sys/src/cmd/sed.c

   1 /*
   2  * sed -- stream editor
   3  */
   4 #include <u.h>
   5 #include <libc.h>
   6 #include <bio.h>
   7 #include <regexp.h>
   8
   9 enum {
  10         DEPTH           = 20,           /* max nesting depth of {} */
  11         MAXCMDS         = 512,          /* max sed commands */
  12         ADDSIZE         = 10000,        /* size of add & read buffer */
  13         MAXADDS         = 20,           /* max pending adds and reads */
  14         LBSIZE          = 8192,         /* input line size */
  15         LABSIZE         = 50,           /* max number of labels */
  16         MAXSUB          = 10,           /* max number of sub reg exp */
  17         MAXFILES        = 120,          /* max output files */
  18 };
  19
  20 /*
  21  * An address is a line #, a R.E., "$", a reference to the last
  22  * R.E., or nothing.
  23  */
  24 typedef struct {
  25         enum {
  26                 A_NONE,
  27                 A_DOL,
  28                 A_LINE,
  29                 A_RE,
  30                 A_LAST,
  31         }type;
  32         union {
  33                 long    line;           /* Line # */
  34                 Reprog  *rp;            /* Compiled R.E. */
  35         };
  36 } Addr;
  37
  38 typedef struct  SEDCOM {
  39         Addr    ad1;                    /* optional start address */
  40         Addr    ad2;                    /* optional end address */
  41         union {
  42                 Reprog  *re1;           /* compiled R.E. */
  43                 Rune    *text;          /* added text or file name */
  44                 struct  SEDCOM  *lb1;   /* destination command of branch */
  45         };
  46         Rune    *rhs;                   /* Right-hand side of substitution */
  47         Biobuf* fcode;                  /* File ID for read and write */
  48         char    command;                /* command code -see below */
  49         char    gfl;                    /* 'Global' flag for substitutions */
  50         char    pfl;                    /* 'print' flag for substitutions */
  51         char    active;                 /* 1 => data between start and end */
  52         char    negfl;                  /* negation flag */
  53 } SedCom;
  54
  55 /* Command Codes for field SedCom.command */
  56 #define ACOM    01
  57 #define BCOM    020
  58 #define CCOM    02
  59 #define CDCOM   025
  60 #define CNCOM   022
  61 #define COCOM   017
  62 #define CPCOM   023
  63 #define DCOM    03
  64 #define ECOM    015
  65 #define EQCOM   013
  66 #define FCOM    016
  67 #define GCOM    027
  68 #define CGCOM   030
  69 #define HCOM    031
  70 #define CHCOM   032
  71 #define ICOM    04
  72 #define LCOM    05
  73 #define NCOM    012
  74 #define PCOM    010
  75 #define QCOM    011
  76 #define RCOM    06
  77 #define SCOM    07
  78 #define TCOM    021
  79 #define WCOM    014
  80 #define CWCOM   024
  81 #define YCOM    026
  82 #define XCOM    033
  83
  84 typedef struct label {                  /* Label symbol table */
  85         Rune    uninm[9];               /* Label name */
  86         SedCom  *chain;
  87         SedCom  *address;               /* Command associated with label */
  88 } Label;
  89
  90 typedef struct  FILE_CACHE {            /* Data file control block */
  91         struct FILE_CACHE *next;        /* Forward Link */
  92         char    *name;                  /* Name of file */
  93 } FileCache;
  94
  95 SedCom pspace[MAXCMDS];                 /* Command storage */
  96 SedCom *pend = pspace+MAXCMDS;          /* End of command storage */
  97 SedCom *rep = pspace;                   /* Current fill point */
  98
  99 Reprog  *lastre = 0;                    /* Last regular expression */
 100 Resub   subexp[MAXSUB];                 /* sub-patterns of pattern match*/
 101
 102 Rune    addspace[ADDSIZE];              /* Buffer for a, c, & i commands */
 103 Rune    *addend = addspace+ADDSIZE;
 104
 105 SedCom  *abuf[MAXADDS];                 /* Queue of pending adds & reads */
 106 SedCom  **aptr = abuf;
 107
 108 struct {                                /* Sed program input control block */
 109         enum PTYPE {                    /* Either on command line or in file */
 110                 P_ARG,
 111                 P_FILE,
 112         } type;
 113         union PCTL {                    /* Pointer to data */
 114                 Biobuf  *bp;
 115                 char    *curr;
 116         };
 117 } prog;
 118
 119 Rune    genbuf[LBSIZE];                 /* Miscellaneous buffer */
 120
 121 FileCache       *fhead = 0;             /* Head of File Cache Chain */
 122 FileCache       *ftail = 0;             /* Tail of File Cache Chain */
 123
 124 Rune    *loc1;                          /* Start of pattern match */
 125 Rune    *loc2;                          /* End of pattern match */
 126 Rune    seof;                           /* Pattern delimiter char */
 127
 128 Rune    linebuf[LBSIZE+1];              /* Input data buffer */
 129 Rune    *lbend = linebuf+LBSIZE;        /* End of buffer */
 130 Rune    *spend = linebuf;               /* End of input data */
 131 Rune    *cp;                            /* Current scan point in linebuf */
 132
 133 Rune    holdsp[LBSIZE+1];               /* Hold buffer */
 134 Rune    *hend = holdsp+LBSIZE;          /* End of hold buffer */
 135 Rune    *hspend = holdsp;               /* End of hold data */
 136
 137 int     nflag;                          /* Command line flags */
 138 int     gflag;
 139
 140 int     dolflag;                        /* Set when at true EOF */
 141 int     sflag;                          /* Set when substitution done */
 142 int     jflag;                          /* Set when jump required */
 143 int     delflag;                        /* Delete current line when set */
 144
 145 long    lnum = 0;                       /* Input line count */
 146
 147 char    fname[MAXFILES][40];            /* File name cache */
 148 Biobuf  *fcode[MAXFILES];               /* File ID cache */
 149 int     nfiles = 0;                     /* Cache fill point */
 150
 151 Biobuf  fout;                           /* Output stream */
 152 Biobuf  stdin;                          /* Default input */
 153 Biobuf* f = 0;                          /* Input data */
 154
 155 Label   ltab[LABSIZE];                  /* Label name symbol table */
 156 Label   *labend = ltab+LABSIZE;         /* End of label table */
 157 Label   *lab = ltab+1;                  /* Current Fill point */
 158
 159 int     depth = 0;                      /* {} stack pointer */
 160
 161 Rune    bad;                            /* Dummy err ptr reference */
 162 Rune    *badp = &bad;
 163
 164
 165 char    CGMES[]  =      "%S command garbled: %S";
 166 char    TMMES[]  =      "Too much text: %S";
 167 char    LTL[]    =      "Label too long: %S";
 168 char    AD0MES[] =      "No addresses allowed: %S";
 169 char    AD1MES[] =      "Only one address allowed: %S";
 170
 171 void    address(Addr *);
 172 void    arout(void);
 173 int     cmp(char *, char *);
 174 int     rcmp(Rune *, Rune *);
 175 void    command(SedCom *);
 176 Reprog  *compile(void);
 177 Rune    *compsub(Rune *, Rune *);
 178 void    dechain(void);
 179 void    dosub(Rune *);
 180 int     ecmp(Rune *, Rune *, int);
 181 void    enroll(char *);
 182 void    errexit(void);
 183 int     executable(SedCom *);
 184 void    execute(void);
 185 void    fcomp(void);
 186 long    getrune(void);
 187 Rune    *gline(Rune *);
 188 int     match(Reprog *, Rune *);
 189 void    newfile(enum PTYPE, char *);
 190 int     opendata(void);
 191 Biobuf  *open_file(char *);
 192 Rune    *place(Rune *, Rune *, Rune *);
 193 void    quit(char *, ...);
 194 int     rline(Rune *, Rune *);
 195 Label   *search(Label *);
 196 int     substitute(SedCom *);
 197 char    *text(char *);
 198 Rune    *stext(Rune *, Rune *);
 199 int     ycomp(SedCom *);
 200 char *  trans(int c);
 201 void    putline(Biobuf *bp, Rune *buf, int n);
 202
 203 void
 204 main(int argc, char **argv)
 205 {
 206         int compfl;
 207
 208         lnum = 0;
 209         Binit(&fout, 1, OWRITE);
 210         Blethal(&fout, nil);
 211         fcode[nfiles++] = &fout;
 212         compfl = 0;
 213
 214         if(argc == 1)
 215                 exits(0);
 216         ARGBEGIN{
 217         case 'e':
 218                 if (argc <= 1)
 219                         quit("missing pattern");
 220                 newfile(P_ARG, ARGF());
 221                 fcomp();
 222                 compfl = 1;
 223                 continue;
 224         case 'f':
 225                 if(argc <= 1)
 226                         quit("no pattern-file");
 227                 newfile(P_FILE, ARGF());
 228                 fcomp();
 229                 compfl = 1;
 230                 continue;
 231         case 'g':
 232                 gflag++;
 233                 continue;
 234         case 'n':
 235                 nflag++;
 236                 continue;
 237         default:
 238                 fprint(2, "sed: Unknown flag: %c\n", ARGC());
 239                 continue;
 240         } ARGEND
 241
 242         if(compfl == 0) {
 243                 if (--argc < 0)
 244                         quit("missing pattern");
 245                 newfile(P_ARG, *argv++);
 246                 fcomp();
 247         }
 248
 249         if(depth)
 250                 quit("Too many {'s");
 251
 252         ltab[0].address = rep;
 253
 254         dechain();
 255
 256         if(argc <= 0)
 257                 enroll(0);              /* Add stdin to cache */
 258         else
 259                 while(--argc >= 0)
 260                         enroll(*argv++);
 261         execute();
 262         exits(0);
 263 }
 264
 265 void
 266 fcomp(void)
 267 {
 268         int     i;
 269         Label   *lpt;
 270         Rune    *tp;
 271         SedCom  *pt, *pt1;
 272         static Rune     *p = addspace;
 273         static SedCom   **cmpend[DEPTH];        /* stack of {} operations */
 274
 275         while (rline(linebuf, lbend) >= 0) {
 276                 cp = linebuf;
 277 comploop:
 278                 while(*cp == L' ' || *cp == L'\t')
 279                         cp++;
 280                 if(*cp == L'\0' || *cp == L'#')
 281                         continue;
 282                 if(*cp == L';') {
 283                         cp++;
 284                         goto comploop;
 285                 }
 286
 287                 address(&rep->ad1);
 288                 if (rep->ad1.type != A_NONE) {
 289                         if (rep->ad1.type == A_LAST) {
 290                                 if (!lastre)
 291                                         quit("First RE may not be null");
 292                                 rep->ad1.type = A_RE;
 293                                 rep->ad1.rp = lastre;
 294                         }
 295                         if(*cp == L',' || *cp == L';') {
 296                                 cp++;
 297                                 address(&rep->ad2);
 298                                 if (rep->ad2.type == A_LAST) {
 299                                         rep->ad2.type = A_RE;
 300                                         rep->ad2.rp = lastre;
 301                                 }
 302                         } else
 303                                 rep->ad2.type = A_NONE;
 304                 }
 305                 while(*cp == L' ' || *cp == L'\t')
 306                         cp++;
 307
 308 swit:
 309                 switch(*cp++) {
 310                 default:
 311                         quit("Unrecognized command: %S", linebuf);
 312
 313                 case '!':
 314                         rep->negfl = 1;
 315                         goto swit;
 316
 317                 case '{':
 318                         rep->command = BCOM;
 319                         rep->negfl = !rep->negfl;
 320                         cmpend[depth++] = &rep->lb1;
 321                         if(++rep >= pend)
 322                                 quit("Too many commands: %S", linebuf);
 323                         if(*cp == '\0')
 324                                 continue;
 325                         goto comploop;
 326
 327                 case '}':
 328                         if(rep->ad1.type != A_NONE)
 329                                 quit(AD0MES, linebuf);
 330                         if(--depth < 0)
 331                                 quit("Too many }'s");
 332                         *cmpend[depth] = rep;
 333                         if(*cp == 0)
 334                                 continue;
 335                         goto comploop;
 336
 337                 case '=':
 338                         rep->command = EQCOM;
 339                         if(rep->ad2.type != A_NONE)
 340                                 quit(AD1MES, linebuf);
 341                         break;
 342
 343                 case ':':
 344                         if(rep->ad1.type != A_NONE)
 345                                 quit(AD0MES, linebuf);
 346
 347                         while(*cp == L' ')
 348                                 cp++;
 349                         tp = lab->uninm;
 350                         while (*cp && *cp != L';' && *cp != L' ' &&
 351                             *cp != L'\t' && *cp != L'#') {
 352                                 *tp++ = *cp++;
 353                                 if(tp >= &lab->uninm[8])
 354                                         quit(LTL, linebuf);
 355                         }
 356                         *tp = L'\0';
 357
 358                         if (*lab->uninm == L'\0')               /* no label? */
 359                                 quit(CGMES, L":", linebuf);
 360                         if(lpt = search(lab)) {
 361                                 if(lpt->address)
 362                                         quit("Duplicate labels: %S", linebuf);
 363                         } else {
 364                                 lab->chain = 0;
 365                                 lpt = lab;
 366                                 if(++lab >= labend)
 367                                         quit("Too many labels: %S", linebuf);
 368                         }
 369                         lpt->address = rep;
 370                         if (*cp == L'#')
 371                                 continue;
 372                         rep--;                  /* reuse this slot */
 373                         break;
 374
 375                 case 'a':
 376                         rep->command = ACOM;
 377                         if(rep->ad2.type != A_NONE)
 378                                 quit(AD1MES, linebuf);
 379                         if(*cp == L'\\')
 380                                 cp++;
 381                         if(*cp++ != L'\n')
 382                                 quit(CGMES, L"a", linebuf);
 383                         rep->text = p;
 384                         p = stext(p, addend);
 385                         break;
 386                 case 'c':
 387                         rep->command = CCOM;
 388                         if(*cp == L'\\')
 389                                 cp++;
 390                         if(*cp++ != L'\n')
 391                                 quit(CGMES, L"c", linebuf);
 392                         rep->text = p;
 393                         p = stext(p, addend);
 394                         break;
 395                 case 'i':
 396                         rep->command = ICOM;
 397                         if(rep->ad2.type != A_NONE)
 398                                 quit(AD1MES, linebuf);
 399                         if(*cp == L'\\')
 400                                 cp++;
 401                         if(*cp++ != L'\n')
 402                                 quit(CGMES, L"i", linebuf);
 403                         rep->text = p;
 404                         p = stext(p, addend);
 405                         break;
 406
 407                 case 'g':
 408                         rep->command = GCOM;
 409                         break;
 410
 411                 case 'G':
 412                         rep->command = CGCOM;
 413                         break;
 414
 415                 case 'h':
 416                         rep->command = HCOM;
 417                         break;
 418
 419                 case 'H':
 420                         rep->command = CHCOM;
 421                         break;
 422
 423                 case 't':
 424                         rep->command = TCOM;
 425                         goto jtcommon;
 426
 427                 case 'b':
 428                         rep->command = BCOM;
 429 jtcommon:
 430                         while(*cp == L' ')
 431                                 cp++;
 432                         if(*cp == L'\0' || *cp == L';') {
 433                                 /* no label; jump to end */
 434                                 if(pt = ltab[0].chain) {
 435                                         while((pt1 = pt->lb1) != nil)
 436                                                 pt = pt1;
 437                                         pt->lb1 = rep;
 438                                 } else
 439                                         ltab[0].chain = rep;
 440                                 break;
 441                         }
 442
 443                         /* copy label into lab->uninm */
 444                         tp = lab->uninm;
 445                         while((*tp = *cp++) != L'\0' && *tp != L';')
 446                                 if(++tp >= &lab->uninm[8])
 447                                         quit(LTL, linebuf);
 448                         cp--;
 449                         *tp = L'\0';
 450
 451                         if (*lab->uninm == L'\0')
 452                                 /* shouldn't get here */
 453                                 quit(CGMES, L"b or t", linebuf);
 454                         if((lpt = search(lab)) != nil) {
 455                                 if(lpt->address)
 456                                         rep->lb1 = lpt->address;
 457                                 else {
 458                                         for(pt = lpt->chain; pt != nil &&
 459                                             (pt1 = pt->lb1) != nil; pt = pt1)
 460                                                 ;
 461                                         if (pt)
 462                                                 pt->lb1 = rep;
 463                                 }
 464                         } else {                        /* add new label */
 465                                 lab->chain = rep;
 466                                 lab->address = 0;
 467                                 if(++lab >= labend)
 468                                         quit("Too many labels: %S", linebuf);
 469                         }
 470                         break;
 471
 472                 case 'n':
 473                         rep->command = NCOM;
 474                         break;
 475
 476                 case 'N':
 477                         rep->command = CNCOM;
 478                         break;
 479
 480                 case 'p':
 481                         rep->command = PCOM;
 482                         break;
 483
 484                 case 'P':
 485                         rep->command = CPCOM;
 486                         break;
 487
 488                 case 'r':
 489                         rep->command = RCOM;
 490                         if(rep->ad2.type != A_NONE)
 491                                 quit(AD1MES, linebuf);
 492                         if(*cp++ != L' ')
 493                                 quit(CGMES, L"r", linebuf);
 494                         rep->text = p;
 495                         p = stext(p, addend);
 496                         break;
 497
 498                 case 'd':
 499                         rep->command = DCOM;
 500                         break;
 501
 502                 case 'D':
 503                         rep->command = CDCOM;
 504                         rep->lb1 = pspace;
 505                         break;
 506
 507                 case 'q':
 508                         rep->command = QCOM;
 509                         if(rep->ad2.type != A_NONE)
 510                                 quit(AD1MES, linebuf);
 511                         break;
 512
 513                 case 'l':
 514                         rep->command = LCOM;
 515                         break;
 516
 517                 case 's':
 518                         rep->command = SCOM;
 519                         seof = *cp++;
 520                         if ((rep->re1 = compile()) == 0) {
 521                                 if(!lastre)
 522                                         quit("First RE may not be null.");
 523                                 rep->re1 = lastre;
 524                         }
 525                         rep->rhs = p;
 526                         if((p = compsub(p, addend)) == 0)
 527                                 quit(CGMES, L"s", linebuf);
 528                         if(*cp == L'g') {
 529                                 cp++;
 530                                 rep->gfl++;
 531                         } else if(gflag)
 532                                 rep->gfl++;
 533
 534                         if(*cp == L'p') {
 535                                 cp++;
 536                                 rep->pfl = 1;
 537                         }
 538
 539                         if(*cp == L'P') {
 540                                 cp++;
 541                                 rep->pfl = 2;
 542                         }
 543
 544                         if(*cp == L'w') {
 545                                 cp++;
 546                                 if(*cp++ !=  L' ')
 547                                         quit(CGMES, L"s", linebuf);
 548                                 text(fname[nfiles]);
 549                                 for(i = nfiles - 1; i >= 0; i--)
 550                                         if(cmp(fname[nfiles], fname[i]) == 0) {
 551                                                 rep->fcode = fcode[i];
 552                                                 goto done;
 553                                         }
 554                                 if(nfiles >= MAXFILES)
 555                                         quit("Too many files in w commands 1");
 556                                 rep->fcode = open_file(fname[nfiles]);
 557                         }
 558                         break;
 559
 560                 case 'w':
 561                         rep->command = WCOM;
 562                         if(*cp++ != L' ')
 563                                 quit(CGMES, L"w", linebuf);
 564                         text(fname[nfiles]);
 565                         for(i = nfiles - 1; i >= 0; i--)
 566                                 if(cmp(fname[nfiles], fname[i]) == 0) {
 567                                         rep->fcode = fcode[i];
 568                                         goto done;
 569                                 }
 570                         if(nfiles >= MAXFILES){
 571                                 fprint(2, "sed: Too many files in w commands 2 \n");
 572                                 fprint(2, "nfiles = %d; MAXF = %d\n",
 573                                         nfiles, MAXFILES);
 574                                 errexit();
 575                         }
 576                         rep->fcode = open_file(fname[nfiles]);
 577                         break;
 578
 579                 case 'x':
 580                         rep->command = XCOM;
 581                         break;
 582
 583                 case 'y':
 584                         rep->command = YCOM;
 585                         seof = *cp++;
 586                         if (ycomp(rep) == 0)
 587                                 quit(CGMES, L"y", linebuf);
 588                         break;
 589
 590                 }
 591 done:
 592                 if(++rep >= pend)
 593                         quit("Too many commands, last: %S", linebuf);
 594                 if(*cp++ != L'\0') {
 595                         if(cp[-1] == L';')
 596                                 goto comploop;
 597                         quit(CGMES, cp - 1, linebuf);
 598                 }
 599         }
 600 }
 601
 602 Biobuf *
 603 open_file(char *name)
 604 {
 605         int fd;
 606         Biobuf *bp;
 607
 608         if ((bp = malloc(sizeof(Biobuf))) == 0)
 609                 quit("Out of memory");
 610         if ((fd = open(name, OWRITE)) < 0 &&
 611             (fd = create(name, OWRITE, 0666)) < 0)
 612                 quit("Cannot create %s", name);
 613         Binit(bp, fd, OWRITE);
 614         Blethal(bp, nil);
 615         Bseek(bp, 0, 2);
 616         fcode[nfiles++] = bp;
 617         return bp;
 618 }
 619
 620 Rune *
 621 compsub(Rune *rhs, Rune *end)
 622 {
 623         Rune r;
 624
 625         while ((r = *cp++) != '\0') {
 626                 if(r == '\\') {
 627                         if (rhs < end)
 628                                 *rhs++ = 0xFFFF;
 629                         else
 630                                 return 0;
 631                         r = *cp++;
 632                         if(r == 'n')
 633                                 r = '\n';
 634                 } else {
 635                         if(r == seof) {
 636                                 if (rhs < end)
 637                                         *rhs++ = '\0';
 638                                 else
 639                                         return 0;
 640                                 return rhs;
 641                         }
 642                 }
 643                 if (rhs < end)
 644                         *rhs++ = r;
 645                 else
 646                         return 0;
 647         }
 648         return 0;
 649 }
 650
 651 Reprog *
 652 compile(void)
 653 {
 654         Rune c;
 655         char *ep;
 656         char expbuf[512];
 657
 658         if((c = *cp++) == seof)         /* L'//' */
 659                 return 0;
 660         ep = expbuf;
 661         do {
 662                 if (c == L'\0' || c == L'\n')
 663                         quit(TMMES, linebuf);
 664                 if (c == L'\\') {
 665                         if (ep >= expbuf+sizeof(expbuf))
 666                                 quit(TMMES, linebuf);
 667                         ep += runetochar(ep, &c);
 668                         if ((c = *cp++) == L'n')
 669                                 c = L'\n';
 670                 }
 671                 if (ep >= expbuf + sizeof(expbuf))
 672                         quit(TMMES, linebuf);
 673                 ep += runetochar(ep, &c);
 674         } while ((c = *cp++) != seof);
 675         *ep = 0;
 676         return lastre = regcomp(expbuf);
 677 }
 678
 679 void
 680 regerror(char *s)
 681 {
 682         USED(s);
 683         quit(CGMES, L"r.e.-using", linebuf);
 684 }
 685
 686 void
 687 newfile(enum PTYPE type, char *name)
 688 {
 689         if (type == P_ARG)
 690                 prog.curr = name;
 691         else if ((prog.bp = Bopen(name, OREAD)) == 0)
 692                 quit("Cannot open pattern-file: %s\n", name);
 693         Blethal(prog.bp, nil);
 694         prog.type = type;
 695 }
 696
 697 int
 698 rline(Rune *buf, Rune *end)
 699 {
 700         long c;
 701         Rune r;
 702
 703         while ((c = getrune()) >= 0) {
 704                 r = c;
 705                 if (r == '\\') {
 706                         if (buf <= end)
 707                                 *buf++ = r;
 708                         if ((c = getrune()) < 0)
 709                                 break;
 710                         r = c;
 711                 } else if (r == '\n') {
 712                         *buf = '\0';
 713                         return 1;
 714                 }
 715                 if (buf <= end)
 716                         *buf++ = r;
 717         }
 718         *buf = '\0';
 719         return -1;
 720 }
 721
 722 long
 723 getrune(void)
 724 {
 725         long c;
 726         Rune r;
 727         char *p;
 728
 729         if (prog.type == P_ARG) {
 730                 if ((p = prog.curr) != 0) {
 731                         if (*p) {
 732                                 prog.curr += chartorune(&r, p);
 733                                 c = r;
 734                         } else {
 735                                 c = '\n';       /* fake an end-of-line */
 736                                 prog.curr = 0;
 737                         }
 738                 } else
 739                         c = -1;
 740         } else if ((c = Bgetrune(prog.bp)) < 0)
 741                 Bterm(prog.bp);
 742         return c;
 743 }
 744
 745 void
 746 address(Addr *ap)
 747 {
 748         int c;
 749         long lno;
 750
 751         if((c = *cp++) == '$')
 752                 ap->type = A_DOL;
 753         else if(c == '/') {
 754                 seof = c;
 755                 if (ap->rp = compile())
 756                         ap->type = A_RE;
 757                 else
 758                         ap->type = A_LAST;
 759         }
 760         else if (c >= '0' && c <= '9') {
 761                 lno = c - '0';
 762                 while ((c = *cp) >= '0' && c <= '9')
 763                         lno = lno*10 + *cp++ - '0';
 764                 if(!lno)
 765                         quit("line number 0 is illegal",0);
 766                 ap->type = A_LINE;
 767                 ap->line = lno;
 768         }
 769         else {
 770                 cp--;
 771                 ap->type = A_NONE;
 772         }
 773 }
 774
 775 cmp(char *a, char *b)           /* compare characters */
 776 {
 777         while(*a == *b++)
 778                 if (*a == '\0')
 779                         return 0;
 780                 else
 781                         a++;
 782         return 1;
 783 }
 784 rcmp(Rune *a, Rune *b)          /* compare runes */
 785 {
 786         while(*a == *b++)
 787                 if (*a == '\0')
 788                         return 0;
 789                 else
 790                         a++;
 791         return 1;
 792 }
 793
 794 char *
 795 text(char *p)           /* extract character string */
 796 {
 797         Rune r;
 798
 799         while(*cp == ' ' || *cp == '\t')
 800                 cp++;
 801         while (*cp) {
 802                 if ((r = *cp++) == '\\' && (r = *cp++) == '\0')
 803                         break;
 804                 if (r == '\n')
 805                         while (*cp == ' ' || *cp == '\t')
 806                                 cp++;
 807                 p += runetochar(p, &r);
 808         }
 809         *p++ = '\0';
 810         return p;
 811 }
 812
 813 Rune *
 814 stext(Rune *p, Rune *end)               /* extract rune string */
 815 {
 816         while(*cp == L' ' || *cp == L'\t')
 817                 cp++;
 818         while (*cp) {
 819                 if (*cp == L'\\' && *++cp == L'\0')
 820                         break;
 821                 if (p >= end-1)
 822                         quit(TMMES, linebuf);
 823                 if ((*p++ = *cp++) == L'\n')
 824                         while(*cp == L' ' || *cp == L'\t')
 825                                 cp++;
 826         }
 827         *p++ = 0;
 828         return p;
 829 }
 830
 831
 832 Label *
 833 search(Label *ptr)
 834 {
 835         Label   *rp;
 836
 837         for (rp = ltab; rp < ptr; rp++)
 838                 if(rcmp(rp->uninm, ptr->uninm) == 0)
 839                         return(rp);
 840         return(0);
 841 }
 842
 843 void
 844 dechain(void)
 845 {
 846         Label   *lptr;
 847         SedCom  *rptr, *trptr;
 848
 849         for(lptr = ltab; lptr < lab; lptr++) {
 850                 if(lptr->address == 0)
 851                         quit("Undefined label: %S", lptr->uninm);
 852                 if(lptr->chain) {
 853                         rptr = lptr->chain;
 854                         while((trptr = rptr->lb1) != nil) {
 855                                 rptr->lb1 = lptr->address;
 856                                 rptr = trptr;
 857                         }
 858                         rptr->lb1 = lptr->address;
 859                 }
 860         }
 861 }
 862
 863 int
 864 ycomp(SedCom *r)
 865 {
 866         int i;
 867         Rune *rp, *sp, *tsp;
 868         Rune c, highc;
 869
 870         highc = 0;
 871         for(tsp = cp; *tsp != seof; tsp++) {
 872                 if(*tsp == L'\\')
 873                         tsp++;
 874                 if(*tsp == L'\n' || *tsp == L'\0')
 875                         return 0;
 876                 if (*tsp > highc)
 877                         highc = *tsp;
 878         }
 879         tsp++;
 880         if ((rp = r->text = (Rune *)malloc(sizeof(Rune) * (highc+2))) == nil)
 881                 quit("Out of memory");
 882         *rp++ = highc;                          /* save upper bound */
 883         for (i = 0; i <= highc; i++)
 884                 rp[i] = i;
 885         sp = cp;
 886         while((c = *sp++) != seof) {
 887                 if(c == L'\\' && *sp == L'n') {
 888                         sp++;
 889                         c = L'\n';
 890                 }
 891                 if((rp[c] = *tsp++) == L'\\' && *tsp == L'n') {
 892                         rp[c] = L'\n';
 893                         tsp++;
 894                 }
 895                 if(rp[c] == seof || rp[c] == L'\0') {
 896                         free(r->re1);
 897                         r->re1 = nil;
 898                         return 0;
 899                 }
 900         }
 901         if(*tsp != seof) {
 902                 free(r->re1);
 903                 r->re1 = nil;
 904                 return 0;
 905         }
 906         cp = tsp+1;
 907         return 1;
 908 }
 909
 910 void
 911 execute(void)
 912 {
 913         SedCom  *ipc;
 914
 915         while (spend = gline(linebuf)){
 916                 for(ipc = pspace; ipc->command; ) {
 917                         if (!executable(ipc)) {
 918                                 ipc++;
 919                                 continue;
 920                         }
 921                         command(ipc);
 922
 923                         if(delflag)
 924                                 break;
 925                         if(jflag) {
 926                                 jflag = 0;
 927                                 if((ipc = ipc->lb1) == 0)
 928                                         break;
 929                         } else
 930                                 ipc++;
 931                 }
 932                 if(!nflag && !delflag)
 933                         putline(&fout, linebuf, spend - linebuf);
 934                 if(aptr > abuf)
 935                         arout();
 936                 delflag = 0;
 937         }
 938 }
 939
 940 /* determine if a statement should be applied to an input line */
 941 int
 942 executable(SedCom *ipc)
 943 {
 944         if (ipc->active) {      /* Addr1 satisfied - accept until Addr2 */
 945                 if (ipc->active == 1)           /* Second line */
 946                         ipc->active = 2;
 947                 switch(ipc->ad2.type) {
 948                 case A_NONE:            /* No second addr; use first */
 949                         ipc->active = 0;
 950                         break;
 951                 case A_DOL:             /* Accept everything */
 952                         return !ipc->negfl;
 953                 case A_LINE:            /* Line at end of range? */
 954                         if (lnum <= ipc->ad2.line) {
 955                                 if (ipc->ad2.line == lnum)
 956                                         ipc->active = 0;
 957                                 return !ipc->negfl;
 958                         }
 959                         ipc->active = 0;        /* out of range */
 960                         return ipc->negfl;
 961                 case A_RE:              /* Check for matching R.E. */
 962                         if (match(ipc->ad2.rp, linebuf))
 963                                 ipc->active = 0;
 964                         return !ipc->negfl;
 965                 default:
 966                         quit("Internal error");
 967                 }
 968         }
 969         switch (ipc->ad1.type) {        /* Check first address */
 970         case A_NONE:                    /* Everything matches */
 971                 return !ipc->negfl;
 972         case A_DOL:                     /* Only last line */
 973                 if (dolflag)
 974                         return !ipc->negfl;
 975                 break;
 976         case A_LINE:                    /* Check line number */
 977                 if (ipc->ad1.line == lnum) {
 978                         ipc->active = 1;        /* In range */
 979                         return !ipc->negfl;
 980                 }
 981                 break;
 982         case A_RE:                      /* Check R.E. */
 983                 if (match(ipc->ad1.rp, linebuf)) {
 984                         ipc->active = 1;        /* In range */
 985                         return !ipc->negfl;
 986                 }
 987                 break;
 988         default:
 989                 quit("Internal error");
 990         }
 991         return ipc->negfl;
 992 }
 993
 994 int
 995 match(Reprog *pattern, Rune *buf)
 996 {
 997         if (!pattern)
 998                 return 0;
 999         subexp[0].rsp = buf;
1000         subexp[0].ep = 0;
1001         if (rregexec(pattern, linebuf, subexp, MAXSUB) > 0) {
1002                 loc1 = subexp[0].rsp;
1003                 loc2 = subexp[0].rep;
1004                 return 1;
1005         }
1006         loc1 = loc2 = 0;
1007         return 0;
1008 }
1009
1010 int
1011 substitute(SedCom *ipc)
1012 {
1013         int len;
1014
1015         if(!match(ipc->re1, linebuf))
1016                 return 0;
1017
1018         /*
1019          * we have at least one match.  some patterns, e.g. '$' or '^', can
1020          * produce 0-length matches, so during a global substitute we must
1021          * bump to the character after a 0-length match to keep from looping.
1022          */
1023         sflag = 1;
1024         if(ipc->gfl == 0)                       /* single substitution */
1025                 dosub(ipc->rhs);
1026         else
1027                 do{                             /* global substitution */
1028                         len = loc2 - loc1;      /* length of match */
1029                         dosub(ipc->rhs);        /* dosub moves loc2 */
1030                         if(*loc2 == 0)          /* end of string */
1031                                 break;
1032                         if(len == 0)            /* zero-length R.E. match */
1033                                 loc2++;         /* bump over 0-length match */
1034                         if(*loc2 == 0)          /* end of string */
1035                                 break;
1036                 } while(match(ipc->re1, loc2));
1037         return 1;
1038 }
1039
1040 void
1041 dosub(Rune *rhsbuf)
1042 {
1043         int c, n;
1044         Rune *lp, *sp, *rp;
1045
1046         lp = linebuf;
1047         sp = genbuf;
1048         rp = rhsbuf;
1049         while (lp < loc1)
1050                 *sp++ = *lp++;
1051         while(c = *rp++) {
1052                 if (c == '&') {
1053                         sp = place(sp, loc1, loc2);
1054                         continue;
1055                 }
1056                 if (c == 0xFFFF && (c = *rp++) >= '1' && c < MAXSUB + '0') {
1057                         n = c-'0';
1058                         if (subexp[n].rsp && subexp[n].rep) {
1059                                 sp = place(sp, subexp[n].rsp, subexp[n].rep);
1060                                 continue;
1061                         }
1062                         else {
1063                                 fprint(2, "sed: Invalid back reference \\%d\n",n);
1064                                 errexit();
1065                         }
1066                 }
1067                 *sp++ = c;
1068                 if (sp >= &genbuf[LBSIZE])
1069                         fprint(2, "sed: Output line too long.\n");
1070         }
1071         lp = loc2;
1072         loc2 = sp - genbuf + linebuf;
1073         while (*sp++ = *lp++)
1074                 if (sp >= &genbuf[LBSIZE])
1075                         fprint(2, "sed: Output line too long.\n");
1076         lp = linebuf;
1077         sp = genbuf;
1078         while (*lp++ = *sp++)
1079                 ;
1080         spend = lp - 1;
1081 }
1082
1083 Rune *
1084 place(Rune *sp, Rune *l1, Rune *l2)
1085 {
1086         while (l1 < l2) {
1087                 *sp++ = *l1++;
1088                 if (sp >= &genbuf[LBSIZE])
1089                         fprint(2, "sed: Output line too long.\n");
1090         }
1091         return sp;
1092 }
1093
1094 char *
1095 trans(int c)
1096 {
1097         static char buf[] = "\\x0000";
1098         static char hex[] = "0123456789abcdef";
1099
1100         switch(c) {
1101         case '\b':
1102                 return "\\b";
1103         case '\n':
1104                 return "\\n";
1105         case '\r':
1106                 return "\\r";
1107         case '\t':
1108                 return "\\t";
1109         case '\\':
1110                 return "\\\\";
1111         }
1112         buf[2] = hex[(c>>12)&0xF];
1113         buf[3] = hex[(c>>8)&0xF];
1114         buf[4] = hex[(c>>4)&0xF];
1115         buf[5] = hex[c&0xF];
1116         return buf;
1117 }
1118
1119 void
1120 command(SedCom *ipc)
1121 {
1122         int i, c;
1123         char *ucp;
1124         Rune *execp, *p1, *p2, *rp;
1125
1126         switch(ipc->command) {
1127         case ACOM:
1128                 *aptr++ = ipc;
1129                 if(aptr >= abuf+MAXADDS)
1130                         quit("sed: Too many appends after line %ld\n",
1131                                 (char *)lnum);
1132                 *aptr = 0;
1133                 break;
1134         case CCOM:
1135                 delflag = 1;
1136                 if(ipc->active == 1) {
1137                         for(rp = ipc->text; *rp; rp++)
1138                                 Bputrune(&fout, *rp);
1139                         Bputc(&fout, '\n');
1140                 }
1141                 break;
1142         case DCOM:
1143                 delflag++;
1144                 break;
1145         case CDCOM:
1146                 p1 = p2 = linebuf;
1147                 while(*p1 != '\n') {
1148                         if(*p1++ == 0) {
1149                                 delflag++;
1150                                 return;
1151                         }
1152                 }
1153                 p1++;
1154                 while(*p2++ = *p1++)
1155                         ;
1156                 spend = p2 - 1;
1157                 jflag++;
1158                 break;
1159         case EQCOM:
1160                 Bprint(&fout, "%ld\n", lnum);
1161                 break;
1162         case GCOM:
1163                 p1 = linebuf;
1164                 p2 = holdsp;
1165                 while(*p1++ = *p2++)
1166                         ;
1167                 spend = p1 - 1;
1168                 break;
1169         case CGCOM:
1170                 *spend++ = '\n';
1171                 p1 = spend;
1172                 p2 = holdsp;
1173                 while(*p1++ = *p2++)
1174                         if(p1 >= lbend)
1175                                 break;
1176                 spend = p1 - 1;
1177                 break;
1178         case HCOM:
1179                 p1 = holdsp;
1180                 p2 = linebuf;
1181                 while(*p1++ = *p2++);
1182                 hspend = p1 - 1;
1183                 break;
1184         case CHCOM:
1185                 *hspend++ = '\n';
1186                 p1 = hspend;
1187                 p2 = linebuf;
1188                 while(*p1++ = *p2++)
1189                         if(p1 >= hend)
1190                                 break;
1191                 hspend = p1 - 1;
1192                 break;
1193         case ICOM:
1194                 for(rp = ipc->text; *rp; rp++)
1195                         Bputrune(&fout, *rp);
1196                 Bputc(&fout, '\n');
1197                 break;
1198         case BCOM:
1199                 jflag = 1;
1200                 break;
1201         case LCOM:
1202                 c = 0;
1203                 for (i = 0, rp = linebuf; *rp; rp++) {
1204                         c = *rp;
1205                         if(c >= 0x20 && c < 0x7F && c != '\\') {
1206                                 Bputc(&fout, c);
1207                                 if(i++ > 71) {
1208                                         Bprint(&fout, "\\\n");
1209                                         i = 0;
1210                                 }
1211                         } else {
1212                                 for (ucp = trans(*rp); *ucp; ucp++){
1213                                         c = *ucp;
1214                                         Bputc(&fout, c);
1215                                         if(i++ > 71) {
1216                                                 Bprint(&fout, "\\\n");
1217                                                 i = 0;
1218                                         }
1219                                 }
1220                         }
1221                 }
1222                 if(c == ' ')
1223                         Bprint(&fout, "\\n");
1224                 Bputc(&fout, '\n');
1225                 break;
1226         case NCOM:
1227                 if(!nflag)
1228                         putline(&fout, linebuf, spend-linebuf);
1229
1230                 if(aptr > abuf)
1231                         arout();
1232                 if((execp = gline(linebuf)) == 0) {
1233                         delflag = 1;
1234                         break;
1235                 }
1236                 spend = execp;
1237                 break;
1238         case CNCOM:
1239                 if(aptr > abuf)
1240                         arout();
1241                 *spend++ = '\n';
1242                 if((execp = gline(spend)) == 0) {
1243                         delflag = 1;
1244                         break;
1245                 }
1246                 spend = execp;
1247                 break;
1248         case PCOM:
1249                 putline(&fout, linebuf, spend-linebuf);
1250                 break;
1251         case CPCOM:
1252 cpcom:
1253                 for(rp = linebuf; *rp && *rp != '\n'; rp++)
1254                         Bputc(&fout, *rp);
1255                 Bputc(&fout, '\n');
1256                 break;
1257         case QCOM:
1258                 if(!nflag)
1259                         putline(&fout, linebuf, spend-linebuf);
1260                 if(aptr > abuf)
1261                         arout();
1262                 exits(0);
1263         case RCOM:
1264                 *aptr++ = ipc;
1265                 if(aptr >= &abuf[MAXADDS])
1266                         quit("sed: Too many reads after line %ld\n",
1267                                 (char *)lnum);
1268                 *aptr = 0;
1269                 break;
1270         case SCOM:
1271                 i = substitute(ipc);
1272                 if(i && ipc->pfl)
1273                         if(ipc->pfl == 1)
1274                                 putline(&fout, linebuf, spend-linebuf);
1275                         else
1276                                 goto cpcom;
1277                 if(i && ipc->fcode)
1278                         goto wcom;
1279                 break;
1280
1281         case TCOM:
1282                 if(sflag) {
1283                         sflag = 0;
1284                         jflag = 1;
1285                 }
1286                 break;
1287
1288         case WCOM:
1289 wcom:
1290                 putline(ipc->fcode,linebuf, spend - linebuf);
1291                 break;
1292         case XCOM:
1293                 p1 = linebuf;
1294                 p2 = genbuf;
1295                 while(*p2++ = *p1++)
1296                         ;
1297                 p1 = holdsp;
1298                 p2 = linebuf;
1299                 while(*p2++ = *p1++)
1300                         ;
1301                 spend = p2 - 1;
1302                 p1 = genbuf;
1303                 p2 = holdsp;
1304                 while(*p2++ = *p1++)
1305                         ;
1306                 hspend = p2 - 1;
1307                 break;
1308         case YCOM:
1309                 p1 = linebuf;
1310                 p2 = ipc->text;
1311                 for (i = *p2++; *p1; p1++)
1312                         if (*p1 <= i)
1313                                 *p1 = p2[*p1];
1314                 break;
1315         }
1316 }
1317
1318 void
1319 putline(Biobuf *bp, Rune *buf, int n)
1320 {
1321         while (n--)
1322                 Bputrune(bp, *buf++);
1323         Bputc(bp, '\n');
1324 }
1325 ecmp(Rune *a, Rune *b, int count)
1326 {
1327         while(count--)
1328                 if(*a++ != *b++)
1329                         return 0;
1330         return 1;
1331 }
1332
1333 void
1334 arout(void)
1335 {
1336         int     c;
1337         char    *s;
1338         char    buf[128];
1339         Rune    *p1;
1340         Biobuf  *fi;
1341
1342         for (aptr = abuf; *aptr; aptr++) {
1343                 if((*aptr)->command == ACOM) {
1344                         for(p1 = (*aptr)->text; *p1; p1++ )
1345                                 Bputrune(&fout, *p1);
1346                         Bputc(&fout, '\n');
1347                 } else {
1348                         for(s = buf, p1 = (*aptr)->text; *p1; p1++)
1349                                 s += runetochar(s, p1);
1350                         *s = '\0';
1351                         if((fi = Bopen(buf, OREAD)) == 0)
1352                                 continue;
1353                         Blethal(fi, nil);
1354                         while((c = Bgetc(fi)) >= 0)
1355                                 Bputc(&fout, c);
1356                         Bterm(fi);
1357                 }
1358         }
1359         aptr = abuf;
1360         *aptr = 0;
1361 }
1362
1363 void
1364 errexit(void)
1365 {
1366         exits("error");
1367 }
1368
1369 void
1370 quit(char *fmt, ...)
1371 {
1372         char *p, *ep;
1373         char msg[256];
1374         va_list arg;
1375
1376         ep = msg + sizeof msg;
1377         p = seprint(msg, ep, "sed: ");
1378         va_start(arg, fmt);
1379         p = vseprint(p, ep, fmt, arg);
1380         va_end(arg);
1381         p = seprint(p, ep, "\n");
1382         write(2, msg, p - msg);
1383         errexit();
1384 }
1385
1386 Rune *
1387 gline(Rune *addr)
1388 {
1389         long c;
1390         Rune *p;
1391         static long peekc = 0;
1392
1393         if (f == 0 && opendata() < 0)
1394                 return 0;
1395         sflag = 0;
1396         lnum++;
1397 /*      Bflush(&fout);********* dumped 4/30/92 - bobf****/
1398         do {
1399                 p = addr;
1400                 for (c = (peekc? peekc: Bgetrune(f)); c >= 0; c = Bgetrune(f)) {
1401                         if (c == '\n') {
1402                                 if ((peekc = Bgetrune(f)) < 0 && fhead == 0)
1403                                         dolflag = 1;
1404                                 *p = '\0';
1405                                 return p;
1406                         }
1407                         if (c && p < lbend)
1408                                 *p++ = c;
1409                 }
1410                 /* return partial final line, adding implicit newline */
1411                 if(p != addr) {
1412                         *p = '\0';
1413                         peekc = -1;
1414                         if (fhead == 0)
1415                                 dolflag = 1;
1416                         return p;
1417                 }
1418                 peekc = 0;
1419                 Bterm(f);
1420         } while (opendata() > 0);               /* Switch to next stream */
1421         f = 0;
1422         return 0;
1423 }
1424
1425 /*
1426  * Data file input section - the intent is to transparently
1427  *      catenate all data input streams.
1428  */
1429 void
1430 enroll(char *filename)          /* Add a file to the input file cache */
1431 {
1432         FileCache *fp;
1433
1434         if ((fp = (FileCache *)malloc(sizeof (FileCache))) == nil)
1435                 quit("Out of memory");
1436         if (ftail == nil)
1437                 fhead = fp;
1438         else
1439                 ftail->next = fp;
1440         ftail = fp;
1441         fp->next = nil;
1442         fp->name = filename;            /* 0 => stdin */
1443 }
1444
1445 int
1446 opendata(void)
1447 {
1448         if (fhead == nil)
1449                 return -1;
1450         if (fhead->name) {
1451                 if ((f = Bopen(fhead->name, OREAD)) == nil)
1452                         quit("Can't open %s", fhead->name);
1453         } else {
1454                 Binit(&stdin, 0, OREAD);
1455                 f = &stdin;
1456         }
1457         Blethal(f, nil);
1458         fhead = fhead->next;
1459         return 1;
1460 }