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 int     dollars;                        /* Number of dollar (first) addresses */
 100
 101 Reprog  *lastre;                        /* Last regular expression */
 102 Resub   subexp[MAXSUB];                 /* sub-patterns of pattern match*/
 103
 104 Rune    addspace[ADDSIZE];              /* Buffer for a, c, & i commands */
 105 Rune    *addend = addspace+ADDSIZE;
 106
 107 SedCom  *abuf[MAXADDS];                 /* Queue of pending adds & reads */
 108 SedCom  **aptr = abuf;
 109
 110 struct {                                /* Sed program input control block */
 111         enum PTYPE {                    /* Either on command line or in file */
 112                 P_ARG,
 113                 P_FILE,
 114         } type;
 115         union PCTL {                    /* Pointer to data */
 116                 Biobuf  *bp;
 117                 char    *curr;
 118         };
 119 } prog;
 120
 121 Rune    genbuf[LBSIZE+1];               /* Miscellaneous buffer */
 122
 123 FileCache       *fhead;                 /* Head of File Cache Chain */
 124 FileCache       *ftail;                 /* Tail of File Cache Chain */
 125
 126 Rune    *loc1;                          /* Start of pattern match */
 127 Rune    *loc2;                          /* End of pattern match */
 128 Rune    seof;                           /* Pattern delimiter char */
 129
 130 Rune    linebuf[LBSIZE+1];              /* Input data buffer */
 131 Rune    *lbend = linebuf+LBSIZE;        /* End of buffer */
 132 Rune    *spend = linebuf;               /* End of input data */
 133 Rune    *cp;                            /* Current scan point in linebuf */
 134
 135 Rune    holdsp[LBSIZE+1];               /* Hold buffer */
 136 Rune    *hend = holdsp+LBSIZE;          /* End of hold buffer */
 137 Rune    *hspend = holdsp;               /* End of hold data */
 138
 139 int     nflag;                          /* Command line flags */
 140 int     gflag;
 141 int     uflag;
 142
 143 int     dolflag;                        /* Set when at true EOF */
 144 int     sflag;                          /* Set when substitution done */
 145 int     jflag;                          /* Set when jump required */
 146 int     delflag;                        /* Delete current line when set */
 147
 148 long    lnum;                           /* Input line count */
 149
 150 char    fname[MAXFILES][40];            /* File name cache */
 151 Biobuf  *fcode[MAXFILES];               /* File ID cache */
 152 int     nfiles;                         /* Cache fill point */
 153
 154 Biobuf  fout;                           /* Output stream */
 155 Biobuf  stdin;                          /* Default input */
 156 Biobuf* f;                              /* Input data */
 157
 158 Label   ltab[LABSIZE];                  /* Label name symbol table */
 159 Label   *labend = ltab+LABSIZE;         /* End of label table */
 160 Label   *lab = ltab+1;                  /* Current Fill point */
 161
 162 int     depth;                          /* {} stack pointer */
 163
 164 Rune    bad;                            /* Dummy err ptr reference */
 165 Rune    *badp = &bad;
 166
 167
 168 char    CGMES[]  =      "%S command garbled: %S";
 169 char    TMMES[]  =      "Too much text: %S";
 170 char    LTL[]    =      "Label too long: %S";
 171 char    AD0MES[] =      "No addresses allowed: %S";
 172 char    AD1MES[] =      "Only one address allowed: %S";
 173
 174 void    address(Addr *);
 175 void    arout(void);
 176 int     cmp(char *, char *);
 177 int     rcmp(Rune *, Rune *);
 178 void    command(SedCom *);
 179 Reprog  *compile(void);
 180 Rune    *compsub(Rune *, Rune *);
 181 void    dechain(void);
 182 void    dosub(Rune *);
 183 void    enroll(char *);
 184 void    errexit(void);
 185 int     executable(SedCom *);
 186 void    execute(void);
 187 void    fcomp(void);
 188 long    getrune(void);
 189 Rune    *gline(Rune *);
 190 int     match(Reprog *, Rune *);
 191 void    newfile(enum PTYPE, char *);
 192 int     opendata(void);
 193 Biobuf  *open_file(char *);
 194 Rune    *place(Rune *, Rune *, Rune *);
 195 void    quit(char *, ...);
 196 int     rline(Rune *, Rune *);
 197 Label   *search(Label *);
 198 int     substitute(SedCom *);
 199 char    *text(char *);
 200 Rune    *stext(Rune *, Rune *);
 201 int     ycomp(SedCom *);
 202 char *  trans(int c);
 203 void    putline(Biobuf *bp, Rune *buf, int n);
 204
 205 void
 206 main(int argc, char **argv)
 207 {
 208         int compfl;
 209
 210         lnum = 0;
 211         Binit(&fout, 1, OWRITE);
 212         Blethal(&fout, nil);
 213         fcode[nfiles++] = &fout;
 214         compfl = 0;
 215
 216         if(argc == 1)
 217                 exits(nil);
 218         ARGBEGIN{
 219         case 'e':
 220                 if (argc <= 1)
 221                         quit("missing pattern");
 222                 newfile(P_ARG, ARGF());
 223                 fcomp();
 224                 compfl = 1;
 225                 continue;
 226         case 'f':
 227                 if(argc <= 1)
 228                         quit("no pattern-file");
 229                 newfile(P_FILE, ARGF());
 230                 fcomp();
 231                 compfl = 1;
 232                 continue;
 233         case 'g':
 234                 gflag++;
 235                 continue;
 236         case 'n':
 237                 nflag++;
 238                 continue;
 239         case 'u':
 240                 uflag++;
 241                 continue;
 242         default:
 243                 quit("Unknown flag: %c", ARGC());
 244         } ARGEND
 245
 246         if(compfl == 0) {
 247                 if (--argc < 0)
 248                         quit("missing pattern");
 249                 newfile(P_ARG, *argv++);
 250                 fcomp();
 251         }
 252
 253         if(depth)
 254                 quit("Too many {'s");
 255
 256         ltab[0].address = rep;
 257
 258         dechain();
 259
 260         if(argc <= 0)
 261                 enroll(nil);            /* Add stdin to cache */
 262         else
 263                 while(--argc >= 0)
 264                         enroll(*argv++);
 265         execute();
 266         exits(nil);
 267 }
 268
 269 void
 270 fcomp(void)
 271 {
 272         int     i;
 273         Label   *lpt;
 274         Rune    *tp;
 275         SedCom  *pt, *pt1;
 276         static Rune     *p = addspace;
 277         static SedCom   **cmpend[DEPTH];        /* stack of {} operations */
 278
 279         while (rline(linebuf, lbend) >= 0) {
 280                 cp = linebuf;
 281 comploop:
 282                 while(*cp == L' ' || *cp == L'\t')
 283                         cp++;
 284                 if(*cp == L'\0' || *cp == L'#')
 285                         continue;
 286                 if(*cp == L';') {
 287                         cp++;
 288                         goto comploop;
 289                 }
 290
 291                 address(&rep->ad1);
 292                 if (rep->ad1.type != A_NONE) {
 293                         if (rep->ad1.type == A_DOL)
 294                                 dollars++;
 295                         if (rep->ad1.type == A_LAST) {
 296                                 if (!lastre)
 297                                         quit("First RE may not be null");
 298                                 rep->ad1.type = A_RE;
 299                                 rep->ad1.rp = lastre;
 300                         }
 301                         if(*cp == L',' || *cp == L';') {
 302                                 cp++;
 303                                 address(&rep->ad2);
 304                                 if (rep->ad2.type == A_LAST) {
 305                                         rep->ad2.type = A_RE;
 306                                         rep->ad2.rp = lastre;
 307                                 }
 308                         } else
 309                                 rep->ad2.type = A_NONE;
 310                 }
 311                 while(*cp == L' ' || *cp == L'\t')
 312                         cp++;
 313
 314 swit:
 315                 switch(*cp++) {
 316                 default:
 317                         quit("Unrecognized command: %S", linebuf);
 318
 319                 case '!':
 320                         rep->negfl = 1;
 321                         goto swit;
 322
 323                 case '{':
 324                         rep->command = BCOM;
 325                         rep->negfl = !rep->negfl;
 326                         cmpend[depth++] = &rep->lb1;
 327                         if(++rep >= pend)
 328                                 quit("Too many commands: %S", linebuf);
 329                         if(*cp == '\0')
 330                                 continue;
 331                         goto comploop;
 332
 333                 case '}':
 334                         if(rep->ad1.type != A_NONE)
 335                                 quit(AD0MES, linebuf);
 336                         if(--depth < 0)
 337                                 quit("Too many }'s");
 338                         *cmpend[depth] = rep;
 339                         if(*cp == 0)
 340                                 continue;
 341                         goto comploop;
 342
 343                 case '=':
 344                         rep->command = EQCOM;
 345                         if(rep->ad2.type != A_NONE)
 346                                 quit(AD1MES, linebuf);
 347                         break;
 348
 349                 case ':':
 350                         if(rep->ad1.type != A_NONE)
 351                                 quit(AD0MES, linebuf);
 352
 353                         while(*cp == L' ')
 354                                 cp++;
 355                         tp = lab->uninm;
 356                         while (*cp && *cp != L';' && *cp != L' ' &&
 357                             *cp != L'\t' && *cp != L'#') {
 358                                 *tp++ = *cp++;
 359                                 if(tp >= &lab->uninm[8])
 360                                         quit(LTL, linebuf);
 361                         }
 362                         *tp = L'\0';
 363
 364                         if (*lab->uninm == L'\0')               /* no label? */
 365                                 quit(CGMES, L":", linebuf);
 366                         if(lpt = search(lab)) {
 367                                 if(lpt->address)
 368                                         quit("Duplicate labels: %S", linebuf);
 369                         } else {
 370                                 lab->chain = 0;
 371                                 lpt = lab;
 372                                 if(++lab >= labend)
 373                                         quit("Too many labels: %S", linebuf);
 374                         }
 375                         lpt->address = rep;
 376                         if (*cp == L'#')
 377                                 continue;
 378                         rep--;                  /* reuse this slot */
 379                         break;
 380
 381                 case 'a':
 382                         rep->command = ACOM;
 383                         if(rep->ad2.type != A_NONE)
 384                                 quit(AD1MES, linebuf);
 385                         if(*cp == L'\\')
 386                                 cp++;
 387                         if(*cp++ != L'\n')
 388                                 quit(CGMES, L"a", linebuf);
 389                         rep->text = p;
 390                         p = stext(p, addend);
 391                         break;
 392                 case 'c':
 393                         rep->command = CCOM;
 394                         if(*cp == L'\\')
 395                                 cp++;
 396                         if(*cp++ != L'\n')
 397                                 quit(CGMES, L"c", linebuf);
 398                         rep->text = p;
 399                         p = stext(p, addend);
 400                         break;
 401                 case 'i':
 402                         rep->command = ICOM;
 403                         if(rep->ad2.type != A_NONE)
 404                                 quit(AD1MES, linebuf);
 405                         if(*cp == L'\\')
 406                                 cp++;
 407                         if(*cp++ != L'\n')
 408                                 quit(CGMES, L"i", linebuf);
 409                         rep->text = p;
 410                         p = stext(p, addend);
 411                         break;
 412
 413                 case 'g':
 414                         rep->command = GCOM;
 415                         break;
 416
 417                 case 'G':
 418                         rep->command = CGCOM;
 419                         break;
 420
 421                 case 'h':
 422                         rep->command = HCOM;
 423                         break;
 424
 425                 case 'H':
 426                         rep->command = CHCOM;
 427                         break;
 428
 429                 case 't':
 430                         rep->command = TCOM;
 431                         goto jtcommon;
 432
 433                 case 'b':
 434                         rep->command = BCOM;
 435 jtcommon:
 436                         while(*cp == L' ')
 437                                 cp++;
 438                         if(*cp == L'\0' || *cp == L';') {
 439                                 /* no label; jump to end */
 440                                 if(pt = ltab[0].chain) {
 441                                         while((pt1 = pt->lb1) != nil)
 442                                                 pt = pt1;
 443                                         pt->lb1 = rep;
 444                                 } else
 445                                         ltab[0].chain = rep;
 446                                 break;
 447                         }
 448
 449                         /* copy label into lab->uninm */
 450                         tp = lab->uninm;
 451                         while((*tp = *cp++) != L'\0' && *tp != L';')
 452                                 if(++tp >= &lab->uninm[8])
 453                                         quit(LTL, linebuf);
 454                         cp--;
 455                         *tp = L'\0';
 456
 457                         if (*lab->uninm == L'\0')
 458                                 /* shouldn't get here */
 459                                 quit(CGMES, L"b or t", linebuf);
 460                         if((lpt = search(lab)) != nil) {
 461                                 if(lpt->address)
 462                                         rep->lb1 = lpt->address;
 463                                 else {
 464                                         for(pt = lpt->chain; pt != nil &&
 465                                             (pt1 = pt->lb1) != nil; pt = pt1)
 466                                                 ;
 467                                         if (pt)
 468                                                 pt->lb1 = rep;
 469                                 }
 470                         } else {                        /* add new label */
 471                                 lab->chain = rep;
 472                                 lab->address = 0;
 473                                 if(++lab >= labend)
 474                                         quit("Too many labels: %S", linebuf);
 475                         }
 476                         break;
 477
 478                 case 'n':
 479                         rep->command = NCOM;
 480                         break;
 481
 482                 case 'N':
 483                         rep->command = CNCOM;
 484                         break;
 485
 486                 case 'p':
 487                         rep->command = PCOM;
 488                         break;
 489
 490                 case 'P':
 491                         rep->command = CPCOM;
 492                         break;
 493
 494                 case 'r':
 495                         rep->command = RCOM;
 496                         if(rep->ad2.type != A_NONE)
 497                                 quit(AD1MES, linebuf);
 498                         if(*cp++ != L' ')
 499                                 quit(CGMES, L"r", linebuf);
 500                         rep->text = p;
 501                         p = stext(p, addend);
 502                         break;
 503
 504                 case 'd':
 505                         rep->command = DCOM;
 506                         break;
 507
 508                 case 'D':
 509                         rep->command = CDCOM;
 510                         rep->lb1 = pspace;
 511                         break;
 512
 513                 case 'q':
 514                         rep->command = QCOM;
 515                         if(rep->ad2.type != A_NONE)
 516                                 quit(AD1MES, linebuf);
 517                         break;
 518
 519                 case 'l':
 520                         rep->command = LCOM;
 521                         break;
 522
 523                 case 's':
 524                         rep->command = SCOM;
 525                         seof = *cp++;
 526                         if ((rep->re1 = compile()) == 0) {
 527                                 if(!lastre)
 528                                         quit("First RE may not be null");
 529                                 rep->re1 = lastre;
 530                         }
 531                         rep->rhs = p;
 532                         if((p = compsub(p, addend)) == 0)
 533                                 quit(CGMES, L"s", linebuf);
 534                         if(*cp == L'g') {
 535                                 cp++;
 536                                 rep->gfl++;
 537                         } else if(gflag)
 538                                 rep->gfl++;
 539
 540                         if(*cp == L'p') {
 541                                 cp++;
 542                                 rep->pfl = 1;
 543                         }
 544
 545                         if(*cp == L'P') {
 546                                 cp++;
 547                                 rep->pfl = 2;
 548                         }
 549
 550                         if(*cp == L'w') {
 551                                 cp++;
 552                                 if(*cp++ !=  L' ')
 553                                         quit(CGMES, L"s", linebuf);
 554                                 text(fname[nfiles]);
 555                                 for(i = nfiles - 1; i >= 0; i--)
 556                                         if(cmp(fname[nfiles], fname[i]) == 0) {
 557                                                 rep->fcode = fcode[i];
 558                                                 goto done;
 559                                         }
 560                                 if(nfiles >= MAXFILES)
 561                                         quit("Too many files in w commands 1");
 562                                 rep->fcode = open_file(fname[nfiles]);
 563                         }
 564                         break;
 565
 566                 case 'w':
 567                         rep->command = WCOM;
 568                         if(*cp++ != L' ')
 569                                 quit(CGMES, L"w", linebuf);
 570                         text(fname[nfiles]);
 571                         for(i = nfiles - 1; i >= 0; i--)
 572                                 if(cmp(fname[nfiles], fname[i]) == 0) {
 573                                         rep->fcode = fcode[i];
 574                                         goto done;
 575                                 }
 576                         if(nfiles >= MAXFILES){
 577                                 fprint(2, "sed: Too many files in w commands 2 \n");
 578                                 fprint(2, "nfiles = %d; MAXF = %d\n",
 579                                         nfiles, MAXFILES);
 580                                 errexit();
 581                         }
 582                         rep->fcode = open_file(fname[nfiles]);
 583                         break;
 584
 585                 case 'x':
 586                         rep->command = XCOM;
 587                         break;
 588
 589                 case 'y':
 590                         rep->command = YCOM;
 591                         seof = *cp++;
 592                         if (ycomp(rep) == 0)
 593                                 quit(CGMES, L"y", linebuf);
 594                         break;
 595
 596                 }
 597 done:
 598                 if(++rep >= pend)
 599                         quit("Too many commands, last: %S", linebuf);
 600                 if(*cp++ != L'\0') {
 601                         if(cp[-1] == L';')
 602                                 goto comploop;
 603                         quit(CGMES, cp - 1, linebuf);
 604                 }
 605         }
 606 }
 607
 608 Biobuf *
 609 open_file(char *name)
 610 {
 611         int fd;
 612         Biobuf *bp;
 613
 614         if ((bp = malloc(sizeof(Biobuf))) == 0)
 615                 quit("Out of memory");
 616         if ((fd = open(name, OWRITE)) < 0 &&
 617             (fd = create(name, OWRITE, 0666)) < 0)
 618                 quit("Cannot create %s", name);
 619         Binit(bp, fd, OWRITE);
 620         Blethal(bp, nil);
 621         Bseek(bp, 0, 2);
 622         fcode[nfiles++] = bp;
 623         return bp;
 624 }
 625
 626 Rune *
 627 compsub(Rune *rhs, Rune *end)
 628 {
 629         Rune r;
 630
 631         while ((r = *cp++) != '\0') {
 632                 if(r == '\\') {
 633                         if (rhs < end)
 634                                 *rhs++ = Runemax;
 635                         else
 636                                 return 0;
 637                         r = *cp++;
 638                         if(r == 'n')
 639                                 r = '\n';
 640                 } else {
 641                         if(r == seof) {
 642                                 if (rhs < end)
 643                                         *rhs++ = '\0';
 644                                 else
 645                                         return 0;
 646                                 return rhs;
 647                         }
 648                 }
 649                 if (rhs < end)
 650                         *rhs++ = r;
 651                 else
 652                         return 0;
 653         }
 654         return 0;
 655 }
 656
 657 Reprog *
 658 compile(void)
 659 {
 660         Rune c;
 661         char *ep;
 662         char expbuf[512];
 663
 664         if((c = *cp++) == seof)         /* L'//' */
 665                 return 0;
 666         ep = expbuf;
 667         do {
 668                 if (c == L'\0' || c == L'\n')
 669                         quit(TMMES, linebuf);
 670                 if (c == L'\\') {
 671                         if (ep >= expbuf+sizeof(expbuf))
 672                                 quit(TMMES, linebuf);
 673                         ep += runetochar(ep, &c);
 674                         if ((c = *cp++) == L'n')
 675                                 c = L'\n';
 676                 }
 677                 if (ep >= expbuf + sizeof(expbuf))
 678                         quit(TMMES, linebuf);
 679                 ep += runetochar(ep, &c);
 680         } while ((c = *cp++) != seof);
 681         *ep = 0;
 682         return lastre = regcomp(expbuf);
 683 }
 684
 685 void
 686 regerror(char *s)
 687 {
 688         USED(s);
 689         quit(CGMES, L"r.e.-using", linebuf);
 690 }
 691
 692 int
 693 flushout(Biobufhdr *bp, void *v, long n)
 694 {
 695         int i;
 696
 697         for(i = 0; i < nfiles; i++)
 698                 Bflush(fcode[i]);
 699         return read(bp->fid, v, n);
 700 }
 701
 702 void
 703 newfile(enum PTYPE type, char *name)
 704 {
 705         if (type == P_ARG)
 706                 prog.curr = name;
 707         else {
 708                 if ((prog.bp = Bopen(name, OREAD)) == 0)
 709                         quit("Cannot open pattern-file: %s\n", name);
 710                 Blethal(prog.bp, nil);
 711                 if(uflag) Biofn(prog.bp, flushout);
 712         }
 713         prog.type = type;
 714 }
 715
 716 int
 717 rline(Rune *buf, Rune *end)
 718 {
 719         long c;
 720         Rune r;
 721
 722         while ((c = getrune()) >= 0) {
 723                 r = c;
 724                 if (r == '\\') {
 725                         if (buf <= end)
 726                                 *buf++ = r;
 727                         if ((c = getrune()) < 0)
 728                                 break;
 729                         r = c;
 730                 } else if (r == '\n') {
 731                         *buf = '\0';
 732                         return 1;
 733                 }
 734                 if (buf <= end)
 735                         *buf++ = r;
 736         }
 737         *buf = '\0';
 738         return -1;
 739 }
 740
 741 long
 742 getrune(void)
 743 {
 744         long c;
 745         Rune r;
 746         char *p;
 747
 748         if (prog.type == P_ARG) {
 749                 if ((p = prog.curr) != 0) {
 750                         if (*p) {
 751                                 prog.curr += chartorune(&r, p);
 752                                 c = r;
 753                         } else {
 754                                 c = '\n';       /* fake an end-of-line */
 755                                 prog.curr = 0;
 756                         }
 757                 } else
 758                         c = -1;
 759         } else if ((c = Bgetrune(prog.bp)) < 0)
 760                 Bterm(prog.bp);
 761         return c;
 762 }
 763
 764 void
 765 address(Addr *ap)
 766 {
 767         int c;
 768         long lno;
 769
 770         if((c = *cp++) == '$')
 771                 ap->type = A_DOL;
 772         else if(c == '/') {
 773                 seof = c;
 774                 if (ap->rp = compile())
 775                         ap->type = A_RE;
 776                 else
 777                         ap->type = A_LAST;
 778         }
 779         else if (c >= '0' && c <= '9') {
 780                 lno = c - '0';
 781                 while ((c = *cp) >= '0' && c <= '9')
 782                         lno = lno*10 + *cp++ - '0';
 783                 if(!lno)
 784                         quit("line number 0 is illegal",0);
 785                 ap->type = A_LINE;
 786                 ap->line = lno;
 787         }
 788         else {
 789                 cp--;
 790                 ap->type = A_NONE;
 791         }
 792 }
 793
 794 cmp(char *a, char *b)           /* compare characters */
 795 {
 796         while(*a == *b++)
 797                 if (*a == '\0')
 798                         return 0;
 799                 else
 800                         a++;
 801         return 1;
 802 }
 803 rcmp(Rune *a, Rune *b)          /* compare runes */
 804 {
 805         while(*a == *b++)
 806                 if (*a == '\0')
 807                         return 0;
 808                 else
 809                         a++;
 810         return 1;
 811 }
 812
 813 char *
 814 text(char *p)           /* extract character string */
 815 {
 816         Rune r;
 817
 818         while(*cp == ' ' || *cp == '\t')
 819                 cp++;
 820         while (*cp) {
 821                 if ((r = *cp++) == '\\' && (r = *cp++) == '\0')
 822                         break;
 823                 if (r == '\n')
 824                         while (*cp == ' ' || *cp == '\t')
 825                                 cp++;
 826                 p += runetochar(p, &r);
 827         }
 828         *p++ = '\0';
 829         return p;
 830 }
 831
 832 Rune *
 833 stext(Rune *p, Rune *end)               /* extract rune string */
 834 {
 835         while(*cp == L' ' || *cp == L'\t')
 836                 cp++;
 837         while (*cp) {
 838                 if (*cp == L'\\' && *++cp == L'\0')
 839                         break;
 840                 if (p >= end-1)
 841                         quit(TMMES, linebuf);
 842                 if ((*p++ = *cp++) == L'\n')
 843                         while(*cp == L' ' || *cp == L'\t')
 844                                 cp++;
 845         }
 846         *p++ = 0;
 847         return p;
 848 }
 849
 850
 851 Label *
 852 search(Label *ptr)
 853 {
 854         Label   *rp;
 855
 856         for (rp = ltab; rp < ptr; rp++)
 857                 if(rcmp(rp->uninm, ptr->uninm) == 0)
 858                         return(rp);
 859         return(0);
 860 }
 861
 862 void
 863 dechain(void)
 864 {
 865         Label   *lptr;
 866         SedCom  *rptr, *trptr;
 867
 868         for(lptr = ltab; lptr < lab; lptr++) {
 869                 if(lptr->address == 0)
 870                         quit("Undefined label: %S", lptr->uninm);
 871                 if(lptr->chain) {
 872                         rptr = lptr->chain;
 873                         while((trptr = rptr->lb1) != nil) {
 874                                 rptr->lb1 = lptr->address;
 875                                 rptr = trptr;
 876                         }
 877                         rptr->lb1 = lptr->address;
 878                 }
 879         }
 880 }
 881
 882 int
 883 ycomp(SedCom *r)
 884 {
 885         int i;
 886         Rune *rp, *sp, *tsp;
 887         Rune c, highc;
 888
 889         highc = 0;
 890         for(tsp = cp; *tsp != seof; tsp++) {
 891                 if(*tsp == L'\\')
 892                         tsp++;
 893                 if(*tsp == L'\n' || *tsp == L'\0')
 894                         return 0;
 895                 if (*tsp > highc)
 896                         highc = *tsp;
 897         }
 898         tsp++;
 899         if ((rp = r->text = (Rune *)malloc(sizeof(Rune) * (highc+2))) == nil)
 900                 quit("Out of memory");
 901         *rp++ = highc;                          /* save upper bound */
 902         for (i = 0; i <= highc; i++)
 903                 rp[i] = i;
 904         sp = cp;
 905         while((c = *sp++) != seof) {
 906                 if(c == L'\\' && *sp == L'n') {
 907                         sp++;
 908                         c = L'\n';
 909                 }
 910                 if((rp[c] = *tsp++) == L'\\' && *tsp == L'n') {
 911                         rp[c] = L'\n';
 912                         tsp++;
 913                 }
 914                 if(rp[c] == seof || rp[c] == L'\0') {
 915                         free(r->re1);
 916                         r->re1 = nil;
 917                         return 0;
 918                 }
 919         }
 920         if(*tsp != seof) {
 921                 free(r->re1);
 922                 r->re1 = nil;
 923                 return 0;
 924         }
 925         cp = tsp+1;
 926         return 1;
 927 }
 928
 929 void
 930 execute(void)
 931 {
 932         SedCom  *ipc;
 933
 934         while (spend = gline(linebuf)){
 935                 for(ipc = pspace; ipc->command; ) {
 936                         if (!executable(ipc)) {
 937                                 ipc++;
 938                                 continue;
 939                         }
 940                         command(ipc);
 941
 942                         if(delflag)
 943                                 break;
 944                         if(jflag) {
 945                                 jflag = 0;
 946                                 if((ipc = ipc->lb1) == 0)
 947                                         break;
 948                         } else
 949                                 ipc++;
 950                 }
 951                 if(!nflag && !delflag)
 952                         putline(&fout, linebuf, spend - linebuf);
 953                 if(aptr > abuf)
 954                         arout();
 955                 delflag = 0;
 956         }
 957 }
 958
 959 /* determine if a statement should be applied to an input line */
 960 int
 961 executable(SedCom *ipc)
 962 {
 963         if (ipc->active) {      /* Addr1 satisfied - accept until Addr2 */
 964                 if (ipc->active == 1)           /* Second line */
 965                         ipc->active = 2;
 966                 switch(ipc->ad2.type) {
 967                 case A_NONE:            /* No second addr; use first */
 968                         ipc->active = 0;
 969                         break;
 970                 case A_DOL:             /* Accept everything */
 971                         return !ipc->negfl;
 972                 case A_LINE:            /* Line at end of range? */
 973                         if (lnum <= ipc->ad2.line) {
 974                                 if (ipc->ad2.line == lnum)
 975                                         ipc->active = 0;
 976                                 return !ipc->negfl;
 977                         }
 978                         ipc->active = 0;        /* out of range */
 979                         return ipc->negfl;
 980                 case A_RE:              /* Check for matching R.E. */
 981                         if (match(ipc->ad2.rp, linebuf))
 982                                 ipc->active = 0;
 983                         return !ipc->negfl;
 984                 default:
 985                         quit("Internal error");
 986                 }
 987         }
 988         switch (ipc->ad1.type) {        /* Check first address */
 989         case A_NONE:                    /* Everything matches */
 990                 return !ipc->negfl;
 991         case A_DOL:                     /* Only last line */
 992                 if (dolflag)
 993                         return !ipc->negfl;
 994                 break;
 995         case A_LINE:                    /* Check line number */
 996                 if (ipc->ad1.line == lnum) {
 997                         ipc->active = 1;        /* In range */
 998                         return !ipc->negfl;
 999                 }
1000                 break;
1001         case A_RE:                      /* Check R.E. */
1002                 if (match(ipc->ad1.rp, linebuf)) {
1003                         ipc->active = 1;        /* In range */
1004                         return !ipc->negfl;
1005                 }
1006                 break;
1007         default:
1008                 quit("Internal error");
1009         }
1010         return ipc->negfl;
1011 }
1012
1013 int
1014 match(Reprog *pattern, Rune *buf)
1015 {
1016         if (!pattern)
1017                 return 0;
1018         subexp[0].rsp = buf;
1019         subexp[0].ep = 0;
1020         if (rregexec(pattern, linebuf, subexp, MAXSUB) > 0) {
1021                 loc1 = subexp[0].rsp;
1022                 loc2 = subexp[0].rep;
1023                 return 1;
1024         }
1025         loc1 = loc2 = 0;
1026         return 0;
1027 }
1028
1029 int
1030 substitute(SedCom *ipc)
1031 {
1032         int len;
1033
1034         if(!match(ipc->re1, linebuf))
1035                 return 0;
1036
1037         /*
1038          * we have at least one match.  some patterns, e.g. '$' or '^', can
1039          * produce 0-length matches, so during a global substitute we must
1040          * bump to the character after a 0-length match to keep from looping.
1041          */
1042         sflag = 1;
1043         if(ipc->gfl == 0)                       /* single substitution */
1044                 dosub(ipc->rhs);
1045         else
1046                 do{                             /* global substitution */
1047                         len = loc2 - loc1;      /* length of match */
1048                         dosub(ipc->rhs);        /* dosub moves loc2 */
1049                         if(*loc2 == 0)          /* end of string */
1050                                 break;
1051                         if(len == 0)            /* zero-length R.E. match */
1052                                 loc2++;         /* bump over 0-length match */
1053                         if(*loc2 == 0)          /* end of string */
1054                                 break;
1055                 } while(match(ipc->re1, loc2));
1056         return 1;
1057 }
1058
1059 void
1060 dosub(Rune *rhsbuf)
1061 {
1062         int c, n;
1063         Rune *lp, *sp, *rp;
1064
1065         lp = linebuf;
1066         sp = genbuf;
1067         rp = rhsbuf;
1068         while (lp < loc1)
1069                 *sp++ = *lp++;
1070         while(c = *rp++) {
1071                 if (c == '&') {
1072                         sp = place(sp, loc1, loc2);
1073                         continue;
1074                 }
1075                 if (c == Runemax && (c = *rp++) >= '1' && c < MAXSUB + '0') {
1076                         n = c-'0';
1077                         if (subexp[n].rsp && subexp[n].rep) {
1078                                 sp = place(sp, subexp[n].rsp, subexp[n].rep);
1079                                 continue;
1080                         }
1081                         else {
1082                                 quit("Invalid back reference \\%d", n);
1083                         }
1084                 }
1085                 *sp++ = c;
1086                 if (sp >= &genbuf[LBSIZE])
1087                         quit("Output line too long");
1088         }
1089         lp = loc2;
1090         loc2 = sp - genbuf + linebuf;
1091         while (*sp++ = *lp++)
1092                 if (sp >= &genbuf[LBSIZE])
1093                         quit("Output line too long");
1094         lp = linebuf;
1095         sp = genbuf;
1096         while (*lp++ = *sp++)
1097                 ;
1098         spend = lp - 1;
1099 }
1100
1101 Rune *
1102 place(Rune *sp, Rune *l1, Rune *l2)
1103 {
1104         while (l1 < l2) {
1105                 *sp++ = *l1++;
1106                 if (sp >= &genbuf[LBSIZE])
1107                         quit("Output line too long");
1108         }
1109         return sp;
1110 }
1111
1112 char *
1113 trans(int c)
1114 {
1115         static char buf[] = "\\x0000";
1116         static char hex[] = "0123456789abcdef";
1117
1118         switch(c) {
1119         case '\b':
1120                 return "\\b";
1121         case '\n':
1122                 return "\\n";
1123         case '\r':
1124                 return "\\r";
1125         case '\t':
1126                 return "\\t";
1127         case '\\':
1128                 return "\\\\";
1129         }
1130         buf[2] = hex[(c>>12)&0xF];
1131         buf[3] = hex[(c>>8)&0xF];
1132         buf[4] = hex[(c>>4)&0xF];
1133         buf[5] = hex[c&0xF];
1134         return buf;
1135 }
1136
1137 void
1138 command(SedCom *ipc)
1139 {
1140         int i, c;
1141         char *ucp;
1142         Rune *execp, *p1, *p2, *rp;
1143
1144         switch(ipc->command) {
1145         case ACOM:
1146                 *aptr++ = ipc;
1147                 if(aptr >= abuf+MAXADDS)
1148                         quit("Too many appends after line %ld", lnum);
1149                 *aptr = 0;
1150                 break;
1151         case CCOM:
1152                 delflag = 1;
1153                 if(ipc->active == 1) {
1154                         for(rp = ipc->text; *rp; rp++)
1155                                 Bputrune(&fout, *rp);
1156                         Bputc(&fout, '\n');
1157                 }
1158                 break;
1159         case DCOM:
1160                 delflag++;
1161                 break;
1162         case CDCOM:
1163                 p1 = p2 = linebuf;
1164                 while(*p1 != '\n') {
1165                         if(*p1++ == 0) {
1166                                 delflag++;
1167                                 return;
1168                         }
1169                 }
1170                 p1++;
1171                 while(*p2++ = *p1++)
1172                         ;
1173                 spend = p2 - 1;
1174                 jflag++;
1175                 break;
1176         case EQCOM:
1177                 Bprint(&fout, "%ld\n", lnum);
1178                 break;
1179         case GCOM:
1180                 p1 = linebuf;
1181                 p2 = holdsp;
1182                 while(*p1++ = *p2++)
1183                         ;
1184                 spend = p1 - 1;
1185                 break;
1186         case CGCOM:
1187                 *spend++ = '\n';
1188                 p1 = spend;
1189                 p2 = holdsp;
1190                 while(*p1++ = *p2++)
1191                         if(p1 >= lbend)
1192                                 break;
1193                 spend = p1 - 1;
1194                 break;
1195         case HCOM:
1196                 p1 = holdsp;
1197                 p2 = linebuf;
1198                 while(*p1++ = *p2++);
1199                 hspend = p1 - 1;
1200                 break;
1201         case CHCOM:
1202                 *hspend++ = '\n';
1203                 p1 = hspend;
1204                 p2 = linebuf;
1205                 while(*p1++ = *p2++)
1206                         if(p1 >= hend)
1207                                 break;
1208                 hspend = p1 - 1;
1209                 break;
1210         case ICOM:
1211                 for(rp = ipc->text; *rp; rp++)
1212                         Bputrune(&fout, *rp);
1213                 Bputc(&fout, '\n');
1214                 break;
1215         case BCOM:
1216                 jflag = 1;
1217                 break;
1218         case LCOM:
1219                 c = 0;
1220                 for (i = 0, rp = linebuf; *rp; rp++) {
1221                         c = *rp;
1222                         if(c >= 0x20 && c < 0x7F && c != '\\') {
1223                                 Bputc(&fout, c);
1224                                 if(i++ > 71) {
1225                                         Bprint(&fout, "\\\n");
1226                                         i = 0;
1227                                 }
1228                         } else {
1229                                 for (ucp = trans(*rp); *ucp; ucp++){
1230                                         c = *ucp;
1231                                         Bputc(&fout, c);
1232                                         if(i++ > 71) {
1233                                                 Bprint(&fout, "\\\n");
1234                                                 i = 0;
1235                                         }
1236                                 }
1237                         }
1238                 }
1239                 if(c == ' ')
1240                         Bprint(&fout, "\\n");
1241                 Bputc(&fout, '\n');
1242                 break;
1243         case NCOM:
1244                 if(!nflag)
1245                         putline(&fout, linebuf, spend-linebuf);
1246
1247                 if(aptr > abuf)
1248                         arout();
1249                 if((execp = gline(linebuf)) == 0) {
1250                         delflag = 1;
1251                         break;
1252                 }
1253                 spend = execp;
1254                 break;
1255         case CNCOM:
1256                 if(aptr > abuf)
1257                         arout();
1258                 *spend++ = '\n';
1259                 if((execp = gline(spend)) == 0) {
1260                         delflag = 1;
1261                         break;
1262                 }
1263                 spend = execp;
1264                 break;
1265         case PCOM:
1266                 putline(&fout, linebuf, spend-linebuf);
1267                 break;
1268         case CPCOM:
1269 cpcom:
1270                 for(rp = linebuf; *rp && *rp != '\n'; rp++)
1271                         Bputc(&fout, *rp);
1272                 Bputc(&fout, '\n');
1273                 break;
1274         case QCOM:
1275                 if(!nflag)
1276                         putline(&fout, linebuf, spend-linebuf);
1277                 if(aptr > abuf)
1278                         arout();
1279                 exits(nil);
1280         case RCOM:
1281                 *aptr++ = ipc;
1282                 if(aptr >= &abuf[MAXADDS])
1283                         quit("Too many reads after line %ld", lnum);
1284                 *aptr = 0;
1285                 break;
1286         case SCOM:
1287                 i = substitute(ipc);
1288                 if(i && ipc->pfl)
1289                         if(ipc->pfl == 1)
1290                                 putline(&fout, linebuf, spend-linebuf);
1291                         else
1292                                 goto cpcom;
1293                 if(i && ipc->fcode)
1294                         goto wcom;
1295                 break;
1296
1297         case TCOM:
1298                 if(sflag) {
1299                         sflag = 0;
1300                         jflag = 1;
1301                 }
1302                 break;
1303
1304         case WCOM:
1305 wcom:
1306                 putline(ipc->fcode,linebuf, spend - linebuf);
1307                 break;
1308         case XCOM:
1309                 p1 = linebuf;
1310                 p2 = genbuf;
1311                 while(*p2++ = *p1++)
1312                         ;
1313                 p1 = holdsp;
1314                 p2 = linebuf;
1315                 while(*p2++ = *p1++)
1316                         ;
1317                 spend = p2 - 1;
1318                 p1 = genbuf;
1319                 p2 = holdsp;
1320                 while(*p2++ = *p1++)
1321                         ;
1322                 hspend = p2 - 1;
1323                 break;
1324         case YCOM:
1325                 p1 = linebuf;
1326                 p2 = ipc->text;
1327                 for (i = *p2++; *p1; p1++)
1328                         if (*p1 <= i)
1329                                 *p1 = p2[*p1];
1330                 break;
1331         }
1332 }
1333
1334 void
1335 putline(Biobuf *bp, Rune *buf, int n)
1336 {
1337         while (n--)
1338                 Bputrune(bp, *buf++);
1339         Bputc(bp, '\n');
1340 }
1341
1342 void
1343 arout(void)
1344 {
1345         int     c;
1346         char    *s, *e;
1347         char    buf[128];
1348         Rune    *p1;
1349         Biobuf  *fi;
1350
1351         for (aptr = abuf; *aptr; aptr++) {
1352                 if((*aptr)->command == ACOM) {
1353                         for(p1 = (*aptr)->text; *p1; p1++ )
1354                                 Bputrune(&fout, *p1);
1355                         Bputc(&fout, '\n');
1356                 } else {
1357                         for(s = buf, e = buf+sizeof(buf)-UTFmax-1, p1 = (*aptr)->text; *p1 && s < e; p1++)
1358                                 s += runetochar(s, p1);
1359                         *s = '\0';
1360                         if((fi = Bopen(buf, OREAD)) == 0)
1361                                 continue;
1362                         Blethal(fi, nil);
1363                         if(uflag) Biofn(fi, flushout);
1364                         while((c = Bgetc(fi)) >= 0)
1365                                 Bputc(&fout, c);
1366                         Bterm(fi);
1367                 }
1368         }
1369         aptr = abuf;
1370         *aptr = 0;
1371 }
1372
1373 void
1374 errexit(void)
1375 {
1376         exits("error");
1377 }
1378
1379 void
1380 quit(char *fmt, ...)
1381 {
1382         char *p, *ep;
1383         char msg[256];
1384         va_list arg;
1385
1386         ep = msg + sizeof msg;
1387         p = seprint(msg, ep, "sed: ");
1388         va_start(arg, fmt);
1389         p = vseprint(p, ep, fmt, arg);
1390         va_end(arg);
1391         p = seprint(p, ep, "\n");
1392         write(2, msg, p - msg);
1393         errexit();
1394 }
1395
1396 Rune *
1397 gline(Rune *addr)
1398 {
1399         long c;
1400         Rune *p;
1401         static long peekc = 0;
1402
1403         if (f == 0 && opendata() < 0)
1404                 return 0;
1405         sflag = 0;
1406         lnum++;
1407 /*      Bflush(&fout);********* dumped 4/30/92 - bobf****/
1408         do {
1409                 p = addr;
1410                 for (c = (peekc? peekc: Bgetrune(f)); c >= 0; c = Bgetrune(f)) {
1411                         if (c == '\n') {
1412                                 if (dollars != 0 && (peekc = Bgetrune(f)) < 0 && fhead == nil)
1413                                         dolflag = 1;
1414                                 *p = '\0';
1415                                 return p;
1416                         }
1417                         if (c && p < lbend)
1418                                 *p++ = c;
1419                 }
1420                 /* return partial final line, adding implicit newline */
1421                 if(p != addr) {
1422                         *p = '\0';
1423                         peekc = -1;
1424                         if (fhead == nil)
1425                                 dolflag = 1;
1426                         return p;
1427                 }
1428                 peekc = 0;
1429                 Bterm(f);
1430         } while (opendata() > 0);               /* Switch to next stream */
1431         f = 0;
1432         return 0;
1433 }
1434
1435 /*
1436  * Data file input section - the intent is to transparently
1437  *      catenate all data input streams.
1438  */
1439 void
1440 enroll(char *filename)          /* Add a file to the input file cache */
1441 {
1442         FileCache *fp;
1443
1444         if ((fp = (FileCache *)malloc(sizeof (FileCache))) == nil)
1445                 quit("Out of memory");
1446         if (ftail == nil)
1447                 fhead = fp;
1448         else
1449                 ftail->next = fp;
1450         ftail = fp;
1451         fp->next = nil;
1452         fp->name = filename;            /* 0 => stdin */
1453 }
1454
1455 int
1456 opendata(void)
1457 {
1458         if (fhead == nil)
1459                 return -1;
1460         if (fhead->name) {
1461                 if ((f = Bopen(fhead->name, OREAD)) == nil)
1462                         quit("Can't open %s", fhead->name);
1463         } else {
1464                 Binit(&stdin, 0, OREAD);
1465                 f = &stdin;
1466         }
1467         Blethal(f, nil);
1468         if(uflag) Biofn(f, flushout);
1469         fhead = fhead->next;
1470         return 1;
1471 }