src/lex.c

   1 /*      $OpenBSD: lex.c,v 1.47 2013/03/03 19:11:34 guenther Exp $       */
   2
   3 /*-
   4  * Copyright (c) 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
   5  *               2011, 2012, 2013
   6  *      Thorsten Glaser <tg@mirbsd.org>
   7  *
   8  * Provided that these terms and disclaimer and all copyright notices
   9  * are retained or reproduced in an accompanying document, permission
  10  * is granted to deal in this work without restriction, including un-
  11  * limited rights to use, publicly perform, distribute, sell, modify,
  12  * merge, give away, or sublicence.
  13  *
  14  * This work is provided "AS IS" and WITHOUT WARRANTY of any kind, to
  15  * the utmost extent permitted by applicable law, neither express nor
  16  * implied; without malicious intent or gross negligence. In no event
  17  * may a licensor, author or contributor be held liable for indirect,
  18  * direct, other damage, loss, or other issues arising in any way out
  19  * of dealing in the work, even if advised of the possibility of such
  20  * damage or existence of a defect, except proven that it results out
  21  * of said person's immediate fault when using the work as intended.
  22  */
  23
  24 #include "sh.h"
  25
  26 __RCSID("$MirOS: src/bin/mksh/lex.c,v 1.188 2013/08/10 13:44:31 tg Exp $");
  27
  28 /*
  29  * states while lexing word
  30  */
  31 #define SBASE           0       /* outside any lexical constructs */
  32 #define SWORD           1       /* implicit quoting for substitute() */
  33 #define SLETPAREN       2       /* inside (( )), implicit quoting */
  34 #define SSQUOTE         3       /* inside '' */
  35 #define SDQUOTE         4       /* inside "" */
  36 #define SEQUOTE         5       /* inside $'' */
  37 #define SBRACE          6       /* inside ${} */
  38 #define SQBRACE         7       /* inside "${}" */
  39 #define SBQUOTE         8       /* inside `` */
  40 #define SASPAREN        9       /* inside $(( )) */
  41 #define SHEREDELIM      10      /* parsing <<,<<-,<<< delimiter */
  42 #define SHEREDQUOTE     11      /* parsing " in <<,<<-,<<< delimiter */
  43 #define SPATTERN        12      /* parsing *(...|...) pattern (*+?@!) */
  44 #define SADELIM         13      /* like SBASE, looking for delimiter */
  45 #define STBRACEKORN     14      /* parsing ${...[#%]...} !FSH */
  46 #define STBRACEBOURNE   15      /* parsing ${...[#%]...} FSH */
  47 #define SINVALID        255     /* invalid state */
  48
  49 struct sretrace_info {
  50         struct sretrace_info *next;
  51         XString xs;
  52         char *xp;
  53 };
  54
  55 /*
  56  * Structure to keep track of the lexing state and the various pieces of info
  57  * needed for each particular state.
  58  */
  59 typedef struct lex_state {
  60         union {
  61                 /* point to the next state block */
  62                 struct lex_state *base;
  63                 /* marks start of state output in output string */
  64                 int start;
  65                 /* SBQUOTE: true if in double quotes: "`...`" */
  66                 /* SEQUOTE: got NUL, ignore rest of string */
  67                 bool abool;
  68                 /* SADELIM information */
  69                 struct {
  70                         /* character to search for */
  71                         unsigned char delimiter;
  72                         /* max. number of delimiters */
  73                         unsigned char num;
  74                 } adelim;
  75         } u;
  76         /* count open parentheses */
  77         short nparen;
  78         /* type of this state */
  79         uint8_t type;
  80 } Lex_state;
  81 #define ls_base         u.base
  82 #define ls_start        u.start
  83 #define ls_bool         u.abool
  84 #define ls_adelim       u.adelim
  85
  86 typedef struct {
  87         Lex_state *base;
  88         Lex_state *end;
  89 } State_info;
  90
  91 static void readhere(struct ioword *);
  92 static void ungetsc(int);
  93 static void ungetsc_i(int);
  94 static int getsc_uu(void);
  95 static void getsc_line(Source *);
  96 static int getsc_bn(void);
  97 static int s_get(void);
  98 static void s_put(int);
  99 static char *get_brace_var(XString *, char *);
 100 static bool arraysub(char **);
 101 static void gethere(bool);
 102 static Lex_state *push_state_i(State_info *, Lex_state *);
 103 static Lex_state *pop_state_i(State_info *, Lex_state *);
 104
 105 static int backslash_skip;
 106 static int ignore_backslash_newline;
 107
 108 /* optimised getsc_bn() */
 109 #define o_getsc()       (*source->str != '\0' && *source->str != '\\' && \
 110                             !backslash_skip ? *source->str++ : getsc_bn())
 111 /* optimised getsc_uu() */
 112 #define o_getsc_u()     ((*source->str != '\0') ? *source->str++ : getsc_uu())
 113
 114 /* retrace helper */
 115 #define o_getsc_r(carg) {                               \
 116         int cev = (carg);                               \
 117         struct sretrace_info *rp = retrace_info;        \
 118                                                         \
 119         while (rp) {                                    \
 120                 Xcheck(rp->xs, rp->xp);                 \
 121                 *rp->xp++ = cev;                        \
 122                 rp = rp->next;                          \
 123         }                                               \
 124                                                         \
 125         return (cev);                                   \
 126 }
 127
 128 #if defined(MKSH_SMALL) && !defined(MKSH_SMALL_BUT_FAST)
 129 static int getsc(void);
 130
 131 static int
 132 getsc(void)
 133 {
 134         o_getsc_r(o_getsc());
 135 }
 136 #else
 137 static int getsc_r(int);
 138
 139 static int
 140 getsc_r(int c)
 141 {
 142         o_getsc_r(c);
 143 }
 144
 145 #define getsc()         getsc_r(o_getsc())
 146 #endif
 147
 148 #define STATE_BSIZE     8
 149
 150 #define PUSH_STATE(s)   do {                                    \
 151         if (++statep == state_info.end)                         \
 152                 statep = push_state_i(&state_info, statep);     \
 153         state = statep->type = (s);                             \
 154 } while (/* CONSTCOND */ 0)
 155
 156 #define POP_STATE()     do {                                    \
 157         if (--statep == state_info.base)                        \
 158                 statep = pop_state_i(&state_info, statep);      \
 159         state = statep->type;                                   \
 160 } while (/* CONSTCOND */ 0)
 161
 162 #define PUSH_SRETRACE() do {                                    \
 163         struct sretrace_info *ri;                               \
 164                                                                 \
 165         statep->ls_start = Xsavepos(ws, wp);                    \
 166         ri = alloc(sizeof(struct sretrace_info), ATEMP);        \
 167         Xinit(ri->xs, ri->xp, 64, ATEMP);                       \
 168         ri->next = retrace_info;                                \
 169         retrace_info = ri;                                      \
 170 } while (/* CONSTCOND */ 0)
 171
 172 #define POP_SRETRACE()  do {                                    \
 173         wp = Xrestpos(ws, wp, statep->ls_start);                \
 174         *retrace_info->xp = '\0';                               \
 175         sp = Xstring(retrace_info->xs, retrace_info->xp);       \
 176         dp = (void *)retrace_info;                              \
 177         retrace_info = retrace_info->next;                      \
 178         afree(dp, ATEMP);                                       \
 179 } while (/* CONSTCOND */ 0)
 180
 181 /**
 182  * Lexical analyser
 183  *
 184  * tokens are not regular expressions, they are LL(1).
 185  * for example, "${var:-${PWD}}", and "$(size $(whence ksh))".
 186  * hence the state stack. Note "$(...)" are now parsed recursively.
 187  */
 188
 189 int
 190 yylex(int cf)
 191 {
 192         Lex_state states[STATE_BSIZE], *statep, *s2, *base;
 193         State_info state_info;
 194         int c, c2, state;
 195         size_t cz;
 196         XString ws;             /* expandable output word */
 197         char *wp;               /* output word pointer */
 198         char *sp, *dp;
 199
 200  Again:
 201         states[0].type = SINVALID;
 202         states[0].ls_base = NULL;
 203         statep = &states[1];
 204         state_info.base = states;
 205         state_info.end = &state_info.base[STATE_BSIZE];
 206
 207         Xinit(ws, wp, 64, ATEMP);
 208
 209         backslash_skip = 0;
 210         ignore_backslash_newline = 0;
 211
 212         if (cf & ONEWORD)
 213                 state = SWORD;
 214         else if (cf & LETEXPR) {
 215                 /* enclose arguments in (double) quotes */
 216                 *wp++ = OQUOTE;
 217                 state = SLETPAREN;
 218                 statep->nparen = 0;
 219         } else {
 220                 /* normal lexing */
 221                 state = (cf & HEREDELIM) ? SHEREDELIM : SBASE;
 222                 while ((c = getsc()) == ' ' || c == '\t')
 223                         ;
 224                 if (c == '#') {
 225                         ignore_backslash_newline++;
 226                         while ((c = getsc()) != '\0' && c != '\n')
 227                                 ;
 228                         ignore_backslash_newline--;
 229                 }
 230                 ungetsc(c);
 231         }
 232         if (source->flags & SF_ALIAS) {
 233                 /* trailing ' ' in alias definition */
 234                 source->flags &= ~SF_ALIAS;
 235                 cf |= ALIAS;
 236         }
 237
 238         /* Initial state: one of SWORD SLETPAREN SHEREDELIM SBASE */
 239         statep->type = state;
 240
 241         /* check for here string */
 242         if (state == SHEREDELIM) {
 243                 c = getsc();
 244                 if (c == '<') {
 245                         state = SHEREDELIM;
 246                         while ((c = getsc()) == ' ' || c == '\t')
 247                                 ;
 248                         ungetsc(c);
 249                         c = '<';
 250                         goto accept_nonword;
 251                 }
 252                 ungetsc(c);
 253         }
 254
 255         /* collect non-special or quoted characters to form word */
 256         while (!((c = getsc()) == 0 ||
 257             ((state == SBASE || state == SHEREDELIM) && ctype(c, C_LEX1)))) {
 258                 if (state == SBASE &&
 259                     subshell_nesting_type == /*{*/ '}' &&
 260                     c == /*{*/ '}')
 261                         /* possibly end ${ :;} */
 262                         break;
 263  accept_nonword:
 264                 Xcheck(ws, wp);
 265                 switch (state) {
 266                 case SADELIM:
 267                         if (c == '(')
 268                                 statep->nparen++;
 269                         else if (c == ')')
 270                                 statep->nparen--;
 271                         else if (statep->nparen == 0 && (c == /*{*/ '}' ||
 272                             c == (int)statep->ls_adelim.delimiter)) {
 273                                 *wp++ = ADELIM;
 274                                 *wp++ = c;
 275                                 if (c == /*{*/ '}' || --statep->ls_adelim.num == 0)
 276                                         POP_STATE();
 277                                 if (c == /*{*/ '}')
 278                                         POP_STATE();
 279                                 break;
 280                         }
 281                         /* FALLTHROUGH */
 282                 case SBASE:
 283                         if (c == '[' && (cf & (VARASN|ARRAYVAR))) {
 284                                 /* temporary */
 285                                 *wp = EOS;
 286                                 if (is_wdvarname(Xstring(ws, wp), false)) {
 287                                         char *p, *tmp;
 288
 289                                         if (arraysub(&tmp)) {
 290                                                 *wp++ = CHAR;
 291                                                 *wp++ = c;
 292                                                 for (p = tmp; *p; ) {
 293                                                         Xcheck(ws, wp);
 294                                                         *wp++ = CHAR;
 295                                                         *wp++ = *p++;
 296                                                 }
 297                                                 afree(tmp, ATEMP);
 298                                                 break;
 299                                         } else {
 300                                                 Source *s;
 301
 302                                                 s = pushs(SREREAD,
 303                                                     source->areap);
 304                                                 s->start = s->str =
 305                                                     s->u.freeme = tmp;
 306                                                 s->next = source;
 307                                                 source = s;
 308                                         }
 309                                 }
 310                                 *wp++ = CHAR;
 311                                 *wp++ = c;
 312                                 break;
 313                         }
 314                         /* FALLTHROUGH */
 315  Sbase1:                /* includes *(...|...) pattern (*+?@!) */
 316                         if (c == '*' || c == '@' || c == '+' || c == '?' ||
 317                             c == '!') {
 318                                 c2 = getsc();
 319                                 if (c2 == '(' /*)*/ ) {
 320                                         *wp++ = OPAT;
 321                                         *wp++ = c;
 322                                         PUSH_STATE(SPATTERN);
 323                                         break;
 324                                 }
 325                                 ungetsc(c2);
 326                         }
 327                         /* FALLTHROUGH */
 328  Sbase2:                /* doesn't include *(...|...) pattern (*+?@!) */
 329                         switch (c) {
 330                         case '\\':
 331  getsc_qchar:
 332                                 if ((c = getsc())) {
 333                                         /* trailing \ is lost */
 334                                         *wp++ = QCHAR;
 335                                         *wp++ = c;
 336                                 }
 337                                 break;
 338                         case '\'':
 339  open_ssquote_unless_heredoc:
 340                                 if ((cf & HEREDOC))
 341                                         goto store_char;
 342                                 *wp++ = OQUOTE;
 343                                 ignore_backslash_newline++;
 344                                 PUSH_STATE(SSQUOTE);
 345                                 break;
 346                         case '"':
 347  open_sdquote:
 348                                 *wp++ = OQUOTE;
 349                                 PUSH_STATE(SDQUOTE);
 350                                 break;
 351                         case '$':
 352                                 /*
 353                                  * processing of dollar sign belongs into
 354                                  * Subst, except for those which can open
 355                                  * a string: $'…' and $"…"
 356                                  */
 357  subst_dollar_ex:
 358                                 c = getsc();
 359                                 switch (c) {
 360                                 case '"':
 361                                         goto open_sdquote;
 362                                 case '\'':
 363                                         goto open_sequote;
 364                                 default:
 365                                         goto SubstS;
 366                                 }
 367                         default:
 368                                 goto Subst;
 369                         }
 370                         break;
 371
 372  Subst:
 373                         switch (c) {
 374                         case '\\':
 375                                 c = getsc();
 376                                 switch (c) {
 377                                 case '"':
 378                                         if ((cf & HEREDOC))
 379                                                 goto heredocquote;
 380                                         /* FALLTHROUGH */
 381                                 case '\\':
 382                                 case '$': case '`':
 383  store_qchar:
 384                                         *wp++ = QCHAR;
 385                                         *wp++ = c;
 386                                         break;
 387                                 default:
 388  heredocquote:
 389                                         Xcheck(ws, wp);
 390                                         if (c) {
 391                                                 /* trailing \ is lost */
 392                                                 *wp++ = CHAR;
 393                                                 *wp++ = '\\';
 394                                                 *wp++ = CHAR;
 395                                                 *wp++ = c;
 396                                         }
 397                                         break;
 398                                 }
 399                                 break;
 400                         case '$':
 401                                 c = getsc();
 402  SubstS:
 403                                 if (c == '(') /*)*/ {
 404                                         c = getsc();
 405                                         if (c == '(') /*)*/ {
 406                                                 *wp++ = EXPRSUB;
 407                                                 PUSH_STATE(SASPAREN);
 408                                                 statep->nparen = 2;
 409                                                 PUSH_SRETRACE();
 410                                                 *retrace_info->xp++ = '(';
 411                                         } else {
 412                                                 ungetsc(c);
 413  subst_command:
 414                                                 c = COMSUB;
 415  subst_command2:
 416                                                 sp = yyrecursive(c);
 417                                                 cz = strlen(sp) + 1;
 418                                                 XcheckN(ws, wp, cz);
 419                                                 *wp++ = c;
 420                                                 memcpy(wp, sp, cz);
 421                                                 wp += cz;
 422                                         }
 423                                 } else if (c == '{') /*}*/ {
 424                                         if ((c = getsc()) == '|') {
 425                                                 /*
 426                                                  * non-subenvironment
 427                                                  * value substitution
 428                                                  */
 429                                                 c = VALSUB;
 430                                                 goto subst_command2;
 431                                         } else if (ctype(c, C_IFSWS)) {
 432                                                 /*
 433                                                  * non-subenvironment
 434                                                  * "command" substitution
 435                                                  */
 436                                                 c = FUNSUB;
 437                                                 goto subst_command2;
 438                                         }
 439                                         ungetsc(c);
 440                                         *wp++ = OSUBST;
 441                                         *wp++ = '{'; /*}*/
 442                                         wp = get_brace_var(&ws, wp);
 443                                         c = getsc();
 444                                         /* allow :# and :% (ksh88 compat) */
 445                                         if (c == ':') {
 446                                                 *wp++ = CHAR;
 447                                                 *wp++ = c;
 448                                                 c = getsc();
 449                                                 if (c == ':') {
 450                                                         *wp++ = CHAR;
 451                                                         *wp++ = '0';
 452                                                         *wp++ = ADELIM;
 453                                                         *wp++ = ':';
 454                                                         PUSH_STATE(SBRACE);
 455                                                         PUSH_STATE(SADELIM);
 456                                                         statep->ls_adelim.delimiter = ':';
 457                                                         statep->ls_adelim.num = 1;
 458                                                         statep->nparen = 0;
 459                                                         break;
 460                                                 } else if (ksh_isdigit(c) ||
 461                                                     c == '('/*)*/ || c == ' ' ||
 462                                                     /*XXX what else? */
 463                                                     c == '$') {
 464                                                         /* substring subst. */
 465                                                         if (c != ' ') {
 466                                                                 *wp++ = CHAR;
 467                                                                 *wp++ = ' ';
 468                                                         }
 469                                                         ungetsc(c);
 470                                                         PUSH_STATE(SBRACE);
 471                                                         PUSH_STATE(SADELIM);
 472                                                         statep->ls_adelim.delimiter = ':';
 473                                                         statep->ls_adelim.num = 2;
 474                                                         statep->nparen = 0;
 475                                                         break;
 476                                                 }
 477                                         } else if (c == '/') {
 478                                                 *wp++ = CHAR;
 479                                                 *wp++ = c;
 480                                                 if ((c = getsc()) == '/') {
 481                                                         *wp++ = ADELIM;
 482                                                         *wp++ = c;
 483                                                 } else
 484                                                         ungetsc(c);
 485                                                 PUSH_STATE(SBRACE);
 486                                                 PUSH_STATE(SADELIM);
 487                                                 statep->ls_adelim.delimiter = '/';
 488                                                 statep->ls_adelim.num = 1;
 489                                                 statep->nparen = 0;
 490                                                 break;
 491                                         }
 492                                         /*
 493                                          * If this is a trim operation,
 494                                          * treat (,|,) specially in STBRACE.
 495                                          */
 496                                         if (ctype(c, C_SUBOP2)) {
 497                                                 ungetsc(c);
 498                                                 if (Flag(FSH))
 499                                                         PUSH_STATE(STBRACEBOURNE);
 500                                                 else
 501                                                         PUSH_STATE(STBRACEKORN);
 502                                         } else {
 503                                                 ungetsc(c);
 504                                                 if (state == SDQUOTE ||
 505                                                     state == SQBRACE)
 506                                                         PUSH_STATE(SQBRACE);
 507                                                 else
 508                                                         PUSH_STATE(SBRACE);
 509                                         }
 510                                 } else if (ksh_isalphx(c)) {
 511                                         *wp++ = OSUBST;
 512                                         *wp++ = 'X';
 513                                         do {
 514                                                 Xcheck(ws, wp);
 515                                                 *wp++ = c;
 516                                                 c = getsc();
 517                                         } while (ksh_isalnux(c));
 518                                         *wp++ = '\0';
 519                                         *wp++ = CSUBST;
 520                                         *wp++ = 'X';
 521                                         ungetsc(c);
 522                                 } else if (ctype(c, C_VAR1 | C_DIGIT)) {
 523                                         Xcheck(ws, wp);
 524                                         *wp++ = OSUBST;
 525                                         *wp++ = 'X';
 526                                         *wp++ = c;
 527                                         *wp++ = '\0';
 528                                         *wp++ = CSUBST;
 529                                         *wp++ = 'X';
 530                                 } else {
 531                                         *wp++ = CHAR;
 532                                         *wp++ = '$';
 533                                         ungetsc(c);
 534                                 }
 535                                 break;
 536                         case '`':
 537  subst_gravis:
 538                                 PUSH_STATE(SBQUOTE);
 539                                 *wp++ = COMSUB;
 540                                 /*
 541                                  * Need to know if we are inside double quotes
 542                                  * since sh/AT&T-ksh translate the \" to " in
 543                                  * "`...\"...`".
 544                                  * This is not done in POSIX mode (section
 545                                  * 3.2.3, Double Quotes: "The backquote shall
 546                                  * retain its special meaning introducing the
 547                                  * other form of command substitution (see
 548                                  * 3.6.3). The portion of the quoted string
 549                                  * from the initial backquote and the
 550                                  * characters up to the next backquote that
 551                                  * is not preceded by a backslash (having
 552                                  * escape characters removed) defines that
 553                                  * command whose output replaces `...` when
 554                                  * the word is expanded."
 555                                  * Section 3.6.3, Command Substitution:
 556                                  * "Within the backquoted style of command
 557                                  * substitution, backslash shall retain its
 558                                  * literal meaning, except when followed by
 559                                  * $ ` \.").
 560                                  */
 561                                 statep->ls_bool = false;
 562                                 s2 = statep;
 563                                 base = state_info.base;
 564                                 while (/* CONSTCOND */ 1) {
 565                                         for (; s2 != base; s2--) {
 566                                                 if (s2->type == SDQUOTE) {
 567                                                         statep->ls_bool = true;
 568                                                         break;
 569                                                 }
 570                                         }
 571                                         if (s2 != base)
 572                                                 break;
 573                                         if (!(s2 = s2->ls_base))
 574                                                 break;
 575                                         base = s2-- - STATE_BSIZE;
 576                                 }
 577                                 break;
 578                         case QCHAR:
 579                                 if (cf & LQCHAR) {
 580                                         *wp++ = QCHAR;
 581                                         *wp++ = getsc();
 582                                         break;
 583                                 }
 584                                 /* FALLTHROUGH */
 585                         default:
 586  store_char:
 587                                 *wp++ = CHAR;
 588                                 *wp++ = c;
 589                         }
 590                         break;
 591
 592                 case SEQUOTE:
 593                         if (c == '\'') {
 594                                 POP_STATE();
 595                                 *wp++ = CQUOTE;
 596                                 ignore_backslash_newline--;
 597                         } else if (c == '\\') {
 598                                 if ((c2 = unbksl(true, s_get, s_put)) == -1)
 599                                         c2 = s_get();
 600                                 if (c2 == 0)
 601                                         statep->ls_bool = true;
 602                                 if (!statep->ls_bool) {
 603                                         char ts[4];
 604
 605                                         if ((unsigned int)c2 < 0x100) {
 606                                                 *wp++ = QCHAR;
 607                                                 *wp++ = c2;
 608                                         } else {
 609                                                 cz = utf_wctomb(ts, c2 - 0x100);
 610                                                 ts[cz] = 0;
 611                                                 for (cz = 0; ts[cz]; ++cz) {
 612                                                         *wp++ = QCHAR;
 613                                                         *wp++ = ts[cz];
 614                                                 }
 615                                         }
 616                                 }
 617                         } else if (!statep->ls_bool) {
 618                                 *wp++ = QCHAR;
 619                                 *wp++ = c;
 620                         }
 621                         break;
 622
 623                 case SSQUOTE:
 624                         if (c == '\'') {
 625                                 POP_STATE();
 626                                 if ((cf & HEREDOC) || state == SQBRACE)
 627                                         goto store_char;
 628                                 *wp++ = CQUOTE;
 629                                 ignore_backslash_newline--;
 630                         } else {
 631                                 *wp++ = QCHAR;
 632                                 *wp++ = c;
 633                         }
 634                         break;
 635
 636                 case SDQUOTE:
 637                         if (c == '"') {
 638                                 POP_STATE();
 639                                 *wp++ = CQUOTE;
 640                         } else
 641                                 goto Subst;
 642                         break;
 643
 644                 /* $(( ... )) */
 645                 case SASPAREN:
 646                         if (c == '(')
 647                                 statep->nparen++;
 648                         else if (c == ')') {
 649                                 statep->nparen--;
 650                                 if (statep->nparen == 1) {
 651                                         /* end of EXPRSUB */
 652                                         POP_SRETRACE();
 653                                         POP_STATE();
 654
 655                                         if ((c2 = getsc()) == /*(*/ ')') {
 656                                                 cz = strlen(sp) - 2;
 657                                                 XcheckN(ws, wp, cz);
 658                                                 memcpy(wp, sp + 1, cz);
 659                                                 wp += cz;
 660                                                 afree(sp, ATEMP);
 661                                                 *wp++ = '\0';
 662                                                 break;
 663                                         } else {
 664                                                 Source *s;
 665
 666                                                 ungetsc(c2);
 667                                                 /*
 668                                                  * mismatched parenthesis -
 669                                                  * assume we were really
 670                                                  * parsing a $(...) expression
 671                                                  */
 672                                                 --wp;
 673                                                 s = pushs(SREREAD,
 674                                                     source->areap);
 675                                                 s->start = s->str =
 676                                                     s->u.freeme = sp;
 677                                                 s->next = source;
 678                                                 source = s;
 679                                                 goto subst_command;
 680                                         }
 681                                 }
 682                         }
 683                         /* reuse existing state machine */
 684                         goto Sbase2;
 685
 686                 case SQBRACE:
 687                         if (c == '\\') {
 688                                 /*
 689                                  * perform POSIX "quote removal" if the back-
 690                                  * slash is "special", i.e. same cases as the
 691                                  * {case '\\':} in Subst: plus closing brace;
 692                                  * in mksh code "quote removal" on '\c' means
 693                                  * write QCHAR+c, otherwise CHAR+\+CHAR+c are
 694                                  * emitted (in heredocquote:)
 695                                  */
 696                                 if ((c = getsc()) == '"' || c == '\\' ||
 697                                     c == '$' || c == '`' || c == /*{*/'}')
 698                                         goto store_qchar;
 699                                 goto heredocquote;
 700                         }
 701                         goto common_SQBRACE;
 702
 703                 case SBRACE:
 704                         if (c == '\'')
 705                                 goto open_ssquote_unless_heredoc;
 706                         else if (c == '\\')
 707                                 goto getsc_qchar;
 708  common_SQBRACE:
 709                         if (c == '"')
 710                                 goto open_sdquote;
 711                         else if (c == '$')
 712                                 goto subst_dollar_ex;
 713                         else if (c == '`')
 714                                 goto subst_gravis;
 715                         else if (c != /*{*/ '}')
 716                                 goto store_char;
 717                         POP_STATE();
 718                         *wp++ = CSUBST;
 719                         *wp++ = /*{*/ '}';
 720                         break;
 721
 722                 /* Same as SBASE, except (,|,) treated specially */
 723                 case STBRACEKORN:
 724                         if (c == '|')
 725                                 *wp++ = SPAT;
 726                         else if (c == '(') {
 727                                 *wp++ = OPAT;
 728                                 /* simile for @ */
 729                                 *wp++ = ' ';
 730                                 PUSH_STATE(SPATTERN);
 731                         } else /* FALLTHROUGH */
 732                 case STBRACEBOURNE:
 733                           if (c == /*{*/ '}') {
 734                                 POP_STATE();
 735                                 *wp++ = CSUBST;
 736                                 *wp++ = /*{*/ '}';
 737                         } else
 738                                 goto Sbase1;
 739                         break;
 740
 741                 case SBQUOTE:
 742                         if (c == '`') {
 743                                 *wp++ = 0;
 744                                 POP_STATE();
 745                         } else if (c == '\\') {
 746                                 switch (c = getsc()) {
 747                                 case 0:
 748                                         /* trailing \ is lost */
 749                                         break;
 750                                 case '\\':
 751                                 case '$': case '`':
 752                                         *wp++ = c;
 753                                         break;
 754                                 case '"':
 755                                         if (statep->ls_bool) {
 756                                                 *wp++ = c;
 757                                                 break;
 758                                         }
 759                                         /* FALLTHROUGH */
 760                                 default:
 761                                         *wp++ = '\\';
 762                                         *wp++ = c;
 763                                         break;
 764                                 }
 765                         } else
 766                                 *wp++ = c;
 767                         break;
 768
 769                 /* ONEWORD */
 770                 case SWORD:
 771                         goto Subst;
 772
 773                 /* LETEXPR: (( ... )) */
 774                 case SLETPAREN:
 775                         if (c == /*(*/ ')') {
 776                                 if (statep->nparen > 0)
 777                                         --statep->nparen;
 778                                 else if ((c2 = getsc()) == /*(*/ ')') {
 779                                         c = 0;
 780                                         *wp++ = CQUOTE;
 781                                         goto Done;
 782                                 } else {
 783                                         Source *s;
 784
 785                                         ungetsc(c2);
 786                                         /*
 787                                          * mismatched parenthesis -
 788                                          * assume we were really
 789                                          * parsing a (...) expression
 790                                          */
 791                                         *wp = EOS;
 792                                         sp = Xstring(ws, wp);
 793                                         dp = wdstrip(sp, WDS_KEEPQ);
 794                                         s = pushs(SREREAD, source->areap);
 795                                         s->start = s->str = s->u.freeme = dp;
 796                                         s->next = source;
 797                                         source = s;
 798                                         return ('('/*)*/);
 799                                 }
 800                         } else if (c == '(')
 801                                 /*
 802                                  * parentheses inside quotes and
 803                                  * backslashes are lost, but AT&T ksh
 804                                  * doesn't count them either
 805                                  */
 806                                 ++statep->nparen;
 807                         goto Sbase2;
 808
 809                 /* <<, <<-, <<< delimiter */
 810                 case SHEREDELIM:
 811                         /*
 812                          * here delimiters need a special case since
 813                          * $ and `...` are not to be treated specially
 814                          */
 815                         switch (c) {
 816                         case '\\':
 817                                 if ((c = getsc())) {
 818                                         /* trailing \ is lost */
 819                                         *wp++ = QCHAR;
 820                                         *wp++ = c;
 821                                 }
 822                                 break;
 823                         case '\'':
 824                                 goto open_ssquote_unless_heredoc;
 825                         case '$':
 826                                 if ((c2 = getsc()) == '\'') {
 827  open_sequote:
 828                                         *wp++ = OQUOTE;
 829                                         ignore_backslash_newline++;
 830                                         PUSH_STATE(SEQUOTE);
 831                                         statep->ls_bool = false;
 832                                         break;
 833                                 } else if (c2 == '"') {
 834                                         /* FALLTHROUGH */
 835                         case '"':
 836                                         state = statep->type = SHEREDQUOTE;
 837                                         PUSH_SRETRACE();
 838                                         break;
 839                                 }
 840                                 ungetsc(c2);
 841                                 /* FALLTHROUGH */
 842                         default:
 843                                 *wp++ = CHAR;
 844                                 *wp++ = c;
 845                         }
 846                         break;
 847
 848                 /* " in <<, <<-, <<< delimiter */
 849                 case SHEREDQUOTE:
 850                         if (c != '"')
 851                                 goto Subst;
 852                         POP_SRETRACE();
 853                         dp = strnul(sp) - 1;
 854                         /* remove the trailing double quote */
 855                         *dp = '\0';
 856                         /* store the quoted string */
 857                         *wp++ = OQUOTE;
 858                         XcheckN(ws, wp, (dp - sp));
 859                         dp = sp;
 860                         while ((c = *dp++)) {
 861                                 if (c == '\\') {
 862                                         switch ((c = *dp++)) {
 863                                         case '\\':
 864                                         case '"':
 865                                         case '$':
 866                                         case '`':
 867                                                 break;
 868                                         default:
 869                                                 *wp++ = CHAR;
 870                                                 *wp++ = '\\';
 871                                                 break;
 872                                         }
 873                                 }
 874                                 *wp++ = CHAR;
 875                                 *wp++ = c;
 876                         }
 877                         afree(sp, ATEMP);
 878                         *wp++ = CQUOTE;
 879                         state = statep->type = SHEREDELIM;
 880                         break;
 881
 882                 /* in *(...|...) pattern (*+?@!) */
 883                 case SPATTERN:
 884                         if (c == /*(*/ ')') {
 885                                 *wp++ = CPAT;
 886                                 POP_STATE();
 887                         } else if (c == '|') {
 888                                 *wp++ = SPAT;
 889                         } else if (c == '(') {
 890                                 *wp++ = OPAT;
 891                                 /* simile for @ */
 892                                 *wp++ = ' ';
 893                                 PUSH_STATE(SPATTERN);
 894                         } else
 895                                 goto Sbase1;
 896                         break;
 897                 }
 898         }
 899  Done:
 900         Xcheck(ws, wp);
 901         if (statep != &states[1])
 902                 /* XXX figure out what is missing */
 903                 yyerror("no closing quote\n");
 904
 905         /* This done to avoid tests for SHEREDELIM wherever SBASE tested */
 906         if (state == SHEREDELIM)
 907                 state = SBASE;
 908
 909         dp = Xstring(ws, wp);
 910         if (state == SBASE && (
 911 #ifndef MKSH_LEGACY_MODE
 912             (c == '&' && !Flag(FSH) && !Flag(FPOSIX)) ||
 913 #endif
 914             c == '<' || c == '>')) {
 915                 struct ioword *iop = alloc(sizeof(struct ioword), ATEMP);
 916
 917                 if (Xlength(ws, wp) == 0)
 918                         iop->unit = c == '<' ? 0 : 1;
 919                 else for (iop->unit = 0, c2 = 0; c2 < Xlength(ws, wp); c2 += 2) {
 920                         if (dp[c2] != CHAR)
 921                                 goto no_iop;
 922                         if (!ksh_isdigit(dp[c2 + 1]))
 923                                 goto no_iop;
 924                         iop->unit = (iop->unit * 10) + dp[c2 + 1] - '0';
 925                 }
 926
 927                 if (iop->unit >= FDBASE)
 928                         goto no_iop;
 929
 930                 if (c == '&') {
 931                         if ((c2 = getsc()) != '>') {
 932                                 ungetsc(c2);
 933                                 goto no_iop;
 934                         }
 935                         c = c2;
 936                         iop->flag = IOBASH;
 937                 } else
 938                         iop->flag = 0;
 939
 940                 c2 = getsc();
 941                 /* <<, >>, <> are ok, >< is not */
 942                 if (c == c2 || (c == '<' && c2 == '>')) {
 943                         iop->flag |= c == c2 ?
 944                             (c == '>' ? IOCAT : IOHERE) : IORDWR;
 945                         if (iop->flag == IOHERE) {
 946                                 if ((c2 = getsc()) == '-') {
 947                                         iop->flag |= IOSKIP;
 948                                         c2 = getsc();
 949                                 } else if (c2 == '<')
 950                                         iop->flag |= IOHERESTR;
 951                                 ungetsc(c2);
 952                                 if (c2 == '\n')
 953                                         iop->flag |= IONDELIM;
 954                         }
 955                 } else if (c2 == '&')
 956                         iop->flag |= IODUP | (c == '<' ? IORDUP : 0);
 957                 else {
 958                         iop->flag |= c == '>' ? IOWRITE : IOREAD;
 959                         if (c == '>' && c2 == '|')
 960                                 iop->flag |= IOCLOB;
 961                         else
 962                                 ungetsc(c2);
 963                 }
 964
 965                 iop->name = NULL;
 966                 iop->delim = NULL;
 967                 iop->heredoc = NULL;
 968                 /* free word */
 969                 Xfree(ws, wp);
 970                 yylval.iop = iop;
 971                 return (REDIR);
 972  no_iop:
 973                 afree(iop, ATEMP);
 974         }
 975
 976         if (wp == dp && state == SBASE) {
 977                 /* free word */
 978                 Xfree(ws, wp);
 979                 /* no word, process LEX1 character */
 980                 if ((c == '|') || (c == '&') || (c == ';') || (c == '('/*)*/)) {
 981                         if ((c2 = getsc()) == c)
 982                                 c = (c == ';') ? BREAK :
 983                                     (c == '|') ? LOGOR :
 984                                     (c == '&') ? LOGAND :
 985                                     /* c == '(' ) */ MDPAREN;
 986                         else if (c == '|' && c2 == '&')
 987                                 c = COPROC;
 988                         else if (c == ';' && c2 == '|')
 989                                 c = BRKEV;
 990                         else if (c == ';' && c2 == '&')
 991                                 c = BRKFT;
 992                         else
 993                                 ungetsc(c2);
 994 #ifndef MKSH_SMALL
 995                         if (c == BREAK) {
 996                                 if ((c2 = getsc()) == '&')
 997                                         c = BRKEV;
 998                                 else
 999                                         ungetsc(c2);
1000                         }
1001 #endif
1002                 } else if (c == '\n') {
1003                         gethere(false);
1004                         if (cf & CONTIN)
1005                                 goto Again;
1006                 } else if (c == '\0')
1007                         /* need here strings at EOF */
1008                         gethere(true);
1009                 return (c);
1010         }
1011
1012         /* terminate word */
1013         *wp++ = EOS;
1014         yylval.cp = Xclose(ws, wp);
1015         if (state == SWORD || state == SLETPAREN
1016             /* XXX ONEWORD? */)
1017                 return (LWORD);
1018
1019         /* unget terminator */
1020         ungetsc(c);
1021
1022         /*
1023          * note: the alias-vs-function code below depends on several
1024          * interna: starting from here, source->str is not modified;
1025          * the way getsc() and ungetsc() operate; etc.
1026          */
1027
1028         /* copy word to unprefixed string ident */
1029         sp = yylval.cp;
1030         dp = ident;
1031         if ((cf & HEREDELIM) && (sp[1] == '<'))
1032                 while ((dp - ident) < IDENT) {
1033                         if ((c = *sp++) == CHAR)
1034                                 *dp++ = *sp++;
1035                         else if ((c != OQUOTE) && (c != CQUOTE))
1036                                 break;
1037                 }
1038         else
1039                 while ((dp - ident) < IDENT && (c = *sp++) == CHAR)
1040                         *dp++ = *sp++;
1041         /* Make sure the ident array stays '\0' padded */
1042         memset(dp, 0, (ident + IDENT) - dp + 1);
1043         if (c != EOS)
1044                 /* word is not unquoted */
1045                 *ident = '\0';
1046
1047         if (*ident != '\0' && (cf & (KEYWORD | ALIAS))) {
1048                 struct tbl *p;
1049                 uint32_t h = hash(ident);
1050
1051                 if ((cf & KEYWORD) && (p = ktsearch(&keywords, ident, h)) &&
1052                     (!(cf & ESACONLY) || p->val.i == ESAC ||
1053                     p->val.i == /*{*/ '}')) {
1054                         afree(yylval.cp, ATEMP);
1055                         return (p->val.i);
1056                 }
1057                 if ((cf & ALIAS) && (p = ktsearch(&aliases, ident, h)) &&
1058                     (p->flag & ISSET)) {
1059                         /*
1060                          * this still points to the same character as the
1061                          * ungetsc'd terminator from above
1062                          */
1063                         const char *cp = source->str;
1064
1065                         /* prefer POSIX but not Korn functions over aliases */
1066                         while (*cp == ' ' || *cp == '\t')
1067                                 /*
1068                                  * this is like getsc() without skipping
1069                                  * over Source boundaries (including not
1070                                  * parsing ungetsc'd characters that got
1071                                  * pushed into an SREREAD) which is what
1072                                  * we want here anyway: find out whether
1073                                  * the alias name is followed by a POSIX
1074                                  * function definition (only the opening
1075                                  * parenthesis is checked though)
1076                                  */
1077                                 ++cp;
1078                         /* prefer functions over aliases */
1079                         if (cp[0] != '(' || cp[1] != ')') {
1080                                 Source *s = source;
1081
1082                                 while (s && (s->flags & SF_HASALIAS))
1083                                         if (s->u.tblp == p)
1084                                                 return (LWORD);
1085                                         else
1086                                                 s = s->next;
1087                                 /* push alias expansion */
1088                                 s = pushs(SALIAS, source->areap);
1089                                 s->start = s->str = p->val.s;
1090                                 s->u.tblp = p;
1091                                 s->flags |= SF_HASALIAS;
1092                                 s->next = source;
1093                                 if (source->type == SEOF) {
1094                                         /* prevent infinite recursion at EOS */
1095                                         source->u.tblp = p;
1096                                         source->flags |= SF_HASALIAS;
1097                                 }
1098                                 source = s;
1099                                 afree(yylval.cp, ATEMP);
1100                                 goto Again;
1101                         }
1102                 }
1103         }
1104
1105         return (LWORD);
1106 }
1107
1108 static void
1109 gethere(bool iseof)
1110 {
1111         struct ioword **p;
1112
1113         for (p = heres; p < herep; p++)
1114                 if (iseof && !((*p)->flag & IOHERESTR))
1115                         /* only here strings at EOF */
1116                         return;
1117                 else
1118                         readhere(*p);
1119         herep = heres;
1120 }
1121
1122 /*
1123  * read "<<word" text into temp file
1124  */
1125
1126 static void
1127 readhere(struct ioword *iop)
1128 {
1129         int c;
1130         const char *eof, *eofp;
1131         XString xs;
1132         char *xp;
1133         int xpos;
1134
1135         if (iop->flag & IOHERESTR) {
1136                 /* process the here string */
1137                 iop->heredoc = xp = evalstr(iop->delim, DOBLANK);
1138                 xpos = strlen(xp) - 1;
1139                 memmove(xp, xp + 1, xpos);
1140                 xp[xpos] = '\n';
1141                 return;
1142         }
1143
1144         eof = iop->flag & IONDELIM ? "<<" : evalstr(iop->delim, 0);
1145
1146         if (!(iop->flag & IOEVAL))
1147                 ignore_backslash_newline++;
1148
1149         Xinit(xs, xp, 256, ATEMP);
1150
1151  heredoc_read_line:
1152         /* beginning of line */
1153         eofp = eof;
1154         xpos = Xsavepos(xs, xp);
1155         if (iop->flag & IOSKIP) {
1156                 /* skip over leading tabs */
1157                 while ((c = getsc()) == '\t')
1158                         /* nothing */;
1159                 goto heredoc_parse_char;
1160         }
1161  heredoc_read_char:
1162         c = getsc();
1163  heredoc_parse_char:
1164         /* compare with here document marker */
1165         if (!*eofp) {
1166                 /* end of here document marker, what to do? */
1167                 switch (c) {
1168                 case /*(*/ ')':
1169                         if (!subshell_nesting_type)
1170                                 /*-
1171                                  * not allowed outside $(...) or (...)
1172                                  * => mismatch
1173                                  */
1174                                 break;
1175                         /* allow $(...) or (...) to close here */
1176                         ungetsc(/*(*/ ')');
1177                         /* FALLTHROUGH */
1178                 case 0:
1179                         /*
1180                          * Allow EOF here to commands without trailing
1181                          * newlines (mksh -c '...') will work as well.
1182                          */
1183                 case '\n':
1184                         /* Newline terminates here document marker */
1185                         goto heredoc_found_terminator;
1186                 }
1187         } else if (c == *eofp++)
1188                 /* store; then read and compare next character */
1189                 goto heredoc_store_and_loop;
1190         /* nope, mismatch; read until end of line */
1191         while (c != '\n') {
1192                 if (!c)
1193                         /* oops, reached EOF */
1194                         yyerror("%s '%s' unclosed\n", "here document", eof);
1195                 /* store character */
1196                 Xcheck(xs, xp);
1197                 Xput(xs, xp, c);
1198                 /* read next character */
1199                 c = getsc();
1200         }
1201         /* we read a newline as last character */
1202  heredoc_store_and_loop:
1203         /* store character */
1204         Xcheck(xs, xp);
1205         Xput(xs, xp, c);
1206         if (c == '\n')
1207                 goto heredoc_read_line;
1208         goto heredoc_read_char;
1209
1210  heredoc_found_terminator:
1211         /* jump back to saved beginning of line */
1212         xp = Xrestpos(xs, xp, xpos);
1213         /* terminate, close and store */
1214         Xput(xs, xp, '\0');
1215         iop->heredoc = Xclose(xs, xp);
1216
1217         if (!(iop->flag & IOEVAL))
1218                 ignore_backslash_newline--;
1219 }
1220
1221 void
1222 yyerror(const char *fmt, ...)
1223 {
1224         va_list va;
1225
1226         /* pop aliases and re-reads */
1227         while (source->type == SALIAS || source->type == SREREAD)
1228                 source = source->next;
1229         /* zap pending input */
1230         source->str = null;
1231
1232         error_prefix(true);
1233         va_start(va, fmt);
1234         shf_vfprintf(shl_out, fmt, va);
1235         va_end(va);
1236         errorfz();
1237 }
1238
1239 /*
1240  * input for yylex with alias expansion
1241  */
1242
1243 Source *
1244 pushs(int type, Area *areap)
1245 {
1246         Source *s;
1247
1248         s = alloc(sizeof(Source), areap);
1249         memset(s, 0, sizeof(Source));
1250         s->type = type;
1251         s->str = null;
1252         s->areap = areap;
1253         if (type == SFILE || type == SSTDIN)
1254                 XinitN(s->xs, 256, s->areap);
1255         return (s);
1256 }
1257
1258 static int
1259 getsc_uu(void)
1260 {
1261         Source *s = source;
1262         int c;
1263
1264         while ((c = *s->str++) == 0) {
1265                 /* return 0 for EOF by default */
1266                 s->str = NULL;
1267                 switch (s->type) {
1268                 case SEOF:
1269                         s->str = null;
1270                         return (0);
1271
1272                 case SSTDIN:
1273                 case SFILE:
1274                         getsc_line(s);
1275                         break;
1276
1277                 case SWSTR:
1278                         break;
1279
1280                 case SSTRING:
1281                 case SSTRINGCMDLINE:
1282                         break;
1283
1284                 case SWORDS:
1285                         s->start = s->str = *s->u.strv++;
1286                         s->type = SWORDSEP;
1287                         break;
1288
1289                 case SWORDSEP:
1290                         if (*s->u.strv == NULL) {
1291                                 s->start = s->str = "\n";
1292                                 s->type = SEOF;
1293                         } else {
1294                                 s->start = s->str = " ";
1295                                 s->type = SWORDS;
1296                         }
1297                         break;
1298
1299                 case SALIAS:
1300                         if (s->flags & SF_ALIASEND) {
1301                                 /* pass on an unused SF_ALIAS flag */
1302                                 source = s->next;
1303                                 source->flags |= s->flags & SF_ALIAS;
1304                                 s = source;
1305                         } else if (*s->u.tblp->val.s &&
1306                             (c = strnul(s->u.tblp->val.s)[-1], ksh_isspace(c))) {
1307                                 /* pop source stack */
1308                                 source = s = s->next;
1309                                 /*
1310                                  * Note that this alias ended with a
1311                                  * space, enabling alias expansion on
1312                                  * the following word.
1313                                  */
1314                                 s->flags |= SF_ALIAS;
1315                         } else {
1316                                 /*
1317                                  * At this point, we need to keep the current
1318                                  * alias in the source list so recursive
1319                                  * aliases can be detected and we also need to
1320                                  * return the next character. Do this by
1321                                  * temporarily popping the alias to get the
1322                                  * next character and then put it back in the
1323                                  * source list with the SF_ALIASEND flag set.
1324                                  */
1325                                 /* pop source stack */
1326                                 source = s->next;
1327                                 source->flags |= s->flags & SF_ALIAS;
1328                                 c = getsc_uu();
1329                                 if (c) {
1330                                         s->flags |= SF_ALIASEND;
1331                                         s->ugbuf[0] = c; s->ugbuf[1] = '\0';
1332                                         s->start = s->str = s->ugbuf;
1333                                         s->next = source;
1334                                         source = s;
1335                                 } else {
1336                                         s = source;
1337                                         /* avoid reading EOF twice */
1338                                         s->str = NULL;
1339                                         break;
1340                                 }
1341                         }
1342                         continue;
1343
1344                 case SREREAD:
1345                         if (s->start != s->ugbuf)
1346                                 /* yuck */
1347                                 afree(s->u.freeme, ATEMP);
1348                         source = s = s->next;
1349                         continue;
1350                 }
1351                 if (s->str == NULL) {
1352                         s->type = SEOF;
1353                         s->start = s->str = null;
1354                         return ('\0');
1355                 }
1356                 if (s->flags & SF_ECHO) {
1357                         shf_puts(s->str, shl_out);
1358                         shf_flush(shl_out);
1359                 }
1360         }
1361         return (c);
1362 }
1363
1364 static void
1365 getsc_line(Source *s)
1366 {
1367         char *xp = Xstring(s->xs, xp), *cp;
1368         bool interactive = Flag(FTALKING) && s->type == SSTDIN;
1369         bool have_tty = tobool(interactive && (s->flags & SF_TTY));
1370
1371         /* Done here to ensure nothing odd happens when a timeout occurs */
1372         XcheckN(s->xs, xp, LINE);
1373         *xp = '\0';
1374         s->start = s->str = xp;
1375
1376         if (have_tty && ksh_tmout) {
1377                 ksh_tmout_state = TMOUT_READING;
1378                 alarm(ksh_tmout);
1379         }
1380         if (interactive)
1381                 change_winsz();
1382 #ifndef MKSH_NO_CMDLINE_EDITING
1383         if (have_tty && (
1384 #if !MKSH_S_NOVI
1385             Flag(FVI) ||
1386 #endif
1387             Flag(FEMACS) || Flag(FGMACS))) {
1388                 int nread;
1389
1390                 nread = x_read(xp);
1391                 if (nread < 0)
1392                         /* read error */
1393                         nread = 0;
1394                 xp[nread] = '\0';
1395                 xp += nread;
1396         } else
1397 #endif
1398           {
1399                 if (interactive)
1400                         pprompt(prompt, 0);
1401                 else
1402                         s->line++;
1403
1404                 while (/* CONSTCOND */ 1) {
1405                         char *p = shf_getse(xp, Xnleft(s->xs, xp), s->u.shf);
1406
1407                         if (!p && shf_error(s->u.shf) &&
1408                             shf_errno(s->u.shf) == EINTR) {
1409                                 shf_clearerr(s->u.shf);
1410                                 if (trap)
1411                                         runtraps(0);
1412                                 continue;
1413                         }
1414                         if (!p || (xp = p, xp[-1] == '\n'))
1415                                 break;
1416                         /* double buffer size */
1417                         /* move past NUL so doubling works... */
1418                         xp++;
1419                         XcheckN(s->xs, xp, Xlength(s->xs, xp));
1420                         /* ...and move back again */
1421                         xp--;
1422                 }
1423                 /*
1424                  * flush any unwanted input so other programs/builtins
1425                  * can read it. Not very optimal, but less error prone
1426                  * than flushing else where, dealing with redirections,
1427                  * etc.
1428                  * TODO: reduce size of shf buffer (~128?) if SSTDIN
1429                  */
1430                 if (s->type == SSTDIN)
1431                         shf_flush(s->u.shf);
1432         }
1433         /*
1434          * XXX: temporary kludge to restore source after a
1435          * trap may have been executed.
1436          */
1437         source = s;
1438         if (have_tty && ksh_tmout) {
1439                 ksh_tmout_state = TMOUT_EXECUTING;
1440                 alarm(0);
1441         }
1442         cp = Xstring(s->xs, xp);
1443         s->start = s->str = cp;
1444         strip_nuls(Xstring(s->xs, xp), Xlength(s->xs, xp));
1445         /* Note: if input is all nulls, this is not eof */
1446         if (Xlength(s->xs, xp) == 0) {
1447                 /* EOF */
1448                 if (s->type == SFILE)
1449                         shf_fdclose(s->u.shf);
1450                 s->str = NULL;
1451         } else if (interactive && *s->str &&
1452             (cur_prompt != PS1 || !ctype(*s->str, C_IFS | C_IFSWS))) {
1453                 histsave(&s->line, s->str, true, true);
1454 #if !defined(MKSH_SMALL) && HAVE_PERSISTENT_HISTORY
1455         } else if (interactive && cur_prompt == PS1) {
1456                 cp = Xstring(s->xs, xp);
1457                 while (*cp && ctype(*cp, C_IFSWS))
1458                         ++cp;
1459                 if (!*cp)
1460                         histsync();
1461 #endif
1462         }
1463         if (interactive)
1464                 set_prompt(PS2, NULL);
1465 }
1466
1467 void
1468 set_prompt(int to, Source *s)
1469 {
1470         cur_prompt = to;
1471
1472         switch (to) {
1473         /* command */
1474         case PS1:
1475                 /*
1476                  * Substitute ! and !! here, before substitutions are done
1477                  * so ! in expanded variables are not expanded.
1478                  * NOTE: this is not what AT&T ksh does (it does it after
1479                  * substitutions, POSIX doesn't say which is to be done.
1480                  */
1481                 {
1482                         struct shf *shf;
1483                         char * volatile ps1;
1484                         Area *saved_atemp;
1485
1486                         ps1 = str_val(global("PS1"));
1487                         shf = shf_sopen(NULL, strlen(ps1) * 2,
1488                             SHF_WR | SHF_DYNAMIC, NULL);
1489                         while (*ps1)
1490                                 if (*ps1 != '!' || *++ps1 == '!')
1491                                         shf_putchar(*ps1++, shf);
1492                                 else
1493                                         shf_fprintf(shf, "%d",
1494                                                 s ? s->line + 1 : 0);
1495                         ps1 = shf_sclose(shf);
1496                         saved_atemp = ATEMP;
1497                         newenv(E_ERRH);
1498                         if (kshsetjmp(e->jbuf)) {
1499                                 prompt = safe_prompt;
1500                                 /*
1501                                  * Don't print an error - assume it has already
1502                                  * been printed. Reason is we may have forked
1503                                  * to run a command and the child may be
1504                                  * unwinding its stack through this code as it
1505                                  * exits.
1506                                  */
1507                         } else {
1508                                 char *cp = substitute(ps1, 0);
1509                                 strdupx(prompt, cp, saved_atemp);
1510                         }
1511                         quitenv(NULL);
1512                 }
1513                 break;
1514         /* command continuation */
1515         case PS2:
1516                 prompt = str_val(global("PS2"));
1517                 break;
1518         }
1519 }
1520
1521 int
1522 pprompt(const char *cp, int ntruncate)
1523 {
1524         int columns = 0, lines = 0;
1525         bool indelimit = false;
1526         char delimiter = 0;
1527
1528         /*
1529          * Undocumented AT&T ksh feature:
1530          * If the second char in the prompt string is \r then the first
1531          * char is taken to be a non-printing delimiter and any chars
1532          * between two instances of the delimiter are not considered to
1533          * be part of the prompt length
1534          */
1535         if (*cp && cp[1] == '\r') {
1536                 delimiter = *cp;
1537                 cp += 2;
1538         }
1539         for (; *cp; cp++) {
1540                 if (indelimit && *cp != delimiter)
1541                         ;
1542                 else if (*cp == '\n' || *cp == '\r') {
1543                         lines += columns / x_cols + ((*cp == '\n') ? 1 : 0);
1544                         columns = 0;
1545                 } else if (*cp == '\t') {
1546                         columns = (columns | 7) + 1;
1547                 } else if (*cp == '\b') {
1548                         if (columns > 0)
1549                                 columns--;
1550                 } else if (*cp == delimiter)
1551                         indelimit = !indelimit;
1552                 else if (UTFMODE && ((unsigned char)*cp > 0x7F)) {
1553                         const char *cp2;
1554                         columns += utf_widthadj(cp, &cp2);
1555                         if (indelimit ||
1556                             (ntruncate < (x_cols * lines + columns)))
1557                                 shf_write(cp, cp2 - cp, shl_out);
1558                         cp = cp2 - /* loop increment */ 1;
1559                         continue;
1560                 } else
1561                         columns++;
1562                 if ((*cp != delimiter) &&
1563                     (indelimit || (ntruncate < (x_cols * lines + columns))))
1564                         shf_putc(*cp, shl_out);
1565         }
1566         shf_flush(shl_out);
1567         return (x_cols * lines + columns);
1568 }
1569
1570 /*
1571  * Read the variable part of a ${...} expression (i.e. up to but not
1572  * including the :[-+?=#%] or close-brace).
1573  */
1574 static char *
1575 get_brace_var(XString *wsp, char *wp)
1576 {
1577         char c;
1578         enum parse_state {
1579                 PS_INITIAL, PS_SAW_HASH, PS_IDENT,
1580                 PS_NUMBER, PS_VAR1
1581         } state = PS_INITIAL;
1582
1583         while (/* CONSTCOND */ 1) {
1584                 c = getsc();
1585                 /* State machine to figure out where the variable part ends. */
1586                 switch (state) {
1587                 case PS_INITIAL:
1588                         if (c == '#' || c == '!' || c == '%') {
1589                                 state = PS_SAW_HASH;
1590                                 break;
1591                         }
1592                         /* FALLTHROUGH */
1593                 case PS_SAW_HASH:
1594                         if (ksh_isalphx(c))
1595                                 state = PS_IDENT;
1596                         else if (ksh_isdigit(c))
1597                                 state = PS_NUMBER;
1598                         else if (c == '#') {
1599                                 if (state == PS_SAW_HASH) {
1600                                         char c2;
1601
1602                                         c2 = getsc();
1603                                         ungetsc(c2);
1604                                         if (c2 != /*{*/ '}') {
1605                                                 ungetsc(c);
1606                                                 goto out;
1607                                         }
1608                                 }
1609                                 state = PS_VAR1;
1610                         } else if (ctype(c, C_VAR1))
1611                                 state = PS_VAR1;
1612                         else
1613                                 goto out;
1614                         break;
1615                 case PS_IDENT:
1616                         if (!ksh_isalnux(c)) {
1617                                 if (c == '[') {
1618                                         char *tmp, *p;
1619
1620                                         if (!arraysub(&tmp))
1621                                                 yyerror("missing ]\n");
1622                                         *wp++ = c;
1623                                         for (p = tmp; *p; ) {
1624                                                 Xcheck(*wsp, wp);
1625                                                 *wp++ = *p++;
1626                                         }
1627                                         afree(tmp, ATEMP);
1628                                         /* the ] */
1629                                         c = getsc();
1630                                 }
1631                                 goto out;
1632                         }
1633                         break;
1634                 case PS_NUMBER:
1635                         if (!ksh_isdigit(c))
1636                                 goto out;
1637                         break;
1638                 case PS_VAR1:
1639                         goto out;
1640                 }
1641                 Xcheck(*wsp, wp);
1642                 *wp++ = c;
1643         }
1644  out:
1645         /* end of variable part */
1646         *wp++ = '\0';
1647         ungetsc(c);
1648         return (wp);
1649 }
1650
1651 /*
1652  * Save an array subscript - returns true if matching bracket found, false
1653  * if eof or newline was found.
1654  * (Returned string double null terminated)
1655  */
1656 static bool
1657 arraysub(char **strp)
1658 {
1659         XString ws;
1660         char *wp, c;
1661         /* we are just past the initial [ */
1662         unsigned int depth = 1;
1663
1664         Xinit(ws, wp, 32, ATEMP);
1665
1666         do {
1667                 c = getsc();
1668                 Xcheck(ws, wp);
1669                 *wp++ = c;
1670                 if (c == '[')
1671                         depth++;
1672                 else if (c == ']')
1673                         depth--;
1674         } while (depth > 0 && c && c != '\n');
1675
1676         *wp++ = '\0';
1677         *strp = Xclose(ws, wp);
1678
1679         return (tobool(depth == 0));
1680 }
1681
1682 /* Unget a char: handles case when we are already at the start of the buffer */
1683 static void
1684 ungetsc(int c)
1685 {
1686         struct sretrace_info *rp = retrace_info;
1687
1688         if (backslash_skip)
1689                 backslash_skip--;
1690         /* Don't unget EOF... */
1691         if (source->str == null && c == '\0')
1692                 return;
1693         while (rp) {
1694                 if (Xlength(rp->xs, rp->xp))
1695                         rp->xp--;
1696                 rp = rp->next;
1697         }
1698         ungetsc_i(c);
1699 }
1700 static void
1701 ungetsc_i(int c)
1702 {
1703         if (source->str > source->start)
1704                 source->str--;
1705         else {
1706                 Source *s;
1707
1708                 s = pushs(SREREAD, source->areap);
1709                 s->ugbuf[0] = c; s->ugbuf[1] = '\0';
1710                 s->start = s->str = s->ugbuf;
1711                 s->next = source;
1712                 source = s;
1713         }
1714 }
1715
1716
1717 /* Called to get a char that isn't a \newline sequence. */
1718 static int
1719 getsc_bn(void)
1720 {
1721         int c, c2;
1722
1723         if (ignore_backslash_newline)
1724                 return (o_getsc_u());
1725
1726         if (backslash_skip == 1) {
1727                 backslash_skip = 2;
1728                 return (o_getsc_u());
1729         }
1730
1731         backslash_skip = 0;
1732
1733         while (/* CONSTCOND */ 1) {
1734                 c = o_getsc_u();
1735                 if (c == '\\') {
1736                         if ((c2 = o_getsc_u()) == '\n')
1737                                 /* ignore the \newline; get the next char... */
1738                                 continue;
1739                         ungetsc_i(c2);
1740                         backslash_skip = 1;
1741                 }
1742                 return (c);
1743         }
1744 }
1745
1746 void
1747 yyskiputf8bom(void)
1748 {
1749         int c;
1750
1751         if ((unsigned char)(c = o_getsc_u()) != 0xEF) {
1752                 ungetsc_i(c);
1753                 return;
1754         }
1755         if ((unsigned char)(c = o_getsc_u()) != 0xBB) {
1756                 ungetsc_i(c);
1757                 ungetsc_i(0xEF);
1758                 return;
1759         }
1760         if ((unsigned char)(c = o_getsc_u()) != 0xBF) {
1761                 ungetsc_i(c);
1762                 ungetsc_i(0xBB);
1763                 ungetsc_i(0xEF);
1764                 return;
1765         }
1766         UTFMODE |= 8;
1767 }
1768
1769 static Lex_state *
1770 push_state_i(State_info *si, Lex_state *old_end)
1771 {
1772         Lex_state *news = alloc2(STATE_BSIZE, sizeof(Lex_state), ATEMP);
1773
1774         news[0].ls_base = old_end;
1775         si->base = &news[0];
1776         si->end = &news[STATE_BSIZE];
1777         return (&news[1]);
1778 }
1779
1780 static Lex_state *
1781 pop_state_i(State_info *si, Lex_state *old_end)
1782 {
1783         Lex_state *old_base = si->base;
1784
1785         si->base = old_end->ls_base - STATE_BSIZE;
1786         si->end = old_end->ls_base;
1787
1788         afree(old_base, ATEMP);
1789
1790         return (si->base + STATE_BSIZE - 1);
1791 }
1792
1793 static int
1794 s_get(void)
1795 {
1796         return (getsc());
1797 }
1798
1799 static void
1800 s_put(int c)
1801 {
1802         ungetsc(c);
1803 }