                }
2717        }
2718}
2719
2720static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2721{
2722        char *names = pprint_rename(p->one->path, p->two->path);
2723
2724        printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2725        free(names);
2726        show_mode_change(p, 0);
2727}
2728
2729static void diff_summary(struct diff_filepair *p)
2730{
2731        switch(p->status) {
2732        case DIFF_STATUS_DELETED:
2733                show_file_mode_name("delete", p->one);
2734                break;
2735        case DIFF_STATUS_ADDED:
2736                show_file_mode_name("create", p->two);
2737                break;
2738        case DIFF_STATUS_COPIED:
2739                show_rename_copy("copy", p);
2740                break;
2741        case DIFF_STATUS_RENAMED:
2742                show_rename_copy("rename", p);
2743                break;
2744        default:
2745                if (p->score) {
2746                        puts(" rewrite ");
2747                        write_name_quoted(p->two->path, stdout, ' ');
2748                        printf("(%d%%)\n", similarity_index(p));
2749                }
2750                show_mode_change(p, !p->score);
2751                break;
2752        }
2753}
2754
2755struct patch_id_t {
2756        struct xdiff_emit_state xm;
2757        SHA_CTX *ctx;
2758        int patchlen;
2759};
2760
2761static int remove_space(char *line, int len)
2762{
2763        int i;
2764        char *dst = line;
2765        unsigned char c;
2766
2767        for (i = 0; i < len; i++)
2768                if (!isspace((c = line[i])))
2769                        *dst++ = c;
2770
2771        return dst - line;
2772}
2773
2774static void patch_id_consume(void *priv, char *line, unsigned long len)
2775{
2776        struct patch_id_t *data = priv;
2777        int new_len;
2778
2779        /* Ignore line numbers when computing the SHA1 of the patch */
2780        if (!prefixcmp(line, "@@ -"))
2781                return;
2782
2783        new_len = remove_space(line, len);
2784
2785        SHA1_Update(data->ctx, line, new_len);
2786        data->patchlen += new_len;
2787}
2788
2789/* returns 0 upon success, and writes result into sha1 */
2790static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2791{
2792        struct diff_queue_struct *q = &diff_queued_diff;
2793        int i;
2794        SHA_CTX ctx;
2795        struct patch_id_t data;
2796        char buffer[PATH_MAX * 4 + 20];
2797
2798        SHA1_Init(&ctx);
2799        memset(&data, 0, sizeof(struct patch_id_t));
2800        data.ctx = &ctx;
2801        data.xm.consume = patch_id_consume;
2802
2803        for (i = 0; i < q->nr; i++) {
2804                xpparam_t xpp;
2805                xdemitconf_t xecfg;
2806                xdemitcb_t ecb;
2807                mmfile_t mf1, mf2;
2808                struct diff_filepair *p = q->queue[i];
2809                int len1, len2;
2810
2811                memset(&xecfg, 0, sizeof(xecfg));
2812                if (p->status == 0)
2813                        return error("internal diff status error");
2814                if (p->status == DIFF_STATUS_UNKNOWN)
2815                        continue;
2816                if (diff_unmodified_pair(p))
2817                        continue;
2818                if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2819                    (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2820                        continue;
2821                if (DIFF_PAIR_UNMERGED(p))
2822                        continue;
2823
2824                diff_fill_sha1_info(p->one);
2825                diff_fill_sha1_info(p->two);
2826                if (fill_mmfile(&mf1, p->one) < 0 ||
2827                                fill_mmfile(&mf2, p->two) < 0)
2828                        return error("unable to read files to diff");
2829
2830                len1 = remove_space(p->one->path, strlen(p->one->path));
2831                len2 = remove_space(p->two->path, strlen(p->two->path));
2832                if (p->one->mode == 0)
2833                        len1 = snprintf(buffer, sizeof(buffer),
2834                                        "diff--gita/%.*sb/%.*s"
2835                                        "newfilemode%06o"
2836                                        "---/dev/null"
2837                                        "+++b/%.*s",
2838                                        len1, p->one->path,
2839                                        len2, p->two->path,
2840                                        p->two->mode,
2841                                        len2, p->two->path);
2842                else if (p->two->mode == 0)
2843                        len1 = snprintf(buffer, sizeof(buffer),
2844                                        "diff--gita/%.*sb/%.*s"
2845                                        "deletedfilemode%06o"
2846                                        "---a/%.*s"
2847                                        "+++/dev/null",
2848                                        len1, p->one->path,
2849                                        len2, p->two->path,
2850                                        p->one->mode,
2851                                        len1, p->one->path);
2852                else
2853                        len1 = snprintf(buffer, sizeof(buffer),
2854                                        "diff--gita/%.*sb/%.*s"
2855                                        "---a/%.*s"
2856                                        "+++b/%.*s",
2857                                        len1, p->one->path,
2858                                        len2, p->two->path,
2859                                        len1, p->one->path,
2860                                        len2, p->two->path);
2861                SHA1_Update(&ctx, buffer, len1);
2862
2863                xpp.flags = XDF_NEED_MINIMAL;
2864                xecfg.ctxlen = 3;
2865                xecfg.flags = XDL_EMIT_FUNCNAMES;
2866                ecb.outf = xdiff_outf;
2867                ecb.priv = &data;
2868                xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2869        }
2870
2871        SHA1_Final(sha1, &ctx);
2872        return 0;
2873}
2874
2875int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2876{
2877        struct diff_queue_struct *q = &diff_queued_diff;
2878        int i;
2879        int result = diff_get_patch_id(options, sha1);
2880
2881        for (i = 0; i < q->nr; i++)
2882                diff_free_filepair(q->queue[i]);
2883
2884        free(q->queue);
2885        q->queue = NULL;
2886        q->nr = q->alloc = 0;
2887
2888        return result;
2889}
2890
2891static int is_summary_empty(const struct diff_queue_struct *q)
2892{
2893        int i;
2894
2895        for (i = 0; i < q->nr; i++) {
2896                const struct diff_filepair *p = q->queue[i];
2897
2898                switch (p->status) {
2899                case DIFF_STATUS_DELETED:
2900                case DIFF_STATUS_ADDED:
2901                case DIFF_STATUS_COPIED:
2902                case DIFF_STATUS_RENAMED:
2903                        return 0;
2904                default:
2905                        if (p->score)
2906                                return 0;
2907                        if (p->one->mode && p->two->mode &&
2908                            p->one->mode != p->two->mode)
2909                                return 0;
2910                        break;
2911                }
2912        }
2913        return 1;
2914}
2915
2916void diff_flush(struct diff_options *options)
2917{
2918        struct diff_queue_struct *q = &diff_queued_diff;
2919        int i, output_format = options->output_format;
2920        int separator = 0;
2921
2922        /*
2923         * Order: raw, stat, summary, patch
2924         * or:    name/name-status/checkdiff (other bits clear)
2925         */
2926        if (!q->nr)
2927                goto free_queue;
2928
2929        if (output_format & (DIFF_FORMAT_RAW |
2930                             DIFF_FORMAT_NAME |
2931                             DIFF_FORMAT_NAME_STATUS |
2932                             DIFF_FORMAT_CHECKDIFF)) {
2933                for (i = 0; i < q->nr; i++) {
2934                        struct diff_filepair *p = q->queue[i];
2935                        if (check_pair_status(p))
2936                                flush_one_pair(p, options);
2937                }
2938                separator++;
2939        }
2940
2941        if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2942                struct diffstat_t diffstat;
2943
2944                memset(&diffstat, 0, sizeof(struct diffstat_t));
2945                diffstat.xm.consume = diffstat_consume;
2946                for (i = 0; i < q->nr; i++) {
2947                        struct diff_filepair *p = q->queue[i];
2948                        if (check_pair_status(p))
2949                                diff_flush_stat(p, options, &diffstat);
2950                }
2951                if (output_format & DIFF_FORMAT_NUMSTAT)
2952                        show_numstat(&diffstat, options);
2953                if (output_format & DIFF_FORMAT_DIFFSTAT)
2954                        show_stats(&diffstat, options);
2955                else if (output_format & DIFF_FORMAT_SHORTSTAT)
2956                        show_shortstats(&diffstat);
2957                separator++;
2958        }
2959
2960        if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2961                for (i = 0; i < q->nr; i++)
2962                        diff_summary(q->queue[i]);
2963                separator++;
2964        }
2965
2966        if (output_format & DIFF_FORMAT_PATCH) {
2967                if (separator) {
2968                        if (options->stat_sep) {
2969                                /* attach patch instead of inline */
2970                                fputs(options->stat_sep, stdout);
2971                        } else {
2972                                putchar(options->line_termination);
2973                        }
2974                }
2975
2976                for (i = 0; i < q->nr; i++) {
2977                        struct diff_filepair *p = q->queue[i];
2978                        if (check_pair_status(p))
2979                                diff_flush_patch(p, options);
2980                }
2981        }
2982
2983        if (output_format & DIFF_FORMAT_CALLBACK)
2984                options->format_callback(q, options, options->format_callback_data);
2985
2986        for (i = 0; i < q->nr; i++)
2987                diff_free_filepair(q->queue[i]);
2988free_queue:
2989        free(q->queue);
2990        q->queue = NULL;
2991        q->nr = q->alloc = 0;
2992}
2993
2994static void diffcore_apply_filter(const char *filter)
2995{
2996        int i;
2997        struct diff_queue_struct *q = &diff_queued_diff;
2998        struct diff_queue_struct outq;
2999        outq.queue = NULL;
3000        outq.nr = outq.alloc = 0;
3001
3002        if (!filter)
3003                return;
3004
3005        if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3006                int found;
3007                for (i = found = 0; !found && i < q->nr; i++) {
3008                        struct diff_filepair *p = q->queue[i];
3009                        if (((p->status == DIFF_STATUS_MODIFIED) &&
3010                             ((p->score &&
3011                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3012                              (!p->score &&
3013                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3014                            ((p->status != DIFF_STATUS_MODIFIED) &&
3015                             strchr(filter, p->status)))
3016                                found++;
3017                }
3018                if (found)
3019                        return;
3020
3021                /* otherwise we will clear the whole queue
3022                 * by copying the empty outq at the end of this
3023                 * function, but first clear the current entries
3024                 * in the queue.
3025                 */
3026                for (i = 0; i < q->nr; i++)
3027                        diff_free_filepair(q->queue[i]);
3028        }
3029        else {
3030                /* Only the matching ones */
3031                for (i = 0; i < q->nr; i++) {
3032                        struct diff_filepair *p = q->queue[i];
3033
3034                        if (((p->status == DIFF_STATUS_MODIFIED) &&
3035                             ((p->score &&
3036                               strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3037                              (!p->score &&
3038                               strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3039                            ((p->status != DIFF_STATUS_MODIFIED) &&
3040                             strchr(filter, p->status)))
3041                                diff_q(&outq, p);
3042                        else
3043                                diff_free_filepair(p);
3044                }
3045        }
3046        free(q->queue);
3047        *q = outq;
3048}
3049
3050/* Check whether two filespecs with the same mode and size are identical */
3051static int diff_filespec_is_identical(struct diff_filespec *one,
3052                                      struct diff_filespec *two)
3053{
3054        if (S_ISGITLINK(one->mode)) {
3055                diff_fill_sha1_info(one);
3056                diff_fill_sha1_info(two);
3057                return !hashcmp(one->sha1, two->sha1);
3058        }
3059        if (diff_populate_filespec(one, 0))
3060                return 0;
3061        if (diff_populate_filespec(two, 0))
3062                return 0;
3063        return !memcmp(one->data, two->data, one->size);
3064}
3065
3066static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3067{
3068        int i;
3069        struct diff_queue_struct *q = &diff_queued_diff;
3070        struct diff_queue_struct outq;
3071        outq.queue = NULL;
3072        outq.nr = outq.alloc = 0;
3073
3074        for (i = 0; i < q->nr; i++) {
3075                struct diff_filepair *p = q->queue[i];
3076
3077                /*
3078                 * 1. Entries that come from stat info dirtyness
3079                 *    always have both sides (iow, not create/delete),
3080                 *    one side of the object name is unknown, with
3081                 *    the same mode and size.  Keep the ones that
3082                 *    do not match these criteria.  They have real
3083                 *    differences.
3084                 *
3085                 * 2. At this point, the file is known to be modified,
3086                 *    with the same mode and size, and the object
3087                 *    name of one side is unknown.  Need to inspect
3088                 *    the identical contents.
3089                 */
3090                if (!DIFF_FILE_VALID(p->one) || /* (1) */
3091                    !DIFF_FILE_VALID(p->two) ||
3092                    (p->one->sha1_valid && p->two->sha1_valid) ||
3093                    (p->one->mode != p->two->mode) ||
3094                    diff_populate_filespec(p->one, 1) ||
3095                    diff_populate_filespec(p->two, 1) ||
3096                    (p->one->size != p->two->size) ||
3097                    !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3098                        diff_q(&outq, p);
3099                else {
3100                        /*
3101                         * The caller can subtract 1 from skip_stat_unmatch
3102                         * to determine how many paths were dirty only
3103                         * due to stat info mismatch.
3104                         */
3105                        if (!diffopt->no_index)
3106                                diffopt->skip_stat_unmatch++;
3107                        diff_free_filepair(p);
3108                }
3109        }
3110        free(q->queue);
3111        *q = outq;
3112}
3113
3114void diffcore_std(struct diff_options *options)
3115{
3116        if (options->quiet)
3117                return;
3118
3119        if (options->skip_stat_unmatch && !options->find_copies_harder)
3120                diffcore_skip_stat_unmatch(options);
3121        if (options->break_opt != -1)
3122                diffcore_break(options->break_opt);
3123        if (options->detect_rename)
3124                diffcore_rename(options);
3125        if (options->break_opt != -1)
3126                diffcore_merge_broken();
3127        if (options->pickaxe)
3128                diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3129        if (options->orderfile)
3130                diffcore_order(options->orderfile);
3131        diff_resolve_rename_copy();
3132        diffcore_apply_filter(options->filter);
3133
3134        options->has_changes = !!diff_queued_diff.nr;
3135}
3136
3137
3138void diff_addremove(struct diff_options *options,
3139                    int addremove, unsigned mode,
3140                    const unsigned char *sha1,
3141                    const char *base, const char *path)
3142{
3143        char concatpath[PATH_MAX];
3144        struct diff_filespec *one, *two;
3145
3146        /* This may look odd, but it is a preparation for
3147         * feeding "there are unchanged files which should
3148         * not produce diffs, but when you are doing copy
3149         * detection you would need them, so here they are"
3150         * entries to the diff-core.  They will be prefixed
3151         * with something like '=' or '*' (I haven't decided
3152         * which but should not make any difference).
3153         * Feeding the same new and old to diff_change()
3154         * also has the same effect.
3155         * Before the final output happens, they are pruned after
3156         * merged into rename/copy pairs as appropriate.
3157         */
3158        if (options->reverse_diff)
3159                addremove = (addremove == '+' ? '-' :
3160                             addremove == '-' ? '+' : addremove);
3161
3162        if (!path) path = "";
3163        sprintf(concatpath, "%s%s", base, path);
3164        one = alloc_filespec(concatpath);
3165        two = alloc_filespec(concatpath);
3166
3167        if (addremove != '+')
3168                fill_filespec(one, sha1, mode);
3169        if (addremove != '-')
3170                fill_filespec(two, sha1, mode);
3171
3172        diff_queue(&diff_queued_diff, one, two);
3173        options->has_changes = 1;
3174}
3175
3176void diff_change(struct diff_options *options,
3177                 unsigned old_mode, unsigned new_mode,
3178                 const unsigned char *old_sha1,
3179                 const unsigned char *new_sha1,
3180                 const char *base, const char *path)
3181{
3182        char concatpath[PATH_MAX];
3183        struct diff_filespec *one, *two;
3184
3185        if (options->reverse_diff) {
3186                unsigned tmp;
3187                const unsigned char *tmp_c;
3188                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3189                tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3190        }
3191        if (!path) path = "";
3192        sprintf(concatpath, "%s%s", base, path);
3193        one = alloc_filespec(concatpath);
3194        two = alloc_filespec(concatpath);
3195        fill_filespec(one, old_sha1, old_mode);
3196        fill_filespec(two, new_sha1, new_mode);
3197
3198        diff_queue(&diff_queued_diff, one, two);
3199        options->has_changes = 1;
3200}
3201
3202void diff_unmerge(struct diff_options *options,
3203                  const char *path,
3204                  unsigned mode, const unsigned char *sha1)
3205{
3206        struct diff_filespec *one, *two;
3207        one = alloc_filespec(path);
3208        two = alloc_filespec(path);
3209        fill_filespec(one, sha1, mode);
3210        diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3211}