1 /* $NetBSD: for.c,v 1.182 2024/06/07 18:57:30 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 have the form: 36 * 37 * .for <varname...> in <value...> 38 * # the body 39 * .endfor 40 * 41 * When a .for line is parsed, the following lines are copied to the body of 42 * the .for loop, until the corresponding .endfor line is reached. In this 43 * phase, the body is not yet evaluated. This also applies to any nested 44 * .for loops. 45 * 46 * After reaching the .endfor, the values from the .for line are grouped 47 * according to the number of variables. For each such group, the unexpanded 48 * body is scanned for expressions, and those that match the 49 * variable names are replaced with expressions of the form ${:U...}. After 50 * that, the body is treated like a file from an .include directive. 51 * 52 * Interface: 53 * For_Eval Evaluate the loop in the passed line. 54 * 55 * For_Run Run accumulated loop 56 */ 57 58 #include "make.h" 59 60 /* "@(#)for.c 8.1 (Berkeley) 6/6/93" */ 61 MAKE_RCSID("$NetBSD: for.c,v 1.182 2024/06/07 18:57:30 rillig Exp $"); 62 63 64 typedef struct ForLoop { 65 Vector /* of 'char *' */ vars; /* Iteration variables */ 66 SubstringWords items; /* Substitution items */ 67 Buffer body; /* Unexpanded body of the loop */ 68 unsigned int nextItem; /* Where to continue iterating */ 69 } ForLoop; 70 71 72 static ForLoop *accumFor; /* Loop being accumulated */ 73 74 75 /* See LK_FOR_BODY. */ 76 static void 77 skip_whitespace_or_line_continuation(const char **pp) 78 { 79 const char *p = *pp; 80 for (;;) { 81 if (ch_isspace(*p)) 82 p++; 83 else if (p[0] == '\\' && p[1] == '\n') 84 p += 2; 85 else 86 break; 87 } 88 *pp = p; 89 } 90 91 static ForLoop * 92 ForLoop_New(void) 93 { 94 ForLoop *f = bmake_malloc(sizeof *f); 95 96 Vector_Init(&f->vars, sizeof(char *)); 97 SubstringWords_Init(&f->items); 98 Buf_Init(&f->body); 99 f->nextItem = 0; 100 101 return f; 102 } 103 104 void 105 ForLoop_Free(ForLoop *f) 106 { 107 while (f->vars.len > 0) 108 free(*(char **)Vector_Pop(&f->vars)); 109 Vector_Done(&f->vars); 110 111 SubstringWords_Free(f->items); 112 Buf_Done(&f->body); 113 114 free(f); 115 } 116 117 char * 118 ForLoop_Details(const ForLoop *f) 119 { 120 size_t i, n; 121 const char **vars; 122 const Substring *items; 123 Buffer buf; 124 125 n = f->vars.len; 126 vars = f->vars.items; 127 assert(f->nextItem >= n); 128 items = f->items.words + f->nextItem - n; 129 130 Buf_Init(&buf); 131 for (i = 0; i < n; i++) { 132 if (i > 0) 133 Buf_AddStr(&buf, ", "); 134 Buf_AddStr(&buf, vars[i]); 135 Buf_AddStr(&buf, " = "); 136 Buf_AddRange(&buf, items[i].start, items[i].end); 137 } 138 return Buf_DoneData(&buf); 139 } 140 141 static bool 142 IsValidInVarname(char c) 143 { 144 return c != '$' && c != ':' && c != '\\' && 145 c != '(' && c != '{' && c != ')' && c != '}'; 146 } 147 148 static void 149 ForLoop_ParseVarnames(ForLoop *f, const char **pp) 150 { 151 const char *p = *pp; 152 153 for (;;) { 154 size_t len; 155 156 cpp_skip_whitespace(&p); 157 if (*p == '\0') { 158 Parse_Error(PARSE_FATAL, "missing `in' in for"); 159 while (f->vars.len > 0) 160 free(*(char **)Vector_Pop(&f->vars)); 161 return; 162 } 163 164 for (len = 0; p[len] != '\0' && !ch_isspace(p[len]); len++) { 165 if (!IsValidInVarname(p[len])) { 166 Parse_Error(PARSE_FATAL, 167 "invalid character '%c' " 168 "in .for loop variable name", 169 p[len]); 170 while (f->vars.len > 0) 171 free(*(char **)Vector_Pop(&f->vars)); 172 return; 173 } 174 } 175 176 if (len == 2 && p[0] == 'i' && p[1] == 'n') { 177 p += 2; 178 break; 179 } 180 181 *(char **)Vector_Push(&f->vars) = bmake_strldup(p, len); 182 p += len; 183 } 184 185 if (f->vars.len == 0) { 186 Parse_Error(PARSE_FATAL, "no iteration variables in for"); 187 return; 188 } 189 190 *pp = p; 191 } 192 193 static bool 194 ForLoop_ParseItems(ForLoop *f, const char *p) 195 { 196 char *items; 197 198 cpp_skip_whitespace(&p); 199 200 items = Var_Subst(p, SCOPE_GLOBAL, VARE_EVAL); 201 /* TODO: handle errors */ 202 203 f->items = Substring_Words(items, false); 204 free(items); 205 206 if (f->items.len == 1 && Substring_IsEmpty(f->items.words[0])) 207 f->items.len = 0; /* .for var in ${:U} */ 208 209 if (f->items.len % f->vars.len != 0) { 210 Parse_Error(PARSE_FATAL, 211 "Wrong number of words (%u) in .for " 212 "substitution list with %u variables", 213 (unsigned)f->items.len, (unsigned)f->vars.len); 214 return false; 215 } 216 217 return true; 218 } 219 220 static bool 221 IsFor(const char *p) 222 { 223 return p[0] == 'f' && p[1] == 'o' && p[2] == 'r' && ch_isspace(p[3]); 224 } 225 226 static bool 227 IsEndfor(const char *p) 228 { 229 return p[0] == 'e' && strncmp(p, "endfor", 6) == 0 && 230 (p[6] == '\0' || ch_isspace(p[6])); 231 } 232 233 /* 234 * Evaluate the for loop in the passed line. The line looks like this: 235 * .for <varname...> in <value...> 236 * 237 * Results: 238 * 0 not a .for directive 239 * 1 found a .for directive 240 * -1 erroneous .for directive 241 */ 242 int 243 For_Eval(const char *line) 244 { 245 const char *p; 246 ForLoop *f; 247 248 p = line + 1; /* skip the '.' */ 249 skip_whitespace_or_line_continuation(&p); 250 251 if (IsFor(p)) { 252 p += 3; 253 254 f = ForLoop_New(); 255 ForLoop_ParseVarnames(f, &p); 256 if (f->vars.len > 0 && !ForLoop_ParseItems(f, p)) 257 f->items.len = 0; /* don't iterate */ 258 259 accumFor = f; 260 return 1; 261 } else if (IsEndfor(p)) { 262 Parse_Error(PARSE_FATAL, "for-less endfor"); 263 return -1; 264 } else 265 return 0; 266 } 267 268 /* 269 * Add another line to the .for loop that is being built up. 270 * Returns false when the matching .endfor is reached. 271 */ 272 bool 273 For_Accum(const char *line, int *forLevel) 274 { 275 const char *p = line; 276 277 if (*p == '.') { 278 p++; 279 skip_whitespace_or_line_continuation(&p); 280 281 if (IsEndfor(p)) { 282 DEBUG1(FOR, "For: end for %d\n", *forLevel); 283 if (--*forLevel == 0) 284 return false; 285 } else if (IsFor(p)) { 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 * When the body of a '.for i' loop is prepared for an iteration, each 298 * occurrence of $i in the body is replaced with ${:U...}, inserting the 299 * value of the item. If this item contains a '$', it may be the start of an 300 * expression. This expression is copied verbatim, its length is 301 * determined here, in a rather naive way, ignoring escape characters and 302 * funny delimiters in modifiers like ':S}from}to}'. 303 */ 304 static size_t 305 ExprLen(const char *s, const char *e) 306 { 307 char expr_open, expr_close; 308 int depth; 309 const char *p; 310 311 if (s == e) 312 return 0; /* just escape the '$' */ 313 314 expr_open = s[0]; 315 if (expr_open == '(') 316 expr_close = ')'; 317 else if (expr_open == '{') 318 expr_close = '}'; 319 else 320 return 1; /* Single char variable */ 321 322 depth = 1; 323 for (p = s + 1; p != e; p++) { 324 if (*p == expr_open) 325 depth++; 326 else if (*p == expr_close && --depth == 0) 327 return (size_t)(p + 1 - s); 328 } 329 330 /* Expression end not found, escape the $ */ 331 return 0; 332 } 333 334 /* 335 * While expanding the body of a .for loop, write the item as a ${:U...} 336 * expression, escaping characters as needed. The result is later unescaped 337 * by ApplyModifier_Defined. 338 */ 339 static void 340 AddEscaped(Buffer *cmds, Substring item, char endc) 341 { 342 const char *p; 343 char ch; 344 345 for (p = item.start; p != item.end;) { 346 ch = *p; 347 if (ch == '$') { 348 size_t len = ExprLen(p + 1, item.end); 349 if (len != 0) { 350 /* 351 * XXX: Should a '\' be added here? 352 * See directive-for-escape.mk, ExprLen. 353 */ 354 Buf_AddBytes(cmds, p, 1 + len); 355 p += 1 + len; 356 continue; 357 } 358 Buf_AddByte(cmds, '\\'); 359 } else if (ch == ':' || ch == '\\' || ch == endc) 360 Buf_AddByte(cmds, '\\'); 361 else if (ch == '\n') { 362 Parse_Error(PARSE_FATAL, "newline in .for value"); 363 ch = ' '; /* prevent newline injection */ 364 } 365 Buf_AddByte(cmds, ch); 366 p++; 367 } 368 } 369 370 /* 371 * While expanding the body of a .for loop, replace the variable name of an 372 * expression like ${i} or ${i:...} or $(i) or $(i:...) with ":Uvalue". 373 */ 374 static void 375 ForLoop_SubstVarLong(ForLoop *f, unsigned int firstItem, Buffer *body, 376 const char **pp, char endc, const char **inout_mark) 377 { 378 size_t i; 379 const char *start = *pp; 380 const char **varnames = Vector_Get(&f->vars, 0); 381 382 for (i = 0; i < f->vars.len; i++) { 383 const char *p = start; 384 385 if (!cpp_skip_string(&p, varnames[i])) 386 continue; 387 /* XXX: why test for backslash here? */ 388 if (*p != ':' && *p != endc && *p != '\\') 389 continue; 390 391 /* 392 * Found a variable match. Skip over the variable name and 393 * instead add ':U<value>' to the current body. 394 */ 395 Buf_AddRange(body, *inout_mark, start); 396 Buf_AddStr(body, ":U"); 397 AddEscaped(body, f->items.words[firstItem + i], endc); 398 399 *inout_mark = p; 400 *pp = p; 401 return; 402 } 403 } 404 405 /* 406 * While expanding the body of a .for loop, replace single-character 407 * expressions like $i with their ${:U...} expansion. 408 */ 409 static void 410 ForLoop_SubstVarShort(ForLoop *f, unsigned int firstItem, Buffer *body, 411 const char *p, const char **inout_mark) 412 { 413 char ch = *p; 414 const char **vars; 415 size_t i; 416 417 /* Skip $$ and stupid ones. */ 418 if (ch == '}' || ch == ')' || ch == ':' || ch == '$') 419 return; 420 421 vars = Vector_Get(&f->vars, 0); 422 for (i = 0; i < f->vars.len; i++) { 423 const char *varname = vars[i]; 424 if (varname[0] == ch && varname[1] == '\0') 425 goto found; 426 } 427 return; 428 429 found: 430 Buf_AddRange(body, *inout_mark, p); 431 *inout_mark = p + 1; 432 433 /* Replace $<ch> with ${:U<value>} */ 434 Buf_AddStr(body, "{:U"); 435 AddEscaped(body, f->items.words[firstItem + i], '}'); 436 Buf_AddByte(body, '}'); 437 } 438 439 /* 440 * Compute the body for the current iteration by copying the unexpanded body, 441 * replacing the expressions for the iteration variables on the way. 442 * 443 * Using expressions ensures that the .for loop can't generate 444 * syntax, and that the later parsing will still see an expression. 445 * This code assumes that the variable with the empty name is never defined, 446 * see unit-tests/varname-empty.mk. 447 * 448 * The detection of substitutions of the loop control variables is naive. 449 * Many of the modifiers use '\$' instead of '$$' to escape '$', so it is 450 * possible to contrive a makefile where an unwanted substitution happens. 451 * See unit-tests/directive-for-escape.mk. 452 */ 453 static void 454 ForLoop_SubstBody(ForLoop *f, unsigned int firstItem, Buffer *body) 455 { 456 const char *p, *end; 457 const char *mark; /* where the last substitution left off */ 458 459 Buf_Clear(body); 460 461 mark = f->body.data; 462 end = f->body.data + f->body.len; 463 for (p = mark; (p = strchr(p, '$')) != NULL;) { 464 if (p[1] == '{' || p[1] == '(') { 465 char endc = p[1] == '{' ? '}' : ')'; 466 p += 2; 467 ForLoop_SubstVarLong(f, firstItem, body, 468 &p, endc, &mark); 469 } else { 470 ForLoop_SubstVarShort(f, firstItem, body, 471 p + 1, &mark); 472 p += 2; 473 } 474 } 475 476 Buf_AddRange(body, mark, end); 477 } 478 479 /* 480 * Compute the body for the current iteration by copying the unexpanded body, 481 * replacing the expressions for the iteration variables on the way. 482 */ 483 bool 484 For_NextIteration(ForLoop *f, Buffer *body) 485 { 486 if (f->nextItem == f->items.len) 487 return false; 488 489 f->nextItem += (unsigned int)f->vars.len; 490 ForLoop_SubstBody(f, f->nextItem - (unsigned int)f->vars.len, body); 491 if (DEBUG(FOR)) { 492 char *details = ForLoop_Details(f); 493 debug_printf("For: loop body with %s:\n%s", 494 details, body->data); 495 free(details); 496 } 497 return true; 498 } 499 500 /* Break out of the .for loop. */ 501 void 502 For_Break(ForLoop *f) 503 { 504 f->nextItem = (unsigned int)f->items.len; 505 } 506 507 /* Run the .for loop, imitating the actions of an include file. */ 508 void 509 For_Run(unsigned headLineno, unsigned bodyReadLines) 510 { 511 Buffer buf; 512 ForLoop *f = accumFor; 513 accumFor = NULL; 514 515 if (f->items.len > 0) { 516 Buf_Init(&buf); 517 Parse_PushInput(NULL, headLineno, bodyReadLines, buf, f); 518 } else 519 ForLoop_Free(f); 520 } 521