src/lex.c

   1 /*      $OpenBSD: lex.c,v 1.49 2013/12/17 16:37:06 deraadt Exp $        */
   2
   3 /*-
   4  * Copyright (c) 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
   5  *               2011, 2012, 2013, 2014
   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 1.193 2014/06/29 11:28:28 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(s) do {                                   \
 163         struct sretrace_info *ri;                               \
 164                                                                 \
 165         PUSH_STATE(s);                                          \
 166         statep->ls_start = Xsavepos(ws, wp);                    \
 167         ri = alloc(sizeof(struct sretrace_info), ATEMP);        \
 168         Xinit(ri->xs, ri->xp, 64, ATEMP);                       \
 169         ri->next = retrace_info;                                \
 170         retrace_info = ri;                                      \
 171 } while (/* CONSTCOND */ 0)
 172
 173 #define POP_SRETRACE()  do {                                    \
 174         wp = Xrestpos(ws, wp, statep->ls_start);                \
 175         *retrace_info->xp = '\0';                               \
 176         sp = Xstring(retrace_info->xs, retrace_info->xp);       \
 177         dp = (void *)retrace_info;                              \
 178         retrace_info = retrace_info->next;                      \
 179         afree(dp, ATEMP);                                       \
 180         POP_STATE();                                            \
 181 } while (/* CONSTCOND */ 0)
 182
 183 /**
 184  * Lexical analyser
 185  *
 186  * tokens are not regular expressions, they are LL(1).
 187  * for example, "${var:-${PWD}}", and "$(size $(whence ksh))".
 188  * hence the state stack. Note "$(...)" are now parsed recursively.
 189  */
 190
 191 int
 192 yylex(int cf)
 193 {
 194         Lex_state states[STATE_BSIZE], *statep, *s2, *base;
 195         State_info state_info;
 196         int c, c2, state;
 197         size_t cz;
 198         XString ws;             /* expandable output word */
 199         char *wp;               /* output word pointer */
 200         char *sp, *dp;
 201
 202  Again:
 203         states[0].type = SINVALID;
 204         states[0].ls_base = NULL;
 205         statep = &states[1];
 206         state_info.base = states;
 207         state_info.end = &state_info.base[STATE_BSIZE];
 208
 209         Xinit(ws, wp, 64, ATEMP);
 210
 211         backslash_skip = 0;
 212         ignore_backslash_newline = 0;
 213
 214         if (cf & ONEWORD)
 215                 state = SWORD;
 216         else if (cf & LETEXPR) {
 217                 /* enclose arguments in (double) quotes */
 218                 *wp++ = OQUOTE;
 219                 state = SLETPAREN;
 220                 statep->nparen = 0;
 221         } else {
 222                 /* normal lexing */
 223                 state = (cf & HEREDELIM) ? SHEREDELIM : SBASE;
 224                 while ((c = getsc()) == ' ' || c == '\t')
 225                         ;
 226                 if (c == '#') {
 227                         ignore_backslash_newline++;
 228                         while ((c = getsc()) != '\0' && c != '\n')
 229                                 ;
 230                         ignore_backslash_newline--;
 231                 }
 232                 ungetsc(c);
 233         }
 234         if (source->flags & SF_ALIAS) {
 235                 /* trailing ' ' in alias definition */
 236                 source->flags &= ~SF_ALIAS;
 237                 cf |= ALIAS;
 238         }
 239
 240         /* Initial state: one of SWORD SLETPAREN SHEREDELIM SBASE */
 241         statep->type = state;
 242
 243         /* check for here string */
 244         if (state == SHEREDELIM) {
 245                 c = getsc();
 246                 if (c == '<') {
 247                         state = SHEREDELIM;
 248                         while ((c = getsc()) == ' ' || c == '\t')
 249                                 ;
 250                         ungetsc(c);
 251                         c = '<';
 252                         goto accept_nonword;
 253                 }
 254                 ungetsc(c);
 255         }
 256
 257         /* collect non-special or quoted characters to form word */
 258         while (!((c = getsc()) == 0 ||
 259             ((state == SBASE || state == SHEREDELIM) && ctype(c, C_LEX1)))) {
 260                 if (state == SBASE &&
 261                     subshell_nesting_type == /*{*/ '}' &&
 262                     c == /*{*/ '}')
 263                         /* possibly end ${ :;} */
 264                         break;
 265  accept_nonword:
 266                 Xcheck(ws, wp);
 267                 switch (state) {
 268                 case SADELIM:
 269                         if (c == '(')
 270                                 statep->nparen++;
 271                         else if (c == ')')
 272                                 statep->nparen--;
 273                         else if (statep->nparen == 0 && (c == /*{*/ '}' ||
 274                             c == (int)statep->ls_adelim.delimiter)) {
 275                                 *wp++ = ADELIM;
 276                                 *wp++ = c;
 277                                 if (c == /*{*/ '}' || --statep->ls_adelim.num == 0)
 278                                         POP_STATE();
 279                                 if (c == /*{*/ '}')
 280                                         POP_STATE();
 281                                 break;
 282                         }
 283                         /* FALLTHROUGH */
 284                 case SBASE:
 285                         if (c == '[' && (cf & (VARASN|ARRAYVAR))) {
 286                                 /* temporary */
 287                                 *wp = EOS;
 288                                 if (is_wdvarname(Xstring(ws, wp), false)) {
 289                                         char *p, *tmp;
 290
 291                                         if (arraysub(&tmp)) {
 292                                                 *wp++ = CHAR;
 293                                                 *wp++ = c;
 294                                                 for (p = tmp; *p; ) {
 295                                                         Xcheck(ws, wp);
 296                                                         *wp++ = CHAR;
 297                                                         *wp++ = *p++;
 298                                                 }
 299                                                 afree(tmp, ATEMP);
 300                                                 break;
 301                                         } else {
 302                                                 Source *s;
 303
 304                                                 s = pushs(SREREAD,
 305                                                     source->areap);
 306                                                 s->start = s->str =
 307                                                     s->u.freeme = tmp;
 308                                                 s->next = source;
 309                                                 source = s;
 310                                         }
 311                                 }
 312                                 *wp++ = CHAR;
 313                                 *wp++ = c;
 314                                 break;
 315                         }
 316                         /* FALLTHROUGH */
 317  Sbase1:                /* includes *(...|...) pattern (*+?@!) */
 318                         if (c == '*' || c == '@' || c == '+' || c == '?' ||
 319                             c == '!') {
 320                                 c2 = getsc();
 321                                 if (c2 == '(' /*)*/ ) {
 322                                         *wp++ = OPAT;
 323                                         *wp++ = c;
 324                                         PUSH_STATE(SPATTERN);
 325                                         break;
 326                                 }
 327                                 ungetsc(c2);
 328                         }
 329                         /* FALLTHROUGH */
 330  Sbase2:                /* doesn't include *(...|...) pattern (*+?@!) */
 331                         switch (c) {
 332                         case '\\':
 333  getsc_qchar:
 334                                 if ((c = getsc())) {
 335                                         /* trailing \ is lost */
 336                                         *wp++ = QCHAR;
 337                                         *wp++ = c;
 338                                 }
 339                                 break;
 340                         case '\'':
 341  open_ssquote_unless_heredoc:
 342                                 if ((cf & HEREDOC))
 343                                         goto store_char;
 344                                 *wp++ = OQUOTE;
 345                                 ignore_backslash_newline++;
 346                                 PUSH_STATE(SSQUOTE);
 347                                 break;
 348                         case '"':
 349  open_sdquote:
 350                                 *wp++ = OQUOTE;
 351                                 PUSH_STATE(SDQUOTE);
 352                                 break;
 353                         case '$':
 354                                 /*
 355                                  * processing of dollar sign belongs into
 356                                  * Subst, except for those which can open
 357                                  * a string: $'…' and $"…"
 358                                  */
 359  subst_dollar_ex:
 360                                 c = getsc();
 361                                 switch (c) {
 362                                 case '"':
 363                                         goto open_sdquote;
 364                                 case '\'':
 365                                         goto open_sequote;
 366                                 default:
 367                                         goto SubstS;
 368                                 }
 369                         default:
 370                                 goto Subst;
 371                         }
 372                         break;
 373
 374  Subst:
 375                         switch (c) {
 376                         case '\\':
 377                                 c = getsc();
 378                                 switch (c) {
 379                                 case '"':
 380                                         if ((cf & HEREDOC))
 381                                                 goto heredocquote;
 382                                         /* FALLTHROUGH */
 383                                 case '\\':
 384                                 case '$': case '`':
 385  store_qchar:
 386                                         *wp++ = QCHAR;
 387                                         *wp++ = c;
 388                                         break;
 389                                 default:
 390  heredocquote:
 391                                         Xcheck(ws, wp);
 392                                         if (c) {
 393                                                 /* trailing \ is lost */
 394                                                 *wp++ = CHAR;
 395                                                 *wp++ = '\\';
 396                                                 *wp++ = CHAR;
 397                                                 *wp++ = c;
 398                                         }
 399                                         break;
 400                                 }
 401                                 break;
 402                         case '$':
 403                                 c = getsc();
 404  SubstS:
 405                                 if (c == '(') /*)*/ {
 406                                         c = getsc();
 407                                         if (c == '(') /*)*/ {
 408                                                 *wp++ = EXPRSUB;
 409                                                 PUSH_SRETRACE(SASPAREN);
 410                                                 statep->nparen = 2;
 411                                                 *retrace_info->xp++ = '(';
 412                                         } else {
 413                                                 ungetsc(c);
 414  subst_command:
 415                                                 c = COMSUB;
 416  subst_command2:
 417                                                 sp = yyrecursive(c);
 418                                                 cz = strlen(sp) + 1;
 419                                                 XcheckN(ws, wp, cz);
 420                                                 *wp++ = c;
 421                                                 memcpy(wp, sp, cz);
 422                                                 wp += cz;
 423                                         }
 424                                 } else if (c == '{') /*}*/ {
 425                                         if ((c = getsc()) == '|') {
 426                                                 /*
 427                                                  * non-subenvironment
 428                                                  * value substitution
 429                                                  */
 430                                                 c = VALSUB;
 431                                                 goto subst_command2;
 432                                         } else if (ctype(c, C_IFSWS)) {
 433                                                 /*
 434                                                  * non-subenvironment
 435                                                  * "command" substitution
 436                                                  */
 437                                                 c = FUNSUB;
 438                                                 goto subst_command2;
 439                                         }
 440                                         ungetsc(c);
 441                                         *wp++ = OSUBST;
 442                                         *wp++ = '{'; /*}*/
 443                                         wp = get_brace_var(&ws, wp);
 444                                         c = getsc();
 445                                         /* allow :# and :% (ksh88 compat) */
 446                                         if (c == ':') {
 447                                                 *wp++ = CHAR;
 448                                                 *wp++ = c;
 449                                                 c = getsc();
 450                                                 if (c == ':') {
 451                                                         *wp++ = CHAR;
 452                                                         *wp++ = '0';
 453                                                         *wp++ = ADELIM;
 454                                                         *wp++ = ':';
 455                                                         PUSH_STATE(SBRACE);
 456                                                         PUSH_STATE(SADELIM);
 457                                                         statep->ls_adelim.delimiter = ':';
 458                                                         statep->ls_adelim.num = 1;
 459                                                         statep->nparen = 0;
 460                                                         break;
 461                                                 } else if (ksh_isdigit(c) ||
 462                                                     c == '('/*)*/ || c == ' ' ||
 463                                                     /*XXX what else? */
 464                                                     c == '$') {
 465                                                         /* substring subst. */
 466                                                         if (c != ' ') {
 467                                                                 *wp++ = CHAR;
 468                                                                 *wp++ = ' ';
 469                                                         }
 470                                                         ungetsc(c);
 471                                                         PUSH_STATE(SBRACE);
 472                                                         PUSH_STATE(SADELIM);
 473                                                         statep->ls_adelim.delimiter = ':';
 474                                                         statep->ls_adelim.num = 2;
 475                                                         statep->nparen = 0;
 476                                                         break;
 477                                                 }
 478                                         } else if (c == '/') {
 479                                                 *wp++ = CHAR;
 480                                                 *wp++ = c;
 481                                                 if ((c = getsc()) == '/') {
 482                                                         *wp++ = ADELIM;
 483                                                         *wp++ = c;
 484                                                 } else
 485                                                         ungetsc(c);
 486                                                 PUSH_STATE(SBRACE);
 487                                                 PUSH_STATE(SADELIM);
 488                                                 statep->ls_adelim.delimiter = '/';
 489                                                 statep->ls_adelim.num = 1;
 490                                                 statep->nparen = 0;
 491                                                 break;
 492                                         }
 493                                         /*
 494                                          * If this is a trim operation,
 495                                          * treat (,|,) specially in STBRACE.
 496                                          */
 497                                         if (ctype(c, C_SUBOP2)) {
 498                                                 ungetsc(c);
 499                                                 if (Flag(FSH))
 500                                                         PUSH_STATE(STBRACEBOURNE);
 501                                                 else
 502                                                         PUSH_STATE(STBRACEKORN);
 503                                         } else {
 504                                                 ungetsc(c);
 505                                                 if (state == SDQUOTE ||
 506                                                     state == SQBRACE)
 507                                                         PUSH_STATE(SQBRACE);
 508                                                 else
 509                                                         PUSH_STATE(SBRACE);
 510                                         }
 511                                 } else if (ksh_isalphx(c)) {
 512                                         *wp++ = OSUBST;
 513                                         *wp++ = 'X';
 514                                         do {
 515                                                 Xcheck(ws, wp);
 516                                                 *wp++ = c;
 517                                                 c = getsc();
 518                                         } while (ksh_isalnux(c));
 519                                         *wp++ = '\0';
 520                                         *wp++ = CSUBST;
 521                                         *wp++ = 'X';
 522                                         ungetsc(c);
 523                                 } else if (ctype(c, C_VAR1 | C_DIGIT)) {
 524                                         Xcheck(ws, wp);
 525                                         *wp++ = OSUBST;
 526                                         *wp++ = 'X';
 527                                         *wp++ = c;
 528                                         *wp++ = '\0';
 529                                         *wp++ = CSUBST;
 530                                         *wp++ = 'X';
 531                                 } else {
 532                                         *wp++ = CHAR;
 533                                         *wp++ = '$';
 534                                         ungetsc(c);
 535                                 }
 536                                 break;
 537                         case '`':
 538  subst_gravis:
 539                                 PUSH_STATE(SBQUOTE);
 540                                 *wp++ = COMSUB;
 541                                 /*
 542                                  * Need to know if we are inside double quotes
 543                                  * since sh/AT&T-ksh translate the \" to " in
 544                                  * "`...\"...`".
 545                                  * This is not done in POSIX mode (section
 546                                  * 3.2.3, Double Quotes: "The backquote shall
 547                                  * retain its special meaning introducing the
 548                                  * other form of command substitution (see
 549                                  * 3.6.3). The portion of the quoted string
 550                                  * from the initial backquote and the
 551                                  * characters up to the next backquote that
 552                                  * is not preceded by a backslash (having
 553                                  * escape characters removed) defines that
 554                                  * command whose output replaces `...` when
 555                                  * the word is expanded."
 556                                  * Section 3.6.3, Command Substitution:
 557                                  * "Within the backquoted style of command
 558                                  * substitution, backslash shall retain its
 559                                  * literal meaning, except when followed by
 560                                  * $ ` \.").
 561                                  */
 562                                 statep->ls_bool = false;
 563                                 s2 = statep;
 564                                 base = state_info.base;
 565                                 while (/* CONSTCOND */ 1) {
 566                                         for (; s2 != base; s2--) {
 567                                                 if (s2->type == SDQUOTE) {
 568                                                         statep->ls_bool = true;
 569                                                         break;
 570                                                 }
 571                                         }
 572                                         if (s2 != base)
 573                                                 break;
 574                                         if (!(s2 = s2->ls_base))
 575                                                 break;
 576                                         base = s2-- - STATE_BSIZE;
 577                                 }
 578                                 break;
 579                         case QCHAR:
 580                                 if (cf & LQCHAR) {
 581                                         *wp++ = QCHAR;
 582                                         *wp++ = getsc();
 583                                         break;
 584                                 }
 585                                 /* FALLTHROUGH */
 586                         default:
 587  store_char:
 588                                 *wp++ = CHAR;
 589                                 *wp++ = c;
 590                         }
 591                         break;
 592
 593                 case SEQUOTE:
 594                         if (c == '\'') {
 595                                 POP_STATE();
 596                                 *wp++ = CQUOTE;
 597                                 ignore_backslash_newline--;
 598                         } else if (c == '\\') {
 599                                 if ((c2 = unbksl(true, s_get, s_put)) == -1)
 600                                         c2 = s_get();
 601                                 if (c2 == 0)
 602                                         statep->ls_bool = true;
 603                                 if (!statep->ls_bool) {
 604                                         char ts[4];
 605
 606                                         if ((unsigned int)c2 < 0x100) {
 607                                                 *wp++ = QCHAR;
 608                                                 *wp++ = c2;
 609                                         } else {
 610                                                 cz = utf_wctomb(ts, c2 - 0x100);
 611                                                 ts[cz] = 0;
 612                                                 for (cz = 0; ts[cz]; ++cz) {
 613                                                         *wp++ = QCHAR;
 614                                                         *wp++ = ts[cz];
 615                                                 }
 616                                         }
 617                                 }
 618                         } else if (!statep->ls_bool) {
 619                                 *wp++ = QCHAR;
 620                                 *wp++ = c;
 621                         }
 622                         break;
 623
 624                 case SSQUOTE:
 625                         if (c == '\'') {
 626                                 POP_STATE();
 627                                 if ((cf & HEREDOC) || state == SQBRACE)
 628                                         goto store_char;
 629                                 *wp++ = CQUOTE;
 630                                 ignore_backslash_newline--;
 631                         } else {
 632                                 *wp++ = QCHAR;
 633                                 *wp++ = c;
 634                         }
 635                         break;
 636
 637                 case SDQUOTE:
 638                         if (c == '"') {
 639                                 POP_STATE();
 640                                 *wp++ = CQUOTE;
 641                         } else
 642                                 goto Subst;
 643                         break;
 644
 645                 /* $(( ... )) */
 646                 case SASPAREN:
 647                         if (c == '(')
 648                                 statep->nparen++;
 649                         else if (c == ')') {
 650                                 statep->nparen--;
 651                                 if (statep->nparen == 1) {
 652                                         /* end of EXPRSUB */
 653                                         POP_SRETRACE();
 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 '$':
 826                                 if ((c2 = getsc()) == '\'') {
 827  open_sequote:
 828                                         *wp++ = OQUOTE;
 829                                         ignore_backslash_newline++;
 830                                         PUSH_STATE(SEQUOTE);
 831                                         statep->ls_bool = false;
 832                                         break;
 833                                 } else if (c2 == '"') {
 834                                         /* FALLTHROUGH */
 835                         case '"':
 836                                         PUSH_SRETRACE(SHEREDQUOTE);
 837                                         break;
 838                                 }
 839                                 ungetsc(c2);
 840                                 /* FALLTHROUGH */
 841                         default:
 842                                 *wp++ = CHAR;
 843                                 *wp++ = c;
 844                         }
 845                         break;
 846
 847                 /* " in <<, <<-, <<< delimiter */
 848                 case SHEREDQUOTE:
 849                         if (c != '"')
 850                                 goto Subst;
 851                         POP_SRETRACE();
 852                         dp = strnul(sp) - 1;
 853                         /* remove the trailing double quote */
 854                         *dp = '\0';
 855                         /* store the quoted string */
 856                         *wp++ = OQUOTE;
 857                         XcheckN(ws, wp, (dp - sp));
 858                         dp = sp;
 859                         while ((c = *dp++)) {
 860                                 if (c == '\\') {
 861                                         switch ((c = *dp++)) {
 862                                         case '\\':
 863                                         case '"':
 864                                         case '$':
 865                                         case '`':
 866                                                 break;
 867                                         default:
 868                                                 *wp++ = CHAR;
 869                                                 *wp++ = '\\';
 870                                                 break;
 871                                         }
 872                                 }
 873                                 *wp++ = CHAR;
 874                                 *wp++ = c;
 875                         }
 876                         afree(sp, ATEMP);
 877                         *wp++ = CQUOTE;
 878                         state = statep->type = SHEREDELIM;
 879                         break;
 880
 881                 /* in *(...|...) pattern (*+?@!) */
 882                 case SPATTERN:
 883                         if (c == /*(*/ ')') {
 884                                 *wp++ = CPAT;
 885                                 POP_STATE();
 886                         } else if (c == '|') {
 887                                 *wp++ = SPAT;
 888                         } else if (c == '(') {
 889                                 *wp++ = OPAT;
 890                                 /* simile for @ */
 891                                 *wp++ = ' ';
 892                                 PUSH_STATE(SPATTERN);
 893                         } else
 894                                 goto Sbase1;
 895                         break;
 896                 }
 897         }
 898  Done:
 899         Xcheck(ws, wp);
 900         if (statep != &states[1])
 901                 /* XXX figure out what is missing */
 902                 yyerror("no closing quote\n");
 903
 904         /* This done to avoid tests for SHEREDELIM wherever SBASE tested */
 905         if (state == SHEREDELIM)
 906                 state = SBASE;
 907
 908         dp = Xstring(ws, wp);
 909         if (state == SBASE && (
 910 #ifndef MKSH_LEGACY_MODE
 911             (c == '&' && !Flag(FSH) && !Flag(FPOSIX)) ||
 912 #endif
 913             c == '<' || c == '>')) {
 914                 struct ioword *iop = alloc(sizeof(struct ioword), ATEMP);
 915
 916                 if (Xlength(ws, wp) == 0)
 917                         iop->unit = c == '<' ? 0 : 1;
 918                 else for (iop->unit = 0, c2 = 0; c2 < Xlength(ws, wp); c2 += 2) {
 919                         if (dp[c2] != CHAR)
 920                                 goto no_iop;
 921                         if (!ksh_isdigit(dp[c2 + 1]))
 922                                 goto no_iop;
 923                         iop->unit = (iop->unit * 10) + dp[c2 + 1] - '0';
 924                 }
 925
 926                 if (iop->unit >= FDBASE)
 927                         goto no_iop;
 928
 929                 if (c == '&') {
 930                         if ((c2 = getsc()) != '>') {
 931                                 ungetsc(c2);
 932                                 goto no_iop;
 933                         }
 934                         c = c2;
 935                         iop->flag = IOBASH;
 936                 } else
 937                         iop->flag = 0;
 938
 939                 c2 = getsc();
 940                 /* <<, >>, <> are ok, >< is not */
 941                 if (c == c2 || (c == '<' && c2 == '>')) {
 942                         iop->flag |= c == c2 ?
 943                             (c == '>' ? IOCAT : IOHERE) : IORDWR;
 944                         if (iop->flag == IOHERE) {
 945                                 if ((c2 = getsc()) == '-') {
 946                                         iop->flag |= IOSKIP;
 947                                         c2 = getsc();
 948                                 } else if (c2 == '<')
 949                                         iop->flag |= IOHERESTR;
 950                                 ungetsc(c2);
 951                                 if (c2 == '\n')
 952                                         iop->flag |= IONDELIM;
 953                         }
 954                 } else if (c2 == '&')
 955                         iop->flag |= IODUP | (c == '<' ? IORDUP : 0);
 956                 else {
 957                         iop->flag |= c == '>' ? IOWRITE : IOREAD;
 958                         if (c == '>' && c2 == '|')
 959                                 iop->flag |= IOCLOB;
 960                         else
 961                                 ungetsc(c2);
 962                 }
 963
 964                 iop->name = NULL;
 965                 iop->delim = NULL;
 966                 iop->heredoc = NULL;
 967                 /* free word */
 968                 Xfree(ws, wp);
 969                 yylval.iop = iop;
 970                 return (REDIR);
 971  no_iop:
 972                 afree(iop, ATEMP);
 973         }
 974
 975         if (wp == dp && state == SBASE) {
 976                 /* free word */
 977                 Xfree(ws, wp);
 978                 /* no word, process LEX1 character */
 979                 if ((c == '|') || (c == '&') || (c == ';') || (c == '('/*)*/)) {
 980                         if ((c2 = getsc()) == c)
 981                                 c = (c == ';') ? BREAK :
 982                                     (c == '|') ? LOGOR :
 983                                     (c == '&') ? LOGAND :
 984                                     /* c == '(' ) */ MDPAREN;
 985                         else if (c == '|' && c2 == '&')
 986                                 c = COPROC;
 987                         else if (c == ';' && c2 == '|')
 988                                 c = BRKEV;
 989                         else if (c == ';' && c2 == '&')
 990                                 c = BRKFT;
 991                         else
 992                                 ungetsc(c2);
 993 #ifndef MKSH_SMALL
 994                         if (c == BREAK) {
 995                                 if ((c2 = getsc()) == '&')
 996                                         c = BRKEV;
 997                                 else
 998                                         ungetsc(c2);
 999                         }
1000 #endif
1001                 } else if (c == '\n') {
1002                         gethere(false);
1003                         if (cf & CONTIN)
1004                                 goto Again;
1005                 } else if (c == '\0')
1006                         /* need here strings at EOF */
1007                         gethere(true);
1008                 return (c);
1009         }
1010
1011         /* terminate word */
1012         *wp++ = EOS;
1013         yylval.cp = Xclose(ws, wp);
1014         if (state == SWORD || state == SLETPAREN
1015             /* XXX ONEWORD? */)
1016                 return (LWORD);
1017
1018         /* unget terminator */
1019         ungetsc(c);
1020
1021         /*
1022          * note: the alias-vs-function code below depends on several
1023          * interna: starting from here, source->str is not modified;
1024          * the way getsc() and ungetsc() operate; etc.
1025          */
1026
1027         /* copy word to unprefixed string ident */
1028         sp = yylval.cp;
1029         dp = ident;
1030         if ((cf & HEREDELIM) && (sp[1] == '<'))
1031                 while ((dp - ident) < IDENT) {
1032                         if ((c = *sp++) == CHAR)
1033                                 *dp++ = *sp++;
1034                         else if ((c != OQUOTE) && (c != CQUOTE))
1035                                 break;
1036                 }
1037         else
1038                 while ((dp - ident) < IDENT && (c = *sp++) == CHAR)
1039                         *dp++ = *sp++;
1040         /* Make sure the ident array stays '\0' padded */
1041         memset(dp, 0, (ident + IDENT) - dp + 1);
1042         if (c != EOS)
1043                 /* word is not unquoted */
1044                 *ident = '\0';
1045
1046         if (*ident != '\0' && (cf & (KEYWORD | ALIAS))) {
1047                 struct tbl *p;
1048                 uint32_t h = hash(ident);
1049
1050                 if ((cf & KEYWORD) && (p = ktsearch(&keywords, ident, h)) &&
1051                     (!(cf & ESACONLY) || p->val.i == ESAC ||
1052                     p->val.i == /*{*/ '}')) {
1053                         afree(yylval.cp, ATEMP);
1054                         return (p->val.i);
1055                 }
1056                 if ((cf & ALIAS) && (p = ktsearch(&aliases, ident, h)) &&
1057                     (p->flag & ISSET)) {
1058                         /*
1059                          * this still points to the same character as the
1060                          * ungetsc'd terminator from above
1061                          */
1062                         const char *cp = source->str;
1063
1064                         /* prefer POSIX but not Korn functions over aliases */
1065                         while (*cp == ' ' || *cp == '\t')
1066                                 /*
1067                                  * this is like getsc() without skipping
1068                                  * over Source boundaries (including not
1069                                  * parsing ungetsc'd characters that got
1070                                  * pushed into an SREREAD) which is what
1071                                  * we want here anyway: find out whether
1072                                  * the alias name is followed by a POSIX
1073                                  * function definition (only the opening
1074                                  * parenthesis is checked though)
1075                                  */
1076                                 ++cp;
1077                         /* prefer functions over aliases */
1078                         if (cp[0] != '(' || cp[1] != ')') {
1079                                 Source *s = source;
1080
1081                                 while (s && (s->flags & SF_HASALIAS))
1082                                         if (s->u.tblp == p)
1083                                                 return (LWORD);
1084                                         else
1085                                                 s = s->next;
1086                                 /* push alias expansion */
1087                                 s = pushs(SALIAS, source->areap);
1088                                 s->start = s->str = p->val.s;
1089                                 s->u.tblp = p;
1090                                 s->flags |= SF_HASALIAS;
1091                                 s->next = source;
1092                                 if (source->type == SEOF) {
1093                                         /* prevent infinite recursion at EOS */
1094                                         source->u.tblp = p;
1095                                         source->flags |= SF_HASALIAS;
1096                                 }
1097                                 source = s;
1098                                 afree(yylval.cp, ATEMP);
1099                                 goto Again;
1100                         }
1101                 }
1102         }
1103
1104         return (LWORD);
1105 }
1106
1107 static void
1108 gethere(bool iseof)
1109 {
1110         struct ioword **p;
1111
1112         for (p = heres; p < herep; p++)
1113                 if (iseof && !((*p)->flag & IOHERESTR))
1114                         /* only here strings at EOF */
1115                         return;
1116                 else
1117                         readhere(*p);
1118         herep = heres;
1119 }
1120
1121 /*
1122  * read "<<word" text into temp file
1123  */
1124
1125 static void
1126 readhere(struct ioword *iop)
1127 {
1128         int c;
1129         const char *eof, *eofp;
1130         XString xs;
1131         char *xp;
1132         int xpos;
1133
1134         if (iop->flag & IOHERESTR) {
1135                 /* process the here string */
1136                 iop->heredoc = xp = evalstr(iop->delim, DOBLANK);
1137                 xpos = strlen(xp) - 1;
1138                 memmove(xp, xp + 1, xpos);
1139                 xp[xpos] = '\n';
1140                 return;
1141         }
1142
1143         eof = iop->flag & IONDELIM ? "<<" : evalstr(iop->delim, 0);
1144
1145         if (!(iop->flag & IOEVAL))
1146                 ignore_backslash_newline++;
1147
1148         Xinit(xs, xp, 256, ATEMP);
1149
1150  heredoc_read_line:
1151         /* beginning of line */
1152         eofp = eof;
1153         xpos = Xsavepos(xs, xp);
1154         if (iop->flag & IOSKIP) {
1155                 /* skip over leading tabs */
1156                 while ((c = getsc()) == '\t')
1157                         /* nothing */;
1158                 goto heredoc_parse_char;
1159         }
1160  heredoc_read_char:
1161         c = getsc();
1162  heredoc_parse_char:
1163         /* compare with here document marker */
1164         if (!*eofp) {
1165                 /* end of here document marker, what to do? */
1166                 switch (c) {
1167                 case /*(*/ ')':
1168                         if (!subshell_nesting_type)
1169                                 /*-
1170                                  * not allowed outside $(...) or (...)
1171                                  * => mismatch
1172                                  */
1173                                 break;
1174                         /* allow $(...) or (...) to close here */
1175                         ungetsc(/*(*/ ')');
1176                         /* FALLTHROUGH */
1177                 case 0:
1178                         /*
1179                          * Allow EOF here to commands without trailing
1180                          * newlines (mksh -c '...') will work as well.
1181                          */
1182                 case '\n':
1183                         /* Newline terminates here document marker */
1184                         goto heredoc_found_terminator;
1185                 }
1186         } else if (c == *eofp++)
1187                 /* store; then read and compare next character */
1188                 goto heredoc_store_and_loop;
1189         /* nope, mismatch; read until end of line */
1190         while (c != '\n') {
1191                 if (!c)
1192                         /* oops, reached EOF */
1193                         yyerror("%s '%s' unclosed\n", "here document", eof);
1194                 /* store character */
1195                 Xcheck(xs, xp);
1196                 Xput(xs, xp, c);
1197                 /* read next character */
1198                 c = getsc();
1199         }
1200         /* we read a newline as last character */
1201  heredoc_store_and_loop:
1202         /* store character */
1203         Xcheck(xs, xp);
1204         Xput(xs, xp, c);
1205         if (c == '\n')
1206                 goto heredoc_read_line;
1207         goto heredoc_read_char;
1208
1209  heredoc_found_terminator:
1210         /* jump back to saved beginning of line */
1211         xp = Xrestpos(xs, xp, xpos);
1212         /* terminate, close and store */
1213         Xput(xs, xp, '\0');
1214         iop->heredoc = Xclose(xs, xp);
1215
1216         if (!(iop->flag & IOEVAL))
1217                 ignore_backslash_newline--;
1218 }
1219
1220 void
1221 yyerror(const char *fmt, ...)
1222 {
1223         va_list va;
1224
1225         /* pop aliases and re-reads */
1226         while (source->type == SALIAS || source->type == SREREAD)
1227                 source = source->next;
1228         /* zap pending input */
1229         source->str = null;
1230
1231         error_prefix(true);
1232         va_start(va, fmt);
1233         shf_vfprintf(shl_out, fmt, va);
1234         va_end(va);
1235         errorfz();
1236 }
1237
1238 /*
1239  * input for yylex with alias expansion
1240  */
1241
1242 Source *
1243 pushs(int type, Area *areap)
1244 {
1245         Source *s;
1246
1247         s = alloc(sizeof(Source), areap);
1248         memset(s, 0, sizeof(Source));
1249         s->type = type;
1250         s->str = null;
1251         s->areap = areap;
1252         if (type == SFILE || type == SSTDIN)
1253                 XinitN(s->xs, 256, s->areap);
1254         return (s);
1255 }
1256
1257 static int
1258 getsc_uu(void)
1259 {
1260         Source *s = source;
1261         int c;
1262
1263         while ((c = *s->str++) == 0) {
1264                 /* return 0 for EOF by default */
1265                 s->str = NULL;
1266                 switch (s->type) {
1267                 case SEOF:
1268                         s->str = null;
1269                         return (0);
1270
1271                 case SSTDIN:
1272                 case SFILE:
1273                         getsc_line(s);
1274                         break;
1275
1276                 case SWSTR:
1277                         break;
1278
1279                 case SSTRING:
1280                 case SSTRINGCMDLINE:
1281                         break;
1282
1283                 case SWORDS:
1284                         s->start = s->str = *s->u.strv++;
1285                         s->type = SWORDSEP;
1286                         break;
1287
1288                 case SWORDSEP:
1289                         if (*s->u.strv == NULL) {
1290                                 s->start = s->str = "\n";
1291                                 s->type = SEOF;
1292                         } else {
1293                                 s->start = s->str = " ";
1294                                 s->type = SWORDS;
1295                         }
1296                         break;
1297
1298                 case SALIAS:
1299                         if (s->flags & SF_ALIASEND) {
1300                                 /* pass on an unused SF_ALIAS flag */
1301                                 source = s->next;
1302                                 source->flags |= s->flags & SF_ALIAS;
1303                                 s = source;
1304                         } else if (*s->u.tblp->val.s &&
1305                             (c = strnul(s->u.tblp->val.s)[-1], ksh_isspace(c))) {
1306                                 /* pop source stack */
1307                                 source = s = s->next;
1308                                 /*
1309                                  * Note that this alias ended with a
1310                                  * space, enabling alias expansion on
1311                                  * the following word.
1312                                  */
1313                                 s->flags |= SF_ALIAS;
1314                         } else {
1315                                 /*
1316                                  * At this point, we need to keep the current
1317                                  * alias in the source list so recursive
1318                                  * aliases can be detected and we also need to
1319                                  * return the next character. Do this by
1320                                  * temporarily popping the alias to get the
1321                                  * next character and then put it back in the
1322                                  * source list with the SF_ALIASEND flag set.
1323                                  */
1324                                 /* pop source stack */
1325                                 source = s->next;
1326                                 source->flags |= s->flags & SF_ALIAS;
1327                                 c = getsc_uu();
1328                                 if (c) {
1329                                         s->flags |= SF_ALIASEND;
1330                                         s->ugbuf[0] = c; s->ugbuf[1] = '\0';
1331                                         s->start = s->str = s->ugbuf;
1332                                         s->next = source;
1333                                         source = s;
1334                                 } else {
1335                                         s = source;
1336                                         /* avoid reading EOF twice */
1337                                         s->str = NULL;
1338                                         break;
1339                                 }
1340                         }
1341                         continue;
1342
1343                 case SREREAD:
1344                         if (s->start != s->ugbuf)
1345                                 /* yuck */
1346                                 afree(s->u.freeme, ATEMP);
1347                         source = s = s->next;
1348                         continue;
1349                 }
1350                 if (s->str == NULL) {
1351                         s->type = SEOF;
1352                         s->start = s->str = null;
1353                         return ('\0');
1354                 }
1355                 if (s->flags & SF_ECHO) {
1356                         shf_puts(s->str, shl_out);
1357                         shf_flush(shl_out);
1358                 }
1359         }
1360         return (c);
1361 }
1362
1363 static void
1364 getsc_line(Source *s)
1365 {
1366         char *xp = Xstring(s->xs, xp), *cp;
1367         bool interactive = Flag(FTALKING) && s->type == SSTDIN;
1368         bool have_tty = tobool(interactive && (s->flags & SF_TTY));
1369
1370         /* Done here to ensure nothing odd happens when a timeout occurs */
1371         XcheckN(s->xs, xp, LINE);
1372         *xp = '\0';
1373         s->start = s->str = xp;
1374
1375         if (have_tty && ksh_tmout) {
1376                 ksh_tmout_state = TMOUT_READING;
1377                 alarm(ksh_tmout);
1378         }
1379         if (interactive)
1380                 change_winsz();
1381 #ifndef MKSH_NO_CMDLINE_EDITING
1382         if (have_tty && (
1383 #if !MKSH_S_NOVI
1384             Flag(FVI) ||
1385 #endif
1386             Flag(FEMACS) || Flag(FGMACS))) {
1387                 int nread;
1388
1389                 nread = x_read(xp);
1390                 if (nread < 0)
1391                         /* read error */
1392                         nread = 0;
1393                 xp[nread] = '\0';
1394                 xp += nread;
1395         } else
1396 #endif
1397           {
1398                 if (interactive)
1399                         pprompt(prompt, 0);
1400                 else
1401                         s->line++;
1402
1403                 while (/* CONSTCOND */ 1) {
1404                         char *p = shf_getse(xp, Xnleft(s->xs, xp), s->u.shf);
1405
1406                         if (!p && shf_error(s->u.shf) &&
1407                             shf_errno(s->u.shf) == EINTR) {
1408                                 shf_clearerr(s->u.shf);
1409                                 if (trap)
1410                                         runtraps(0);
1411                                 continue;
1412                         }
1413                         if (!p || (xp = p, xp[-1] == '\n'))
1414                                 break;
1415                         /* double buffer size */
1416                         /* move past NUL so doubling works... */
1417                         xp++;
1418                         XcheckN(s->xs, xp, Xlength(s->xs, xp));
1419                         /* ...and move back again */
1420                         xp--;
1421                 }
1422                 /*
1423                  * flush any unwanted input so other programs/builtins
1424                  * can read it. Not very optimal, but less error prone
1425                  * than flushing else where, dealing with redirections,
1426                  * etc.
1427                  * TODO: reduce size of shf buffer (~128?) if SSTDIN
1428                  */
1429                 if (s->type == SSTDIN)
1430                         shf_flush(s->u.shf);
1431         }
1432         /*
1433          * XXX: temporary kludge to restore source after a
1434          * trap may have been executed.
1435          */
1436         source = s;
1437         if (have_tty && ksh_tmout) {
1438                 ksh_tmout_state = TMOUT_EXECUTING;
1439                 alarm(0);
1440         }
1441         cp = Xstring(s->xs, xp);
1442         rndpush(cp);
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         char delimiter = 0;
1525         bool doprint = (ntruncate != -1);
1526         bool indelimit = false;
1527         int columns = 0, lines = 0;
1528
1529         /*
1530          * Undocumented AT&T ksh feature:
1531          * If the second char in the prompt string is \r then the first
1532          * char is taken to be a non-printing delimiter and any chars
1533          * between two instances of the delimiter are not considered to
1534          * be part of the prompt length
1535          */
1536         if (*cp && cp[1] == '\r') {
1537                 delimiter = *cp;
1538                 cp += 2;
1539         }
1540         for (; *cp; cp++) {
1541                 if (indelimit && *cp != delimiter)
1542                         ;
1543                 else if (*cp == '\n' || *cp == '\r') {
1544                         lines += columns / x_cols + ((*cp == '\n') ? 1 : 0);
1545                         columns = 0;
1546                 } else if (*cp == '\t') {
1547                         columns = (columns | 7) + 1;
1548                 } else if (*cp == '\b') {
1549                         if (columns > 0)
1550                                 columns--;
1551                 } else if (*cp == delimiter)
1552                         indelimit = !indelimit;
1553                 else if (UTFMODE && ((unsigned char)*cp > 0x7F)) {
1554                         const char *cp2;
1555                         columns += utf_widthadj(cp, &cp2);
1556                         if (doprint && (indelimit ||
1557                             (ntruncate < (x_cols * lines + columns))))
1558                                 shf_write(cp, cp2 - cp, shl_out);
1559                         cp = cp2 - /* loop increment */ 1;
1560                         continue;
1561                 } else
1562                         columns++;
1563                 if (doprint && (*cp != delimiter) &&
1564                     (indelimit || (ntruncate < (x_cols * lines + columns))))
1565                         shf_putc(*cp, shl_out);
1566         }
1567         if (doprint)
1568                 shf_flush(shl_out);
1569         return (x_cols * lines + columns);
1570 }
1571
1572 /*
1573  * Read the variable part of a ${...} expression (i.e. up to but not
1574  * including the :[-+?=#%] or close-brace).
1575  */
1576 static char *
1577 get_brace_var(XString *wsp, char *wp)
1578 {
1579         char c;
1580         enum parse_state {
1581                 PS_INITIAL, PS_SAW_HASH, PS_IDENT,
1582                 PS_NUMBER, PS_VAR1
1583         } state = PS_INITIAL;
1584
1585         while (/* CONSTCOND */ 1) {
1586                 c = getsc();
1587                 /* State machine to figure out where the variable part ends. */
1588                 switch (state) {
1589                 case PS_INITIAL:
1590                         if (c == '#' || c == '!' || c == '%') {
1591                                 state = PS_SAW_HASH;
1592                                 break;
1593                         }
1594                         /* FALLTHROUGH */
1595                 case PS_SAW_HASH:
1596                         if (ksh_isalphx(c))
1597                                 state = PS_IDENT;
1598                         else if (ksh_isdigit(c))
1599                                 state = PS_NUMBER;
1600                         else if (c == '#') {
1601                                 if (state == PS_SAW_HASH) {
1602                                         char c2;
1603
1604                                         c2 = getsc();
1605                                         ungetsc(c2);
1606                                         if (c2 != /*{*/ '}') {
1607                                                 ungetsc(c);
1608                                                 goto out;
1609                                         }
1610                                 }
1611                                 state = PS_VAR1;
1612                         } else if (ctype(c, C_VAR1))
1613                                 state = PS_VAR1;
1614                         else
1615                                 goto out;
1616                         break;
1617                 case PS_IDENT:
1618                         if (!ksh_isalnux(c)) {
1619                                 if (c == '[') {
1620                                         char *tmp, *p;
1621
1622                                         if (!arraysub(&tmp))
1623                                                 yyerror("missing ]\n");
1624                                         *wp++ = c;
1625                                         for (p = tmp; *p; ) {
1626                                                 Xcheck(*wsp, wp);
1627                                                 *wp++ = *p++;
1628                                         }
1629                                         afree(tmp, ATEMP);
1630                                         /* the ] */
1631                                         c = getsc();
1632                                 }
1633                                 goto out;
1634                         }
1635                         break;
1636                 case PS_NUMBER:
1637                         if (!ksh_isdigit(c))
1638                                 goto out;
1639                         break;
1640                 case PS_VAR1:
1641                         goto out;
1642                 }
1643                 Xcheck(*wsp, wp);
1644                 *wp++ = c;
1645         }
1646  out:
1647         /* end of variable part */
1648         *wp++ = '\0';
1649         ungetsc(c);
1650         return (wp);
1651 }
1652
1653 /*
1654  * Save an array subscript - returns true if matching bracket found, false
1655  * if eof or newline was found.
1656  * (Returned string double null terminated)
1657  */
1658 static bool
1659 arraysub(char **strp)
1660 {
1661         XString ws;
1662         char *wp, c;
1663         /* we are just past the initial [ */
1664         unsigned int depth = 1;
1665
1666         Xinit(ws, wp, 32, ATEMP);
1667
1668         do {
1669                 c = getsc();
1670                 Xcheck(ws, wp);
1671                 *wp++ = c;
1672                 if (c == '[')
1673                         depth++;
1674                 else if (c == ']')
1675                         depth--;
1676         } while (depth > 0 && c && c != '\n');
1677
1678         *wp++ = '\0';
1679         *strp = Xclose(ws, wp);
1680
1681         return (tobool(depth == 0));
1682 }
1683
1684 /* Unget a char: handles case when we are already at the start of the buffer */
1685 static void
1686 ungetsc(int c)
1687 {
1688         struct sretrace_info *rp = retrace_info;
1689
1690         if (backslash_skip)
1691                 backslash_skip--;
1692         /* Don't unget EOF... */
1693         if (source->str == null && c == '\0')
1694                 return;
1695         while (rp) {
1696                 if (Xlength(rp->xs, rp->xp))
1697                         rp->xp--;
1698                 rp = rp->next;
1699         }
1700         ungetsc_i(c);
1701 }
1702 static void
1703 ungetsc_i(int c)
1704 {
1705         if (source->str > source->start)
1706                 source->str--;
1707         else {
1708                 Source *s;
1709
1710                 s = pushs(SREREAD, source->areap);
1711                 s->ugbuf[0] = c; s->ugbuf[1] = '\0';
1712                 s->start = s->str = s->ugbuf;
1713                 s->next = source;
1714                 source = s;
1715         }
1716 }
1717
1718
1719 /* Called to get a char that isn't a \newline sequence. */
1720 static int
1721 getsc_bn(void)
1722 {
1723         int c, c2;
1724
1725         if (ignore_backslash_newline)
1726                 return (o_getsc_u());
1727
1728         if (backslash_skip == 1) {
1729                 backslash_skip = 2;
1730                 return (o_getsc_u());
1731         }
1732
1733         backslash_skip = 0;
1734
1735         while (/* CONSTCOND */ 1) {
1736                 c = o_getsc_u();
1737                 if (c == '\\') {
1738                         if ((c2 = o_getsc_u()) == '\n')
1739                                 /* ignore the \newline; get the next char... */
1740                                 continue;
1741                         ungetsc_i(c2);
1742                         backslash_skip = 1;
1743                 }
1744                 return (c);
1745         }
1746 }
1747
1748 void
1749 yyskiputf8bom(void)
1750 {
1751         int c;
1752
1753         if ((unsigned char)(c = o_getsc_u()) != 0xEF) {
1754                 ungetsc_i(c);
1755                 return;
1756         }
1757         if ((unsigned char)(c = o_getsc_u()) != 0xBB) {
1758                 ungetsc_i(c);
1759                 ungetsc_i(0xEF);
1760                 return;
1761         }
1762         if ((unsigned char)(c = o_getsc_u()) != 0xBF) {
1763                 ungetsc_i(c);
1764                 ungetsc_i(0xBB);
1765                 ungetsc_i(0xEF);
1766                 return;
1767         }
1768         UTFMODE |= 8;
1769 }
1770
1771 static Lex_state *
1772 push_state_i(State_info *si, Lex_state *old_end)
1773 {
1774         Lex_state *news = alloc2(STATE_BSIZE, sizeof(Lex_state), ATEMP);
1775
1776         news[0].ls_base = old_end;
1777         si->base = &news[0];
1778         si->end = &news[STATE_BSIZE];
1779         return (&news[1]);
1780 }
1781
1782 static Lex_state *
1783 pop_state_i(State_info *si, Lex_state *old_end)
1784 {
1785         Lex_state *old_base = si->base;
1786
1787         si->base = old_end->ls_base - STATE_BSIZE;
1788         si->end = old_end->ls_base;
1789
1790         afree(old_base, ATEMP);
1791
1792         return (si->base + STATE_BSIZE - 1);
1793 }
1794
1795 static int
1796 s_get(void)
1797 {
1798         return (getsc());
1799 }
1800
1801 static void
1802 s_put(int c)
1803 {
1804         ungetsc(c);
1805 }