xsv.c

   1 /*
   2  *      The Swiss-Army Knife for CSV-like Files
   3  *
   4  *      (c) 2012 Martin Mares <mj@ucw.cz>
   5  */
   6
   7 #define _GNU_SOURCE
   8
   9 #include <stdio.h>
  10 #include <stdlib.h>
  11 #include <string.h>
  12 #include <stdarg.h>
  13 #include <getopt.h>
  14 #include <wchar.h>
  15 #include <locale.h>
  16
  17 #include <pcre.h>
  18
  19 #ifdef __GNUC__
  20 #define NONRET __attribute__((noreturn))
  21 #define UNUSED __attribute__((unused))
  22 #else
  23 #define NONRET
  24 #define UNUSED
  25 #endif
  26
  27 static void select_fields(void);
  28 static void select_all_fields(void);
  29
  30 /*** General functions ***/
  31
  32 static void NONRET die(char *msg, ...)
  33 {
  34         va_list args;
  35         va_start(args, msg);
  36         fprintf(stderr, "xsv: ");
  37         vfprintf(stderr, msg, args);
  38         fputc('\n', stderr);
  39         va_end(args);
  40         exit(1);
  41 }
  42
  43 /*** Memory allocation ***/
  44
  45 static void *xmalloc(size_t bytes)
  46 {
  47         void *p = malloc(bytes);
  48         if (!p)
  49                 die("Out of memory (cannot allocate %zu bytes)", bytes);
  50         return p;
  51 }
  52
  53 static void *xmalloc_zero(size_t bytes)
  54 {
  55         void *p = xmalloc(bytes);
  56         memset(p, 0, bytes);
  57         return p;
  58 }
  59
  60 static void *xrealloc(void *old, size_t bytes)
  61 {
  62         void *p = realloc(old, bytes);
  63         if (!p)
  64                 die("Out of memory (cannot allocate %zu bytes)", bytes);
  65         return p;
  66 }
  67
  68 #define DECLARE_BUF(name, type) \
  69         typedef struct { type *start; int count; int max; } name##_t;                           \
  70         static inline void name##_init(name##_t *b) { b->start = NULL; b->count = b->max = 0; } \
  71         static inline void name##_reset(name##_t *b) { b->count = 0; }                          \
  72         static inline int name##_count(name##_t *b) { return b->count; }                        \
  73         static void name##_extend(name##_t *b) {                                                \
  74                 b->max = b->max ? 2*b->max : 16;                                                \
  75                 b->start = xrealloc(b->start, b->max * sizeof(type));                           \
  76         }                                                                                       \
  77         static inline type *name##_push(name##_t *b) {                                          \
  78                 if (b->count >= b->max) name##_extend(b);                                       \
  79                 return &b->start[b->count++];                                                   \
  80         }                                                                                       \
  81         static inline type *name##_first(name##_t *b) { return b->start; }                      \
  82         static inline type *name##_nth(name##_t *b, int n) { return &b->start[n]; }             \
  83         // end
  84
  85 DECLARE_BUF(intarray, int);
  86 DECLARE_BUF(stringarray, char *);
  87
  88 /*** Formats and their parameters ***/
  89
  90 enum format_id {
  91         FORM_UNSPEC,
  92         FORM_TSV,
  93         FORM_CSV,
  94         FORM_WS,
  95         FORM_REGEX,
  96         FORM_TMP,
  97         FORM_TABLE,
  98 };
  99
 100 struct format {
 101         enum format_id id;
 102         int fs;
 103         int quote;
 104         int quiet;
 105         int sloppy;
 106         int (*read_line)(struct format *fmt);
 107         void (*write_line)(struct format *fmt);
 108         void (*write_grid)(struct format *fmt, int pos);        // -1=above, 1=below, 0=after header
 109         int needs_stats;
 110
 111         // Field names
 112         int has_header;
 113         char *set_field_names;
 114         struct field_names *field_names;
 115
 116         // CSV backend:
 117         int always_quote;
 118
 119         // regex backend:
 120         pcre *pcre;
 121         pcre_extra *pcre_extra;
 122
 123         // Temporary file backend:
 124         FILE *tmp_file;
 125
 126         // Table backend:
 127         int table_sep;
 128         int table_grid;
 129 };
 130
 131 static struct format *in_format, *out_format;
 132 static int want_trim, want_equalize, want_stats;
 133
 134 struct field {
 135         int start_pos;
 136         int len;
 137 };
 138
 139 DECLARE_BUF(fields, struct field);
 140 DECLARE_BUF(line, unsigned char);
 141
 142 static fields_t in_fields, out_fields;
 143 static struct field *in_field;
 144 static line_t in_line;
 145 static int line_number;
 146
 147 static int read_line(void)
 148 {
 149         fields_reset(&in_fields);
 150         line_reset(&in_line);
 151         in_field = NULL;
 152         if (!in_format->read_line(in_format))
 153                 return 0;
 154         if (ferror_unlocked(stdin))
 155                 die("I/O error when reading standard input");
 156         return 1;
 157 }
 158
 159 static void write_line(void)
 160 {
 161         out_format->write_line(out_format);
 162         if (ferror_unlocked(stdout))
 163                 die("I/O error when writing standard input");
 164 }
 165
 166 static void write_grid(int pos)
 167 {
 168         if (out_format->write_grid) {
 169                 out_format->write_grid(out_format, pos);
 170                 if (ferror_unlocked(stdout))
 171                         die("I/O error when writing standard input");
 172         }
 173 }
 174
 175 static void new_field(int pos)
 176 {
 177         in_field = fields_push(&in_fields);
 178         in_field->start_pos = pos;
 179         in_field->len = 0;
 180 }
 181
 182 static void ensure_field(int pos)
 183 {
 184         if (!in_field)
 185                 new_field(pos);
 186 }
 187
 188 static unsigned char *get_field(fields_t *fields, int i, int *len)
 189 {
 190         struct field *f = fields_nth(fields, i);
 191         *len = f->len;
 192         return line_nth(&in_line, f->start_pos);
 193 }
 194
 195 static void warn(struct format *fmt, char *msg, ...)
 196 {
 197         if (!fmt->quiet) {
 198                 fprintf(stderr, "Warning at line %d: ", line_number);
 199                 va_list args;
 200                 va_start(args, msg);
 201                 vfprintf(stderr, msg, args);
 202                 va_end(args);
 203                 fputc('\n', stderr);
 204         }
 205 }
 206
 207 static int next_line(void)
 208 {
 209         for (;;) {
 210                 int c = getchar_unlocked();
 211                 if (c == '\r')
 212                         continue;
 213                 if (c < 0)
 214                         return !!line_count(&in_line);
 215                 if (c == '\n')
 216                         return 1;
 217                 *line_push(&in_line) = c;
 218         }
 219 }
 220
 221 static int field_chars(struct field *f)
 222 {
 223         unsigned char *s = line_nth(&in_line, f->start_pos);
 224         int i = 0;
 225         mbstate_t mbs;
 226         memset(&mbs, 0, sizeof(mbs));
 227
 228         int chars = 0;
 229         while (i < f->len) {
 230                 size_t k = mbrlen((char *) s + i, f->len - i, &mbs);
 231                 if ((int) k <= 0)
 232                         break;
 233                 i += k;
 234                 chars++;
 235         }
 236
 237         return chars;
 238 }
 239
 240 /*** Field statistics ***/
 241
 242 static intarray_t column_widths;
 243
 244 static void update_stats(void)
 245 {
 246         if (!want_stats)
 247                 return;
 248
 249         for (int i = 0; i < fields_count(&out_fields); i++) {
 250                 struct field *f = fields_nth(&out_fields, i);
 251                 intarray_t *w = &column_widths;
 252
 253                 while (i >= intarray_count(w))
 254                         *intarray_push(w) = 0;
 255                 int fw = field_chars(f);
 256                 if (*intarray_nth(w, i) < fw)
 257                         *intarray_nth(w, i) = fw;
 258         }
 259 }
 260
 261 /*** CSV/TSV back-end */
 262
 263 static int csv_read(struct format *fmt)
 264 {
 265         int quoted = 0;
 266         for (;;) {
 267                 int c = getchar_unlocked();
 268                 int i = line_count(&in_line);
 269 restart:
 270                 if (c == '\r')
 271                         continue;
 272                 if (c < 0 || c == '\n') {
 273                         if (quoted)
 274                                 warn(fmt, "Missing closing quote.");
 275                         if (c < 0)
 276                                 return !!fields_count(&in_fields);
 277                         else
 278                                 return 1;
 279                 }
 280                 if (quoted) {
 281                         if (c == fmt->quote) {
 282                                 c = getchar_unlocked();
 283                                 if (c != fmt->quote) {
 284                                         quoted = 0;
 285                                         goto restart;
 286                                 }
 287                                 // Two quotes assimilate to one
 288                         }
 289                         // Fall through to pushing the character
 290                 } else if (c == fmt->quote) {
 291                         quoted = 1;
 292                         continue;
 293                 } else if (c == fmt->fs && !quoted) {
 294                         ensure_field(i);
 295                         new_field(i);
 296                         continue;
 297                 }
 298                 ensure_field(i);
 299                 *line_push(&in_line) = c;
 300                 in_field->len++;
 301         }
 302 }
 303
 304 static int is_ws(int c)
 305 {
 306         return (c == ' ' || c == '\t' || c == '\f');
 307 }
 308
 309 static void csv_write(struct format *fmt)
 310 {
 311         for (int i=0; i < fields_count(&out_fields); i++) {
 312                 int len;
 313                 unsigned char *p = get_field(&out_fields, i, &len);
 314
 315                 int need_quotes = 0;
 316                 if (fmt->quote >= 0) {
 317                         need_quotes = fmt->always_quote;
 318                         for (int j=0; !need_quotes && j < len; j++) {
 319                                 if (p[j] == fmt->fs || p[j] == fmt->quote)
 320                                         need_quotes = 1;
 321                         }
 322                 }
 323                 if (i)
 324                         putchar_unlocked(fmt->fs);
 325                 if (need_quotes)
 326                         putchar_unlocked(fmt->quote);
 327                 for (int j=0; j < len; j++) {
 328                         int c = p[j];
 329                         if (c == fmt->fs && !need_quotes)
 330                                 warn(fmt, "Field separator found inside field and quoting is turned off.");
 331                         if (c == fmt->quote)
 332                                 putchar_unlocked(c);
 333                         putchar_unlocked(c);
 334                 }
 335                 if (need_quotes)
 336                         putchar_unlocked(fmt->quote);
 337         }
 338         putchar_unlocked('\n');
 339 }
 340
 341 /*** White-space back-end ***/
 342
 343 static int ws_read(struct format *fmt)
 344 {
 345         if (!next_line())
 346                 return 0;
 347
 348         unsigned char *line = line_first(&in_line);
 349         int n = line_count(&in_line);
 350         if (!n)
 351                 return 1;
 352
 353         int ws = 0;
 354         new_field(0);
 355         for (int i=0; i<n; i++) {
 356                 int c = line[i];
 357                 if (is_ws(c)) {
 358                         ws++;
 359                 } else {
 360                         if (ws) {
 361                                 if (!in_field->start_pos &&
 362                                     !in_field->len &&
 363                                     fmt->sloppy)
 364                                         in_field->start_pos = i;
 365                                 else
 366                                         new_field(i);
 367                                 ws = 0;
 368                         }
 369                         in_field->len++;
 370                 }
 371         }
 372
 373         if (ws && !fmt->sloppy)
 374                 new_field(n);
 375         return 1;
 376 }
 377
 378 /*** Regex back-end ***/
 379
 380 static const char *regex_set(struct format *f, char *rx)
 381 {
 382         const char *err;
 383         int errpos;
 384         f->pcre = pcre_compile(rx, PCRE_DOLLAR_ENDONLY, &err, &errpos, NULL);
 385         if (!f->pcre)
 386                 return err;
 387
 388         f->pcre_extra = pcre_study(f->pcre, 0, &err);
 389         if (!f->pcre_extra)
 390                 return err;
 391
 392         return NULL;
 393 }
 394
 395 static int regex_read(struct format *fmt)
 396 {
 397         if (!next_line())
 398                 return 0;
 399
 400         unsigned char *c = line_first(&in_line);
 401         int n = line_count(&in_line);
 402         if (!n)
 403                 return 1;
 404
 405         int i = 0;
 406         for (;;) {
 407                 int ovec[3];
 408                 int err = pcre_exec(fmt->pcre, fmt->pcre_extra, (char *) c, n, i, 0, ovec, 3);
 409                 if (err < 0) {
 410                         if (err != PCRE_ERROR_NOMATCH)
 411                                 warn(fmt, "PCRE matching error %d", err);
 412                         // No further occurrence of the separator: the rest is a single field
 413                         if (!fmt->sloppy || i < n) {
 414                                 new_field(i);
 415                                 in_field->len = n - i;
 416                         }
 417                         return 1;
 418                 }
 419                 if (!fmt->sloppy || ovec[0]) {
 420                         new_field(i);
 421                         in_field->len = ovec[0] - i;
 422                 }
 423                 i = ovec[1];
 424         }
 425 }
 426
 427 /*** Table back-end ***/
 428
 429 static void table_write(struct format *fmt)
 430 {
 431         for (int i = 0; i < intarray_count(&column_widths); i++) {
 432                 if (fmt->table_grid) {
 433                         putchar_unlocked('|');
 434                         printf("%*s", fmt->table_sep / 2, "");
 435                 } else if (i)
 436                         printf("%*s", fmt->table_sep, "");
 437
 438                 int cw = *intarray_nth(&column_widths, i);
 439                 int fw = 0;
 440                 if (i < fields_count(&out_fields)) {
 441                         int len;
 442                         unsigned char *p = get_field(&out_fields, i, &len);
 443                         fw = field_chars(fields_nth(&out_fields, i));
 444                         if (fw > cw) {
 445                                 warn(fmt, "Internal error: Wrongly calculated width of column %d (%d > %d)", i, fw, cw);
 446                                 cw = fw;
 447                         }
 448                         while (len--)
 449                                 putchar(*p++);
 450                 }
 451                 while (fw < cw) {
 452                         putchar_unlocked(' ');
 453                         fw++;
 454                 }
 455
 456                 if (fmt->table_grid)
 457                         printf("%*s", fmt->table_sep - fmt->table_sep / 2, "");
 458         }
 459
 460         if (fmt->table_grid)
 461                 putchar_unlocked('|');
 462         putchar_unlocked('\n');
 463 }
 464
 465 static void table_write_grid(struct format *fmt, int pos UNUSED)
 466 {
 467         if (!fmt->table_grid)
 468                 return;
 469
 470         for (int i = 0; i < intarray_count(&column_widths); i++) {
 471                 putchar_unlocked('+');
 472                 int w = fmt->table_sep + *intarray_nth(&column_widths, i);
 473                 while (w--)
 474                         putchar('-');
 475         }
 476         putchar_unlocked('+');
 477         putchar_unlocked('\n');
 478 }
 479
 480 /*** Temporary file back-end ***/
 481
 482 static int tmp_read(struct format *fmt)
 483 {
 484         FILE *tf = fmt->tmp_file;
 485
 486         for (;;) {
 487                 int c = getc_unlocked(tf);
 488                 if (c < 0)
 489                         return 0;
 490                 if (c == 0xff)
 491                         return 1;
 492                 if (c == 0xfe) {
 493                         c = getc_unlocked(tf);
 494                         c = (c << 8) | getc_unlocked(tf);
 495                         c = (c << 8) | getc_unlocked(tf);
 496                         c = (c << 8) | getc_unlocked(tf);
 497                 }
 498                 new_field(line_count(&in_line));
 499                 in_field->len = c;
 500                 while (c--) {
 501                         int x = getc_unlocked(tf);
 502                         if (x < 0)
 503                                 die("Truncated temporary file");
 504                         *line_push(&in_line) = x;
 505                 }
 506         }
 507
 508         if (ferror_unlocked(tf))
 509                 die("I/O error when reading temporary file");
 510 }
 511
 512 static void tmp_write(struct format *fmt)
 513 {
 514         FILE *tf = fmt->tmp_file;
 515
 516         for (int i = 0; i < fields_count(&out_fields); i++) {
 517                 int len;
 518                 unsigned char *p = get_field(&out_fields, i, &len);
 519
 520                 if (len < 0xfe)
 521                         putc_unlocked(len, tf);
 522                 else {
 523                         putc_unlocked(0xfe, tf);
 524                         putc_unlocked((len >> 24) & 0xff, tf);
 525                         putc_unlocked((len >> 16) & 0xff, tf);
 526                         putc_unlocked((len >> 8) & 0xff, tf);
 527                         putc_unlocked(len & 0xff, tf);
 528                 }
 529
 530                 while (len--)
 531                         putc_unlocked(*p++, tf);
 532         }
 533         putc_unlocked(0xff, tf);
 534
 535         if (ferror_unlocked(tf))
 536                 die("I/O error when writing temporary file");
 537 }
 538
 539 /*** Transforms ***/
 540
 541 static void trim_fields(void)
 542 {
 543         unsigned char *line = line_first(&in_line);
 544         for (int i = 0; i < fields_count(&in_fields); i++) {
 545                 struct field *f = fields_nth(&in_fields, i);
 546                 while (f->len && is_ws(line[f->start_pos]))
 547                         f->start_pos++, f->len--;
 548                 while (f->len && is_ws(line[f->start_pos + f->len - 1]))
 549                         f->len--;
 550         }
 551 }
 552
 553 static void equalize_fields(void)
 554 {
 555         while (fields_count(&out_fields) < intarray_count(&column_widths)) {
 556                 struct field *f = fields_push(&out_fields);
 557                 f->start_pos = f->len = 0;
 558         }
 559 }
 560
 561 /*** Field names and headers ***/
 562
 563 struct field_names {
 564         stringarray_t names;
 565 };
 566
 567 static void add_field(struct field_names *fn, char *name, int namelen)
 568 {
 569         char *n = xmalloc(namelen + 1);
 570         memcpy(n, name, namelen);
 571         n[namelen] = 0;
 572         *stringarray_push(&fn->names) = n;
 573 }
 574
 575 static void add_field_names(struct field_names *fn, char *names)
 576 {
 577         char *p = names;
 578         while (p) {
 579                 char *q = strchr(p, ',');
 580                 int len = q ? q-p : (int) strlen(p);
 581                 add_field(fn, p, len);
 582                 p = q ? q+1 : NULL;
 583         }
 584 }
 585
 586 static void read_header(void)
 587 {
 588         if (!(in_format->has_header || in_format->set_field_names))
 589                 return;
 590
 591         struct field_names *fn = xmalloc_zero(sizeof(*fn));
 592         in_format->field_names = fn;
 593
 594         if (in_format->has_header) {
 595                 if (!read_line())
 596                         die("Missing input header");
 597         }
 598
 599         if (in_format->set_field_names) {
 600                 add_field_names(fn, in_format->set_field_names);
 601         } else {
 602                 for (int i = 0; i < fields_count(&in_fields); i++) {
 603                         int len;
 604                         char *s = (char *) get_field(&in_fields, i, &len);
 605                         add_field(fn, s, len);
 606                 }
 607         }
 608 }
 609
 610 static void write_header(void)
 611 {
 612         if (!out_format->has_header) {
 613                 write_grid(-1);
 614                 return;
 615         }
 616
 617         int want_select_fields = 0;
 618         if (out_format->set_field_names) {
 619                 struct field_names *fn = xmalloc_zero(sizeof(*fn));
 620                 out_format->field_names = fn;
 621                 add_field_names(fn, out_format->set_field_names);
 622         } else if (in_format->field_names) {
 623                 out_format->field_names = in_format->field_names;
 624                 want_select_fields = 1;
 625         } else
 626                 die("Output header requested, but no field names specified");
 627
 628         line_reset(&in_line);
 629         fields_reset(&in_fields);
 630         struct field_names *fn = out_format->field_names;
 631         for (int i = 0; i < stringarray_count(&fn->names); i++) {
 632                 struct field *f = fields_push(&in_fields);
 633                 f->start_pos = line_count(&in_line);
 634                 f->len = 0;
 635                 char *s = *stringarray_nth(&fn->names, i);
 636                 while (*s) {
 637                         *line_push(&in_line) = *s++;
 638                         f->len++;
 639                 }
 640         }
 641
 642         fields_reset(&out_fields);
 643         if (want_select_fields)
 644                 select_fields();
 645         else
 646                 select_all_fields();
 647
 648         // This is tricky: when we are formatting a table, field names are normally
 649         // calculated in pass 1, but the header is written in pass 2, so we have to
 650         // update column statistics, because field name can be too wide to fit.
 651         want_stats++;
 652         update_stats();
 653         want_stats--;
 654         if (want_equalize)
 655                 equalize_fields();
 656         write_grid(-1);
 657         write_line();
 658         write_grid(0);
 659 }
 660
 661 static void write_footer(void)
 662 {
 663         write_grid(1);
 664 }
 665
 666 static int find_field_by_name(struct field_names *fn, char *name)
 667 {
 668         for (int i = 0; i < stringarray_count(&fn->names); i++)
 669                 if (!strcmp(*stringarray_nth(&fn->names, i), name))
 670                         return i + 1;
 671         return -1;
 672 }
 673
 674 /*** Field selection ***/
 675
 676 struct selector {
 677         int first_field, last_field;            // 0 means "boundary"
 678 };
 679
 680 DECLARE_BUF(selectors, struct selector);
 681 static selectors_t selectors;
 682
 683 static int parse_field_num(char *str)
 684 {
 685         int f = 0;
 686
 687         while (*str) {
 688                 if (*str < '0' || *str > '9')
 689                         return -1;
 690                 if (f >= 100000000)
 691                         return -1;
 692                 f = 10*f + *str - '0';
 693                 str++;
 694         }
 695         return f;
 696 }
 697
 698 static int parse_field(char *str)
 699 {
 700         if (!*str)
 701                 return 0;
 702
 703         int f = parse_field_num(str);
 704         if (f > 0)
 705                 return f;
 706
 707         if (in_format->field_names && (f = find_field_by_name(in_format->field_names, str)) > 0)
 708                 return f;
 709
 710         die("Unknown field `%s'", str);
 711 }
 712
 713 static char *parse_selector(char *str)
 714 {
 715         char buf[strlen(str) + 1];
 716         strcpy(buf, str);
 717
 718         struct selector *s = selectors_push(&selectors);
 719         char *sep = strchr(buf, '-');
 720         if (sep) {
 721                 *sep++ = 0;
 722                 s->first_field = parse_field(buf);
 723                 s->last_field = parse_field(sep);
 724         } else
 725                 s->first_field = s->last_field = parse_field(buf);
 726
 727         return NULL;
 728 }
 729
 730 static void finish_parse_selectors(void)
 731 {
 732         if (!selectors_count(&selectors))
 733                 parse_selector("-");
 734 }
 735
 736 static void select_fields(void)
 737 {
 738         for (int i = 0; i < selectors_count(&selectors); i++) {
 739                 struct selector *s = selectors_nth(&selectors, i);
 740                 int first = s->first_field;
 741                 if (first <= 0)
 742                         first = 1;
 743                 int last = s->last_field;
 744                 if (last <= 0)
 745                         last = fields_count(&in_fields);
 746                 for (int j = first; j <= last; j++) {
 747                         struct field *f = fields_push(&out_fields);
 748                         if (j >= 1 && j <= fields_count(&in_fields))
 749                                 *f = *fields_nth(&in_fields, j-1);
 750                         else
 751                                 f->start_pos = f->len = 0;
 752                 }
 753         }
 754 }
 755
 756 static void select_all_fields(void)
 757 {
 758         for (int i = 0; i < fields_count(&in_fields); i++)
 759                 *fields_push(&out_fields) = *fields_nth(&in_fields, i);
 760 }
 761
 762 /*** Processing of files ***/
 763
 764 static void one_pass(int pass)
 765 {
 766         if (pass & 2)
 767                 write_header();
 768
 769         for (;;) {
 770                 line_number++;
 771                 if (!read_line())
 772                         break;
 773
 774                 if (want_trim && (pass & 1))
 775                         trim_fields();
 776
 777                 fields_reset(&out_fields);
 778                 if (pass & 1)
 779                         select_fields();
 780                 else
 781                         select_all_fields();
 782
 783                 if (want_equalize && (pass & 2))
 784                         equalize_fields();
 785                 update_stats();
 786                 write_line();
 787         }
 788
 789         if (pass & 2)
 790                 write_footer();
 791 }
 792
 793 static void two_pass(void)
 794 {
 795         struct format *final_format = out_format;
 796
 797         // We need to use character set info from the current locale
 798         setlocale(LC_CTYPE, "");
 799
 800         // Pass 1: Set up writer of intermediate format
 801         out_format = xmalloc_zero(sizeof(*out_format));
 802         out_format->id = FORM_TMP;
 803         out_format->read_line = tmp_read;
 804         out_format->write_line = tmp_write;
 805         out_format->tmp_file = tmpfile();
 806         out_format->field_names = in_format->field_names;
 807         one_pass(1);
 808
 809         // Pass 2: Set up reader of intermediate format
 810         in_format = out_format;
 811         rewind(in_format->tmp_file);
 812         line_number = 0;
 813         out_format = final_format;
 814         want_stats = 0;
 815         one_pass(2);
 816         fclose(in_format->tmp_file);
 817 }
 818
 819 /*** Parsing of arguments ***/
 820
 821 static void NONRET usage(void)
 822 {
 823         printf("\
 824 Usage: xsv <in-format> [<out-format>] <options> [<fields>]\n\
 825 \n\
 826 Formats:\n\
 827 -t, --tsv               Tab-separated values (default)\n\
 828 -c, --csv               Comma-separated values\n\
 829 -w, --ws                Values separated by arbitrary whitespace\n\
 830 -r, --regex=<rx>        Separator given by Perl regular expression (input only)\n\
 831     --table             Format a table (output only)\n\
 832 \n\
 833 Format parameters:\n\
 834 -d, --fs=<char>         Delimiter of fields\n\
 835 -f, --fields=<f>,...    Set field names\n\
 836 -h, --header            The first line contains field names\n\
 837 -q, --quiet             Do not show warnings\n\
 838     --always-quote      Put quotes around all fields (CSV output only)\n\
 839     --table-sep=<n>     Separate table columns by <n> spaces (default: 2)\n\
 840     --grid              Separate table columns by grid lines\n\
 841 -s, --sloppy            Ignore separators at the start/end of line (ws/regex only)\n\
 842 \n\
 843 Other options:\n\
 844     --trim              Trim leading and trailing whitespaces in fields\n\
 845     --equalize          Pad all lines to the maximum number of fields\n\
 846 ");
 847         exit(0);
 848 }
 849
 850 static void NONRET bad_args(const char *msg, ...)
 851 {
 852         if (msg) {
 853                 va_list args;
 854                 va_start(args, msg);
 855                 fprintf(stderr, "xsv: ");
 856                 vfprintf(stderr, msg, args);
 857                 fputc('\n', stderr);
 858                 va_end(args);
 859         }
 860         fprintf(stderr, "Try `xsv --help' for more information.\n");
 861         exit(1);
 862 }
 863
 864 static const char short_options[] = "cd:f:hqr:twW";
 865
 866 enum long_options {
 867         OPT_HELP = 256,
 868         OPT_TRIM,
 869         OPT_ALWAYS_QUOTE,
 870         OPT_TABLE,
 871         OPT_TABLE_SEP,
 872         OPT_GRID,
 873         OPT_EQUALIZE,
 874 };
 875
 876 static const struct option long_options[] = {
 877         { "always-quote",       0,      NULL,   OPT_ALWAYS_QUOTE },
 878         { "csv",                0,      NULL,   'c' },
 879         { "equalize",           0,      NULL,   OPT_EQUALIZE },
 880         { "fields",             1,      NULL,   'f' },
 881         { "fs",                 1,      NULL,   'd' },
 882         { "grid",               0,      NULL,   OPT_GRID },
 883         { "header",             0,      NULL,   'h' },
 884         { "quiet",              0,      NULL,   'q' },
 885         { "regex",              1,      NULL,   'r' },
 886         { "sloppy",             0,      NULL,   's' },
 887         { "table",              0,      NULL,   OPT_TABLE },
 888         { "table-sep",          1,      NULL,   OPT_TABLE_SEP },
 889         { "trim",               0,      NULL,   OPT_TRIM },
 890         { "tsv",                0,      NULL,   't' },
 891         { "ws",                 0,      NULL,   'w' },
 892         { "help",               0,      NULL,   OPT_HELP },
 893         { NULL,                 0,      NULL,   0 },
 894 };
 895
 896 static void set_format(int format_id)
 897 {
 898         struct format *f = xmalloc_zero(sizeof(*f));
 899         f->id = format_id;
 900
 901         switch (format_id) {
 902                 case FORM_TSV:
 903                         f->fs = '\t';
 904                         f->quote = -1;
 905                         f->read_line = csv_read;
 906                         f->write_line = csv_write;
 907                         break;
 908                 case FORM_CSV:
 909                         f->fs = ',';
 910                         f->quote = '"';
 911                         f->read_line = csv_read;
 912                         f->write_line = csv_write;
 913                         break;
 914                 case FORM_WS:
 915                         f->fs = ' ';
 916                         f->quote = -1;
 917                         f->read_line = ws_read;
 918                         f->write_line = csv_write;
 919                         break;
 920                 case FORM_REGEX:
 921                         f->read_line = regex_read;
 922                         break;
 923                 case FORM_TABLE:
 924                         f->write_line = table_write;
 925                         f->write_grid = table_write_grid;
 926                         f->needs_stats = 1;
 927                         f->table_sep = 2;
 928                         break;
 929         }
 930
 931         if (!in_format)
 932                 in_format = f;
 933         else if (!out_format)
 934                 out_format = f;
 935         else
 936                 bad_args("At most two formats may be given.");
 937 }
 938
 939 static struct format *current_format(void)
 940 {
 941         if (out_format)
 942                 return out_format;
 943         if (in_format)
 944                 return in_format;
 945         set_format(FORM_TSV);
 946         return in_format;
 947 }
 948
 949 int main(int argc, char **argv)
 950 {
 951         int opt;
 952         const char *err;
 953
 954         while ((opt = getopt_long(argc, argv, short_options, long_options, NULL)) >= 0)
 955                 switch (opt) {
 956                         case 'c':
 957                                 set_format(FORM_CSV);
 958                                 break;
 959                         case 'd':
 960                                 if (optarg[0])
 961                                         current_format()->fs = optarg[0];
 962                                 else
 963                                         bad_args("No field delimiter given.");
 964                                 break;
 965                         case 'f':
 966                                 current_format()->set_field_names = optarg;
 967                                 break;
 968                         case 'h':
 969                                 current_format()->has_header = 1;
 970                                 break;
 971                         case 'q':
 972                                 current_format()->quiet = 1;
 973                                 break;
 974                         case 'r':
 975                                 set_format(FORM_REGEX);
 976                                 err = regex_set(current_format(), optarg);
 977                                 if (err)
 978                                         bad_args("Error compiling regex: %s", err);
 979                                 break;
 980                         case 's':
 981                                 if (current_format()->id != FORM_WS && current_format()->id != FORM_REGEX)
 982                                         bad_args("--sloppy makes sense only for --ws or --regex.");
 983                                 current_format()->sloppy = 1;
 984                                 break;
 985                         case 't':
 986                                 set_format(FORM_TSV);
 987                                 break;
 988                         case 'w':
 989                                 set_format(FORM_WS);
 990                                 break;
 991                         case OPT_ALWAYS_QUOTE:
 992                                 if (current_format()->id != FORM_CSV)
 993                                         bad_args("--always-quote makes sense only for --csv.");
 994                                 current_format()->always_quote = 1;
 995                                 break;
 996                         case OPT_HELP:
 997                                 usage();
 998                         case OPT_TRIM:
 999                                 want_trim = 1;
1000                                 break;
1001                         case OPT_TABLE:
1002                                 set_format(FORM_TABLE);
1003                                 break;
1004                         case OPT_TABLE_SEP:
1005                                 current_format()->table_sep = atoi(optarg);
1006                                 break;
1007                         case OPT_GRID:
1008                                 current_format()->table_grid = 1;
1009                                 break;
1010                         case OPT_EQUALIZE:
1011                                 want_equalize = 1;
1012                                 break;
1013                         default:
1014                                 bad_args(NULL);
1015                 }
1016
1017         current_format();
1018         if (!out_format)
1019                 out_format = in_format;
1020         if (!in_format->read_line)
1021                 bad_args("Write-only format selected for input.");
1022         if (!out_format->write_line)
1023                 bad_args("Read-only format selected for output.");
1024         read_header();
1025
1026         for (int i = optind; i < argc; i++) {
1027                 err = parse_selector(argv[i]);
1028                 if (err)
1029                         bad_args(err);
1030         }
1031         finish_parse_selectors();
1032
1033         want_stats = out_format->needs_stats | want_equalize;
1034         if (want_stats)
1035                 two_pass();
1036         else
1037                 one_pass(3);
1038         return 0;
1039 }