Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1 | /* vi: set sw=4 ts=4: */ |
| 2 | /* |
| 3 | * awk implementation for busybox |
| 4 | * |
| 5 | * Copyright (C) 2002 by Dmitry Zakharov <dmit@crp.bank.gov.ua> |
| 6 | * |
| 7 | * This program is free software; you can redistribute it and/or modify |
| 8 | * it under the terms of the GNU General Public License as published by |
| 9 | * the Free Software Foundation; either version 2 of the License, or |
| 10 | * (at your option) any later version. |
| 11 | * |
| 12 | * This program is distributed in the hope that it will be useful, |
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 15 | * General Public License for more details. |
| 16 | * |
| 17 | * You should have received a copy of the GNU General Public License |
| 18 | * along with this program; if not, write to the Free Software |
| 19 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
| 20 | * |
| 21 | */ |
| 22 | |
| 23 | #include <stdio.h> |
| 24 | #include <stdlib.h> |
| 25 | #include <unistd.h> |
| 26 | #include <errno.h> |
| 27 | #include <string.h> |
| 28 | #include <time.h> |
| 29 | #include <math.h> |
| 30 | #include <ctype.h> |
| 31 | #include <getopt.h> |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 32 | |
"Vladimir N. Oleynik" | 23f62fc | 2005-09-14 16:59:11 +0000 | [diff] [blame] | 33 | #include "xregex.h" |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 34 | #include "busybox.h" |
| 35 | |
| 36 | |
| 37 | #define MAXVARFMT 240 |
| 38 | #define MINNVBLOCK 64 |
| 39 | |
| 40 | /* variable flags */ |
| 41 | #define VF_NUMBER 0x0001 /* 1 = primary type is number */ |
| 42 | #define VF_ARRAY 0x0002 /* 1 = it's an array */ |
| 43 | |
| 44 | #define VF_CACHED 0x0100 /* 1 = num/str value has cached str/num eq */ |
| 45 | #define VF_USER 0x0200 /* 1 = user input (may be numeric string) */ |
| 46 | #define VF_SPECIAL 0x0400 /* 1 = requires extra handling when changed */ |
| 47 | #define VF_WALK 0x0800 /* 1 = variable has alloc'd x.walker list */ |
| 48 | #define VF_FSTR 0x1000 /* 1 = string points to fstring buffer */ |
| 49 | #define VF_CHILD 0x2000 /* 1 = function arg; x.parent points to source */ |
| 50 | #define VF_DIRTY 0x4000 /* 1 = variable was set explicitly */ |
| 51 | |
| 52 | /* these flags are static, don't change them when value is changed */ |
| 53 | #define VF_DONTTOUCH (VF_ARRAY | VF_SPECIAL | VF_WALK | VF_CHILD | VF_DIRTY) |
| 54 | |
| 55 | /* Variable */ |
| 56 | typedef struct var_s { |
| 57 | unsigned short type; /* flags */ |
| 58 | double number; |
| 59 | char *string; |
| 60 | union { |
| 61 | int aidx; /* func arg index (on compilation stage) */ |
| 62 | struct xhash_s *array; /* array ptr */ |
| 63 | struct var_s *parent; /* for func args, ptr to actual parameter */ |
| 64 | char **walker; /* list of array elements (for..in) */ |
| 65 | } x; |
| 66 | } var; |
| 67 | |
| 68 | /* Node chain (pattern-action chain, BEGIN, END, function bodies) */ |
| 69 | typedef struct chain_s { |
| 70 | struct node_s *first; |
| 71 | struct node_s *last; |
| 72 | char *programname; |
| 73 | } chain; |
| 74 | |
| 75 | /* Function */ |
| 76 | typedef struct func_s { |
| 77 | unsigned short nargs; |
| 78 | struct chain_s body; |
| 79 | } func; |
| 80 | |
| 81 | /* I/O stream */ |
| 82 | typedef struct rstream_s { |
| 83 | FILE *F; |
| 84 | char *buffer; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 85 | int adv; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 86 | int size; |
| 87 | int pos; |
| 88 | unsigned short is_pipe; |
| 89 | } rstream; |
| 90 | |
| 91 | typedef struct hash_item_s { |
| 92 | union { |
| 93 | struct var_s v; /* variable/array hash */ |
| 94 | struct rstream_s rs; /* redirect streams hash */ |
| 95 | struct func_s f; /* functions hash */ |
| 96 | } data; |
| 97 | struct hash_item_s *next; /* next in chain */ |
| 98 | char name[1]; /* really it's longer */ |
| 99 | } hash_item; |
| 100 | |
| 101 | typedef struct xhash_s { |
| 102 | unsigned int nel; /* num of elements */ |
| 103 | unsigned int csize; /* current hash size */ |
| 104 | unsigned int nprime; /* next hash size in PRIMES[] */ |
| 105 | unsigned int glen; /* summary length of item names */ |
| 106 | struct hash_item_s **items; |
| 107 | } xhash; |
| 108 | |
| 109 | /* Tree node */ |
| 110 | typedef struct node_s { |
| 111 | unsigned long info; |
| 112 | unsigned short lineno; |
| 113 | union { |
| 114 | struct node_s *n; |
| 115 | var *v; |
| 116 | int i; |
| 117 | char *s; |
| 118 | regex_t *re; |
| 119 | } l; |
| 120 | union { |
| 121 | struct node_s *n; |
| 122 | regex_t *ire; |
| 123 | func *f; |
| 124 | int argno; |
| 125 | } r; |
| 126 | union { |
| 127 | struct node_s *n; |
| 128 | } a; |
| 129 | } node; |
| 130 | |
| 131 | /* Block of temporary variables */ |
| 132 | typedef struct nvblock_s { |
| 133 | int size; |
| 134 | var *pos; |
| 135 | struct nvblock_s *prev; |
| 136 | struct nvblock_s *next; |
| 137 | var nv[0]; |
| 138 | } nvblock; |
| 139 | |
| 140 | typedef struct tsplitter_s { |
| 141 | node n; |
| 142 | regex_t re[2]; |
| 143 | } tsplitter; |
| 144 | |
| 145 | /* simple token classes */ |
| 146 | /* Order and hex values are very important!!! See next_token() */ |
| 147 | #define TC_SEQSTART 1 /* ( */ |
| 148 | #define TC_SEQTERM (1 << 1) /* ) */ |
| 149 | #define TC_REGEXP (1 << 2) /* /.../ */ |
| 150 | #define TC_OUTRDR (1 << 3) /* | > >> */ |
| 151 | #define TC_UOPPOST (1 << 4) /* unary postfix operator */ |
| 152 | #define TC_UOPPRE1 (1 << 5) /* unary prefix operator */ |
| 153 | #define TC_BINOPX (1 << 6) /* two-opnd operator */ |
| 154 | #define TC_IN (1 << 7) |
| 155 | #define TC_COMMA (1 << 8) |
| 156 | #define TC_PIPE (1 << 9) /* input redirection pipe */ |
| 157 | #define TC_UOPPRE2 (1 << 10) /* unary prefix operator */ |
| 158 | #define TC_ARRTERM (1 << 11) /* ] */ |
| 159 | #define TC_GRPSTART (1 << 12) /* { */ |
| 160 | #define TC_GRPTERM (1 << 13) /* } */ |
| 161 | #define TC_SEMICOL (1 << 14) |
| 162 | #define TC_NEWLINE (1 << 15) |
| 163 | #define TC_STATX (1 << 16) /* ctl statement (for, next...) */ |
| 164 | #define TC_WHILE (1 << 17) |
| 165 | #define TC_ELSE (1 << 18) |
| 166 | #define TC_BUILTIN (1 << 19) |
| 167 | #define TC_GETLINE (1 << 20) |
| 168 | #define TC_FUNCDECL (1 << 21) /* `function' `func' */ |
| 169 | #define TC_BEGIN (1 << 22) |
| 170 | #define TC_END (1 << 23) |
| 171 | #define TC_EOF (1 << 24) |
| 172 | #define TC_VARIABLE (1 << 25) |
| 173 | #define TC_ARRAY (1 << 26) |
| 174 | #define TC_FUNCTION (1 << 27) |
| 175 | #define TC_STRING (1 << 28) |
| 176 | #define TC_NUMBER (1 << 29) |
| 177 | |
| 178 | #define TC_UOPPRE (TC_UOPPRE1 | TC_UOPPRE2) |
| 179 | |
| 180 | /* combined token classes */ |
| 181 | #define TC_BINOP (TC_BINOPX | TC_COMMA | TC_PIPE | TC_IN) |
| 182 | #define TC_UNARYOP (TC_UOPPRE | TC_UOPPOST) |
| 183 | #define TC_OPERAND (TC_VARIABLE | TC_ARRAY | TC_FUNCTION | \ |
| 184 | TC_BUILTIN | TC_GETLINE | TC_SEQSTART | TC_STRING | TC_NUMBER) |
| 185 | |
| 186 | #define TC_STATEMNT (TC_STATX | TC_WHILE) |
| 187 | #define TC_OPTERM (TC_SEMICOL | TC_NEWLINE) |
| 188 | |
| 189 | /* word tokens, cannot mean something else if not expected */ |
| 190 | #define TC_WORD (TC_IN | TC_STATEMNT | TC_ELSE | TC_BUILTIN | \ |
| 191 | TC_GETLINE | TC_FUNCDECL | TC_BEGIN | TC_END) |
| 192 | |
| 193 | /* discard newlines after these */ |
| 194 | #define TC_NOTERM (TC_COMMA | TC_GRPSTART | TC_GRPTERM | \ |
| 195 | TC_BINOP | TC_OPTERM) |
| 196 | |
| 197 | /* what can expression begin with */ |
| 198 | #define TC_OPSEQ (TC_OPERAND | TC_UOPPRE | TC_REGEXP) |
| 199 | /* what can group begin with */ |
| 200 | #define TC_GRPSEQ (TC_OPSEQ | TC_OPTERM | TC_STATEMNT | TC_GRPSTART) |
| 201 | |
| 202 | /* if previous token class is CONCAT1 and next is CONCAT2, concatenation */ |
| 203 | /* operator is inserted between them */ |
| 204 | #define TC_CONCAT1 (TC_VARIABLE | TC_ARRTERM | TC_SEQTERM | \ |
| 205 | TC_STRING | TC_NUMBER | TC_UOPPOST) |
| 206 | #define TC_CONCAT2 (TC_OPERAND | TC_UOPPRE) |
| 207 | |
| 208 | #define OF_RES1 0x010000 |
| 209 | #define OF_RES2 0x020000 |
| 210 | #define OF_STR1 0x040000 |
| 211 | #define OF_STR2 0x080000 |
| 212 | #define OF_NUM1 0x100000 |
| 213 | #define OF_CHECKED 0x200000 |
| 214 | |
| 215 | /* combined operator flags */ |
| 216 | #define xx 0 |
| 217 | #define xV OF_RES2 |
| 218 | #define xS (OF_RES2 | OF_STR2) |
| 219 | #define Vx OF_RES1 |
| 220 | #define VV (OF_RES1 | OF_RES2) |
| 221 | #define Nx (OF_RES1 | OF_NUM1) |
| 222 | #define NV (OF_RES1 | OF_NUM1 | OF_RES2) |
| 223 | #define Sx (OF_RES1 | OF_STR1) |
| 224 | #define SV (OF_RES1 | OF_STR1 | OF_RES2) |
| 225 | #define SS (OF_RES1 | OF_STR1 | OF_RES2 | OF_STR2) |
| 226 | |
| 227 | #define OPCLSMASK 0xFF00 |
| 228 | #define OPNMASK 0x007F |
| 229 | |
| 230 | /* operator priority is a highest byte (even: r->l, odd: l->r grouping) |
| 231 | * For builtins it has different meaning: n n s3 s2 s1 v3 v2 v1, |
| 232 | * n - min. number of args, vN - resolve Nth arg to var, sN - resolve to string |
| 233 | */ |
| 234 | #define P(x) (x << 24) |
| 235 | #define PRIMASK 0x7F000000 |
| 236 | #define PRIMASK2 0x7E000000 |
| 237 | |
| 238 | /* Operation classes */ |
| 239 | |
| 240 | #define SHIFT_TIL_THIS 0x0600 |
| 241 | #define RECUR_FROM_THIS 0x1000 |
| 242 | |
| 243 | enum { |
| 244 | OC_DELETE=0x0100, OC_EXEC=0x0200, OC_NEWSOURCE=0x0300, |
| 245 | OC_PRINT=0x0400, OC_PRINTF=0x0500, OC_WALKINIT=0x0600, |
| 246 | |
| 247 | OC_BR=0x0700, OC_BREAK=0x0800, OC_CONTINUE=0x0900, |
| 248 | OC_EXIT=0x0a00, OC_NEXT=0x0b00, OC_NEXTFILE=0x0c00, |
| 249 | OC_TEST=0x0d00, OC_WALKNEXT=0x0e00, |
| 250 | |
| 251 | OC_BINARY=0x1000, OC_BUILTIN=0x1100, OC_COLON=0x1200, |
| 252 | OC_COMMA=0x1300, OC_COMPARE=0x1400, OC_CONCAT=0x1500, |
| 253 | OC_FBLTIN=0x1600, OC_FIELD=0x1700, OC_FNARG=0x1800, |
| 254 | OC_FUNC=0x1900, OC_GETLINE=0x1a00, OC_IN=0x1b00, |
| 255 | OC_LAND=0x1c00, OC_LOR=0x1d00, OC_MATCH=0x1e00, |
| 256 | OC_MOVE=0x1f00, OC_PGETLINE=0x2000, OC_REGEXP=0x2100, |
| 257 | OC_REPLACE=0x2200, OC_RETURN=0x2300, OC_SPRINTF=0x2400, |
| 258 | OC_TERNARY=0x2500, OC_UNARY=0x2600, OC_VAR=0x2700, |
| 259 | OC_DONE=0x2800, |
| 260 | |
| 261 | ST_IF=0x3000, ST_DO=0x3100, ST_FOR=0x3200, |
| 262 | ST_WHILE=0x3300 |
| 263 | }; |
| 264 | |
| 265 | /* simple builtins */ |
| 266 | enum { |
| 267 | F_in=0, F_rn, F_co, F_ex, F_lg, F_si, F_sq, F_sr, |
| 268 | F_ti, F_le, F_sy, F_ff, F_cl |
| 269 | }; |
| 270 | |
| 271 | /* builtins */ |
| 272 | enum { |
| 273 | B_a2=0, B_ix, B_ma, B_sp, B_ss, B_ti, B_lo, B_up, |
| 274 | B_ge, B_gs, B_su |
| 275 | }; |
| 276 | |
| 277 | /* tokens and their corresponding info values */ |
| 278 | |
| 279 | #define NTC "\377" /* switch to next token class (tc<<1) */ |
| 280 | #define NTCC '\377' |
| 281 | |
| 282 | #define OC_B OC_BUILTIN |
| 283 | |
| 284 | static char * const tokenlist = |
| 285 | "\1(" NTC |
| 286 | "\1)" NTC |
| 287 | "\1/" NTC /* REGEXP */ |
| 288 | "\2>>" "\1>" "\1|" NTC /* OUTRDR */ |
| 289 | "\2++" "\2--" NTC /* UOPPOST */ |
| 290 | "\2++" "\2--" "\1$" NTC /* UOPPRE1 */ |
| 291 | "\2==" "\1=" "\2+=" "\2-=" /* BINOPX */ |
| 292 | "\2*=" "\2/=" "\2%=" "\2^=" |
| 293 | "\1+" "\1-" "\3**=" "\2**" |
| 294 | "\1/" "\1%" "\1^" "\1*" |
| 295 | "\2!=" "\2>=" "\2<=" "\1>" |
| 296 | "\1<" "\2!~" "\1~" "\2&&" |
| 297 | "\2||" "\1?" "\1:" NTC |
| 298 | "\2in" NTC |
| 299 | "\1," NTC |
| 300 | "\1|" NTC |
| 301 | "\1+" "\1-" "\1!" NTC /* UOPPRE2 */ |
| 302 | "\1]" NTC |
| 303 | "\1{" NTC |
| 304 | "\1}" NTC |
| 305 | "\1;" NTC |
| 306 | "\1\n" NTC |
| 307 | "\2if" "\2do" "\3for" "\5break" /* STATX */ |
| 308 | "\10continue" "\6delete" "\5print" |
| 309 | "\6printf" "\4next" "\10nextfile" |
| 310 | "\6return" "\4exit" NTC |
| 311 | "\5while" NTC |
| 312 | "\4else" NTC |
| 313 | |
| 314 | "\5close" "\6system" "\6fflush" "\5atan2" /* BUILTIN */ |
| 315 | "\3cos" "\3exp" "\3int" "\3log" |
| 316 | "\4rand" "\3sin" "\4sqrt" "\5srand" |
| 317 | "\6gensub" "\4gsub" "\5index" "\6length" |
| 318 | "\5match" "\5split" "\7sprintf" "\3sub" |
| 319 | "\6substr" "\7systime" "\10strftime" |
| 320 | "\7tolower" "\7toupper" NTC |
| 321 | "\7getline" NTC |
| 322 | "\4func" "\10function" NTC |
| 323 | "\5BEGIN" NTC |
| 324 | "\3END" "\0" |
| 325 | ; |
| 326 | |
| 327 | static unsigned long tokeninfo[] = { |
| 328 | |
| 329 | 0, |
| 330 | 0, |
| 331 | OC_REGEXP, |
| 332 | xS|'a', xS|'w', xS|'|', |
| 333 | OC_UNARY|xV|P(9)|'p', OC_UNARY|xV|P(9)|'m', |
| 334 | OC_UNARY|xV|P(9)|'P', OC_UNARY|xV|P(9)|'M', |
| 335 | OC_FIELD|xV|P(5), |
| 336 | OC_COMPARE|VV|P(39)|5, OC_MOVE|VV|P(74), |
| 337 | OC_REPLACE|NV|P(74)|'+', OC_REPLACE|NV|P(74)|'-', |
| 338 | OC_REPLACE|NV|P(74)|'*', OC_REPLACE|NV|P(74)|'/', |
| 339 | OC_REPLACE|NV|P(74)|'%', OC_REPLACE|NV|P(74)|'&', |
| 340 | OC_BINARY|NV|P(29)|'+', OC_BINARY|NV|P(29)|'-', |
| 341 | OC_REPLACE|NV|P(74)|'&', OC_BINARY|NV|P(15)|'&', |
| 342 | OC_BINARY|NV|P(25)|'/', OC_BINARY|NV|P(25)|'%', |
| 343 | OC_BINARY|NV|P(15)|'&', OC_BINARY|NV|P(25)|'*', |
| 344 | OC_COMPARE|VV|P(39)|4, OC_COMPARE|VV|P(39)|3, |
| 345 | OC_COMPARE|VV|P(39)|0, OC_COMPARE|VV|P(39)|1, |
| 346 | OC_COMPARE|VV|P(39)|2, OC_MATCH|Sx|P(45)|'!', |
| 347 | OC_MATCH|Sx|P(45)|'~', OC_LAND|Vx|P(55), |
| 348 | OC_LOR|Vx|P(59), OC_TERNARY|Vx|P(64)|'?', |
| 349 | OC_COLON|xx|P(67)|':', |
| 350 | OC_IN|SV|P(49), |
| 351 | OC_COMMA|SS|P(80), |
| 352 | OC_PGETLINE|SV|P(37), |
| 353 | OC_UNARY|xV|P(19)|'+', OC_UNARY|xV|P(19)|'-', |
| 354 | OC_UNARY|xV|P(19)|'!', |
| 355 | 0, |
| 356 | 0, |
| 357 | 0, |
| 358 | 0, |
| 359 | 0, |
| 360 | ST_IF, ST_DO, ST_FOR, OC_BREAK, |
| 361 | OC_CONTINUE, OC_DELETE|Vx, OC_PRINT, |
| 362 | OC_PRINTF, OC_NEXT, OC_NEXTFILE, |
| 363 | OC_RETURN|Vx, OC_EXIT|Nx, |
| 364 | ST_WHILE, |
| 365 | 0, |
| 366 | |
| 367 | OC_FBLTIN|Sx|F_cl, OC_FBLTIN|Sx|F_sy, OC_FBLTIN|Sx|F_ff, OC_B|B_a2|P(0x83), |
| 368 | OC_FBLTIN|Nx|F_co, OC_FBLTIN|Nx|F_ex, OC_FBLTIN|Nx|F_in, OC_FBLTIN|Nx|F_lg, |
| 369 | OC_FBLTIN|F_rn, OC_FBLTIN|Nx|F_si, OC_FBLTIN|Nx|F_sq, OC_FBLTIN|Nx|F_sr, |
| 370 | OC_B|B_ge|P(0xd6), OC_B|B_gs|P(0xb6), OC_B|B_ix|P(0x9b), OC_FBLTIN|Sx|F_le, |
| 371 | OC_B|B_ma|P(0x89), OC_B|B_sp|P(0x8b), OC_SPRINTF, OC_B|B_su|P(0xb6), |
| 372 | OC_B|B_ss|P(0x8f), OC_FBLTIN|F_ti, OC_B|B_ti|P(0x0b), |
| 373 | OC_B|B_lo|P(0x49), OC_B|B_up|P(0x49), |
| 374 | OC_GETLINE|SV|P(0), |
| 375 | 0, 0, |
| 376 | 0, |
| 377 | 0 |
| 378 | }; |
| 379 | |
| 380 | /* internal variable names and their initial values */ |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 381 | /* asterisk marks SPECIAL vars; $ is just no-named Field0 */ |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 382 | enum { |
| 383 | CONVFMT=0, OFMT, FS, OFS, |
| 384 | ORS, RS, RT, FILENAME, |
| 385 | SUBSEP, ARGIND, ARGC, ARGV, |
| 386 | ERRNO, FNR, |
| 387 | NR, NF, IGNORECASE, |
| 388 | ENVIRON, F0, _intvarcount_ |
| 389 | }; |
| 390 | |
| 391 | static char * vNames = |
| 392 | "CONVFMT\0" "OFMT\0" "FS\0*" "OFS\0" |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 393 | "ORS\0" "RS\0*" "RT\0" "FILENAME\0" |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 394 | "SUBSEP\0" "ARGIND\0" "ARGC\0" "ARGV\0" |
| 395 | "ERRNO\0" "FNR\0" |
| 396 | "NR\0" "NF\0*" "IGNORECASE\0*" |
| 397 | "ENVIRON\0" "$\0*" "\0"; |
| 398 | |
| 399 | static char * vValues = |
| 400 | "%.6g\0" "%.6g\0" " \0" " \0" |
| 401 | "\n\0" "\n\0" "\0" "\0" |
| 402 | "\034\0" |
| 403 | "\377"; |
| 404 | |
| 405 | /* hash size may grow to these values */ |
| 406 | #define FIRST_PRIME 61; |
| 407 | static const unsigned int PRIMES[] = { 251, 1021, 4093, 16381, 65521 }; |
| 408 | static const unsigned int NPRIMES = sizeof(PRIMES) / sizeof(unsigned int); |
| 409 | |
| 410 | /* globals */ |
| 411 | |
| 412 | extern char **environ; |
| 413 | |
| 414 | static var * V[_intvarcount_]; |
| 415 | static chain beginseq, mainseq, endseq, *seq; |
| 416 | static int nextrec, nextfile; |
| 417 | static node *break_ptr, *continue_ptr; |
| 418 | static rstream *iF; |
| 419 | static xhash *vhash, *ahash, *fdhash, *fnhash; |
| 420 | static char *programname; |
| 421 | static short lineno; |
| 422 | static int is_f0_split; |
| 423 | static int nfields = 0; |
| 424 | static var *Fields = NULL; |
| 425 | static tsplitter fsplitter, rsplitter; |
| 426 | static nvblock *cb = NULL; |
| 427 | static char *pos; |
| 428 | static char *buf; |
| 429 | static int icase = FALSE; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 430 | static int exiting = FALSE; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 431 | |
| 432 | static struct { |
| 433 | unsigned long tclass; |
| 434 | unsigned long info; |
| 435 | char *string; |
| 436 | double number; |
| 437 | short lineno; |
| 438 | int rollback; |
| 439 | } t; |
| 440 | |
| 441 | /* function prototypes */ |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 442 | static void handle_special(var *); |
| 443 | static node *parse_expr(unsigned long); |
| 444 | static void chain_group(void); |
| 445 | static var *evaluate(node *, var *); |
| 446 | static rstream *next_input_file(void); |
| 447 | static int fmt_num(char *, int, char *, double, int); |
| 448 | static int awk_exit(int); |
| 449 | |
| 450 | /* ---- error handling ---- */ |
| 451 | |
| 452 | static const char EMSG_INTERNAL_ERROR[] = "Internal error"; |
| 453 | static const char EMSG_UNEXP_EOS[] = "Unexpected end of string"; |
| 454 | static const char EMSG_UNEXP_TOKEN[] = "Unexpected token"; |
| 455 | static const char EMSG_DIV_BY_ZERO[] = "Division by zero"; |
| 456 | static const char EMSG_INV_FMT[] = "Invalid format specifier"; |
| 457 | static const char EMSG_TOO_FEW_ARGS[] = "Too few arguments for builtin"; |
| 458 | static const char EMSG_NOT_ARRAY[] = "Not an array"; |
| 459 | static const char EMSG_POSSIBLE_ERROR[] = "Possible syntax error"; |
| 460 | static const char EMSG_UNDEF_FUNC[] = "Call to undefined function"; |
| 461 | #ifndef CONFIG_FEATURE_AWK_MATH |
| 462 | static const char EMSG_NO_MATH[] = "Math support is not compiled in"; |
| 463 | #endif |
| 464 | |
Glenn L McGrath | d4036f8 | 2002-11-28 09:30:40 +0000 | [diff] [blame] | 465 | static void syntax_error(const char * const message) |
| 466 | { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 467 | bb_error_msg("%s:%i: %s", programname, lineno, message); |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 468 | exit(1); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 469 | } |
| 470 | |
| 471 | #define runtime_error(x) syntax_error(x) |
| 472 | |
| 473 | |
| 474 | /* ---- hash stuff ---- */ |
| 475 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 476 | static unsigned int hashidx(char *name) |
| 477 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 478 | register unsigned int idx=0; |
| 479 | |
| 480 | while (*name) idx = *name++ + (idx << 6) - idx; |
| 481 | return idx; |
| 482 | } |
| 483 | |
| 484 | /* create new hash */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 485 | static xhash *hash_init(void) |
| 486 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 487 | xhash *newhash; |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 488 | |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 489 | newhash = (xhash *)xcalloc(1, sizeof(xhash)); |
| 490 | newhash->csize = FIRST_PRIME; |
| 491 | newhash->items = (hash_item **)xcalloc(newhash->csize, sizeof(hash_item *)); |
| 492 | |
| 493 | return newhash; |
| 494 | } |
| 495 | |
| 496 | /* find item in hash, return ptr to data, NULL if not found */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 497 | static void *hash_search(xhash *hash, char *name) |
| 498 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 499 | hash_item *hi; |
| 500 | |
| 501 | hi = hash->items [ hashidx(name) % hash->csize ]; |
| 502 | while (hi) { |
| 503 | if (strcmp(hi->name, name) == 0) |
| 504 | return &(hi->data); |
| 505 | hi = hi->next; |
| 506 | } |
| 507 | return NULL; |
| 508 | } |
| 509 | |
| 510 | /* grow hash if it becomes too big */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 511 | static void hash_rebuild(xhash *hash) |
| 512 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 513 | unsigned int newsize, i, idx; |
| 514 | hash_item **newitems, *hi, *thi; |
| 515 | |
| 516 | if (hash->nprime == NPRIMES) |
| 517 | return; |
| 518 | |
| 519 | newsize = PRIMES[hash->nprime++]; |
| 520 | newitems = (hash_item **)xcalloc(newsize, sizeof(hash_item *)); |
| 521 | |
| 522 | for (i=0; i<hash->csize; i++) { |
| 523 | hi = hash->items[i]; |
| 524 | while (hi) { |
| 525 | thi = hi; |
| 526 | hi = thi->next; |
| 527 | idx = hashidx(thi->name) % newsize; |
| 528 | thi->next = newitems[idx]; |
| 529 | newitems[idx] = thi; |
| 530 | } |
| 531 | } |
| 532 | |
| 533 | free(hash->items); |
| 534 | hash->csize = newsize; |
| 535 | hash->items = newitems; |
| 536 | } |
| 537 | |
| 538 | /* find item in hash, add it if necessary. Return ptr to data */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 539 | static void *hash_find(xhash *hash, char *name) |
| 540 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 541 | hash_item *hi; |
| 542 | unsigned int idx; |
| 543 | int l; |
| 544 | |
| 545 | hi = hash_search(hash, name); |
| 546 | if (! hi) { |
| 547 | if (++hash->nel / hash->csize > 10) |
| 548 | hash_rebuild(hash); |
| 549 | |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 550 | l = bb_strlen(name) + 1; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 551 | hi = xcalloc(sizeof(hash_item) + l, 1); |
| 552 | memcpy(hi->name, name, l); |
| 553 | |
| 554 | idx = hashidx(name) % hash->csize; |
| 555 | hi->next = hash->items[idx]; |
| 556 | hash->items[idx] = hi; |
| 557 | hash->glen += l; |
| 558 | } |
| 559 | return &(hi->data); |
| 560 | } |
| 561 | |
| 562 | #define findvar(hash, name) (var *) hash_find ( (hash) , (name) ) |
| 563 | #define newvar(name) (var *) hash_find ( vhash , (name) ) |
| 564 | #define newfile(name) (rstream *) hash_find ( fdhash , (name) ) |
| 565 | #define newfunc(name) (func *) hash_find ( fnhash , (name) ) |
| 566 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 567 | static void hash_remove(xhash *hash, char *name) |
| 568 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 569 | hash_item *hi, **phi; |
| 570 | |
| 571 | phi = &(hash->items[ hashidx(name) % hash->csize ]); |
| 572 | while (*phi) { |
| 573 | hi = *phi; |
| 574 | if (strcmp(hi->name, name) == 0) { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 575 | hash->glen -= (bb_strlen(name) + 1); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 576 | hash->nel--; |
| 577 | *phi = hi->next; |
| 578 | free(hi); |
| 579 | break; |
| 580 | } |
| 581 | phi = &(hi->next); |
| 582 | } |
| 583 | } |
| 584 | |
| 585 | /* ------ some useful functions ------ */ |
| 586 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 587 | static void skip_spaces(char **s) |
| 588 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 589 | register char *p = *s; |
| 590 | |
| 591 | while(*p == ' ' || *p == '\t' || |
| 592 | (*p == '\\' && *(p+1) == '\n' && (++p, ++t.lineno))) { |
| 593 | p++; |
| 594 | } |
| 595 | *s = p; |
| 596 | } |
| 597 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 598 | static char *nextword(char **s) |
| 599 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 600 | register char *p = *s; |
| 601 | |
| 602 | while (*(*s)++) ; |
| 603 | |
| 604 | return p; |
| 605 | } |
| 606 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 607 | static char nextchar(char **s) |
| 608 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 609 | register char c, *pps; |
| 610 | |
| 611 | c = *((*s)++); |
| 612 | pps = *s; |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 613 | if (c == '\\') c = bb_process_escape_sequence((const char**)s); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 614 | if (c == '\\' && *s == pps) c = *((*s)++); |
| 615 | return c; |
| 616 | } |
| 617 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 618 | static inline int isalnum_(int c) |
| 619 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 620 | return (isalnum(c) || c == '_'); |
| 621 | } |
| 622 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 623 | static FILE *afopen(const char *path, const char *mode) |
| 624 | { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 625 | return (*path == '-' && *(path+1) == '\0') ? stdin : bb_xfopen(path, mode); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 626 | } |
| 627 | |
| 628 | /* -------- working with variables (set/get/copy/etc) -------- */ |
| 629 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 630 | static xhash *iamarray(var *v) |
| 631 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 632 | var *a = v; |
| 633 | |
| 634 | while (a->type & VF_CHILD) |
| 635 | a = a->x.parent; |
| 636 | |
| 637 | if (! (a->type & VF_ARRAY)) { |
| 638 | a->type |= VF_ARRAY; |
| 639 | a->x.array = hash_init(); |
| 640 | } |
| 641 | return a->x.array; |
| 642 | } |
| 643 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 644 | static void clear_array(xhash *array) |
| 645 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 646 | unsigned int i; |
| 647 | hash_item *hi, *thi; |
| 648 | |
| 649 | for (i=0; i<array->csize; i++) { |
| 650 | hi = array->items[i]; |
| 651 | while (hi) { |
| 652 | thi = hi; |
| 653 | hi = hi->next; |
Aaron Lehmann | a170e1c | 2002-11-28 11:27:31 +0000 | [diff] [blame] | 654 | free(thi->data.v.string); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 655 | free(thi); |
| 656 | } |
| 657 | array->items[i] = NULL; |
| 658 | } |
| 659 | array->glen = array->nel = 0; |
| 660 | } |
| 661 | |
| 662 | /* clear a variable */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 663 | static var *clrvar(var *v) |
| 664 | { |
Aaron Lehmann | a170e1c | 2002-11-28 11:27:31 +0000 | [diff] [blame] | 665 | if (!(v->type & VF_FSTR)) |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 666 | free(v->string); |
| 667 | |
| 668 | v->type &= VF_DONTTOUCH; |
| 669 | v->type |= VF_DIRTY; |
| 670 | v->string = NULL; |
| 671 | return v; |
| 672 | } |
| 673 | |
| 674 | /* assign string value to variable */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 675 | static var *setvar_p(var *v, char *value) |
| 676 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 677 | clrvar(v); |
| 678 | v->string = value; |
| 679 | handle_special(v); |
| 680 | |
| 681 | return v; |
| 682 | } |
| 683 | |
| 684 | /* same as setvar_p but make a copy of string */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 685 | static var *setvar_s(var *v, char *value) |
| 686 | { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 687 | return setvar_p(v, (value && *value) ? bb_xstrdup(value) : NULL); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 688 | } |
| 689 | |
| 690 | /* same as setvar_s but set USER flag */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 691 | static var *setvar_u(var *v, char *value) |
| 692 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 693 | setvar_s(v, value); |
| 694 | v->type |= VF_USER; |
| 695 | return v; |
| 696 | } |
| 697 | |
| 698 | /* set array element to user string */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 699 | static void setari_u(var *a, int idx, char *s) |
| 700 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 701 | register var *v; |
| 702 | static char sidx[12]; |
| 703 | |
| 704 | sprintf(sidx, "%d", idx); |
| 705 | v = findvar(iamarray(a), sidx); |
| 706 | setvar_u(v, s); |
| 707 | } |
| 708 | |
| 709 | /* assign numeric value to variable */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 710 | static var *setvar_i(var *v, double value) |
| 711 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 712 | clrvar(v); |
| 713 | v->type |= VF_NUMBER; |
| 714 | v->number = value; |
| 715 | handle_special(v); |
| 716 | return v; |
| 717 | } |
| 718 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 719 | static char *getvar_s(var *v) |
| 720 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 721 | /* if v is numeric and has no cached string, convert it to string */ |
| 722 | if ((v->type & (VF_NUMBER | VF_CACHED)) == VF_NUMBER) { |
| 723 | fmt_num(buf, MAXVARFMT, getvar_s(V[CONVFMT]), v->number, TRUE); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 724 | v->string = bb_xstrdup(buf); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 725 | v->type |= VF_CACHED; |
| 726 | } |
| 727 | return (v->string == NULL) ? "" : v->string; |
| 728 | } |
| 729 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 730 | static double getvar_i(var *v) |
| 731 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 732 | char *s; |
| 733 | |
| 734 | if ((v->type & (VF_NUMBER | VF_CACHED)) == 0) { |
| 735 | v->number = 0; |
| 736 | s = v->string; |
| 737 | if (s && *s) { |
| 738 | v->number = strtod(s, &s); |
| 739 | if (v->type & VF_USER) { |
| 740 | skip_spaces(&s); |
| 741 | if (*s != '\0') |
| 742 | v->type &= ~VF_USER; |
| 743 | } |
| 744 | } else { |
| 745 | v->type &= ~VF_USER; |
| 746 | } |
| 747 | v->type |= VF_CACHED; |
| 748 | } |
| 749 | return v->number; |
| 750 | } |
| 751 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 752 | static var *copyvar(var *dest, var *src) |
| 753 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 754 | if (dest != src) { |
| 755 | clrvar(dest); |
| 756 | dest->type |= (src->type & ~VF_DONTTOUCH); |
| 757 | dest->number = src->number; |
| 758 | if (src->string) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 759 | dest->string = bb_xstrdup(src->string); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 760 | } |
| 761 | handle_special(dest); |
| 762 | return dest; |
| 763 | } |
| 764 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 765 | static var *incvar(var *v) |
| 766 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 767 | return setvar_i(v, getvar_i(v)+1.); |
| 768 | } |
| 769 | |
| 770 | /* return true if v is number or numeric string */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 771 | static int is_numeric(var *v) |
| 772 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 773 | getvar_i(v); |
| 774 | return ((v->type ^ VF_DIRTY) & (VF_NUMBER | VF_USER | VF_DIRTY)); |
| 775 | } |
| 776 | |
| 777 | /* return 1 when value of v corresponds to true, 0 otherwise */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 778 | static int istrue(var *v) |
| 779 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 780 | if (is_numeric(v)) |
| 781 | return (v->number == 0) ? 0 : 1; |
| 782 | else |
| 783 | return (v->string && *(v->string)) ? 1 : 0; |
| 784 | } |
| 785 | |
Eric Andersen | aff114c | 2004-04-14 17:51:38 +0000 | [diff] [blame] | 786 | /* temporary variables allocator. Last allocated should be first freed */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 787 | static var *nvalloc(int n) |
| 788 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 789 | nvblock *pb = NULL; |
| 790 | var *v, *r; |
| 791 | int size; |
| 792 | |
| 793 | while (cb) { |
| 794 | pb = cb; |
| 795 | if ((cb->pos - cb->nv) + n <= cb->size) break; |
| 796 | cb = cb->next; |
| 797 | } |
| 798 | |
| 799 | if (! cb) { |
| 800 | size = (n <= MINNVBLOCK) ? MINNVBLOCK : n; |
| 801 | cb = (nvblock *)xmalloc(sizeof(nvblock) + size * sizeof(var)); |
| 802 | cb->size = size; |
| 803 | cb->pos = cb->nv; |
| 804 | cb->prev = pb; |
| 805 | cb->next = NULL; |
| 806 | if (pb) pb->next = cb; |
| 807 | } |
| 808 | |
| 809 | v = r = cb->pos; |
| 810 | cb->pos += n; |
| 811 | |
| 812 | while (v < cb->pos) { |
| 813 | v->type = 0; |
| 814 | v->string = NULL; |
| 815 | v++; |
| 816 | } |
| 817 | |
| 818 | return r; |
| 819 | } |
| 820 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 821 | static void nvfree(var *v) |
| 822 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 823 | var *p; |
| 824 | |
| 825 | if (v < cb->nv || v >= cb->pos) |
| 826 | runtime_error(EMSG_INTERNAL_ERROR); |
| 827 | |
| 828 | for (p=v; p<cb->pos; p++) { |
| 829 | if ((p->type & (VF_ARRAY|VF_CHILD)) == VF_ARRAY) { |
| 830 | clear_array(iamarray(p)); |
| 831 | free(p->x.array->items); |
| 832 | free(p->x.array); |
| 833 | } |
| 834 | if (p->type & VF_WALK) |
| 835 | free(p->x.walker); |
| 836 | |
| 837 | clrvar(p); |
| 838 | } |
| 839 | |
| 840 | cb->pos = v; |
| 841 | while (cb->prev && cb->pos == cb->nv) { |
| 842 | cb = cb->prev; |
| 843 | } |
| 844 | } |
| 845 | |
| 846 | /* ------- awk program text parsing ------- */ |
| 847 | |
| 848 | /* Parse next token pointed by global pos, place results into global t. |
| 849 | * If token isn't expected, give away. Return token class |
| 850 | */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 851 | static unsigned long next_token(unsigned long expected) |
| 852 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 853 | char *p, *pp, *s; |
| 854 | char *tl; |
| 855 | unsigned long tc, *ti; |
| 856 | int l; |
| 857 | static int concat_inserted = FALSE; |
| 858 | static unsigned long save_tclass, save_info; |
| 859 | static unsigned long ltclass = TC_OPTERM; |
| 860 | |
| 861 | if (t.rollback) { |
| 862 | |
| 863 | t.rollback = FALSE; |
| 864 | |
| 865 | } else if (concat_inserted) { |
| 866 | |
| 867 | concat_inserted = FALSE; |
| 868 | t.tclass = save_tclass; |
| 869 | t.info = save_info; |
| 870 | |
| 871 | } else { |
| 872 | |
| 873 | p = pos; |
| 874 | |
| 875 | readnext: |
| 876 | skip_spaces(&p); |
| 877 | lineno = t.lineno; |
| 878 | if (*p == '#') |
| 879 | while (*p != '\n' && *p != '\0') p++; |
| 880 | |
| 881 | if (*p == '\n') |
| 882 | t.lineno++; |
| 883 | |
| 884 | if (*p == '\0') { |
| 885 | tc = TC_EOF; |
| 886 | |
| 887 | } else if (*p == '\"') { |
| 888 | /* it's a string */ |
| 889 | t.string = s = ++p; |
| 890 | while (*p != '\"') { |
| 891 | if (*p == '\0' || *p == '\n') |
| 892 | syntax_error(EMSG_UNEXP_EOS); |
| 893 | *(s++) = nextchar(&p); |
| 894 | } |
| 895 | p++; |
| 896 | *s = '\0'; |
| 897 | tc = TC_STRING; |
| 898 | |
| 899 | } else if ((expected & TC_REGEXP) && *p == '/') { |
| 900 | /* it's regexp */ |
| 901 | t.string = s = ++p; |
| 902 | while (*p != '/') { |
| 903 | if (*p == '\0' || *p == '\n') |
| 904 | syntax_error(EMSG_UNEXP_EOS); |
| 905 | if ((*s++ = *p++) == '\\') { |
| 906 | pp = p; |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 907 | *(s-1) = bb_process_escape_sequence((const char **)&p); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 908 | if (*pp == '\\') *s++ = '\\'; |
| 909 | if (p == pp) *s++ = *p++; |
| 910 | } |
| 911 | } |
| 912 | p++; |
| 913 | *s = '\0'; |
| 914 | tc = TC_REGEXP; |
| 915 | |
| 916 | } else if (*p == '.' || isdigit(*p)) { |
| 917 | /* it's a number */ |
| 918 | t.number = strtod(p, &p); |
| 919 | if (*p == '.') |
| 920 | syntax_error(EMSG_UNEXP_TOKEN); |
| 921 | tc = TC_NUMBER; |
| 922 | |
| 923 | } else { |
| 924 | /* search for something known */ |
| 925 | tl = tokenlist; |
| 926 | tc = 0x00000001; |
| 927 | ti = tokeninfo; |
| 928 | while (*tl) { |
| 929 | l = *(tl++); |
| 930 | if (l == NTCC) { |
| 931 | tc <<= 1; |
| 932 | continue; |
| 933 | } |
| 934 | /* if token class is expected, token |
| 935 | * matches and it's not a longer word, |
| 936 | * then this is what we are looking for |
| 937 | */ |
| 938 | if ((tc & (expected | TC_WORD | TC_NEWLINE)) && |
| 939 | *tl == *p && strncmp(p, tl, l) == 0 && |
| 940 | !((tc & TC_WORD) && isalnum_(*(p + l)))) { |
| 941 | t.info = *ti; |
| 942 | p += l; |
| 943 | break; |
| 944 | } |
| 945 | ti++; |
| 946 | tl += l; |
| 947 | } |
| 948 | |
| 949 | if (! *tl) { |
| 950 | /* it's a name (var/array/function), |
| 951 | * otherwise it's something wrong |
| 952 | */ |
| 953 | if (! isalnum_(*p)) |
| 954 | syntax_error(EMSG_UNEXP_TOKEN); |
| 955 | |
| 956 | t.string = --p; |
| 957 | while(isalnum_(*(++p))) { |
| 958 | *(p-1) = *p; |
| 959 | } |
| 960 | *(p-1) = '\0'; |
| 961 | tc = TC_VARIABLE; |
| 962 | if (*p == '(') { |
| 963 | tc = TC_FUNCTION; |
| 964 | } else { |
| 965 | skip_spaces(&p); |
| 966 | if (*p == '[') { |
| 967 | p++; |
| 968 | tc = TC_ARRAY; |
| 969 | } |
| 970 | } |
| 971 | } |
| 972 | } |
| 973 | pos = p; |
| 974 | |
| 975 | /* skipping newlines in some cases */ |
| 976 | if ((ltclass & TC_NOTERM) && (tc & TC_NEWLINE)) |
| 977 | goto readnext; |
| 978 | |
| 979 | /* insert concatenation operator when needed */ |
| 980 | if ((ltclass&TC_CONCAT1) && (tc&TC_CONCAT2) && (expected&TC_BINOP)) { |
| 981 | concat_inserted = TRUE; |
| 982 | save_tclass = tc; |
| 983 | save_info = t.info; |
| 984 | tc = TC_BINOP; |
| 985 | t.info = OC_CONCAT | SS | P(35); |
| 986 | } |
| 987 | |
| 988 | t.tclass = tc; |
| 989 | } |
| 990 | ltclass = t.tclass; |
| 991 | |
| 992 | /* Are we ready for this? */ |
| 993 | if (! (ltclass & expected)) |
| 994 | syntax_error((ltclass & (TC_NEWLINE | TC_EOF)) ? |
| 995 | EMSG_UNEXP_EOS : EMSG_UNEXP_TOKEN); |
| 996 | |
| 997 | return ltclass; |
| 998 | } |
| 999 | |
| 1000 | static void rollback_token(void) { t.rollback = TRUE; } |
| 1001 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1002 | static node *new_node(unsigned long info) |
| 1003 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1004 | register node *n; |
| 1005 | |
| 1006 | n = (node *)xcalloc(sizeof(node), 1); |
| 1007 | n->info = info; |
| 1008 | n->lineno = lineno; |
| 1009 | return n; |
| 1010 | } |
| 1011 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1012 | static node *mk_re_node(char *s, node *n, regex_t *re) |
| 1013 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1014 | n->info = OC_REGEXP; |
| 1015 | n->l.re = re; |
| 1016 | n->r.ire = re + 1; |
| 1017 | xregcomp(re, s, REG_EXTENDED); |
| 1018 | xregcomp(re+1, s, REG_EXTENDED | REG_ICASE); |
| 1019 | |
| 1020 | return n; |
| 1021 | } |
| 1022 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1023 | static node *condition(void) |
| 1024 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1025 | next_token(TC_SEQSTART); |
| 1026 | return parse_expr(TC_SEQTERM); |
| 1027 | } |
| 1028 | |
| 1029 | /* parse expression terminated by given argument, return ptr |
| 1030 | * to built subtree. Terminator is eaten by parse_expr */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1031 | static node *parse_expr(unsigned long iexp) |
| 1032 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1033 | node sn; |
| 1034 | node *cn = &sn; |
| 1035 | node *vn, *glptr; |
| 1036 | unsigned long tc, xtc; |
| 1037 | var *v; |
| 1038 | |
| 1039 | sn.info = PRIMASK; |
| 1040 | sn.r.n = glptr = NULL; |
| 1041 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP | iexp; |
| 1042 | |
| 1043 | while (! ((tc = next_token(xtc)) & iexp)) { |
| 1044 | if (glptr && (t.info == (OC_COMPARE|VV|P(39)|2))) { |
| 1045 | /* input redirection (<) attached to glptr node */ |
| 1046 | cn = glptr->l.n = new_node(OC_CONCAT|SS|P(37)); |
Glenn L McGrath | 4bded58 | 2004-02-22 11:55:09 +0000 | [diff] [blame] | 1047 | cn->a.n = glptr; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1048 | xtc = TC_OPERAND | TC_UOPPRE; |
| 1049 | glptr = NULL; |
| 1050 | |
| 1051 | } else if (tc & (TC_BINOP | TC_UOPPOST)) { |
| 1052 | /* for binary and postfix-unary operators, jump back over |
| 1053 | * previous operators with higher priority */ |
| 1054 | vn = cn; |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 1055 | while ( ((t.info & PRIMASK) > (vn->a.n->info & PRIMASK2)) || |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1056 | ((t.info == vn->info) && ((t.info & OPCLSMASK) == OC_COLON)) ) |
| 1057 | vn = vn->a.n; |
| 1058 | if ((t.info & OPCLSMASK) == OC_TERNARY) |
| 1059 | t.info += P(6); |
| 1060 | cn = vn->a.n->r.n = new_node(t.info); |
| 1061 | cn->a.n = vn->a.n; |
| 1062 | if (tc & TC_BINOP) { |
| 1063 | cn->l.n = vn; |
| 1064 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
| 1065 | if ((t.info & OPCLSMASK) == OC_PGETLINE) { |
| 1066 | /* it's a pipe */ |
| 1067 | next_token(TC_GETLINE); |
| 1068 | /* give maximum priority to this pipe */ |
| 1069 | cn->info &= ~PRIMASK; |
| 1070 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
| 1071 | } |
| 1072 | } else { |
| 1073 | cn->r.n = vn; |
| 1074 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
| 1075 | } |
| 1076 | vn->a.n = cn; |
| 1077 | |
| 1078 | } else { |
| 1079 | /* for operands and prefix-unary operators, attach them |
| 1080 | * to last node */ |
| 1081 | vn = cn; |
| 1082 | cn = vn->r.n = new_node(t.info); |
| 1083 | cn->a.n = vn; |
| 1084 | xtc = TC_OPERAND | TC_UOPPRE | TC_REGEXP; |
| 1085 | if (tc & (TC_OPERAND | TC_REGEXP)) { |
Rob Landley | ed830e8 | 2005-06-07 02:43:52 +0000 | [diff] [blame] | 1086 | xtc = TC_UOPPRE | TC_UOPPOST | TC_BINOP | TC_OPERAND | iexp; |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 1087 | /* one should be very careful with switch on tclass - |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1088 | * only simple tclasses should be used! */ |
| 1089 | switch (tc) { |
| 1090 | case TC_VARIABLE: |
| 1091 | case TC_ARRAY: |
| 1092 | cn->info = OC_VAR; |
| 1093 | if ((v = hash_search(ahash, t.string)) != NULL) { |
| 1094 | cn->info = OC_FNARG; |
| 1095 | cn->l.i = v->x.aidx; |
| 1096 | } else { |
| 1097 | cn->l.v = newvar(t.string); |
| 1098 | } |
| 1099 | if (tc & TC_ARRAY) { |
| 1100 | cn->info |= xS; |
| 1101 | cn->r.n = parse_expr(TC_ARRTERM); |
| 1102 | } |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1103 | break; |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 1104 | |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1105 | case TC_NUMBER: |
| 1106 | case TC_STRING: |
| 1107 | cn->info = OC_VAR; |
| 1108 | v = cn->l.v = xcalloc(sizeof(var), 1); |
| 1109 | if (tc & TC_NUMBER) |
| 1110 | setvar_i(v, t.number); |
| 1111 | else |
| 1112 | setvar_s(v, t.string); |
| 1113 | break; |
| 1114 | |
| 1115 | case TC_REGEXP: |
| 1116 | mk_re_node(t.string, cn, |
| 1117 | (regex_t *)xcalloc(sizeof(regex_t),2)); |
| 1118 | break; |
| 1119 | |
| 1120 | case TC_FUNCTION: |
| 1121 | cn->info = OC_FUNC; |
| 1122 | cn->r.f = newfunc(t.string); |
| 1123 | cn->l.n = condition(); |
| 1124 | break; |
| 1125 | |
| 1126 | case TC_SEQSTART: |
| 1127 | cn = vn->r.n = parse_expr(TC_SEQTERM); |
| 1128 | cn->a.n = vn; |
| 1129 | break; |
| 1130 | |
| 1131 | case TC_GETLINE: |
| 1132 | glptr = cn; |
| 1133 | xtc = TC_OPERAND | TC_UOPPRE | TC_BINOP | iexp; |
| 1134 | break; |
| 1135 | |
| 1136 | case TC_BUILTIN: |
| 1137 | cn->l.n = condition(); |
| 1138 | break; |
| 1139 | } |
| 1140 | } |
| 1141 | } |
| 1142 | } |
| 1143 | return sn.r.n; |
| 1144 | } |
| 1145 | |
| 1146 | /* add node to chain. Return ptr to alloc'd node */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1147 | static node *chain_node(unsigned long info) |
| 1148 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1149 | register node *n; |
| 1150 | |
| 1151 | if (! seq->first) |
| 1152 | seq->first = seq->last = new_node(0); |
| 1153 | |
| 1154 | if (seq->programname != programname) { |
| 1155 | seq->programname = programname; |
| 1156 | n = chain_node(OC_NEWSOURCE); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1157 | n->l.s = bb_xstrdup(programname); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1158 | } |
| 1159 | |
| 1160 | n = seq->last; |
| 1161 | n->info = info; |
| 1162 | seq->last = n->a.n = new_node(OC_DONE); |
| 1163 | |
| 1164 | return n; |
| 1165 | } |
| 1166 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1167 | static void chain_expr(unsigned long info) |
| 1168 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1169 | node *n; |
| 1170 | |
| 1171 | n = chain_node(info); |
| 1172 | n->l.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
| 1173 | if (t.tclass & TC_GRPTERM) |
| 1174 | rollback_token(); |
| 1175 | } |
| 1176 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1177 | static node *chain_loop(node *nn) |
| 1178 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1179 | node *n, *n2, *save_brk, *save_cont; |
| 1180 | |
| 1181 | save_brk = break_ptr; |
| 1182 | save_cont = continue_ptr; |
| 1183 | |
| 1184 | n = chain_node(OC_BR | Vx); |
| 1185 | continue_ptr = new_node(OC_EXEC); |
| 1186 | break_ptr = new_node(OC_EXEC); |
| 1187 | chain_group(); |
| 1188 | n2 = chain_node(OC_EXEC | Vx); |
| 1189 | n2->l.n = nn; |
| 1190 | n2->a.n = n; |
| 1191 | continue_ptr->a.n = n2; |
| 1192 | break_ptr->a.n = n->r.n = seq->last; |
| 1193 | |
| 1194 | continue_ptr = save_cont; |
| 1195 | break_ptr = save_brk; |
| 1196 | |
| 1197 | return n; |
| 1198 | } |
| 1199 | |
| 1200 | /* parse group and attach it to chain */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1201 | static void chain_group(void) |
| 1202 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1203 | unsigned long c; |
| 1204 | node *n, *n2, *n3; |
| 1205 | |
| 1206 | do { |
| 1207 | c = next_token(TC_GRPSEQ); |
| 1208 | } while (c & TC_NEWLINE); |
| 1209 | |
| 1210 | if (c & TC_GRPSTART) { |
| 1211 | while(next_token(TC_GRPSEQ | TC_GRPTERM) != TC_GRPTERM) { |
Glenn L McGrath | 1fbc909 | 2004-07-23 02:11:25 +0000 | [diff] [blame] | 1212 | if (t.tclass & TC_NEWLINE) continue; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1213 | rollback_token(); |
| 1214 | chain_group(); |
| 1215 | } |
| 1216 | } else if (c & (TC_OPSEQ | TC_OPTERM)) { |
| 1217 | rollback_token(); |
| 1218 | chain_expr(OC_EXEC | Vx); |
| 1219 | } else { /* TC_STATEMNT */ |
| 1220 | switch (t.info & OPCLSMASK) { |
| 1221 | case ST_IF: |
| 1222 | n = chain_node(OC_BR | Vx); |
| 1223 | n->l.n = condition(); |
| 1224 | chain_group(); |
| 1225 | n2 = chain_node(OC_EXEC); |
| 1226 | n->r.n = seq->last; |
| 1227 | if (next_token(TC_GRPSEQ | TC_GRPTERM | TC_ELSE)==TC_ELSE) { |
| 1228 | chain_group(); |
| 1229 | n2->a.n = seq->last; |
| 1230 | } else { |
| 1231 | rollback_token(); |
| 1232 | } |
| 1233 | break; |
| 1234 | |
| 1235 | case ST_WHILE: |
| 1236 | n2 = condition(); |
| 1237 | n = chain_loop(NULL); |
| 1238 | n->l.n = n2; |
| 1239 | break; |
| 1240 | |
| 1241 | case ST_DO: |
| 1242 | n2 = chain_node(OC_EXEC); |
| 1243 | n = chain_loop(NULL); |
| 1244 | n2->a.n = n->a.n; |
| 1245 | next_token(TC_WHILE); |
| 1246 | n->l.n = condition(); |
| 1247 | break; |
| 1248 | |
| 1249 | case ST_FOR: |
| 1250 | next_token(TC_SEQSTART); |
| 1251 | n2 = parse_expr(TC_SEMICOL | TC_SEQTERM); |
| 1252 | if (t.tclass & TC_SEQTERM) { /* for-in */ |
| 1253 | if ((n2->info & OPCLSMASK) != OC_IN) |
| 1254 | syntax_error(EMSG_UNEXP_TOKEN); |
| 1255 | n = chain_node(OC_WALKINIT | VV); |
| 1256 | n->l.n = n2->l.n; |
| 1257 | n->r.n = n2->r.n; |
| 1258 | n = chain_loop(NULL); |
| 1259 | n->info = OC_WALKNEXT | Vx; |
| 1260 | n->l.n = n2->l.n; |
| 1261 | } else { /* for(;;) */ |
| 1262 | n = chain_node(OC_EXEC | Vx); |
| 1263 | n->l.n = n2; |
| 1264 | n2 = parse_expr(TC_SEMICOL); |
| 1265 | n3 = parse_expr(TC_SEQTERM); |
| 1266 | n = chain_loop(n3); |
| 1267 | n->l.n = n2; |
| 1268 | if (! n2) |
| 1269 | n->info = OC_EXEC; |
| 1270 | } |
| 1271 | break; |
| 1272 | |
| 1273 | case OC_PRINT: |
| 1274 | case OC_PRINTF: |
| 1275 | n = chain_node(t.info); |
| 1276 | n->l.n = parse_expr(TC_OPTERM | TC_OUTRDR | TC_GRPTERM); |
| 1277 | if (t.tclass & TC_OUTRDR) { |
| 1278 | n->info |= t.info; |
| 1279 | n->r.n = parse_expr(TC_OPTERM | TC_GRPTERM); |
| 1280 | } |
| 1281 | if (t.tclass & TC_GRPTERM) |
| 1282 | rollback_token(); |
| 1283 | break; |
| 1284 | |
| 1285 | case OC_BREAK: |
| 1286 | n = chain_node(OC_EXEC); |
| 1287 | n->a.n = break_ptr; |
| 1288 | break; |
| 1289 | |
| 1290 | case OC_CONTINUE: |
| 1291 | n = chain_node(OC_EXEC); |
| 1292 | n->a.n = continue_ptr; |
| 1293 | break; |
| 1294 | |
| 1295 | /* delete, next, nextfile, return, exit */ |
| 1296 | default: |
| 1297 | chain_expr(t.info); |
| 1298 | |
| 1299 | } |
| 1300 | } |
| 1301 | } |
| 1302 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1303 | static void parse_program(char *p) |
| 1304 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1305 | unsigned long tclass; |
| 1306 | node *cn; |
| 1307 | func *f; |
| 1308 | var *v; |
| 1309 | |
| 1310 | pos = p; |
| 1311 | t.lineno = 1; |
| 1312 | while((tclass = next_token(TC_EOF | TC_OPSEQ | TC_GRPSTART | |
| 1313 | TC_OPTERM | TC_BEGIN | TC_END | TC_FUNCDECL)) != TC_EOF) { |
| 1314 | |
| 1315 | if (tclass & TC_OPTERM) |
| 1316 | continue; |
| 1317 | |
| 1318 | seq = &mainseq; |
| 1319 | if (tclass & TC_BEGIN) { |
| 1320 | seq = &beginseq; |
| 1321 | chain_group(); |
| 1322 | |
| 1323 | } else if (tclass & TC_END) { |
| 1324 | seq = &endseq; |
| 1325 | chain_group(); |
| 1326 | |
| 1327 | } else if (tclass & TC_FUNCDECL) { |
| 1328 | next_token(TC_FUNCTION); |
| 1329 | pos++; |
| 1330 | f = newfunc(t.string); |
| 1331 | f->body.first = NULL; |
| 1332 | f->nargs = 0; |
| 1333 | while(next_token(TC_VARIABLE | TC_SEQTERM) & TC_VARIABLE) { |
| 1334 | v = findvar(ahash, t.string); |
| 1335 | v->x.aidx = (f->nargs)++; |
| 1336 | |
| 1337 | if (next_token(TC_COMMA | TC_SEQTERM) & TC_SEQTERM) |
| 1338 | break; |
| 1339 | } |
| 1340 | seq = &(f->body); |
| 1341 | chain_group(); |
| 1342 | clear_array(ahash); |
| 1343 | |
| 1344 | } else if (tclass & TC_OPSEQ) { |
| 1345 | rollback_token(); |
| 1346 | cn = chain_node(OC_TEST); |
| 1347 | cn->l.n = parse_expr(TC_OPTERM | TC_EOF | TC_GRPSTART); |
| 1348 | if (t.tclass & TC_GRPSTART) { |
| 1349 | rollback_token(); |
| 1350 | chain_group(); |
| 1351 | } else { |
| 1352 | chain_node(OC_PRINT); |
| 1353 | } |
| 1354 | cn->r.n = mainseq.last; |
| 1355 | |
| 1356 | } else /* if (tclass & TC_GRPSTART) */ { |
| 1357 | rollback_token(); |
| 1358 | chain_group(); |
| 1359 | } |
| 1360 | } |
| 1361 | } |
| 1362 | |
| 1363 | |
| 1364 | /* -------- program execution part -------- */ |
| 1365 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1366 | static node *mk_splitter(char *s, tsplitter *spl) |
| 1367 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1368 | register regex_t *re, *ire; |
| 1369 | node *n; |
| 1370 | |
| 1371 | re = &spl->re[0]; |
| 1372 | ire = &spl->re[1]; |
| 1373 | n = &spl->n; |
| 1374 | if ((n->info && OPCLSMASK) == OC_REGEXP) { |
| 1375 | regfree(re); |
| 1376 | regfree(ire); |
| 1377 | } |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1378 | if (bb_strlen(s) > 1) { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1379 | mk_re_node(s, n, re); |
| 1380 | } else { |
| 1381 | n->info = (unsigned long) *s; |
| 1382 | } |
| 1383 | |
| 1384 | return n; |
| 1385 | } |
| 1386 | |
| 1387 | /* use node as a regular expression. Supplied with node ptr and regex_t |
Eric Andersen | aff114c | 2004-04-14 17:51:38 +0000 | [diff] [blame] | 1388 | * storage space. Return ptr to regex (if result points to preg, it should |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1389 | * be later regfree'd manually |
| 1390 | */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1391 | static regex_t *as_regex(node *op, regex_t *preg) |
| 1392 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1393 | var *v; |
| 1394 | char *s; |
| 1395 | |
| 1396 | if ((op->info & OPCLSMASK) == OC_REGEXP) { |
| 1397 | return icase ? op->r.ire : op->l.re; |
| 1398 | } else { |
| 1399 | v = nvalloc(1); |
| 1400 | s = getvar_s(evaluate(op, v)); |
| 1401 | xregcomp(preg, s, icase ? REG_EXTENDED | REG_ICASE : REG_EXTENDED); |
| 1402 | nvfree(v); |
| 1403 | return preg; |
| 1404 | } |
| 1405 | } |
| 1406 | |
| 1407 | /* gradually increasing buffer */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1408 | static void qrealloc(char **b, int n, int *size) |
| 1409 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1410 | if (! *b || n >= *size) |
| 1411 | *b = xrealloc(*b, *size = n + (n>>1) + 80); |
| 1412 | } |
| 1413 | |
| 1414 | /* resize field storage space */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1415 | static void fsrealloc(int size) |
| 1416 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1417 | static int maxfields = 0; |
| 1418 | int i; |
| 1419 | |
| 1420 | if (size >= maxfields) { |
| 1421 | i = maxfields; |
| 1422 | maxfields = size + 16; |
| 1423 | Fields = (var *)xrealloc(Fields, maxfields * sizeof(var)); |
| 1424 | for (; i<maxfields; i++) { |
| 1425 | Fields[i].type = VF_SPECIAL; |
| 1426 | Fields[i].string = NULL; |
| 1427 | } |
| 1428 | } |
| 1429 | |
| 1430 | if (size < nfields) { |
| 1431 | for (i=size; i<nfields; i++) { |
| 1432 | clrvar(Fields+i); |
| 1433 | } |
| 1434 | } |
| 1435 | nfields = size; |
| 1436 | } |
| 1437 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1438 | static int awk_split(char *s, node *spl, char **slist) |
| 1439 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1440 | int l, n=0; |
| 1441 | char c[4]; |
| 1442 | char *s1; |
| 1443 | regmatch_t pmatch[2]; |
| 1444 | |
| 1445 | /* in worst case, each char would be a separate field */ |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1446 | *slist = s1 = bb_xstrndup(s, bb_strlen(s) * 2 + 3); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1447 | |
| 1448 | c[0] = c[1] = (char)spl->info; |
| 1449 | c[2] = c[3] = '\0'; |
| 1450 | if (*getvar_s(V[RS]) == '\0') c[2] = '\n'; |
| 1451 | |
| 1452 | if ((spl->info & OPCLSMASK) == OC_REGEXP) { /* regex split */ |
| 1453 | while (*s) { |
| 1454 | l = strcspn(s, c+2); |
| 1455 | if (regexec(icase ? spl->r.ire : spl->l.re, s, 1, pmatch, 0) == 0 && |
| 1456 | pmatch[0].rm_so <= l) { |
| 1457 | l = pmatch[0].rm_so; |
| 1458 | if (pmatch[0].rm_eo == 0) { l++; pmatch[0].rm_eo++; } |
| 1459 | } else { |
| 1460 | pmatch[0].rm_eo = l; |
| 1461 | if (*(s+l)) pmatch[0].rm_eo++; |
| 1462 | } |
| 1463 | |
| 1464 | memcpy(s1, s, l); |
| 1465 | *(s1+l) = '\0'; |
| 1466 | nextword(&s1); |
| 1467 | s += pmatch[0].rm_eo; |
| 1468 | n++; |
| 1469 | } |
| 1470 | } else if (c[0] == '\0') { /* null split */ |
| 1471 | while(*s) { |
| 1472 | *(s1++) = *(s++); |
| 1473 | *(s1++) = '\0'; |
| 1474 | n++; |
| 1475 | } |
| 1476 | } else if (c[0] != ' ') { /* single-character split */ |
| 1477 | if (icase) { |
| 1478 | c[0] = toupper(c[0]); |
| 1479 | c[1] = tolower(c[1]); |
| 1480 | } |
| 1481 | if (*s1) n++; |
| 1482 | while ((s1 = strpbrk(s1, c))) { |
| 1483 | *(s1++) = '\0'; |
| 1484 | n++; |
| 1485 | } |
| 1486 | } else { /* space split */ |
| 1487 | while (*s) { |
| 1488 | while (isspace(*s)) s++; |
| 1489 | if (! *s) break; |
| 1490 | n++; |
| 1491 | while (*s && !isspace(*s)) |
| 1492 | *(s1++) = *(s++); |
| 1493 | *(s1++) = '\0'; |
| 1494 | } |
| 1495 | } |
| 1496 | return n; |
| 1497 | } |
| 1498 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1499 | static void split_f0(void) |
| 1500 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1501 | static char *fstrings = NULL; |
| 1502 | int i, n; |
| 1503 | char *s; |
| 1504 | |
| 1505 | if (is_f0_split) |
| 1506 | return; |
| 1507 | |
| 1508 | is_f0_split = TRUE; |
Aaron Lehmann | a170e1c | 2002-11-28 11:27:31 +0000 | [diff] [blame] | 1509 | free(fstrings); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1510 | fsrealloc(0); |
| 1511 | n = awk_split(getvar_s(V[F0]), &fsplitter.n, &fstrings); |
| 1512 | fsrealloc(n); |
| 1513 | s = fstrings; |
| 1514 | for (i=0; i<n; i++) { |
| 1515 | Fields[i].string = nextword(&s); |
| 1516 | Fields[i].type |= (VF_FSTR | VF_USER | VF_DIRTY); |
| 1517 | } |
| 1518 | |
| 1519 | /* set NF manually to avoid side effects */ |
| 1520 | clrvar(V[NF]); |
| 1521 | V[NF]->type = VF_NUMBER | VF_SPECIAL; |
| 1522 | V[NF]->number = nfields; |
| 1523 | } |
| 1524 | |
| 1525 | /* perform additional actions when some internal variables changed */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1526 | static void handle_special(var *v) |
| 1527 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1528 | int n; |
| 1529 | char *b, *sep, *s; |
| 1530 | int sl, l, len, i, bsize; |
| 1531 | |
| 1532 | if (! (v->type & VF_SPECIAL)) |
| 1533 | return; |
| 1534 | |
| 1535 | if (v == V[NF]) { |
| 1536 | n = (int)getvar_i(v); |
| 1537 | fsrealloc(n); |
| 1538 | |
| 1539 | /* recalculate $0 */ |
| 1540 | sep = getvar_s(V[OFS]); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1541 | sl = bb_strlen(sep); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1542 | b = NULL; |
| 1543 | len = 0; |
| 1544 | for (i=0; i<n; i++) { |
| 1545 | s = getvar_s(&Fields[i]); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1546 | l = bb_strlen(s); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1547 | if (b) { |
| 1548 | memcpy(b+len, sep, sl); |
| 1549 | len += sl; |
| 1550 | } |
| 1551 | qrealloc(&b, len+l+sl, &bsize); |
| 1552 | memcpy(b+len, s, l); |
| 1553 | len += l; |
| 1554 | } |
Glenn L McGrath | ca29ffc | 2004-09-24 09:24:27 +0000 | [diff] [blame] | 1555 | if (b) b[len] = '\0'; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1556 | setvar_p(V[F0], b); |
| 1557 | is_f0_split = TRUE; |
| 1558 | |
| 1559 | } else if (v == V[F0]) { |
| 1560 | is_f0_split = FALSE; |
| 1561 | |
| 1562 | } else if (v == V[FS]) { |
| 1563 | mk_splitter(getvar_s(v), &fsplitter); |
| 1564 | |
| 1565 | } else if (v == V[RS]) { |
| 1566 | mk_splitter(getvar_s(v), &rsplitter); |
| 1567 | |
| 1568 | } else if (v == V[IGNORECASE]) { |
| 1569 | icase = istrue(v); |
| 1570 | |
| 1571 | } else { /* $n */ |
| 1572 | n = getvar_i(V[NF]); |
| 1573 | setvar_i(V[NF], n > v-Fields ? n : v-Fields+1); |
| 1574 | /* right here v is invalid. Just to note... */ |
| 1575 | } |
| 1576 | } |
| 1577 | |
| 1578 | /* step through func/builtin/etc arguments */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1579 | static node *nextarg(node **pn) |
| 1580 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1581 | node *n; |
| 1582 | |
| 1583 | n = *pn; |
| 1584 | if (n && (n->info & OPCLSMASK) == OC_COMMA) { |
| 1585 | *pn = n->r.n; |
| 1586 | n = n->l.n; |
| 1587 | } else { |
| 1588 | *pn = NULL; |
| 1589 | } |
| 1590 | return n; |
| 1591 | } |
| 1592 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1593 | static void hashwalk_init(var *v, xhash *array) |
| 1594 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1595 | char **w; |
| 1596 | hash_item *hi; |
| 1597 | int i; |
| 1598 | |
| 1599 | if (v->type & VF_WALK) |
| 1600 | free(v->x.walker); |
| 1601 | |
| 1602 | v->type |= VF_WALK; |
| 1603 | w = v->x.walker = (char **)xcalloc(2 + 2*sizeof(char *) + array->glen, 1); |
| 1604 | *w = *(w+1) = (char *)(w + 2); |
| 1605 | for (i=0; i<array->csize; i++) { |
| 1606 | hi = array->items[i]; |
| 1607 | while(hi) { |
| 1608 | strcpy(*w, hi->name); |
| 1609 | nextword(w); |
| 1610 | hi = hi->next; |
| 1611 | } |
| 1612 | } |
| 1613 | } |
| 1614 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1615 | static int hashwalk_next(var *v) |
| 1616 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1617 | char **w; |
| 1618 | |
| 1619 | w = v->x.walker; |
| 1620 | if (*(w+1) == *w) |
| 1621 | return FALSE; |
| 1622 | |
| 1623 | setvar_s(v, nextword(w+1)); |
| 1624 | return TRUE; |
| 1625 | } |
| 1626 | |
| 1627 | /* evaluate node, return 1 when result is true, 0 otherwise */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1628 | static int ptest(node *pattern) |
| 1629 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1630 | static var v; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1631 | return istrue(evaluate(pattern, &v)); |
| 1632 | } |
| 1633 | |
| 1634 | /* read next record from stream rsm into a variable v */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1635 | static int awk_getline(rstream *rsm, var *v) |
| 1636 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1637 | char *b; |
| 1638 | regmatch_t pmatch[2]; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 1639 | int a, p, pp=0, size; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1640 | int fd, so, eo, r, rp; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 1641 | char c, *m, *s; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1642 | |
| 1643 | /* we're using our own buffer since we need access to accumulating |
| 1644 | * characters |
| 1645 | */ |
| 1646 | fd = fileno(rsm->F); |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 1647 | m = rsm->buffer; |
| 1648 | a = rsm->adv; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1649 | p = rsm->pos; |
| 1650 | size = rsm->size; |
| 1651 | c = (char) rsplitter.n.info; |
| 1652 | rp = 0; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 1653 | |
| 1654 | if (! m) qrealloc(&m, 256, &size); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1655 | do { |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 1656 | b = m + a; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1657 | so = eo = p; |
| 1658 | r = 1; |
| 1659 | if (p > 0) { |
| 1660 | if ((rsplitter.n.info & OPCLSMASK) == OC_REGEXP) { |
| 1661 | if (regexec(icase ? rsplitter.n.r.ire : rsplitter.n.l.re, |
| 1662 | b, 1, pmatch, 0) == 0) { |
| 1663 | so = pmatch[0].rm_so; |
| 1664 | eo = pmatch[0].rm_eo; |
| 1665 | if (b[eo] != '\0') |
| 1666 | break; |
| 1667 | } |
| 1668 | } else if (c != '\0') { |
| 1669 | s = strchr(b+pp, c); |
| 1670 | if (s) { |
| 1671 | so = eo = s-b; |
| 1672 | eo++; |
| 1673 | break; |
| 1674 | } |
| 1675 | } else { |
| 1676 | while (b[rp] == '\n') |
| 1677 | rp++; |
| 1678 | s = strstr(b+rp, "\n\n"); |
| 1679 | if (s) { |
| 1680 | so = eo = s-b; |
| 1681 | while (b[eo] == '\n') eo++; |
| 1682 | if (b[eo] != '\0') |
| 1683 | break; |
| 1684 | } |
| 1685 | } |
| 1686 | } |
| 1687 | |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 1688 | if (a > 0) { |
| 1689 | memmove(m, (const void *)(m+a), p+1); |
| 1690 | b = m; |
| 1691 | a = 0; |
| 1692 | } |
| 1693 | |
| 1694 | qrealloc(&m, a+p+128, &size); |
| 1695 | b = m + a; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1696 | pp = p; |
| 1697 | p += safe_read(fd, b+p, size-p-1); |
| 1698 | if (p < pp) { |
| 1699 | p = 0; |
| 1700 | r = 0; |
| 1701 | setvar_i(V[ERRNO], errno); |
| 1702 | } |
| 1703 | b[p] = '\0'; |
| 1704 | |
| 1705 | } while (p > pp); |
| 1706 | |
| 1707 | if (p == 0) { |
| 1708 | r--; |
| 1709 | } else { |
| 1710 | c = b[so]; b[so] = '\0'; |
| 1711 | setvar_s(v, b+rp); |
| 1712 | v->type |= VF_USER; |
| 1713 | b[so] = c; |
| 1714 | c = b[eo]; b[eo] = '\0'; |
| 1715 | setvar_s(V[RT], b+so); |
| 1716 | b[eo] = c; |
| 1717 | } |
| 1718 | |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 1719 | rsm->buffer = m; |
| 1720 | rsm->adv = a + eo; |
| 1721 | rsm->pos = p - eo; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1722 | rsm->size = size; |
| 1723 | |
| 1724 | return r; |
| 1725 | } |
| 1726 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1727 | static int fmt_num(char *b, int size, char *format, double n, int int_as_int) |
| 1728 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1729 | int r=0; |
| 1730 | char c, *s=format; |
| 1731 | |
| 1732 | if (int_as_int && n == (int)n) { |
| 1733 | r = snprintf(b, size, "%d", (int)n); |
| 1734 | } else { |
| 1735 | do { c = *s; } while (*s && *++s); |
| 1736 | if (strchr("diouxX", c)) { |
| 1737 | r = snprintf(b, size, format, (int)n); |
| 1738 | } else if (strchr("eEfgG", c)) { |
| 1739 | r = snprintf(b, size, format, n); |
| 1740 | } else { |
| 1741 | runtime_error(EMSG_INV_FMT); |
| 1742 | } |
| 1743 | } |
| 1744 | return r; |
| 1745 | } |
| 1746 | |
| 1747 | |
| 1748 | /* formatted output into an allocated buffer, return ptr to buffer */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1749 | static char *awk_printf(node *n) |
| 1750 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1751 | char *b = NULL; |
| 1752 | char *fmt, *s, *s1, *f; |
| 1753 | int i, j, incr, bsize; |
| 1754 | char c, c1; |
| 1755 | var *v, *arg; |
| 1756 | |
| 1757 | v = nvalloc(1); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1758 | fmt = f = bb_xstrdup(getvar_s(evaluate(nextarg(&n), v))); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1759 | |
| 1760 | i = 0; |
| 1761 | while (*f) { |
| 1762 | s = f; |
| 1763 | while (*f && (*f != '%' || *(++f) == '%')) |
| 1764 | f++; |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 1765 | while (*f && !isalpha(*f)) |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1766 | f++; |
| 1767 | |
| 1768 | incr = (f - s) + MAXVARFMT; |
| 1769 | qrealloc(&b, incr+i, &bsize); |
| 1770 | c = *f; if (c != '\0') f++; |
| 1771 | c1 = *f ; *f = '\0'; |
| 1772 | arg = evaluate(nextarg(&n), v); |
| 1773 | |
| 1774 | j = i; |
| 1775 | if (c == 'c' || !c) { |
| 1776 | i += sprintf(b+i, s, |
| 1777 | is_numeric(arg) ? (char)getvar_i(arg) : *getvar_s(arg)); |
| 1778 | |
| 1779 | } else if (c == 's') { |
| 1780 | s1 = getvar_s(arg); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1781 | qrealloc(&b, incr+i+bb_strlen(s1), &bsize); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1782 | i += sprintf(b+i, s, s1); |
| 1783 | |
| 1784 | } else { |
| 1785 | i += fmt_num(b+i, incr, s, getvar_i(arg), FALSE); |
| 1786 | } |
| 1787 | *f = c1; |
| 1788 | |
| 1789 | /* if there was an error while sprintf, return value is negative */ |
| 1790 | if (i < j) i = j; |
| 1791 | |
| 1792 | } |
| 1793 | |
| 1794 | b = xrealloc(b, i+1); |
| 1795 | free(fmt); |
| 1796 | nvfree(v); |
| 1797 | b[i] = '\0'; |
| 1798 | return b; |
| 1799 | } |
| 1800 | |
| 1801 | /* common substitution routine |
| 1802 | * replace (nm) substring of (src) that match (n) with (repl), store |
| 1803 | * result into (dest), return number of substitutions. If nm=0, replace |
| 1804 | * all matches. If src or dst is NULL, use $0. If ex=TRUE, enable |
| 1805 | * subexpression matching (\1-\9) |
| 1806 | */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1807 | static int awk_sub(node *rn, char *repl, int nm, var *src, var *dest, int ex) |
| 1808 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1809 | char *ds = NULL; |
| 1810 | char *sp, *s; |
| 1811 | int c, i, j, di, rl, so, eo, nbs, n, dssize; |
| 1812 | regmatch_t pmatch[10]; |
| 1813 | regex_t sreg, *re; |
| 1814 | |
| 1815 | re = as_regex(rn, &sreg); |
| 1816 | if (! src) src = V[F0]; |
| 1817 | if (! dest) dest = V[F0]; |
| 1818 | |
| 1819 | i = di = 0; |
| 1820 | sp = getvar_s(src); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1821 | rl = bb_strlen(repl); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1822 | while (regexec(re, sp, 10, pmatch, sp==getvar_s(src) ? 0:REG_NOTBOL) == 0) { |
| 1823 | so = pmatch[0].rm_so; |
| 1824 | eo = pmatch[0].rm_eo; |
| 1825 | |
| 1826 | qrealloc(&ds, di + eo + rl, &dssize); |
| 1827 | memcpy(ds + di, sp, eo); |
| 1828 | di += eo; |
| 1829 | if (++i >= nm) { |
| 1830 | /* replace */ |
| 1831 | di -= (eo - so); |
| 1832 | nbs = 0; |
| 1833 | for (s = repl; *s; s++) { |
| 1834 | ds[di++] = c = *s; |
| 1835 | if (c == '\\') { |
| 1836 | nbs++; |
| 1837 | continue; |
| 1838 | } |
| 1839 | if (c == '&' || (ex && c >= '0' && c <= '9')) { |
| 1840 | di -= ((nbs + 3) >> 1); |
| 1841 | j = 0; |
| 1842 | if (c != '&') { |
| 1843 | j = c - '0'; |
| 1844 | nbs++; |
| 1845 | } |
| 1846 | if (nbs % 2) { |
| 1847 | ds[di++] = c; |
| 1848 | } else { |
| 1849 | n = pmatch[j].rm_eo - pmatch[j].rm_so; |
| 1850 | qrealloc(&ds, di + rl + n, &dssize); |
| 1851 | memcpy(ds + di, sp + pmatch[j].rm_so, n); |
| 1852 | di += n; |
| 1853 | } |
| 1854 | } |
| 1855 | nbs = 0; |
| 1856 | } |
| 1857 | } |
| 1858 | |
| 1859 | sp += eo; |
| 1860 | if (i == nm) break; |
| 1861 | if (eo == so) { |
| 1862 | if (! (ds[di++] = *sp++)) break; |
| 1863 | } |
| 1864 | } |
| 1865 | |
| 1866 | qrealloc(&ds, di + strlen(sp), &dssize); |
| 1867 | strcpy(ds + di, sp); |
| 1868 | setvar_p(dest, ds); |
| 1869 | if (re == &sreg) regfree(re); |
| 1870 | return i; |
| 1871 | } |
| 1872 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 1873 | static var *exec_builtin(node *op, var *res) |
| 1874 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1875 | int (*to_xxx)(int); |
| 1876 | var *tv; |
| 1877 | node *an[4]; |
| 1878 | var *av[4]; |
| 1879 | char *as[4]; |
| 1880 | regmatch_t pmatch[2]; |
| 1881 | regex_t sreg, *re; |
| 1882 | static tsplitter tspl; |
| 1883 | node *spl; |
| 1884 | unsigned long isr, info; |
| 1885 | int nargs; |
| 1886 | time_t tt; |
| 1887 | char *s, *s1; |
| 1888 | int i, l, ll, n; |
| 1889 | |
| 1890 | tv = nvalloc(4); |
| 1891 | isr = info = op->info; |
| 1892 | op = op->l.n; |
| 1893 | |
| 1894 | av[2] = av[3] = NULL; |
| 1895 | for (i=0 ; i<4 && op ; i++) { |
| 1896 | an[i] = nextarg(&op); |
| 1897 | if (isr & 0x09000000) av[i] = evaluate(an[i], &tv[i]); |
| 1898 | if (isr & 0x08000000) as[i] = getvar_s(av[i]); |
| 1899 | isr >>= 1; |
| 1900 | } |
| 1901 | |
| 1902 | nargs = i; |
| 1903 | if (nargs < (info >> 30)) |
| 1904 | runtime_error(EMSG_TOO_FEW_ARGS); |
| 1905 | |
| 1906 | switch (info & OPNMASK) { |
| 1907 | |
| 1908 | case B_a2: |
| 1909 | #ifdef CONFIG_FEATURE_AWK_MATH |
| 1910 | setvar_i(res, atan2(getvar_i(av[i]), getvar_i(av[1]))); |
| 1911 | #else |
| 1912 | runtime_error(EMSG_NO_MATH); |
| 1913 | #endif |
| 1914 | break; |
| 1915 | |
| 1916 | case B_sp: |
| 1917 | if (nargs > 2) { |
| 1918 | spl = (an[2]->info & OPCLSMASK) == OC_REGEXP ? |
| 1919 | an[2] : mk_splitter(getvar_s(evaluate(an[2], &tv[2])), &tspl); |
| 1920 | } else { |
| 1921 | spl = &fsplitter.n; |
| 1922 | } |
| 1923 | |
| 1924 | n = awk_split(as[0], spl, &s); |
| 1925 | s1 = s; |
| 1926 | clear_array(iamarray(av[1])); |
| 1927 | for (i=1; i<=n; i++) |
| 1928 | setari_u(av[1], i, nextword(&s1)); |
| 1929 | free(s); |
| 1930 | setvar_i(res, n); |
| 1931 | break; |
| 1932 | |
| 1933 | case B_ss: |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1934 | l = bb_strlen(as[0]); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1935 | i = getvar_i(av[1]) - 1; |
| 1936 | if (i>l) i=l; if (i<0) i=0; |
| 1937 | n = (nargs > 2) ? getvar_i(av[2]) : l-i; |
| 1938 | if (n<0) n=0; |
| 1939 | s = xmalloc(n+1); |
| 1940 | strncpy(s, as[0]+i, n); |
| 1941 | s[n] = '\0'; |
| 1942 | setvar_p(res, s); |
| 1943 | break; |
| 1944 | |
| 1945 | case B_lo: |
| 1946 | to_xxx = tolower; |
| 1947 | goto lo_cont; |
| 1948 | |
| 1949 | case B_up: |
| 1950 | to_xxx = toupper; |
| 1951 | lo_cont: |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1952 | s1 = s = bb_xstrdup(as[0]); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1953 | while (*s1) { |
| 1954 | *s1 = (*to_xxx)(*s1); |
| 1955 | s1++; |
| 1956 | } |
| 1957 | setvar_p(res, s); |
| 1958 | break; |
| 1959 | |
| 1960 | case B_ix: |
| 1961 | n = 0; |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1962 | ll = bb_strlen(as[1]); |
| 1963 | l = bb_strlen(as[0]) - ll; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 1964 | if (ll > 0 && l >= 0) { |
| 1965 | if (! icase) { |
| 1966 | s = strstr(as[0], as[1]); |
| 1967 | if (s) n = (s - as[0]) + 1; |
| 1968 | } else { |
| 1969 | /* this piece of code is terribly slow and |
| 1970 | * really should be rewritten |
| 1971 | */ |
| 1972 | for (i=0; i<=l; i++) { |
| 1973 | if (strncasecmp(as[0]+i, as[1], ll) == 0) { |
| 1974 | n = i+1; |
| 1975 | break; |
| 1976 | } |
| 1977 | } |
| 1978 | } |
| 1979 | } |
| 1980 | setvar_i(res, n); |
| 1981 | break; |
| 1982 | |
| 1983 | case B_ti: |
| 1984 | if (nargs > 1) |
| 1985 | tt = getvar_i(av[1]); |
| 1986 | else |
| 1987 | time(&tt); |
| 1988 | s = (nargs > 0) ? as[0] : "%a %b %d %H:%M:%S %Z %Y"; |
| 1989 | i = strftime(buf, MAXVARFMT, s, localtime(&tt)); |
| 1990 | buf[i] = '\0'; |
| 1991 | setvar_s(res, buf); |
| 1992 | break; |
| 1993 | |
| 1994 | case B_ma: |
| 1995 | re = as_regex(an[1], &sreg); |
| 1996 | n = regexec(re, as[0], 1, pmatch, 0); |
| 1997 | if (n == 0) { |
| 1998 | pmatch[0].rm_so++; |
| 1999 | pmatch[0].rm_eo++; |
| 2000 | } else { |
| 2001 | pmatch[0].rm_so = 0; |
| 2002 | pmatch[0].rm_eo = -1; |
| 2003 | } |
| 2004 | setvar_i(newvar("RSTART"), pmatch[0].rm_so); |
| 2005 | setvar_i(newvar("RLENGTH"), pmatch[0].rm_eo - pmatch[0].rm_so); |
| 2006 | setvar_i(res, pmatch[0].rm_so); |
| 2007 | if (re == &sreg) regfree(re); |
| 2008 | break; |
| 2009 | |
| 2010 | case B_ge: |
| 2011 | awk_sub(an[0], as[1], getvar_i(av[2]), av[3], res, TRUE); |
| 2012 | break; |
| 2013 | |
| 2014 | case B_gs: |
| 2015 | setvar_i(res, awk_sub(an[0], as[1], 0, av[2], av[2], FALSE)); |
| 2016 | break; |
| 2017 | |
| 2018 | case B_su: |
| 2019 | setvar_i(res, awk_sub(an[0], as[1], 1, av[2], av[2], FALSE)); |
| 2020 | break; |
| 2021 | } |
| 2022 | |
| 2023 | nvfree(tv); |
| 2024 | return res; |
| 2025 | } |
| 2026 | |
| 2027 | /* |
| 2028 | * Evaluate node - the heart of the program. Supplied with subtree |
| 2029 | * and place where to store result. returns ptr to result. |
| 2030 | */ |
| 2031 | #define XC(n) ((n) >> 8) |
| 2032 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 2033 | static var *evaluate(node *op, var *res) |
| 2034 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2035 | /* This procedure is recursive so we should count every byte */ |
| 2036 | static var *fnargs = NULL; |
| 2037 | static unsigned int seed = 1; |
| 2038 | static regex_t sreg; |
| 2039 | node *op1; |
| 2040 | var *v1; |
| 2041 | union { |
| 2042 | var *v; |
| 2043 | char *s; |
| 2044 | double d; |
| 2045 | int i; |
| 2046 | } L, R; |
| 2047 | unsigned long opinfo; |
| 2048 | short opn; |
| 2049 | union { |
| 2050 | char *s; |
| 2051 | rstream *rsm; |
| 2052 | FILE *F; |
| 2053 | var *v; |
| 2054 | regex_t *re; |
| 2055 | unsigned long info; |
| 2056 | } X; |
| 2057 | |
| 2058 | if (! op) |
| 2059 | return setvar_s(res, NULL); |
| 2060 | |
| 2061 | v1 = nvalloc(2); |
| 2062 | |
| 2063 | while (op) { |
| 2064 | |
| 2065 | opinfo = op->info; |
| 2066 | opn = (short)(opinfo & OPNMASK); |
| 2067 | lineno = op->lineno; |
| 2068 | |
| 2069 | /* execute inevitable things */ |
| 2070 | op1 = op->l.n; |
| 2071 | if (opinfo & OF_RES1) X.v = L.v = evaluate(op1, v1); |
| 2072 | if (opinfo & OF_RES2) R.v = evaluate(op->r.n, v1+1); |
| 2073 | if (opinfo & OF_STR1) L.s = getvar_s(L.v); |
| 2074 | if (opinfo & OF_STR2) R.s = getvar_s(R.v); |
| 2075 | if (opinfo & OF_NUM1) L.d = getvar_i(L.v); |
| 2076 | |
| 2077 | switch (XC(opinfo & OPCLSMASK)) { |
| 2078 | |
| 2079 | /* -- iterative node type -- */ |
| 2080 | |
| 2081 | /* test pattern */ |
| 2082 | case XC( OC_TEST ): |
| 2083 | if ((op1->info & OPCLSMASK) == OC_COMMA) { |
| 2084 | /* it's range pattern */ |
| 2085 | if ((opinfo & OF_CHECKED) || ptest(op1->l.n)) { |
| 2086 | op->info |= OF_CHECKED; |
| 2087 | if (ptest(op1->r.n)) |
| 2088 | op->info &= ~OF_CHECKED; |
| 2089 | |
| 2090 | op = op->a.n; |
| 2091 | } else { |
| 2092 | op = op->r.n; |
| 2093 | } |
| 2094 | } else { |
| 2095 | op = (ptest(op1)) ? op->a.n : op->r.n; |
| 2096 | } |
| 2097 | break; |
| 2098 | |
| 2099 | /* just evaluate an expression, also used as unconditional jump */ |
| 2100 | case XC( OC_EXEC ): |
| 2101 | break; |
| 2102 | |
| 2103 | /* branch, used in if-else and various loops */ |
| 2104 | case XC( OC_BR ): |
| 2105 | op = istrue(L.v) ? op->a.n : op->r.n; |
| 2106 | break; |
| 2107 | |
| 2108 | /* initialize for-in loop */ |
| 2109 | case XC( OC_WALKINIT ): |
| 2110 | hashwalk_init(L.v, iamarray(R.v)); |
| 2111 | break; |
| 2112 | |
| 2113 | /* get next array item */ |
| 2114 | case XC( OC_WALKNEXT ): |
| 2115 | op = hashwalk_next(L.v) ? op->a.n : op->r.n; |
| 2116 | break; |
| 2117 | |
| 2118 | case XC( OC_PRINT ): |
| 2119 | case XC( OC_PRINTF ): |
| 2120 | X.F = stdout; |
| 2121 | if (op->r.n) { |
| 2122 | X.rsm = newfile(R.s); |
| 2123 | if (! X.rsm->F) { |
| 2124 | if (opn == '|') { |
| 2125 | if((X.rsm->F = popen(R.s, "w")) == NULL) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2126 | bb_perror_msg_and_die("popen"); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2127 | X.rsm->is_pipe = 1; |
| 2128 | } else { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2129 | X.rsm->F = bb_xfopen(R.s, opn=='w' ? "w" : "a"); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2130 | } |
| 2131 | } |
| 2132 | X.F = X.rsm->F; |
| 2133 | } |
| 2134 | |
| 2135 | if ((opinfo & OPCLSMASK) == OC_PRINT) { |
| 2136 | if (! op1) { |
| 2137 | fputs(getvar_s(V[F0]), X.F); |
| 2138 | } else { |
| 2139 | while (op1) { |
| 2140 | L.v = evaluate(nextarg(&op1), v1); |
| 2141 | if (L.v->type & VF_NUMBER) { |
| 2142 | fmt_num(buf, MAXVARFMT, getvar_s(V[OFMT]), |
| 2143 | getvar_i(L.v), TRUE); |
| 2144 | fputs(buf, X.F); |
| 2145 | } else { |
| 2146 | fputs(getvar_s(L.v), X.F); |
| 2147 | } |
| 2148 | |
| 2149 | if (op1) fputs(getvar_s(V[OFS]), X.F); |
| 2150 | } |
| 2151 | } |
| 2152 | fputs(getvar_s(V[ORS]), X.F); |
| 2153 | |
| 2154 | } else { /* OC_PRINTF */ |
| 2155 | L.s = awk_printf(op1); |
| 2156 | fputs(L.s, X.F); |
| 2157 | free(L.s); |
| 2158 | } |
| 2159 | fflush(X.F); |
| 2160 | break; |
| 2161 | |
| 2162 | case XC( OC_DELETE ): |
| 2163 | X.info = op1->info & OPCLSMASK; |
| 2164 | if (X.info == OC_VAR) { |
| 2165 | R.v = op1->l.v; |
| 2166 | } else if (X.info == OC_FNARG) { |
| 2167 | R.v = &fnargs[op1->l.i]; |
| 2168 | } else { |
| 2169 | runtime_error(EMSG_NOT_ARRAY); |
| 2170 | } |
| 2171 | |
| 2172 | if (op1->r.n) { |
| 2173 | clrvar(L.v); |
| 2174 | L.s = getvar_s(evaluate(op1->r.n, v1)); |
| 2175 | hash_remove(iamarray(R.v), L.s); |
| 2176 | } else { |
| 2177 | clear_array(iamarray(R.v)); |
| 2178 | } |
| 2179 | break; |
| 2180 | |
| 2181 | case XC( OC_NEWSOURCE ): |
| 2182 | programname = op->l.s; |
| 2183 | break; |
| 2184 | |
| 2185 | case XC( OC_RETURN ): |
| 2186 | copyvar(res, L.v); |
| 2187 | break; |
| 2188 | |
| 2189 | case XC( OC_NEXTFILE ): |
| 2190 | nextfile = TRUE; |
| 2191 | case XC( OC_NEXT ): |
| 2192 | nextrec = TRUE; |
| 2193 | case XC( OC_DONE ): |
| 2194 | clrvar(res); |
| 2195 | break; |
| 2196 | |
| 2197 | case XC( OC_EXIT ): |
| 2198 | awk_exit(L.d); |
| 2199 | |
| 2200 | /* -- recursive node type -- */ |
| 2201 | |
| 2202 | case XC( OC_VAR ): |
| 2203 | L.v = op->l.v; |
| 2204 | if (L.v == V[NF]) |
| 2205 | split_f0(); |
| 2206 | goto v_cont; |
| 2207 | |
| 2208 | case XC( OC_FNARG ): |
| 2209 | L.v = &fnargs[op->l.i]; |
| 2210 | |
| 2211 | v_cont: |
| 2212 | res = (op->r.n) ? findvar(iamarray(L.v), R.s) : L.v; |
| 2213 | break; |
| 2214 | |
| 2215 | case XC( OC_IN ): |
| 2216 | setvar_i(res, hash_search(iamarray(R.v), L.s) ? 1 : 0); |
| 2217 | break; |
| 2218 | |
| 2219 | case XC( OC_REGEXP ): |
| 2220 | op1 = op; |
| 2221 | L.s = getvar_s(V[F0]); |
| 2222 | goto re_cont; |
| 2223 | |
| 2224 | case XC( OC_MATCH ): |
| 2225 | op1 = op->r.n; |
| 2226 | re_cont: |
| 2227 | X.re = as_regex(op1, &sreg); |
| 2228 | R.i = regexec(X.re, L.s, 0, NULL, 0); |
| 2229 | if (X.re == &sreg) regfree(X.re); |
| 2230 | setvar_i(res, (R.i == 0 ? 1 : 0) ^ (opn == '!' ? 1 : 0)); |
| 2231 | break; |
| 2232 | |
| 2233 | case XC( OC_MOVE ): |
| 2234 | /* if source is a temporary string, jusk relink it to dest */ |
| 2235 | if (R.v == v1+1 && R.v->string) { |
| 2236 | res = setvar_p(L.v, R.v->string); |
| 2237 | R.v->string = NULL; |
| 2238 | } else { |
| 2239 | res = copyvar(L.v, R.v); |
| 2240 | } |
| 2241 | break; |
| 2242 | |
| 2243 | case XC( OC_TERNARY ): |
| 2244 | if ((op->r.n->info & OPCLSMASK) != OC_COLON) |
| 2245 | runtime_error(EMSG_POSSIBLE_ERROR); |
| 2246 | res = evaluate(istrue(L.v) ? op->r.n->l.n : op->r.n->r.n, res); |
| 2247 | break; |
| 2248 | |
| 2249 | case XC( OC_FUNC ): |
| 2250 | if (! op->r.f->body.first) |
| 2251 | runtime_error(EMSG_UNDEF_FUNC); |
| 2252 | |
| 2253 | X.v = R.v = nvalloc(op->r.f->nargs+1); |
| 2254 | while (op1) { |
| 2255 | L.v = evaluate(nextarg(&op1), v1); |
| 2256 | copyvar(R.v, L.v); |
| 2257 | R.v->type |= VF_CHILD; |
| 2258 | R.v->x.parent = L.v; |
| 2259 | if (++R.v - X.v >= op->r.f->nargs) |
| 2260 | break; |
| 2261 | } |
| 2262 | |
| 2263 | R.v = fnargs; |
| 2264 | fnargs = X.v; |
| 2265 | |
| 2266 | L.s = programname; |
| 2267 | res = evaluate(op->r.f->body.first, res); |
| 2268 | programname = L.s; |
| 2269 | |
| 2270 | nvfree(fnargs); |
| 2271 | fnargs = R.v; |
| 2272 | break; |
| 2273 | |
| 2274 | case XC( OC_GETLINE ): |
| 2275 | case XC( OC_PGETLINE ): |
| 2276 | if (op1) { |
| 2277 | X.rsm = newfile(L.s); |
| 2278 | if (! X.rsm->F) { |
| 2279 | if ((opinfo & OPCLSMASK) == OC_PGETLINE) { |
| 2280 | X.rsm->F = popen(L.s, "r"); |
| 2281 | X.rsm->is_pipe = TRUE; |
| 2282 | } else { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2283 | X.rsm->F = fopen(L.s, "r"); /* not bb_xfopen! */ |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2284 | } |
| 2285 | } |
| 2286 | } else { |
| 2287 | if (! iF) iF = next_input_file(); |
| 2288 | X.rsm = iF; |
| 2289 | } |
| 2290 | |
| 2291 | if (! X.rsm->F) { |
| 2292 | setvar_i(V[ERRNO], errno); |
| 2293 | setvar_i(res, -1); |
| 2294 | break; |
| 2295 | } |
| 2296 | |
| 2297 | if (! op->r.n) |
| 2298 | R.v = V[F0]; |
| 2299 | |
| 2300 | L.i = awk_getline(X.rsm, R.v); |
| 2301 | if (L.i > 0) { |
| 2302 | if (! op1) { |
| 2303 | incvar(V[FNR]); |
| 2304 | incvar(V[NR]); |
| 2305 | } |
| 2306 | } |
| 2307 | setvar_i(res, L.i); |
| 2308 | break; |
| 2309 | |
| 2310 | /* simple builtins */ |
| 2311 | case XC( OC_FBLTIN ): |
| 2312 | switch (opn) { |
| 2313 | |
| 2314 | case F_in: |
| 2315 | R.d = (int)L.d; |
| 2316 | break; |
| 2317 | |
| 2318 | case F_rn: |
| 2319 | R.d = (double)rand() / (double)RAND_MAX; |
| 2320 | break; |
| 2321 | |
| 2322 | #ifdef CONFIG_FEATURE_AWK_MATH |
| 2323 | case F_co: |
| 2324 | R.d = cos(L.d); |
| 2325 | break; |
| 2326 | |
| 2327 | case F_ex: |
| 2328 | R.d = exp(L.d); |
| 2329 | break; |
| 2330 | |
| 2331 | case F_lg: |
| 2332 | R.d = log(L.d); |
| 2333 | break; |
| 2334 | |
| 2335 | case F_si: |
| 2336 | R.d = sin(L.d); |
| 2337 | break; |
| 2338 | |
| 2339 | case F_sq: |
| 2340 | R.d = sqrt(L.d); |
| 2341 | break; |
| 2342 | #else |
| 2343 | case F_co: |
| 2344 | case F_ex: |
| 2345 | case F_lg: |
| 2346 | case F_si: |
| 2347 | case F_sq: |
| 2348 | runtime_error(EMSG_NO_MATH); |
| 2349 | break; |
| 2350 | #endif |
| 2351 | |
| 2352 | case F_sr: |
| 2353 | R.d = (double)seed; |
| 2354 | seed = op1 ? (unsigned int)L.d : (unsigned int)time(NULL); |
| 2355 | srand(seed); |
| 2356 | break; |
| 2357 | |
| 2358 | case F_ti: |
| 2359 | R.d = time(NULL); |
| 2360 | break; |
| 2361 | |
| 2362 | case F_le: |
| 2363 | if (! op1) |
| 2364 | L.s = getvar_s(V[F0]); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2365 | R.d = bb_strlen(L.s); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2366 | break; |
| 2367 | |
| 2368 | case F_sy: |
| 2369 | fflush(NULL); |
| 2370 | R.d = (L.s && *L.s) ? system(L.s) : 0; |
| 2371 | break; |
| 2372 | |
| 2373 | case F_ff: |
| 2374 | if (! op1) |
| 2375 | fflush(stdout); |
| 2376 | else { |
| 2377 | if (L.s && *L.s) { |
| 2378 | X.rsm = newfile(L.s); |
| 2379 | fflush(X.rsm->F); |
| 2380 | } else { |
| 2381 | fflush(NULL); |
| 2382 | } |
| 2383 | } |
| 2384 | break; |
| 2385 | |
| 2386 | case F_cl: |
| 2387 | X.rsm = (rstream *)hash_search(fdhash, L.s); |
| 2388 | if (X.rsm) { |
| 2389 | R.i = X.rsm->is_pipe ? pclose(X.rsm->F) : fclose(X.rsm->F); |
Aaron Lehmann | a170e1c | 2002-11-28 11:27:31 +0000 | [diff] [blame] | 2390 | free(X.rsm->buffer); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2391 | hash_remove(fdhash, L.s); |
| 2392 | } |
| 2393 | if (R.i != 0) |
| 2394 | setvar_i(V[ERRNO], errno); |
| 2395 | R.d = (double)R.i; |
| 2396 | break; |
| 2397 | } |
| 2398 | setvar_i(res, R.d); |
| 2399 | break; |
| 2400 | |
| 2401 | case XC( OC_BUILTIN ): |
| 2402 | res = exec_builtin(op, res); |
| 2403 | break; |
| 2404 | |
| 2405 | case XC( OC_SPRINTF ): |
| 2406 | setvar_p(res, awk_printf(op1)); |
| 2407 | break; |
| 2408 | |
| 2409 | case XC( OC_UNARY ): |
| 2410 | X.v = R.v; |
| 2411 | L.d = R.d = getvar_i(R.v); |
| 2412 | switch (opn) { |
| 2413 | case 'P': |
| 2414 | L.d = ++R.d; |
| 2415 | goto r_op_change; |
| 2416 | case 'p': |
| 2417 | R.d++; |
| 2418 | goto r_op_change; |
| 2419 | case 'M': |
| 2420 | L.d = --R.d; |
| 2421 | goto r_op_change; |
| 2422 | case 'm': |
| 2423 | R.d--; |
| 2424 | goto r_op_change; |
| 2425 | case '!': |
| 2426 | L.d = istrue(X.v) ? 0 : 1; |
| 2427 | break; |
| 2428 | case '-': |
| 2429 | L.d = -R.d; |
| 2430 | break; |
| 2431 | r_op_change: |
| 2432 | setvar_i(X.v, R.d); |
| 2433 | } |
| 2434 | setvar_i(res, L.d); |
| 2435 | break; |
| 2436 | |
| 2437 | case XC( OC_FIELD ): |
| 2438 | R.i = (int)getvar_i(R.v); |
| 2439 | if (R.i == 0) { |
| 2440 | res = V[F0]; |
| 2441 | } else { |
| 2442 | split_f0(); |
| 2443 | if (R.i > nfields) |
| 2444 | fsrealloc(R.i); |
| 2445 | |
| 2446 | res = &Fields[R.i-1]; |
| 2447 | } |
| 2448 | break; |
| 2449 | |
| 2450 | /* concatenation (" ") and index joining (",") */ |
| 2451 | case XC( OC_CONCAT ): |
| 2452 | case XC( OC_COMMA ): |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2453 | opn = bb_strlen(L.s) + bb_strlen(R.s) + 2; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2454 | X.s = (char *)xmalloc(opn); |
| 2455 | strcpy(X.s, L.s); |
| 2456 | if ((opinfo & OPCLSMASK) == OC_COMMA) { |
| 2457 | L.s = getvar_s(V[SUBSEP]); |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2458 | X.s = (char *)xrealloc(X.s, opn + bb_strlen(L.s)); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2459 | strcat(X.s, L.s); |
| 2460 | } |
| 2461 | strcat(X.s, R.s); |
| 2462 | setvar_p(res, X.s); |
| 2463 | break; |
| 2464 | |
| 2465 | case XC( OC_LAND ): |
| 2466 | setvar_i(res, istrue(L.v) ? ptest(op->r.n) : 0); |
| 2467 | break; |
| 2468 | |
| 2469 | case XC( OC_LOR ): |
| 2470 | setvar_i(res, istrue(L.v) ? 1 : ptest(op->r.n)); |
| 2471 | break; |
| 2472 | |
| 2473 | case XC( OC_BINARY ): |
| 2474 | case XC( OC_REPLACE ): |
| 2475 | R.d = getvar_i(R.v); |
| 2476 | switch (opn) { |
| 2477 | case '+': |
| 2478 | L.d += R.d; |
| 2479 | break; |
| 2480 | case '-': |
| 2481 | L.d -= R.d; |
| 2482 | break; |
| 2483 | case '*': |
| 2484 | L.d *= R.d; |
| 2485 | break; |
| 2486 | case '/': |
| 2487 | if (R.d == 0) runtime_error(EMSG_DIV_BY_ZERO); |
| 2488 | L.d /= R.d; |
| 2489 | break; |
| 2490 | case '&': |
| 2491 | #ifdef CONFIG_FEATURE_AWK_MATH |
| 2492 | L.d = pow(L.d, R.d); |
| 2493 | #else |
| 2494 | runtime_error(EMSG_NO_MATH); |
| 2495 | #endif |
| 2496 | break; |
| 2497 | case '%': |
| 2498 | if (R.d == 0) runtime_error(EMSG_DIV_BY_ZERO); |
| 2499 | L.d -= (int)(L.d / R.d) * R.d; |
| 2500 | break; |
| 2501 | } |
| 2502 | res = setvar_i(((opinfo&OPCLSMASK) == OC_BINARY) ? res : X.v, L.d); |
| 2503 | break; |
| 2504 | |
| 2505 | case XC( OC_COMPARE ): |
| 2506 | if (is_numeric(L.v) && is_numeric(R.v)) { |
| 2507 | L.d = getvar_i(L.v) - getvar_i(R.v); |
| 2508 | } else { |
| 2509 | L.s = getvar_s(L.v); |
| 2510 | R.s = getvar_s(R.v); |
| 2511 | L.d = icase ? strcasecmp(L.s, R.s) : strcmp(L.s, R.s); |
| 2512 | } |
| 2513 | switch (opn & 0xfe) { |
| 2514 | case 0: |
| 2515 | R.i = (L.d > 0); |
| 2516 | break; |
| 2517 | case 2: |
| 2518 | R.i = (L.d >= 0); |
| 2519 | break; |
| 2520 | case 4: |
| 2521 | R.i = (L.d == 0); |
| 2522 | break; |
| 2523 | } |
| 2524 | setvar_i(res, (opn & 0x1 ? R.i : !R.i) ? 1 : 0); |
| 2525 | break; |
| 2526 | |
| 2527 | default: |
| 2528 | runtime_error(EMSG_POSSIBLE_ERROR); |
| 2529 | } |
| 2530 | if ((opinfo & OPCLSMASK) <= SHIFT_TIL_THIS) |
| 2531 | op = op->a.n; |
| 2532 | if ((opinfo & OPCLSMASK) >= RECUR_FROM_THIS) |
| 2533 | break; |
| 2534 | if (nextrec) |
| 2535 | break; |
| 2536 | } |
| 2537 | nvfree(v1); |
| 2538 | return res; |
| 2539 | } |
| 2540 | |
| 2541 | |
| 2542 | /* -------- main & co. -------- */ |
| 2543 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 2544 | static int awk_exit(int r) |
| 2545 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2546 | unsigned int i; |
| 2547 | hash_item *hi; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 2548 | static var tv; |
| 2549 | |
| 2550 | if (! exiting) { |
| 2551 | exiting = TRUE; |
Glenn L McGrath | ca29ffc | 2004-09-24 09:24:27 +0000 | [diff] [blame] | 2552 | nextrec = FALSE; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 2553 | evaluate(endseq.first, &tv); |
| 2554 | } |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2555 | |
| 2556 | /* waiting for children */ |
| 2557 | for (i=0; i<fdhash->csize; i++) { |
| 2558 | hi = fdhash->items[i]; |
| 2559 | while(hi) { |
| 2560 | if (hi->data.rs.F && hi->data.rs.is_pipe) |
| 2561 | pclose(hi->data.rs.F); |
| 2562 | hi = hi->next; |
| 2563 | } |
| 2564 | } |
| 2565 | |
| 2566 | exit(r); |
| 2567 | } |
| 2568 | |
| 2569 | /* if expr looks like "var=value", perform assignment and return 1, |
| 2570 | * otherwise return 0 */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 2571 | static int is_assignment(char *expr) |
| 2572 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2573 | char *exprc, *s, *s0, *s1; |
| 2574 | |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2575 | exprc = bb_xstrdup(expr); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2576 | if (!isalnum_(*exprc) || (s = strchr(exprc, '=')) == NULL) { |
| 2577 | free(exprc); |
| 2578 | return FALSE; |
| 2579 | } |
| 2580 | |
| 2581 | *(s++) = '\0'; |
| 2582 | s0 = s1 = s; |
| 2583 | while (*s) |
| 2584 | *(s1++) = nextchar(&s); |
| 2585 | |
| 2586 | *s1 = '\0'; |
| 2587 | setvar_u(newvar(exprc), s0); |
| 2588 | free(exprc); |
| 2589 | return TRUE; |
| 2590 | } |
| 2591 | |
| 2592 | /* switch to next input file */ |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 2593 | static rstream *next_input_file(void) |
| 2594 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2595 | static rstream rsm; |
| 2596 | FILE *F = NULL; |
| 2597 | char *fname, *ind; |
| 2598 | static int files_happen = FALSE; |
| 2599 | |
| 2600 | if (rsm.F) fclose(rsm.F); |
| 2601 | rsm.F = NULL; |
Glenn L McGrath | 00ed36f | 2003-10-30 13:36:39 +0000 | [diff] [blame] | 2602 | rsm.pos = rsm.adv = 0; |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2603 | |
| 2604 | do { |
| 2605 | if (getvar_i(V[ARGIND])+1 >= getvar_i(V[ARGC])) { |
| 2606 | if (files_happen) |
| 2607 | return NULL; |
| 2608 | fname = "-"; |
| 2609 | F = stdin; |
| 2610 | } else { |
| 2611 | ind = getvar_s(incvar(V[ARGIND])); |
| 2612 | fname = getvar_s(findvar(iamarray(V[ARGV]), ind)); |
| 2613 | if (fname && *fname && !is_assignment(fname)) |
| 2614 | F = afopen(fname, "r"); |
| 2615 | } |
| 2616 | } while (!F); |
| 2617 | |
| 2618 | files_happen = TRUE; |
| 2619 | setvar_s(V[FILENAME], fname); |
| 2620 | rsm.F = F; |
| 2621 | return &rsm; |
| 2622 | } |
| 2623 | |
Mike Frysinger | 10a11e2 | 2005-09-27 02:23:02 +0000 | [diff] [blame^] | 2624 | extern int awk_main(int argc, char **argv) |
| 2625 | { |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2626 | char *s, *s1; |
| 2627 | int i, j, c; |
| 2628 | var *v; |
| 2629 | static var tv; |
| 2630 | char **envp; |
| 2631 | static int from_file = FALSE; |
| 2632 | rstream *rsm; |
| 2633 | FILE *F, *stdfiles[3]; |
| 2634 | static char * stdnames = "/dev/stdin\0/dev/stdout\0/dev/stderr"; |
| 2635 | |
| 2636 | /* allocate global buffer */ |
| 2637 | buf = xmalloc(MAXVARFMT+1); |
| 2638 | |
| 2639 | vhash = hash_init(); |
| 2640 | ahash = hash_init(); |
| 2641 | fdhash = hash_init(); |
| 2642 | fnhash = hash_init(); |
| 2643 | |
| 2644 | /* initialize variables */ |
| 2645 | for (i=0; *vNames; i++) { |
| 2646 | V[i] = v = newvar(nextword(&vNames)); |
| 2647 | if (*vValues != '\377') |
| 2648 | setvar_s(v, nextword(&vValues)); |
| 2649 | else |
| 2650 | setvar_i(v, 0); |
| 2651 | |
| 2652 | if (*vNames == '*') { |
| 2653 | v->type |= VF_SPECIAL; |
| 2654 | vNames++; |
| 2655 | } |
| 2656 | } |
| 2657 | |
| 2658 | handle_special(V[FS]); |
| 2659 | handle_special(V[RS]); |
| 2660 | |
| 2661 | stdfiles[0] = stdin; |
| 2662 | stdfiles[1] = stdout; |
| 2663 | stdfiles[2] = stderr; |
| 2664 | for (i=0; i<3; i++) { |
| 2665 | rsm = newfile(nextword(&stdnames)); |
| 2666 | rsm->F = stdfiles[i]; |
| 2667 | } |
| 2668 | |
| 2669 | for (envp=environ; *envp; envp++) { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2670 | s = bb_xstrdup(*envp); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2671 | s1 = strchr(s, '='); |
Eric Andersen | 67776be | 2004-07-30 23:52:08 +0000 | [diff] [blame] | 2672 | if (!s1) { |
| 2673 | goto keep_going; |
| 2674 | } |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2675 | *(s1++) = '\0'; |
| 2676 | setvar_u(findvar(iamarray(V[ENVIRON]), s), s1); |
Eric Andersen | 67776be | 2004-07-30 23:52:08 +0000 | [diff] [blame] | 2677 | keep_going: |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2678 | free(s); |
| 2679 | } |
| 2680 | |
| 2681 | while((c = getopt(argc, argv, "F:v:f:W:")) != EOF) { |
| 2682 | switch (c) { |
| 2683 | case 'F': |
| 2684 | setvar_s(V[FS], optarg); |
| 2685 | break; |
| 2686 | case 'v': |
| 2687 | if (! is_assignment(optarg)) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2688 | bb_show_usage(); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2689 | break; |
| 2690 | case 'f': |
| 2691 | from_file = TRUE; |
| 2692 | F = afopen(programname = optarg, "r"); |
| 2693 | s = NULL; |
| 2694 | /* one byte is reserved for some trick in next_token */ |
| 2695 | for (i=j=1; j>0; i+=j) { |
| 2696 | s = (char *)xrealloc(s, i+4096); |
| 2697 | j = fread(s+i, 1, 4094, F); |
| 2698 | } |
| 2699 | s[i] = '\0'; |
| 2700 | fclose(F); |
| 2701 | parse_program(s+1); |
| 2702 | free(s); |
| 2703 | break; |
| 2704 | case 'W': |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2705 | bb_error_msg("Warning: unrecognized option '-W %s' ignored\n", optarg); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2706 | break; |
| 2707 | |
| 2708 | default: |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2709 | bb_show_usage(); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2710 | } |
| 2711 | } |
| 2712 | |
| 2713 | if (!from_file) { |
| 2714 | if (argc == optind) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 2715 | bb_show_usage(); |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2716 | programname="cmd. line"; |
| 2717 | parse_program(argv[optind++]); |
| 2718 | |
| 2719 | } |
| 2720 | |
| 2721 | /* fill in ARGV array */ |
| 2722 | setvar_i(V[ARGC], argc - optind + 1); |
| 2723 | setari_u(V[ARGV], 0, "awk"); |
| 2724 | for(i=optind; i < argc; i++) |
| 2725 | setari_u(V[ARGV], i+1-optind, argv[i]); |
| 2726 | |
| 2727 | evaluate(beginseq.first, &tv); |
| 2728 | if (! mainseq.first && ! endseq.first) |
| 2729 | awk_exit(EXIT_SUCCESS); |
| 2730 | |
| 2731 | /* input file could already be opened in BEGIN block */ |
| 2732 | if (! iF) iF = next_input_file(); |
| 2733 | |
| 2734 | /* passing through input files */ |
| 2735 | while (iF) { |
| 2736 | |
| 2737 | nextfile = FALSE; |
| 2738 | setvar_i(V[FNR], 0); |
| 2739 | |
| 2740 | while ((c = awk_getline(iF, V[F0])) > 0) { |
| 2741 | |
| 2742 | nextrec = FALSE; |
| 2743 | incvar(V[NR]); |
| 2744 | incvar(V[FNR]); |
| 2745 | evaluate(mainseq.first, &tv); |
| 2746 | |
| 2747 | if (nextfile) |
| 2748 | break; |
| 2749 | } |
| 2750 | |
| 2751 | if (c < 0) |
| 2752 | runtime_error(strerror(errno)); |
| 2753 | |
| 2754 | iF = next_input_file(); |
| 2755 | |
| 2756 | } |
| 2757 | |
Glenn L McGrath | 545106f | 2002-11-11 06:21:00 +0000 | [diff] [blame] | 2758 | awk_exit(EXIT_SUCCESS); |
| 2759 | |
| 2760 | return 0; |
| 2761 | } |
| 2762 | |