| 1 |
#ifndef lint |
| 2 |
static const char RCSid[] = "$Id: rcollate.c,v 2.5 2013/09/06 21:43:29 greg Exp $"; |
| 3 |
#endif |
| 4 |
/* |
| 5 |
* Utility to re-order records in a binary or ASCII data file (matrix) |
| 6 |
*/ |
| 7 |
|
| 8 |
#include <stdlib.h> |
| 9 |
#include <string.h> |
| 10 |
#include <ctype.h> |
| 11 |
#include "platform.h" |
| 12 |
#include "rtio.h" |
| 13 |
#include "resolu.h" |
| 14 |
#ifndef _WIN32 |
| 15 |
#include <sys/mman.h> |
| 16 |
#endif |
| 17 |
|
| 18 |
#ifdef getc_unlocked /* avoid horrendous overhead of flockfile */ |
| 19 |
#undef getc |
| 20 |
#undef putc |
| 21 |
#define getc getc_unlocked |
| 22 |
#define putc putc_unlocked |
| 23 |
#endif |
| 24 |
|
| 25 |
typedef struct { |
| 26 |
void *base; /* pointer to base memory */ |
| 27 |
size_t len; /* allocated memory length */ |
| 28 |
int mapped; /* memory-mapped file? */ |
| 29 |
} MEMLOAD; /* file loaded/mapped into memory */ |
| 30 |
|
| 31 |
typedef struct { |
| 32 |
int nw_rec; /* number of words per record */ |
| 33 |
int nrecs; /* number of records we found */ |
| 34 |
char *rec[1]; /* record array (extends struct) */ |
| 35 |
} RECINDEX; |
| 36 |
|
| 37 |
int warnings = 1; /* report warnings? */ |
| 38 |
|
| 39 |
/* free loaded file */ |
| 40 |
static void |
| 41 |
free_load(MEMLOAD *mp) |
| 42 |
{ |
| 43 |
if (mp == NULL || (mp->base == NULL) | (mp->len <= 0)) |
| 44 |
return; |
| 45 |
#ifdef MAP_FILE |
| 46 |
if (mp->mapped) |
| 47 |
munmap(mp->base, mp->len); |
| 48 |
else |
| 49 |
#endif |
| 50 |
free(mp->base); |
| 51 |
mp->base = NULL; |
| 52 |
mp->len = 0; |
| 53 |
} |
| 54 |
|
| 55 |
/* load a file into memory */ |
| 56 |
static int |
| 57 |
load_file(MEMLOAD *mp, FILE *fp) |
| 58 |
{ |
| 59 |
int fd; |
| 60 |
off_t skip, flen; |
| 61 |
|
| 62 |
if (mp == NULL) |
| 63 |
return(-1); |
| 64 |
mp->base = NULL; |
| 65 |
mp->len = 0; |
| 66 |
mp->mapped = 0; |
| 67 |
if (fp == NULL) |
| 68 |
return(-1); |
| 69 |
fd = fileno(fp); |
| 70 |
skip = ftello(fp); |
| 71 |
flen = lseek(fd, 0, SEEK_END); |
| 72 |
if (flen <= skip) |
| 73 |
return((int)(flen - skip)); |
| 74 |
mp->len = (size_t)(flen - skip); |
| 75 |
#ifdef MAP_FILE |
| 76 |
if (mp->len > 1L<<20) { /* map file if > 1 MByte */ |
| 77 |
mp->base = mmap(NULL, mp->len, PROT_READ, MAP_PRIVATE, fd, skip); |
| 78 |
if (mp->base != MAP_FAILED) { |
| 79 |
mp->mapped = 1; |
| 80 |
return(1); /* mmap() success */ |
| 81 |
} |
| 82 |
mp->base = NULL; /* fall back to reading it in... */ |
| 83 |
} |
| 84 |
#endif |
| 85 |
if (lseek(fd, skip, SEEK_SET) != skip || |
| 86 |
(mp->base = malloc(mp->len)) == NULL) { |
| 87 |
mp->len = 0; |
| 88 |
return(-1); |
| 89 |
} |
| 90 |
if (read(fd, (char *)mp->base, mp->len) != mp->len) { |
| 91 |
free_load(mp); |
| 92 |
return(-1); |
| 93 |
} |
| 94 |
return(1); |
| 95 |
} |
| 96 |
|
| 97 |
/* load memory from an input stream, starting from current position */ |
| 98 |
static int |
| 99 |
load_stream(MEMLOAD *mp, FILE *fp) |
| 100 |
{ |
| 101 |
char buf[8192]; |
| 102 |
size_t nr; |
| 103 |
|
| 104 |
if (mp == NULL) |
| 105 |
return(-1); |
| 106 |
mp->base = NULL; |
| 107 |
mp->len = 0; |
| 108 |
mp->mapped = 0; |
| 109 |
if (fp == NULL) |
| 110 |
return(-1); |
| 111 |
while ((nr = fread(buf, 1, sizeof(buf), fp)) > 0) { |
| 112 |
if (!mp->len) |
| 113 |
mp->base = malloc(nr); |
| 114 |
else |
| 115 |
mp->base = realloc(mp->base, mp->len+nr); |
| 116 |
if (mp->base == NULL) |
| 117 |
return(-1); |
| 118 |
memcpy((char *)mp->base + mp->len, buf, nr); |
| 119 |
mp->len += nr; |
| 120 |
} |
| 121 |
if (ferror(fp)) { |
| 122 |
free_load(mp); |
| 123 |
return(-1); |
| 124 |
} |
| 125 |
return(mp->len > 0); |
| 126 |
} |
| 127 |
|
| 128 |
/* free a record index */ |
| 129 |
#define free_records(rp) free(rp) |
| 130 |
|
| 131 |
/* compute record index */ |
| 132 |
static RECINDEX * |
| 133 |
index_records(const MEMLOAD *mp, int nw_rec) |
| 134 |
{ |
| 135 |
RECINDEX *rp; |
| 136 |
char *cp, *mend; |
| 137 |
int n; |
| 138 |
|
| 139 |
if (mp == NULL || (mp->base == NULL) | (mp->len <= 0)) |
| 140 |
return(NULL); |
| 141 |
if (nw_rec <= 0) |
| 142 |
return(NULL); |
| 143 |
rp = (RECINDEX *)malloc(sizeof(RECINDEX) + mp->len/(2*nw_rec)*sizeof(char *)); |
| 144 |
if (rp == NULL) |
| 145 |
return(NULL); |
| 146 |
rp->nw_rec = nw_rec; |
| 147 |
rp->nrecs = 0; |
| 148 |
cp = (char *)mp->base; |
| 149 |
mend = cp + mp->len; |
| 150 |
for ( ; ; ) { /* whitespace-separated words */ |
| 151 |
while (cp < mend && !*cp | isspace(*cp)) |
| 152 |
++cp; |
| 153 |
if (cp >= mend) |
| 154 |
break; |
| 155 |
rp->rec[rp->nrecs++] = cp; /* point to first non-white */ |
| 156 |
n = rp->nw_rec; |
| 157 |
while (++cp < mend) /* find end of record */ |
| 158 |
if (!*cp | isspace(*cp)) { |
| 159 |
if (--n <= 0) |
| 160 |
break; /* got requisite # words */ |
| 161 |
do { /* else find next word */ |
| 162 |
if (*cp == '\n') { |
| 163 |
fprintf(stderr, |
| 164 |
"Unexpected EOL in record!\n"); |
| 165 |
free_records(rp); |
| 166 |
return(NULL); |
| 167 |
} |
| 168 |
if (++cp >= mend) |
| 169 |
break; |
| 170 |
} while (!*cp | isspace(*cp)); |
| 171 |
} |
| 172 |
} |
| 173 |
rp->rec[rp->nrecs] = mend; /* reallocate to save space */ |
| 174 |
rp = (RECINDEX *)realloc(rp, |
| 175 |
sizeof(RECINDEX) + rp->nrecs*sizeof(char *)); |
| 176 |
return(rp); |
| 177 |
} |
| 178 |
|
| 179 |
/* count number of columns based on first EOL */ |
| 180 |
static int |
| 181 |
count_columns(const RECINDEX *rp) |
| 182 |
{ |
| 183 |
char *cp = rp->rec[0]; |
| 184 |
char *mend = rp->rec[rp->nrecs]; |
| 185 |
int i; |
| 186 |
|
| 187 |
while (*cp != '\n') |
| 188 |
if (++cp >= mend) |
| 189 |
return(0); |
| 190 |
for (i = 0; i < rp->nrecs; i++) |
| 191 |
if (rp->rec[i] >= cp) |
| 192 |
break; |
| 193 |
return(i); |
| 194 |
} |
| 195 |
|
| 196 |
/* copy nth record from index to stdout */ |
| 197 |
static int |
| 198 |
print_record(const RECINDEX *rp, int n) |
| 199 |
{ |
| 200 |
int words2go = rp->nw_rec; |
| 201 |
char *scp; |
| 202 |
|
| 203 |
if ((n < 0) | (n >= rp->nrecs)) |
| 204 |
return(0); |
| 205 |
scp = rp->rec[n]; |
| 206 |
do { |
| 207 |
putc(*scp++, stdout); |
| 208 |
if (!*scp | isspace(*scp)) { |
| 209 |
if (--words2go <= 0) |
| 210 |
break; |
| 211 |
putc(' ', stdout); /* single space btwn. words */ |
| 212 |
do |
| 213 |
if (++scp >= rp->rec[n+1]) |
| 214 |
break; |
| 215 |
while (!*scp | isspace(*scp)); |
| 216 |
} |
| 217 |
} while (scp < rp->rec[n+1]); |
| 218 |
/* caller adds record sep. */ |
| 219 |
return(1); |
| 220 |
} |
| 221 |
|
| 222 |
/* copy a stream to stdout */ |
| 223 |
static int |
| 224 |
output_stream(FILE *fp) |
| 225 |
{ |
| 226 |
char buf[8192]; |
| 227 |
ssize_t n; |
| 228 |
|
| 229 |
if (fp == NULL) |
| 230 |
return(0); |
| 231 |
fflush(stdout); /* assumes nothing in input buffer */ |
| 232 |
while ((n = read(fileno(fp), buf, sizeof(buf))) > 0) |
| 233 |
if (write(fileno(stdout), buf, n) != n) |
| 234 |
return(0); |
| 235 |
return(n >= 0); |
| 236 |
} |
| 237 |
|
| 238 |
/* get next word from stream, leaving stream on EOL or start of next word */ |
| 239 |
static char * |
| 240 |
fget_word(char buf[256], FILE *fp) |
| 241 |
{ |
| 242 |
int c; |
| 243 |
char *cp; |
| 244 |
/* skip nul's and white space */ |
| 245 |
while (!(c = getc(fp)) || isspace(c)) |
| 246 |
; |
| 247 |
if (c == EOF) |
| 248 |
return(NULL); |
| 249 |
cp = buf; |
| 250 |
do |
| 251 |
*cp++ = c; |
| 252 |
while ((c = getc(fp)) != EOF && !isspace(c) && cp < buf+255); |
| 253 |
*cp = '\0'; |
| 254 |
while (isspace(c) & (c != '\n')) |
| 255 |
c = getc(fp); |
| 256 |
if (c != EOF) |
| 257 |
ungetc(c, fp); |
| 258 |
return(buf); |
| 259 |
} |
| 260 |
|
| 261 |
char *fmtid = "ascii"; /* format id */ |
| 262 |
int record_width = 3; /* words/record (<0 binary) */ |
| 263 |
int ni_columns = 0; /* number of input columns */ |
| 264 |
int ni_rows = 0; /* number of input rows */ |
| 265 |
int no_columns = 0; /* number of output columns */ |
| 266 |
int no_rows = 0; /* number of output rows */ |
| 267 |
|
| 268 |
/* output transposed ASCII or binary data from memory */ |
| 269 |
static int |
| 270 |
do_transpose(const MEMLOAD *mp) |
| 271 |
{ |
| 272 |
static const char tabEOL[2] = {'\t','\n'}; |
| 273 |
RECINDEX *rp = NULL; |
| 274 |
long nrecords; |
| 275 |
int i, j; |
| 276 |
/* propogate sizes */ |
| 277 |
if (ni_rows <= 0) |
| 278 |
ni_rows = no_columns; |
| 279 |
if (ni_columns <= 0) |
| 280 |
ni_columns = no_rows; |
| 281 |
/* get # records (& index) */ |
| 282 |
if (record_width > 0) { |
| 283 |
if ((rp = index_records(mp, record_width)) == NULL) |
| 284 |
return(0); |
| 285 |
if (ni_columns <= 0) |
| 286 |
ni_columns = count_columns(rp); |
| 287 |
nrecords = rp->nrecs; |
| 288 |
} else if ((ni_rows > 0) & (ni_columns > 0)) { |
| 289 |
nrecords = ni_rows*ni_columns; |
| 290 |
if (nrecords > mp->len / -record_width) { |
| 291 |
fprintf(stderr, |
| 292 |
"Input too small for specified size and type\n"); |
| 293 |
return(0); |
| 294 |
} |
| 295 |
} else |
| 296 |
nrecords = mp->len / -record_width; |
| 297 |
/* check sizes */ |
| 298 |
if ((ni_rows <= 0) & (ni_columns > 0)) |
| 299 |
ni_rows = nrecords/ni_columns; |
| 300 |
if ((ni_columns <= 0) & (ni_rows > 0)) |
| 301 |
ni_columns = nrecords/ni_rows; |
| 302 |
if (nrecords != ni_rows*ni_columns) |
| 303 |
goto badspec; |
| 304 |
if (no_columns <= 0) |
| 305 |
no_columns = ni_rows; |
| 306 |
if (no_rows <= 0) |
| 307 |
no_rows = ni_columns; |
| 308 |
if ((no_rows != ni_columns) | (no_columns != ni_rows)) |
| 309 |
goto badspec; |
| 310 |
/* transpose records */ |
| 311 |
for (i = 0; i < no_rows; i++) { |
| 312 |
for (j = 0; j < no_columns; j++) |
| 313 |
if (rp != NULL) { /* ASCII output */ |
| 314 |
print_record(rp, j*ni_columns + i); |
| 315 |
putc(tabEOL[j >= no_columns-1], stdout); |
| 316 |
} else { /* binary output */ |
| 317 |
fwrite((char *)mp->base + |
| 318 |
-record_width*(j*ni_columns + i), |
| 319 |
-record_width, 1, stdout); |
| 320 |
} |
| 321 |
if (ferror(stdout)) { |
| 322 |
fprintf(stderr, "Error writing to stdout\n"); |
| 323 |
return(0); |
| 324 |
} |
| 325 |
} |
| 326 |
if (rp != NULL) |
| 327 |
free_records(rp); |
| 328 |
return(1); |
| 329 |
badspec: |
| 330 |
fprintf(stderr, "Bad transpose specification -- check dimension(s)\n"); |
| 331 |
return(0); |
| 332 |
} |
| 333 |
|
| 334 |
/* resize ASCII stream input by ignoring EOLs between records */ |
| 335 |
static int |
| 336 |
do_resize(FILE *fp) |
| 337 |
{ |
| 338 |
long records2go = ni_rows*ni_columns; |
| 339 |
int columns2go = no_columns; |
| 340 |
char word[256]; |
| 341 |
/* sanity checks */ |
| 342 |
if (record_width <= 0) { |
| 343 |
fprintf(stderr, "Bad call to do_resize (record_width = %d)\n", |
| 344 |
record_width); |
| 345 |
return(0); |
| 346 |
} |
| 347 |
if (no_columns <= 0) { |
| 348 |
fprintf(stderr, "Missing -oc specification\n"); |
| 349 |
return(0); |
| 350 |
} |
| 351 |
if ((records2go <= 0) & (no_rows > 0)) |
| 352 |
records2go = no_rows*no_columns; |
| 353 |
else if (no_rows*no_columns != records2go) { |
| 354 |
fprintf(stderr, |
| 355 |
"Input and output data sizes disagree (%dx%d != %dx%d)\n", |
| 356 |
ni_rows, ni_columns, no_rows, no_columns); |
| 357 |
return(0); |
| 358 |
} |
| 359 |
do { /* reshape records */ |
| 360 |
int n; |
| 361 |
|
| 362 |
for (n = record_width; n--; ) { |
| 363 |
if (fget_word(word, fp) == NULL) { |
| 364 |
if (records2go > 0 || n < record_width-1) |
| 365 |
break; |
| 366 |
goto done; /* normal EOD */ |
| 367 |
} |
| 368 |
fputs(word, stdout); |
| 369 |
if (n) { /* mid-record? */ |
| 370 |
int c = getc(fp); |
| 371 |
if ((c == '\n') | (c == EOF)) |
| 372 |
break; |
| 373 |
ungetc(c, fp); |
| 374 |
putc(' ', stdout); |
| 375 |
} |
| 376 |
} |
| 377 |
if (n >= 0) { |
| 378 |
fprintf(stderr, "Incomplete record / unexpected EOF\n"); |
| 379 |
return(0); |
| 380 |
} |
| 381 |
if (--columns2go <= 0) { /* time to end output row? */ |
| 382 |
putc('\n', stdout); |
| 383 |
columns2go = no_columns; |
| 384 |
} else /* else separate records */ |
| 385 |
putc('\t', stdout); |
| 386 |
} while (--records2go); /* expected EOD? */ |
| 387 |
done: |
| 388 |
if (warnings && columns2go != no_columns) |
| 389 |
fprintf(stderr, "Warning -- incomplete final row\n"); |
| 390 |
if (warnings && fget_word(word, fp) != NULL) |
| 391 |
fprintf(stderr, "Warning -- characters beyond expected EOD\n"); |
| 392 |
return(1); |
| 393 |
} |
| 394 |
|
| 395 |
/* process a header line and copy to stdout */ |
| 396 |
static int |
| 397 |
headline(char *s, void *p) |
| 398 |
{ |
| 399 |
char fmt[32]; |
| 400 |
|
| 401 |
if (formatval(fmt, s)) { |
| 402 |
if (!strcmp(fmt, fmtid)) |
| 403 |
return(0); |
| 404 |
fprintf(stderr, "Input format '%s' != '%s'\n", fmt, fmtid); |
| 405 |
return(-1); |
| 406 |
} |
| 407 |
fputs(s, stdout); /* copy header info. */ |
| 408 |
return(0); |
| 409 |
} |
| 410 |
|
| 411 |
/* main routine for converting rows/columns in data file */ |
| 412 |
int |
| 413 |
main(int argc, char *argv[]) |
| 414 |
{ |
| 415 |
int do_header = 1; /* header i/o? */ |
| 416 |
int transpose = 0; /* transpose rows & cols? */ |
| 417 |
int i; |
| 418 |
|
| 419 |
for (i = 1; i < argc && argv[i][0] == '-'; i++) |
| 420 |
switch (argv[i][1]) { |
| 421 |
case 'i': /* input */ |
| 422 |
if (argv[i][2] == 'c') /* columns */ |
| 423 |
ni_columns = atoi(argv[++i]); |
| 424 |
else if (argv[i][2] == 'r') |
| 425 |
ni_rows = atoi(argv[++i]); |
| 426 |
else |
| 427 |
goto userr; |
| 428 |
break; |
| 429 |
case 'o': /* output */ |
| 430 |
if (argv[i][2] == 'c') /* columns */ |
| 431 |
no_columns = atoi(argv[++i]); |
| 432 |
else if (argv[i][2] == 'r') |
| 433 |
no_rows = atoi(argv[++i]); |
| 434 |
else |
| 435 |
goto userr; |
| 436 |
break; |
| 437 |
case 'h': /* header on/off */ |
| 438 |
do_header = !do_header; |
| 439 |
break; |
| 440 |
case 't': /* transpose on/off */ |
| 441 |
transpose = !transpose; |
| 442 |
break; |
| 443 |
case 'f': /* format */ |
| 444 |
switch (argv[i][2]) { |
| 445 |
case 'a': /* ASCII */ |
| 446 |
case 'A': |
| 447 |
fmtid = "ascii"; |
| 448 |
record_width = 1; |
| 449 |
break; |
| 450 |
case 'f': /* float */ |
| 451 |
case 'F': |
| 452 |
fmtid = "float"; |
| 453 |
record_width = -(int)sizeof(float); |
| 454 |
break; |
| 455 |
case 'd': /* double */ |
| 456 |
case 'D': |
| 457 |
fmtid = "double"; |
| 458 |
record_width = -(int)sizeof(double); |
| 459 |
break; |
| 460 |
case 'b': /* binary (bytes) */ |
| 461 |
case 'B': |
| 462 |
fmtid = "byte"; |
| 463 |
record_width = -1; |
| 464 |
break; |
| 465 |
default: |
| 466 |
goto userr; |
| 467 |
} |
| 468 |
if (argv[i][3]) { |
| 469 |
if (!isdigit(argv[i][3])) |
| 470 |
goto userr; |
| 471 |
record_width *= atoi(argv[i]+3); |
| 472 |
} |
| 473 |
break; |
| 474 |
case 'w': /* warnings on/off */ |
| 475 |
warnings = !warnings; |
| 476 |
break; |
| 477 |
default: |
| 478 |
goto userr; |
| 479 |
} |
| 480 |
if (!record_width) |
| 481 |
goto userr; |
| 482 |
if (i < argc-1) /* arg count OK? */ |
| 483 |
goto userr; |
| 484 |
/* open input file? */ |
| 485 |
if (i == argc-1 && freopen(argv[i], "r", stdin) == NULL) { |
| 486 |
fprintf(stderr, "%s: cannot open for reading\n", argv[i]); |
| 487 |
return(1); |
| 488 |
} |
| 489 |
if (record_width < 0) { |
| 490 |
SET_FILE_BINARY(stdin); |
| 491 |
SET_FILE_BINARY(stdout); |
| 492 |
} |
| 493 |
/* check for no-op */ |
| 494 |
if (!transpose && (record_width < 0 || |
| 495 |
(no_columns == ni_columns) & (no_rows == ni_rows))) { |
| 496 |
if (warnings) |
| 497 |
fprintf(stderr, "%s: no-op -- copying input verbatim\n", |
| 498 |
argv[0]); |
| 499 |
if (!output_stream(stdin)) |
| 500 |
return(1); |
| 501 |
return(0); |
| 502 |
} |
| 503 |
if (do_header) { /* read/write header */ |
| 504 |
if (getheader(stdin, &headline, NULL) < 0) |
| 505 |
return(1); |
| 506 |
printargs(argc, argv, stdout); |
| 507 |
fputformat(fmtid, stdout); |
| 508 |
fputc('\n', stdout); /* finish new header */ |
| 509 |
} |
| 510 |
if (transpose) { /* transposing rows & columns? */ |
| 511 |
MEMLOAD myMem; /* need to load into memory */ |
| 512 |
if (i == argc-1) { |
| 513 |
if (load_file(&myMem, stdin) <= 0) { |
| 514 |
fprintf(stderr, "%s: error loading file into memory\n", |
| 515 |
argv[i]); |
| 516 |
return(1); |
| 517 |
} |
| 518 |
} else if (load_stream(&myMem, stdin) <= 0) { |
| 519 |
fprintf(stderr, "%s: error loading stdin into memory\n", |
| 520 |
argv[0]); |
| 521 |
return(1); |
| 522 |
} |
| 523 |
if (!do_transpose(&myMem)) |
| 524 |
return(1); |
| 525 |
/* free_load(&myMem); */ |
| 526 |
} else if (!do_resize(stdin)) /* just reshaping input */ |
| 527 |
return(1); |
| 528 |
return(0); |
| 529 |
userr: |
| 530 |
fprintf(stderr, |
| 531 |
"Usage: %s [-h][-w][-f[afdb][N]][-t][-ic in_col][-ir in_row][-oc out_col][-or out_row] [input.dat]\n", |
| 532 |
argv[0]); |
| 533 |
return(1); |
| 534 |
} |
