xsv.c

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