1 /* $NetBSD: join.c,v 1.20 2002/02/14 03:21:07 jonb Exp $ */ 2 3 /*- 4 * Copyright (c) 1991 The Regents of the University of California. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Steve Hayman of Indiana University, Michiro Hikida and David 9 * Goodenough. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the University of 22 * California, Berkeley and its contributors. 23 * 4. Neither the name of the University nor the names of its contributors 24 * may be used to endorse or promote products derived from this software 25 * without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 37 * SUCH DAMAGE. 38 */ 39 40 #include <sys/cdefs.h> 41 #ifndef lint 42 __COPYRIGHT( 43 "@(#) Copyright (c) 1991 The Regents of the University of California.\n\ 44 All rights reserved.\n"); 45 #endif /* not lint */ 46 47 #ifndef lint 48 #if 0 49 static char sccsid[] = "from: @(#)join.c 5.1 (Berkeley) 11/18/91"; 50 #else 51 __RCSID("$NetBSD: join.c,v 1.20 2002/02/14 03:21:07 jonb Exp $"); 52 #endif 53 #endif /* not lint */ 54 55 #include <sys/types.h> 56 #include <ctype.h> 57 #include <err.h> 58 #include <errno.h> 59 #include <stdio.h> 60 #include <stdlib.h> 61 #include <unistd.h> 62 #include <string.h> 63 #include <unistd.h> 64 65 /* 66 * There's a structure per input file which encapsulates the state of the 67 * file. We repeatedly read lines from each file until we've read in all 68 * the consecutive lines from the file with a common join field. Then we 69 * compare the set of lines with an equivalent set from the other file. 70 */ 71 typedef struct { 72 char *line; /* line */ 73 u_long linealloc; /* line allocated count */ 74 char **fields; /* line field(s) */ 75 u_long fieldcnt; /* line field(s) count */ 76 u_long fieldalloc; /* line field(s) allocated count */ 77 } LINE; 78 79 typedef struct { 80 FILE *fp; /* file descriptor */ 81 u_long joinf; /* join field (-1, -2, -j) */ 82 int unpair; /* output unpairable lines (-a) */ 83 int number; /* 1 for file 1, 2 for file 2 */ 84 85 LINE *set; /* set of lines with same field */ 86 u_long pushback; /* line on the stack */ 87 u_long setcnt; /* set count */ 88 u_long setalloc; /* set allocated count */ 89 } INPUT; 90 INPUT input1 = { NULL, 0, 0, 1, NULL, -1, 0, 0, }, 91 input2 = { NULL, 0, 0, 2, NULL, -1, 0, 0, }; 92 93 typedef struct { 94 u_long fileno; /* file number */ 95 u_long fieldno; /* field number */ 96 } OLIST; 97 OLIST *olist; /* output field list */ 98 u_long olistcnt; /* output field list count */ 99 u_long olistalloc; /* output field allocated count */ 100 101 int joinout = 1; /* show lines with matched join fields (-v) */ 102 int needsep; /* need separator character */ 103 int spans = 1; /* span multiple delimiters (-t) */ 104 char *empty; /* empty field replacement string (-e) */ 105 char *tabchar = " \t"; /* delimiter characters (-t) */ 106 107 int cmp __P((LINE *, u_long, LINE *, u_long)); 108 void enomem __P((void)); 109 void fieldarg __P((char *)); 110 void joinlines __P((INPUT *, INPUT *)); 111 int main __P((int, char **)); 112 void obsolete __P((char **)); 113 void outfield __P((LINE *, u_long)); 114 void outoneline __P((INPUT *, LINE *)); 115 void outtwoline __P((INPUT *, LINE *, INPUT *, LINE *)); 116 void slurp __P((INPUT *)); 117 void usage __P((void)); 118 119 int 120 main(argc, argv) 121 int argc; 122 char *argv[]; 123 { 124 INPUT *F1, *F2; 125 int aflag, ch, cval, vflag; 126 char *end; 127 128 F1 = &input1; 129 F2 = &input2; 130 131 aflag = vflag = 0; 132 obsolete(argv); 133 while ((ch = getopt(argc, argv, "\01a:e:j:1:2:o:t:v:")) != -1) { 134 switch (ch) { 135 case '\01': 136 aflag = 1; 137 F1->unpair = F2->unpair = 1; 138 break; 139 case '1': 140 if ((F1->joinf = strtol(optarg, &end, 10)) < 1) { 141 warnx("-1 option field number less than 1"); 142 usage(); 143 } 144 if (*end) { 145 warnx("illegal field number -- %s", optarg); 146 usage(); 147 } 148 --F1->joinf; 149 break; 150 case '2': 151 if ((F2->joinf = strtol(optarg, &end, 10)) < 1) { 152 warnx("-2 option field number less than 1"); 153 usage(); 154 } 155 if (*end) { 156 warnx("illegal field number -- %s", optarg); 157 usage(); 158 } 159 --F2->joinf; 160 break; 161 case 'a': 162 aflag = 1; 163 switch(strtol(optarg, &end, 10)) { 164 case 1: 165 F1->unpair = 1; 166 break; 167 case 2: 168 F2->unpair = 1; 169 break; 170 default: 171 warnx("-a option file number not 1 or 2"); 172 usage(); 173 break; 174 } 175 if (*end) { 176 warnx("illegal file number -- %s", optarg); 177 usage(); 178 } 179 break; 180 case 'e': 181 empty = optarg; 182 break; 183 case 'j': 184 if ((F1->joinf = F2->joinf = 185 strtol(optarg, &end, 10)) < 1) { 186 warnx("-j option field number less than 1"); 187 usage(); 188 } 189 if (*end) { 190 warnx("illegal field number -- %s", optarg); 191 usage(); 192 } 193 --F1->joinf; 194 --F2->joinf; 195 break; 196 case 'o': 197 fieldarg(optarg); 198 break; 199 case 't': 200 spans = 0; 201 if (strlen(tabchar = optarg) != 1) { 202 warnx("illegal tab character specification"); 203 usage(); 204 } 205 break; 206 case 'v': 207 vflag = 1; 208 joinout = 0; 209 switch(strtol(optarg, &end, 10)) { 210 case 1: 211 F1->unpair = 1; 212 break; 213 case 2: 214 F2->unpair = 1; 215 break; 216 default: 217 warnx("-v option file number not 1 or 2"); 218 usage(); 219 break; 220 } 221 if (*end) { 222 warnx("illegal file number -- %s", optarg); 223 usage(); 224 } 225 break; 226 case '?': 227 default: 228 usage(); 229 } 230 } 231 argc -= optind; 232 argv += optind; 233 234 if (aflag && vflag) 235 errx(1, "-a and -v options mutually exclusive"); 236 237 if (argc != 2) 238 usage(); 239 240 /* Open the files; "-" means stdin. */ 241 if (!strcmp(*argv, "-")) 242 F1->fp = stdin; 243 else if ((F1->fp = fopen(*argv, "r")) == NULL) 244 err(1, "%s", *argv); 245 ++argv; 246 if (!strcmp(*argv, "-")) 247 F2->fp = stdin; 248 else if ((F2->fp = fopen(*argv, "r")) == NULL) 249 err(1, "%s", *argv); 250 if (F1->fp == stdin && F2->fp == stdin) 251 errx(1, "only one input file may be stdin"); 252 253 slurp(F1); 254 slurp(F2); 255 while (F1->setcnt && F2->setcnt) { 256 cval = cmp(F1->set, F1->joinf, F2->set, F2->joinf); 257 if (cval == 0) { 258 /* Oh joy, oh rapture, oh beauty divine! */ 259 if (joinout) 260 joinlines(F1, F2); 261 slurp(F1); 262 slurp(F2); 263 } else if (cval < 0) { 264 /* File 1 takes the lead... */ 265 if (F1->unpair) 266 joinlines(F1, NULL); 267 slurp(F1); 268 } else { 269 /* File 2 takes the lead... */ 270 if (F2->unpair) 271 joinlines(F2, NULL); 272 slurp(F2); 273 } 274 } 275 276 /* 277 * Now that one of the files is used up, optionally output any 278 * remaining lines from the other file. 279 */ 280 if (F1->unpair) 281 while (F1->setcnt) { 282 joinlines(F1, NULL); 283 slurp(F1); 284 } 285 if (F2->unpair) 286 while (F2->setcnt) { 287 joinlines(F2, NULL); 288 slurp(F2); 289 } 290 exit(0); 291 } 292 293 void 294 slurp(F) 295 INPUT *F; 296 { 297 LINE *lp; 298 LINE tmp; 299 size_t len; 300 int cnt; 301 char *bp, *fieldp; 302 303 /* 304 * Read all of the lines from an input file that have the same 305 * join field. 306 */ 307 for (F->setcnt = 0;; ++F->setcnt) { 308 /* 309 * If we're out of space to hold line structures, allocate 310 * more. Initialize the structure so that we know that this 311 * is new space. 312 */ 313 if (F->setcnt == F->setalloc) { 314 cnt = F->setalloc; 315 if (F->setalloc == 0) 316 F->setalloc = 64; 317 else 318 F->setalloc <<= 1; 319 if ((F->set = realloc(F->set, 320 F->setalloc * sizeof(LINE))) == NULL) 321 enomem(); 322 memset(F->set + cnt, 0, 323 (F->setalloc - cnt) * sizeof(LINE)); 324 } 325 326 /* 327 * Get any pushed back line, else get the next line. Allocate 328 * space as necessary. If taking the line from the stack swap 329 * the two structures so that we don't lose the allocated space. 330 * This could be avoided by doing another level of indirection, 331 * but it's probably okay as is. 332 */ 333 lp = &F->set[F->setcnt]; 334 if (F->pushback != -1) { 335 tmp = F->set[F->setcnt]; 336 F->set[F->setcnt] = F->set[F->pushback]; 337 F->set[F->pushback] = tmp; 338 F->pushback = -1; 339 continue; 340 } 341 if ((bp = fgetln(F->fp, &len)) == NULL) 342 return; 343 if (lp->linealloc <= len + 1) { 344 if (lp->linealloc == 0) 345 lp->linealloc = 128; 346 while (lp->linealloc <= len + 1) 347 lp->linealloc <<= 1; 348 if ((lp->line = realloc(lp->line, 349 lp->linealloc * sizeof(char))) == NULL) 350 enomem(); 351 } 352 memmove(lp->line, bp, len); 353 354 /* Replace trailing newline, if it exists. */ 355 if (bp[len - 1] == '\n') 356 lp->line[len - 1] = '\0'; 357 else 358 lp->line[len] = '\0'; 359 bp = lp->line; 360 361 /* Split the line into fields, allocate space as necessary. */ 362 lp->fieldcnt = 0; 363 while ((fieldp = strsep(&bp, tabchar)) != NULL) { 364 if (spans && *fieldp == '\0') 365 continue; 366 if (lp->fieldcnt == lp->fieldalloc) { 367 if (lp->fieldalloc == 0) 368 lp->fieldalloc = 16; 369 else 370 lp->fieldalloc <<= 1; 371 if ((lp->fields = realloc(lp->fields, 372 lp->fieldalloc * sizeof(char *))) == NULL) 373 enomem(); 374 } 375 lp->fields[lp->fieldcnt++] = fieldp; 376 } 377 378 /* See if the join field value has changed. */ 379 if (F->setcnt && cmp(lp, F->joinf, lp - 1, F->joinf)) { 380 F->pushback = F->setcnt; 381 break; 382 } 383 } 384 } 385 386 int 387 cmp(lp1, fieldno1, lp2, fieldno2) 388 LINE *lp1, *lp2; 389 u_long fieldno1, fieldno2; 390 { 391 392 if (lp1->fieldcnt <= fieldno1) 393 return (lp2->fieldcnt <= fieldno2 ? 0 : 1); 394 if (lp2->fieldcnt <= fieldno2) 395 return (-1); 396 return (strcmp(lp1->fields[fieldno1], lp2->fields[fieldno2])); 397 } 398 399 void 400 joinlines(F1, F2) 401 INPUT *F1, *F2; 402 { 403 int cnt1, cnt2; 404 405 /* 406 * Output the results of a join comparison. The output may be from 407 * either file 1 or file 2 (in which case the first argument is the 408 * file from which to output) or from both. 409 */ 410 if (F2 == NULL) { 411 for (cnt1 = 0; cnt1 < F1->setcnt; ++cnt1) 412 outoneline(F1, &F1->set[cnt1]); 413 return; 414 } 415 for (cnt1 = 0; cnt1 < F1->setcnt; ++cnt1) 416 for (cnt2 = 0; cnt2 < F2->setcnt; ++cnt2) 417 outtwoline(F1, &F1->set[cnt1], F2, &F2->set[cnt2]); 418 } 419 420 void 421 outoneline(F, lp) 422 INPUT *F; 423 LINE *lp; 424 { 425 int cnt; 426 427 /* 428 * Output a single line from one of the files, according to the 429 * join rules. This happens when we are writing unmatched single 430 * lines. Output empty fields in the right places. 431 */ 432 if (olist) 433 for (cnt = 0; cnt < olistcnt; ++cnt) { 434 if (olist[cnt].fileno == F->number) 435 outfield(lp, olist[cnt].fieldno); 436 } 437 else 438 for (cnt = 0; cnt < lp->fieldcnt; ++cnt) 439 outfield(lp, cnt); 440 (void)printf("\n"); 441 if (ferror(stdout)) 442 err(1, "stdout"); 443 needsep = 0; 444 } 445 446 void 447 outtwoline(F1, lp1, F2, lp2) 448 INPUT *F1, *F2; 449 LINE *lp1, *lp2; 450 { 451 int cnt; 452 453 /* Output a pair of lines according to the join list (if any). */ 454 if (olist) { 455 for (cnt = 0; cnt < olistcnt; ++cnt) 456 if (olist[cnt].fileno == 1) 457 outfield(lp1, olist[cnt].fieldno); 458 else /* if (olist[cnt].fileno == 2) */ 459 outfield(lp2, olist[cnt].fieldno); 460 } else { 461 /* 462 * Output the join field, then the remaining fields from F1 463 * and F2. 464 */ 465 outfield(lp1, F1->joinf); 466 for (cnt = 0; cnt < lp1->fieldcnt; ++cnt) 467 if (F1->joinf != cnt) 468 outfield(lp1, cnt); 469 for (cnt = 0; cnt < lp2->fieldcnt; ++cnt) 470 if (F2->joinf != cnt) 471 outfield(lp2, cnt); 472 } 473 (void)printf("\n"); 474 if (ferror(stdout)) 475 err(1, "stdout"); 476 needsep = 0; 477 } 478 479 void 480 outfield(lp, fieldno) 481 LINE *lp; 482 u_long fieldno; 483 { 484 if (needsep++) 485 (void)printf("%c", *tabchar); 486 if (!ferror(stdout)) { 487 if (lp->fieldcnt <= fieldno) { 488 if (empty != NULL) 489 (void)printf("%s", empty); 490 } else { 491 if (*lp->fields[fieldno] == '\0') 492 return; 493 (void)printf("%s", lp->fields[fieldno]); 494 } 495 } 496 if (ferror(stdout)) 497 err(1, "stdout"); 498 } 499 500 /* 501 * Convert an output list argument "2.1, 1.3, 2.4" into an array of output 502 * fields. 503 */ 504 void 505 fieldarg(option) 506 char *option; 507 { 508 u_long fieldno; 509 char *end, *token; 510 511 while ((token = strsep(&option, ", \t")) != NULL) { 512 if (*token == '\0') 513 continue; 514 if ((token[0] != '1' && token[0] != '2') || token[1] != '.') 515 errx(1, "malformed -o option field"); 516 fieldno = strtol(token + 2, &end, 10); 517 if (*end) 518 errx(1, "malformed -o option field"); 519 if (fieldno == 0) 520 errx(1, "field numbers are 1 based"); 521 if (olistcnt == olistalloc) { 522 olistalloc += 50; 523 if ((olist = realloc(olist, 524 olistalloc * sizeof(OLIST))) == NULL) 525 enomem(); 526 } 527 olist[olistcnt].fileno = token[0] - '0'; 528 olist[olistcnt].fieldno = fieldno - 1; 529 ++olistcnt; 530 } 531 } 532 533 void 534 obsolete(argv) 535 char **argv; 536 { 537 int len; 538 char **p, *ap, *t; 539 540 while ((ap = *++argv) != NULL) { 541 /* Return if "--". */ 542 if (ap[0] == '-' && ap[1] == '-') 543 return; 544 switch (ap[1]) { 545 case 'a': 546 /* 547 * The original join allowed "-a", which meant the 548 * same as -a1 plus -a2. POSIX 1003.2, Draft 11.2 549 * only specifies this as "-a 1" and "a -2", so we 550 * have to use another option flag, one that is 551 * unlikely to ever be used or accidentally entered 552 * on the command line. (Well, we could reallocate 553 * the argv array, but that hardly seems worthwhile.) 554 */ 555 if (ap[2] == '\0') 556 ap[1] = '\01'; 557 break; 558 case 'j': 559 /* 560 * The original join allowed "-j[12] arg" and "-j arg". 561 * Convert the former to "-[12] arg". Don't convert 562 * the latter since getopt(3) can handle it. 563 */ 564 switch(ap[2]) { 565 case '1': 566 if (ap[3] != '\0') 567 goto jbad; 568 ap[1] = '1'; 569 ap[2] = '\0'; 570 break; 571 case '2': 572 if (ap[3] != '\0') 573 goto jbad; 574 ap[1] = '2'; 575 ap[2] = '\0'; 576 break; 577 case '\0': 578 break; 579 default: 580 jbad: errx(1, "illegal option -- %s", ap); 581 usage(); 582 } 583 break; 584 case 'o': 585 /* 586 * The original join allowed "-o arg arg". Convert to 587 * "-o arg -o arg". 588 */ 589 if (ap[2] != '\0') 590 break; 591 for (p = argv + 2; *p; ++p) { 592 if ((p[0][0] != '1' && p[0][0] != '2') || 593 p[0][1] != '.') 594 break; 595 len = strlen(*p); 596 if (len - 2 != strspn(*p + 2, "0123456789")) 597 break; 598 if ((t = malloc(len + 3)) == NULL) 599 enomem(); 600 t[0] = '-'; 601 t[1] = 'o'; 602 memmove(t + 2, *p, len + 1); 603 *p = t; 604 } 605 argv = p - 1; 606 break; 607 } 608 } 609 } 610 611 void 612 enomem() 613 { 614 errx(1, "no memory"); 615 } 616 617 void 618 usage() 619 { 620 (void)fprintf(stderr, "%s%s\n", 621 "usage: join [-a fileno | -v fileno ] [-e string] [-1 field] ", 622 "[-2 field]\n [-o list] [-t char] file1 file2"); 623 exit(1); 624 } 625