1 /* $NetBSD: for.c,v 1.112 2020/10/31 18:41:07 rillig Exp $ */ 2 3 /* 4 * Copyright (c) 1992, The Regents of the University of California. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Handling of .for/.endfor loops in a makefile. 34 * 35 * For loops are of the form: 36 * 37 * .for <varname...> in <value...> 38 * ... 39 * .endfor 40 * 41 * When a .for line is parsed, all following lines are accumulated into a 42 * buffer, up to but excluding the corresponding .endfor line. To find the 43 * corresponding .endfor, the number of nested .for and .endfor directives 44 * are counted. 45 * 46 * During parsing, any nested .for loops are just passed through; they get 47 * handled recursively in For_Eval when the enclosing .for loop is evaluated 48 * in For_Run. 49 * 50 * When the .for loop has been parsed completely, the variable expressions 51 * for the iteration variables are replaced with expressions of the form 52 * ${:Uvalue}, and then this modified body is "included" as a special file. 53 * 54 * Interface: 55 * For_Eval Evaluate the loop in the passed line. 56 * 57 * For_Run Run accumulated loop 58 */ 59 60 #include "make.h" 61 62 /* "@(#)for.c 8.1 (Berkeley) 6/6/93" */ 63 MAKE_RCSID("$NetBSD: for.c,v 1.112 2020/10/31 18:41:07 rillig Exp $"); 64 65 /* The .for loop substitutes the items as ${:U<value>...}, which means 66 * that characters that break this syntax must be backslash-escaped. */ 67 typedef enum ForEscapes { 68 FOR_SUB_ESCAPE_CHAR = 0x0001, 69 FOR_SUB_ESCAPE_BRACE = 0x0002, 70 FOR_SUB_ESCAPE_PAREN = 0x0004 71 } ForEscapes; 72 73 static int forLevel = 0; /* Nesting level */ 74 75 /* One of the variables to the left of the "in" in a .for loop. */ 76 typedef struct ForVar { 77 char *name; 78 size_t len; 79 } ForVar; 80 81 /* 82 * State of a for loop. 83 */ 84 typedef struct For { 85 Buffer body; /* Unexpanded body of the loop */ 86 Vector /* of ForVar */ vars; /* Iteration variables */ 87 Words items; /* Substitution items */ 88 Buffer curBody; /* Expanded body of the current iteration */ 89 /* Is any of the names 1 character long? If so, when the variable values 90 * are substituted, the parser must handle $V expressions as well, not 91 * only ${V} and $(V). */ 92 Boolean short_var; 93 unsigned int sub_next; /* Where to continue iterating */ 94 } For; 95 96 static For *accumFor; /* Loop being accumulated */ 97 98 static void 99 ForAddVar(For *f, const char *name, size_t len) 100 { 101 ForVar *var = Vector_Push(&f->vars); 102 var->name = bmake_strldup(name, len); 103 var->len = len; 104 } 105 106 static void 107 For_Free(For *f) 108 { 109 Buf_Destroy(&f->body, TRUE); 110 111 while (f->vars.len > 0) { 112 ForVar *var = Vector_Pop(&f->vars); 113 free(var->name); 114 } 115 Vector_Done(&f->vars); 116 117 Words_Free(f->items); 118 Buf_Destroy(&f->curBody, TRUE); 119 120 free(f); 121 } 122 123 static ForEscapes 124 GetEscapes(const char *word) 125 { 126 const char *p; 127 ForEscapes escapes = 0; 128 129 for (p = word; *p != '\0'; p++) { 130 switch (*p) { 131 case ':': 132 case '$': 133 case '\\': 134 escapes |= FOR_SUB_ESCAPE_CHAR; 135 break; 136 case ')': 137 escapes |= FOR_SUB_ESCAPE_PAREN; 138 break; 139 case '}': 140 escapes |= FOR_SUB_ESCAPE_BRACE; 141 break; 142 } 143 } 144 return escapes; 145 } 146 147 static Boolean 148 IsFor(const char *p) 149 { 150 return p[0] == 'f' && p[1] == 'o' && p[2] == 'r' && ch_isspace(p[3]); 151 } 152 153 static Boolean 154 IsEndfor(const char *p) 155 { 156 return p[0] == 'e' && strncmp(p, "endfor", 6) == 0 && 157 (p[6] == '\0' || ch_isspace(p[6])); 158 } 159 160 /* Evaluate the for loop in the passed line. The line looks like this: 161 * .for <varname...> in <value...> 162 * 163 * Input: 164 * line Line to parse 165 * 166 * Results: 167 * 0: Not a .for statement, parse the line 168 * 1: We found a for loop 169 * -1: A .for statement with a bad syntax error, discard. 170 */ 171 int 172 For_Eval(const char *line) 173 { 174 For *f; 175 const char *p; 176 177 p = line + 1; /* skip the '.' */ 178 cpp_skip_whitespace(&p); 179 180 if (!IsFor(p)) { 181 if (IsEndfor(p)) { 182 Parse_Error(PARSE_FATAL, "for-less endfor"); 183 return -1; 184 } 185 return 0; 186 } 187 p += 3; 188 189 /* 190 * we found a for loop, and now we are going to parse it. 191 */ 192 193 f = bmake_malloc(sizeof *f); 194 Buf_Init(&f->body, 0); 195 Vector_Init(&f->vars, sizeof(ForVar)); 196 f->items.words = NULL; 197 f->items.freeIt = NULL; 198 Buf_Init(&f->curBody, 0); 199 f->short_var = FALSE; 200 f->sub_next = 0; 201 202 /* Grab the variables. Terminate on "in". */ 203 for (;;) { 204 size_t len; 205 206 cpp_skip_whitespace(&p); 207 if (*p == '\0') { 208 Parse_Error(PARSE_FATAL, "missing `in' in for"); 209 For_Free(f); 210 return -1; 211 } 212 213 /* XXX: This allows arbitrary variable names; see directive-for.mk. */ 214 for (len = 1; p[len] != '\0' && !ch_isspace(p[len]); len++) 215 continue; 216 217 if (len == 2 && p[0] == 'i' && p[1] == 'n') { 218 p += 2; 219 break; 220 } 221 if (len == 1) 222 f->short_var = TRUE; 223 224 ForAddVar(f, p, len); 225 p += len; 226 } 227 228 if (f->vars.len == 0) { 229 Parse_Error(PARSE_FATAL, "no iteration variables in for"); 230 For_Free(f); 231 return -1; 232 } 233 234 cpp_skip_whitespace(&p); 235 236 { 237 char *items; 238 (void)Var_Subst(p, VAR_GLOBAL, VARE_WANTRES, &items); 239 /* TODO: handle errors */ 240 f->items = Str_Words(items, FALSE); 241 free(items); 242 243 if (f->items.len == 1 && f->items.words[0][0] == '\0') 244 f->items.len = 0; /* .for var in ${:U} */ 245 } 246 247 { 248 size_t nitems, nvars; 249 250 if ((nitems = f->items.len) > 0 && nitems % (nvars = f->vars.len)) { 251 Parse_Error(PARSE_FATAL, 252 "Wrong number of words (%zu) in .for substitution list" 253 " with %zu variables", nitems, nvars); 254 /* 255 * Return 'success' so that the body of the .for loop is 256 * accumulated. 257 * Remove all items so that the loop doesn't iterate. 258 */ 259 f->items.len = 0; 260 } 261 } 262 263 accumFor = f; 264 forLevel = 1; 265 return 1; 266 } 267 268 /* 269 * Add another line to a .for loop. 270 * Returns FALSE when the matching .endfor is reached. 271 */ 272 Boolean 273 For_Accum(const char *line) 274 { 275 const char *ptr = line; 276 277 if (*ptr == '.') { 278 ptr++; 279 cpp_skip_whitespace(&ptr); 280 281 if (IsEndfor(ptr)) { 282 DEBUG1(FOR, "For: end for %d\n", forLevel); 283 if (--forLevel <= 0) 284 return FALSE; 285 } else if (IsFor(ptr)) { 286 forLevel++; 287 DEBUG1(FOR, "For: new loop %d\n", forLevel); 288 } 289 } 290 291 Buf_AddStr(&accumFor->body, line); 292 Buf_AddByte(&accumFor->body, '\n'); 293 return TRUE; 294 } 295 296 297 static size_t 298 for_var_len(const char *var) 299 { 300 char ch, var_start, var_end; 301 int depth; 302 size_t len; 303 304 var_start = *var; 305 if (var_start == 0) 306 /* just escape the $ */ 307 return 0; 308 309 if (var_start == '(') 310 var_end = ')'; 311 else if (var_start == '{') 312 var_end = '}'; 313 else 314 /* Single char variable */ 315 return 1; 316 317 depth = 1; 318 for (len = 1; (ch = var[len++]) != 0;) { 319 if (ch == var_start) 320 depth++; 321 else if (ch == var_end && --depth == 0) 322 return len; 323 } 324 325 /* Variable end not found, escape the $ */ 326 return 0; 327 } 328 329 /* While expanding the body of a .for loop, write the item in the ${:U...} 330 * expression, escaping characters as needed. See ApplyModifier_Defined. */ 331 static void 332 Buf_AddEscaped(Buffer *cmds, const char *item, char ech) 333 { 334 ForEscapes escapes = GetEscapes(item); 335 char ch; 336 337 /* If there were no escapes, or the only escape is the other variable 338 * terminator, then just substitute the full string */ 339 if (!(escapes & (ech == ')' ? ~(unsigned)FOR_SUB_ESCAPE_BRACE 340 : ~(unsigned)FOR_SUB_ESCAPE_PAREN))) { 341 Buf_AddStr(cmds, item); 342 return; 343 } 344 345 /* Escape ':', '$', '\\' and 'ech' - these will be removed later by 346 * :U processing, see ApplyModifier_Defined. */ 347 while ((ch = *item++) != '\0') { 348 if (ch == '$') { 349 size_t len = for_var_len(item); 350 if (len != 0) { 351 Buf_AddBytes(cmds, item - 1, len + 1); 352 item += len; 353 continue; 354 } 355 Buf_AddByte(cmds, '\\'); 356 } else if (ch == ':' || ch == '\\' || ch == ech) 357 Buf_AddByte(cmds, '\\'); 358 Buf_AddByte(cmds, ch); 359 } 360 } 361 362 /* While expanding the body of a .for loop, replace expressions like 363 * ${i}, ${i:...}, $(i) or $(i:...) with their ${:U...} expansion. */ 364 static void 365 SubstVarLong(For *f, const char **pp, const char **inout_mark, char ech) 366 { 367 size_t i; 368 const char *p = *pp; 369 370 for (i = 0; i < f->vars.len; i++) { 371 ForVar *forVar = Vector_Get(&f->vars, i); 372 char *var = forVar->name; 373 size_t vlen = forVar->len; 374 375 /* XXX: undefined behavior for p if vlen is longer than p? */ 376 if (memcmp(p, var, vlen) != 0) 377 continue; 378 /* XXX: why test for backslash here? */ 379 if (p[vlen] != ':' && p[vlen] != ech && p[vlen] != '\\') 380 continue; 381 382 /* Found a variable match. Replace with :U<value> */ 383 Buf_AddBytesBetween(&f->curBody, *inout_mark, p); 384 Buf_AddStr(&f->curBody, ":U"); 385 Buf_AddEscaped(&f->curBody, f->items.words[f->sub_next + i], ech); 386 387 p += vlen; 388 *inout_mark = p; 389 break; 390 } 391 392 *pp = p; 393 } 394 395 /* While expanding the body of a .for loop, replace single-character 396 * variable expressions like $i with their ${:U...} expansion. */ 397 static void 398 SubstVarShort(For *f, char const ch, const char **pp, const char **inout_mark) 399 { 400 const char *p = *pp; 401 size_t i; 402 403 /* Probably a single character name, ignore $$ and stupid ones. */ 404 if (!f->short_var || strchr("}):$", ch) != NULL) { 405 p++; 406 *pp = p; 407 return; 408 } 409 410 for (i = 0; i < f->vars.len; i++) { 411 ForVar *var = Vector_Get(&f->vars, i); 412 const char *varname = var->name; 413 if (varname[0] != ch || varname[1] != '\0') 414 continue; 415 416 /* Found a variable match. Replace with ${:U<value>} */ 417 Buf_AddBytesBetween(&f->curBody, *inout_mark, p); 418 Buf_AddStr(&f->curBody, "{:U"); 419 Buf_AddEscaped(&f->curBody, f->items.words[f->sub_next + i], '}'); 420 Buf_AddByte(&f->curBody, '}'); 421 422 *inout_mark = ++p; 423 break; 424 } 425 426 *pp = p; 427 } 428 429 /* 430 * Scan the for loop body and replace references to the loop variables 431 * with variable references that expand to the required text. 432 * 433 * Using variable expansions ensures that the .for loop can't generate 434 * syntax, and that the later parsing will still see a variable. 435 * We assume that the null variable will never be defined. 436 * 437 * The detection of substitutions of the loop control variable is naive. 438 * Many of the modifiers use \ to escape $ (not $) so it is possible 439 * to contrive a makefile where an unwanted substitution happens. 440 */ 441 static char * 442 ForIterate(void *v_arg, size_t *out_len) 443 { 444 For *f = v_arg; 445 const char *p; 446 const char *mark; /* where the last replacement left off */ 447 const char *body_end; 448 char *cmds_str; 449 450 if (f->sub_next + f->vars.len > f->items.len) { 451 /* No more iterations */ 452 For_Free(f); 453 return NULL; 454 } 455 456 Buf_Empty(&f->curBody); 457 458 mark = Buf_GetAll(&f->body, NULL); 459 body_end = mark + Buf_Len(&f->body); 460 for (p = mark; (p = strchr(p, '$')) != NULL;) { 461 char ch, ech; 462 ch = *++p; 463 if ((ch == '(' && (ech = ')', 1)) || (ch == '{' && (ech = '}', 1))) { 464 p++; 465 /* Check variable name against the .for loop variables */ 466 SubstVarLong(f, &p, &mark, ech); 467 continue; 468 } 469 if (ch == '\0') 470 break; 471 472 SubstVarShort(f, ch, &p, &mark); 473 } 474 Buf_AddBytesBetween(&f->curBody, mark, body_end); 475 476 *out_len = Buf_Len(&f->curBody); 477 cmds_str = Buf_GetAll(&f->curBody, NULL); 478 DEBUG1(FOR, "For: loop body:\n%s", cmds_str); 479 480 f->sub_next += f->vars.len; 481 482 return cmds_str; 483 } 484 485 /* Run the for loop, imitating the actions of an include file. */ 486 void 487 For_Run(int lineno) 488 { 489 For *f = accumFor; 490 accumFor = NULL; 491 492 if (f->items.len == 0) { 493 /* Nothing to expand - possibly due to an earlier syntax error. */ 494 For_Free(f); 495 return; 496 } 497 498 Parse_SetInput(NULL, lineno, -1, ForIterate, f); 499 } 500