1 /* $NetBSD: for.c,v 1.185 2025/04/22 19:28:50 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 1.185 2025/04/22 19:28:50 rillig Exp $");
62
63
64 typedef struct ForLoop {
65 Vector /* of 'char *' */ vars; /* Iteration variables */
66 SubstringWords items; /* Substitution items */
67 Buffer body; /* Unexpanded body of the loop */
68 unsigned nextItem; /* Where to continue iterating */
69 } ForLoop;
70
71
72 static ForLoop *accumFor; /* Loop being accumulated */
73
74
75 /* See LK_FOR_BODY. */
76 static void
skip_whitespace_or_line_continuation(const char ** pp)77 skip_whitespace_or_line_continuation(const char **pp)
78 {
79 const char *p = *pp;
80 for (;;) {
81 if (ch_isspace(*p))
82 p++;
83 else if (p[0] == '\\' && p[1] == '\n')
84 p += 2;
85 else
86 break;
87 }
88 *pp = p;
89 }
90
91 static ForLoop *
ForLoop_New(void)92 ForLoop_New(void)
93 {
94 ForLoop *f = bmake_malloc(sizeof *f);
95
96 Vector_Init(&f->vars, sizeof(char *));
97 SubstringWords_Init(&f->items);
98 Buf_Init(&f->body);
99 f->nextItem = 0;
100
101 return f;
102 }
103
104 void
ForLoop_Free(ForLoop * f)105 ForLoop_Free(ForLoop *f)
106 {
107 while (f->vars.len > 0)
108 free(*(char **)Vector_Pop(&f->vars));
109 Vector_Done(&f->vars);
110
111 SubstringWords_Free(f->items);
112 Buf_Done(&f->body);
113
114 free(f);
115 }
116
117 char *
ForLoop_Details(const ForLoop * f)118 ForLoop_Details(const ForLoop *f)
119 {
120 size_t i, n;
121 const char **vars;
122 const Substring *items;
123 Buffer buf;
124
125 n = f->vars.len;
126 vars = f->vars.items;
127 assert(f->nextItem >= n);
128 items = f->items.words + f->nextItem - n;
129
130 Buf_Init(&buf);
131 for (i = 0; i < n; i++) {
132 if (i > 0)
133 Buf_AddStr(&buf, ", ");
134 Buf_AddStr(&buf, vars[i]);
135 Buf_AddStr(&buf, " = ");
136 Buf_AddRange(&buf, items[i].start, items[i].end);
137 }
138 return Buf_DoneData(&buf);
139 }
140
141 static bool
IsValidInVarname(char c)142 IsValidInVarname(char c)
143 {
144 return c != '$' && c != ':' && c != '\\' &&
145 c != '(' && c != '{' && c != ')' && c != '}';
146 }
147
148 static void
ForLoop_ParseVarnames(ForLoop * f,const char ** pp)149 ForLoop_ParseVarnames(ForLoop *f, const char **pp)
150 {
151 const char *p = *pp, *start;
152
153 for (;;) {
154 cpp_skip_whitespace(&p);
155 if (*p == '\0') {
156 Parse_Error(PARSE_FATAL, "missing `in' in for");
157 goto cleanup;
158 }
159
160 for (start = p; *p != '\0' && !ch_isspace(*p); p++)
161 if (!IsValidInVarname(*p))
162 goto invalid_variable_name;
163
164 if (p - start == 2 && memcmp(start, "in", 2) == 0)
165 break;
166
167 *(char **)Vector_Push(&f->vars) = bmake_strsedup(start, p);
168 }
169
170 if (f->vars.len == 0) {
171 Parse_Error(PARSE_FATAL, "no iteration variables in for");
172 return;
173 }
174
175 *pp = p;
176 return;
177
178 invalid_variable_name:
179 Parse_Error(PARSE_FATAL,
180 "invalid character '%c' in .for loop variable name", *p);
181 cleanup:
182 while (f->vars.len > 0)
183 free(*(char **)Vector_Pop(&f->vars));
184 }
185
186 static bool
ForLoop_ParseItems(ForLoop * f,const char * p)187 ForLoop_ParseItems(ForLoop *f, const char *p)
188 {
189 char *items;
190 int parseErrorsBefore = parseErrors;
191
192 cpp_skip_whitespace(&p);
193
194 items = Var_Subst(p, SCOPE_GLOBAL, VARE_EVAL);
195 f->items = Substring_Words(
196 parseErrors == parseErrorsBefore ? items : "", false);
197 free(items);
198
199 if (f->items.len == 1 && Substring_IsEmpty(f->items.words[0]))
200 f->items.len = 0; /* .for var in ${:U} */
201
202 if (f->items.len % f->vars.len != 0) {
203 Parse_Error(PARSE_FATAL,
204 "Wrong number of words (%u) in .for "
205 "substitution list with %u variables",
206 (unsigned)f->items.len, (unsigned)f->vars.len);
207 return false;
208 }
209
210 return true;
211 }
212
213 static bool
IsFor(const char * p)214 IsFor(const char *p)
215 {
216 return p[0] == 'f' && p[1] == 'o' && p[2] == 'r' && ch_isspace(p[3]);
217 }
218
219 static bool
IsEndfor(const char * p)220 IsEndfor(const char *p)
221 {
222 return p[0] == 'e' && strncmp(p, "endfor", 6) == 0 &&
223 (p[6] == '\0' || ch_isspace(p[6]));
224 }
225
226 /*
227 * Evaluate the for loop in the passed line. The line looks like this:
228 * .for <varname...> in <value...>
229 *
230 * Results:
231 * 0 not a .for directive
232 * 1 found a .for directive
233 * -1 erroneous .for directive
234 */
235 int
For_Eval(const char * line)236 For_Eval(const char *line)
237 {
238 const char *p;
239 ForLoop *f;
240
241 p = line + 1; /* skip the '.' */
242 skip_whitespace_or_line_continuation(&p);
243
244 if (IsFor(p)) {
245 p += 3;
246
247 f = ForLoop_New();
248 ForLoop_ParseVarnames(f, &p);
249 if (f->vars.len > 0 && !ForLoop_ParseItems(f, p))
250 f->items.len = 0; /* don't iterate */
251
252 accumFor = f;
253 return 1;
254 } else if (IsEndfor(p)) {
255 Parse_Error(PARSE_FATAL, "for-less endfor");
256 return -1;
257 } else
258 return 0;
259 }
260
261 /*
262 * Add another line to the .for loop that is being built up.
263 * Returns false when the matching .endfor is reached.
264 */
265 bool
For_Accum(const char * line,int * forLevel)266 For_Accum(const char *line, int *forLevel)
267 {
268 const char *p = line;
269
270 if (*p == '.') {
271 p++;
272 skip_whitespace_or_line_continuation(&p);
273
274 if (IsEndfor(p)) {
275 DEBUG1(FOR, "For: end for %d\n", *forLevel);
276 if (--*forLevel == 0)
277 return false;
278 } else if (IsFor(p)) {
279 (*forLevel)++;
280 DEBUG1(FOR, "For: new loop %d\n", *forLevel);
281 }
282 }
283
284 Buf_AddStr(&accumFor->body, line);
285 Buf_AddByte(&accumFor->body, '\n');
286 return true;
287 }
288
289 /*
290 * When the body of a '.for i' loop is prepared for an iteration, each
291 * occurrence of $i in the body is replaced with ${:U...}, inserting the
292 * value of the item. If this item contains a '$', it may be the start of an
293 * expression. This expression is copied verbatim, its length is
294 * determined here, in a rather naive way, ignoring escape characters and
295 * funny delimiters in modifiers like ':S}from}to}'.
296 */
297 static size_t
ExprLen(const char * s,const char * e)298 ExprLen(const char *s, const char *e)
299 {
300 char expr_open, expr_close;
301 int depth;
302 const char *p;
303
304 if (s == e)
305 return 0; /* just escape the '$' */
306
307 expr_open = s[0];
308 if (expr_open == '(')
309 expr_close = ')';
310 else if (expr_open == '{')
311 expr_close = '}';
312 else
313 return 1; /* Single char variable */
314
315 depth = 1;
316 for (p = s + 1; p != e; p++) {
317 if (*p == expr_open)
318 depth++;
319 else if (*p == expr_close && --depth == 0)
320 return (size_t)(p + 1 - s);
321 }
322
323 /* Expression end not found, escape the $ */
324 return 0;
325 }
326
327 /*
328 * While expanding the body of a .for loop, write the item as a ${:U...}
329 * expression, escaping characters as needed. The result is later unescaped
330 * by ApplyModifier_Defined.
331 */
332 static void
AddEscaped(Buffer * body,Substring item,char endc)333 AddEscaped(Buffer *body, Substring item, char endc)
334 {
335 const char *p;
336 char ch;
337
338 for (p = item.start; p != item.end;) {
339 ch = *p;
340 if (ch == '$') {
341 size_t len = ExprLen(p + 1, item.end);
342 if (len != 0) {
343 /*
344 * XXX: Should a '\' be added here?
345 * See directive-for-escape.mk, ExprLen.
346 */
347 Buf_AddBytes(body, p, 1 + len);
348 p += 1 + len;
349 continue;
350 }
351 Buf_AddByte(body, '\\');
352 } else if (ch == ':' || ch == '\\' || ch == endc)
353 Buf_AddByte(body, '\\');
354 else if (ch == '\n') {
355 Parse_Error(PARSE_FATAL, "newline in .for value");
356 ch = ' '; /* prevent newline injection */
357 }
358 Buf_AddByte(body, ch);
359 p++;
360 }
361 }
362
363 /*
364 * While expanding the body of a .for loop, replace the variable name of an
365 * expression like ${i} or ${i:...} or $(i) or $(i:...) with ":Uvalue".
366 */
367 static void
ForLoop_SubstVarLong(ForLoop * f,unsigned firstItem,Buffer * body,const char ** pp,char endc,const char ** inout_mark)368 ForLoop_SubstVarLong(ForLoop *f, unsigned firstItem, Buffer *body,
369 const char **pp, char endc, const char **inout_mark)
370 {
371 size_t i;
372 const char *start = *pp;
373 const char **varnames = Vector_Get(&f->vars, 0);
374
375 for (i = 0; i < f->vars.len; i++) {
376 const char *p = start;
377
378 if (!cpp_skip_string(&p, varnames[i]))
379 continue;
380 /* XXX: why test for backslash here? */
381 if (*p != ':' && *p != endc && *p != '\\')
382 continue;
383
384 /*
385 * Found a variable match. Skip over the variable name and
386 * instead add ':U<value>' to the current body.
387 */
388 Buf_AddRange(body, *inout_mark, start);
389 Buf_AddStr(body, ":U");
390 AddEscaped(body, f->items.words[firstItem + i], endc);
391
392 *inout_mark = p;
393 *pp = p;
394 return;
395 }
396 }
397
398 /*
399 * While expanding the body of a .for loop, replace single-character
400 * expressions like $i with their ${:U...} expansion.
401 */
402 static void
ForLoop_SubstVarShort(ForLoop * f,unsigned firstItem,Buffer * body,const char * p,const char ** inout_mark)403 ForLoop_SubstVarShort(ForLoop *f, unsigned firstItem, Buffer *body,
404 const char *p, const char **inout_mark)
405 {
406 char ch = *p;
407 const char **vars;
408 size_t i;
409
410 /* Skip $$ and stupid ones. */
411 if (ch == '}' || ch == ')' || ch == ':' || ch == '$')
412 return;
413
414 vars = Vector_Get(&f->vars, 0);
415 for (i = 0; i < f->vars.len; i++) {
416 const char *varname = vars[i];
417 if (varname[0] == ch && varname[1] == '\0')
418 goto found;
419 }
420 return;
421
422 found:
423 Buf_AddRange(body, *inout_mark, p);
424 *inout_mark = p + 1;
425
426 /* Replace $<ch> with ${:U<value>} */
427 Buf_AddStr(body, "{:U");
428 AddEscaped(body, f->items.words[firstItem + i], '}');
429 Buf_AddByte(body, '}');
430 }
431
432 /*
433 * Compute the body for the current iteration by copying the unexpanded body,
434 * replacing the expressions for the iteration variables on the way.
435 *
436 * Using expressions ensures that the .for loop can't generate
437 * syntax, and that the later parsing will still see an expression.
438 * This code assumes that the variable with the empty name is never defined,
439 * see unit-tests/varname-empty.mk.
440 *
441 * The detection of substitutions of the loop control variables is naive.
442 * Many of the modifiers use '\$' instead of '$$' to escape '$', so it is
443 * possible to contrive a makefile where an unwanted substitution happens.
444 * See unit-tests/directive-for-escape.mk.
445 */
446 static void
ForLoop_SubstBody(ForLoop * f,unsigned firstItem,Buffer * body)447 ForLoop_SubstBody(ForLoop *f, unsigned firstItem, Buffer *body)
448 {
449 const char *p, *end;
450 const char *mark; /* where the last substitution left off */
451
452 Buf_Clear(body);
453
454 mark = f->body.data;
455 end = f->body.data + f->body.len;
456 for (p = mark; (p = strchr(p, '$')) != NULL;) {
457 if (p[1] == '{' || p[1] == '(') {
458 char endc = p[1] == '{' ? '}' : ')';
459 p += 2;
460 ForLoop_SubstVarLong(f, firstItem, body,
461 &p, endc, &mark);
462 } else {
463 ForLoop_SubstVarShort(f, firstItem, body,
464 p + 1, &mark);
465 p += 2;
466 }
467 }
468
469 Buf_AddRange(body, mark, end);
470 }
471
472 /*
473 * Compute the body for the current iteration by copying the unexpanded body,
474 * replacing the expressions for the iteration variables on the way.
475 */
476 bool
For_NextIteration(ForLoop * f,Buffer * body)477 For_NextIteration(ForLoop *f, Buffer *body)
478 {
479 if (f->nextItem == f->items.len)
480 return false;
481
482 f->nextItem += (unsigned)f->vars.len;
483 ForLoop_SubstBody(f, f->nextItem - (unsigned)f->vars.len, body);
484 if (DEBUG(FOR)) {
485 char *details = ForLoop_Details(f);
486 debug_printf("For: loop body with %s:\n%s",
487 details, body->data);
488 free(details);
489 }
490 return true;
491 }
492
493 /* Break out of the .for loop. */
494 void
For_Break(ForLoop * f)495 For_Break(ForLoop *f)
496 {
497 f->nextItem = (unsigned)f->items.len;
498 }
499
500 /* Run the .for loop, imitating the actions of an include file. */
501 void
For_Run(unsigned headLineno,unsigned bodyReadLines)502 For_Run(unsigned headLineno, unsigned bodyReadLines)
503 {
504 Buffer buf;
505 ForLoop *f = accumFor;
506 accumFor = NULL;
507
508 if (f->items.len > 0) {
509 Buf_Init(&buf);
510 Parse_PushInput(NULL, headLineno, bodyReadLines, buf, f);
511 } else
512 ForLoop_Free(f);
513 }
514