]>
Commit | Line | Data |
---|---|---|
447924ef | 1 | /* Data and functions related to line maps and input files. |
8d9254fc | 2 | Copyright (C) 2004-2020 Free Software Foundation, Inc. |
447924ef JM |
3 | |
4 | This file is part of GCC. | |
5 | ||
6 | GCC is free software; you can redistribute it and/or modify it under | |
7 | the terms of the GNU General Public License as published by the Free | |
8 | Software Foundation; either version 3, or (at your option) any later | |
9 | version. | |
10 | ||
11 | GCC is distributed in the hope that it will be useful, but WITHOUT ANY | |
12 | WARRANTY; without even the implied warranty of MERCHANTABILITY or | |
13 | FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License | |
14 | for more details. | |
15 | ||
16 | You should have received a copy of the GNU General Public License | |
17 | along with GCC; see the file COPYING3. If not see | |
18 | <http://www.gnu.org/licenses/>. */ | |
19 | ||
20 | #include "config.h" | |
21 | #include "system.h" | |
22 | #include "coretypes.h" | |
23 | #include "intl.h" | |
bc65bad2 | 24 | #include "diagnostic.h" |
ba4ad400 | 25 | #include "diagnostic-core.h" |
d9b950dd | 26 | #include "selftest.h" |
741d3be5 | 27 | #include "cpplib.h" |
7ecc3eb9 | 28 | |
a7d79e5c DM |
29 | #ifndef HAVE_ICONV |
30 | #define HAVE_ICONV 0 | |
31 | #endif | |
32 | ||
7ecc3eb9 DS |
33 | /* This is a cache used by get_next_line to store the content of a |
34 | file to be searched for file lines. */ | |
6c1dae73 | 35 | class fcache |
7ecc3eb9 | 36 | { |
6c1dae73 | 37 | public: |
7ecc3eb9 | 38 | /* These are information used to store a line boundary. */ |
6c1dae73 | 39 | class line_info |
7ecc3eb9 | 40 | { |
6c1dae73 | 41 | public: |
7ecc3eb9 DS |
42 | /* The line number. It starts from 1. */ |
43 | size_t line_num; | |
44 | ||
45 | /* The position (byte count) of the beginning of the line, | |
46 | relative to the file data pointer. This starts at zero. */ | |
47 | size_t start_pos; | |
48 | ||
49 | /* The position (byte count) of the last byte of the line. This | |
50 | normally points to the '\n' character, or to one byte after the | |
51 | last byte of the file, if the file doesn't contain a '\n' | |
52 | character. */ | |
53 | size_t end_pos; | |
54 | ||
55 | line_info (size_t l, size_t s, size_t e) | |
56 | : line_num (l), start_pos (s), end_pos (e) | |
57 | {} | |
58 | ||
59 | line_info () | |
60 | :line_num (0), start_pos (0), end_pos (0) | |
61 | {} | |
62 | }; | |
63 | ||
64 | /* The number of time this file has been accessed. This is used | |
65 | to designate which file cache to evict from the cache | |
66 | array. */ | |
67 | unsigned use_count; | |
68 | ||
f5ea989d DM |
69 | /* The file_path is the key for identifying a particular file in |
70 | the cache. | |
71 | For libcpp-using code, the underlying buffer for this field is | |
72 | owned by the corresponding _cpp_file within the cpp_reader. */ | |
7ecc3eb9 DS |
73 | const char *file_path; |
74 | ||
75 | FILE *fp; | |
76 | ||
77 | /* This points to the content of the file that we've read so | |
78 | far. */ | |
79 | char *data; | |
80 | ||
81 | /* The size of the DATA array above.*/ | |
82 | size_t size; | |
83 | ||
84 | /* The number of bytes read from the underlying file so far. This | |
85 | must be less (or equal) than SIZE above. */ | |
86 | size_t nb_read; | |
87 | ||
88 | /* The index of the beginning of the current line. */ | |
89 | size_t line_start_idx; | |
90 | ||
91 | /* The number of the previous line read. This starts at 1. Zero | |
92 | means we've read no line so far. */ | |
93 | size_t line_num; | |
94 | ||
95 | /* This is the total number of lines of the current file. At the | |
96 | moment, we try to get this information from the line map | |
97 | subsystem. Note that this is just a hint. When using the C++ | |
98 | front-end, this hint is correct because the input file is then | |
99 | completely tokenized before parsing starts; so the line map knows | |
100 | the number of lines before compilation really starts. For e.g, | |
101 | the C front-end, it can happen that we start emitting diagnostics | |
102 | before the line map has seen the end of the file. */ | |
103 | size_t total_lines; | |
104 | ||
c65236d6 DM |
105 | /* Could this file be missing a trailing newline on its final line? |
106 | Initially true (to cope with empty files), set to true/false | |
107 | as each line is read. */ | |
108 | bool missing_trailing_newline; | |
109 | ||
7ecc3eb9 DS |
110 | /* This is a record of the beginning and end of the lines we've seen |
111 | while reading the file. This is useful to avoid walking the data | |
112 | from the beginning when we are asked to read a line that is | |
113 | before LINE_START_IDX above. Note that the maximum size of this | |
114 | record is fcache_line_record_size, so that the memory consumption | |
115 | doesn't explode. We thus scale total_lines down to | |
116 | fcache_line_record_size. */ | |
117 | vec<line_info, va_heap> line_record; | |
118 | ||
119 | fcache (); | |
120 | ~fcache (); | |
121 | }; | |
447924ef JM |
122 | |
123 | /* Current position in real source file. */ | |
124 | ||
3edf64aa | 125 | location_t input_location = UNKNOWN_LOCATION; |
447924ef | 126 | |
99b1c316 | 127 | class line_maps *line_table; |
447924ef | 128 | |
f87e22c5 DM |
129 | /* A stashed copy of "line_table" for use by selftest::line_table_test. |
130 | This needs to be a global so that it can be a GC root, and thus | |
131 | prevent the stashed copy from being garbage-collected if the GC runs | |
132 | during a line_table_test. */ | |
133 | ||
99b1c316 | 134 | class line_maps *saved_line_table; |
f87e22c5 | 135 | |
7ecc3eb9 DS |
136 | static fcache *fcache_tab; |
137 | static const size_t fcache_tab_size = 16; | |
138 | static const size_t fcache_buffer_size = 4 * 1024; | |
139 | static const size_t fcache_line_record_size = 100; | |
140 | ||
84756fd4 DS |
141 | /* Expand the source location LOC into a human readable location. If |
142 | LOC resolves to a builtin location, the file name of the readable | |
7eb918cc DS |
143 | location is set to the string "<built-in>". If EXPANSION_POINT_P is |
144 | TRUE and LOC is virtual, then it is resolved to the expansion | |
145 | point of the involved macro. Otherwise, it is resolved to the | |
c4ca1a09 DS |
146 | spelling location of the token. |
147 | ||
148 | When resolving to the spelling location of the token, if the | |
149 | resulting location is for a built-in location (that is, it has no | |
150 | associated line/column) in the context of a macro expansion, the | |
151 | returned location is the first one (while unwinding the macro | |
152 | location towards its expansion point) that is in real source | |
c471c6ed DM |
153 | code. |
154 | ||
155 | ASPECT controls which part of the location to use. */ | |
7eb918cc DS |
156 | |
157 | static expanded_location | |
620e594b | 158 | expand_location_1 (location_t loc, |
c471c6ed DM |
159 | bool expansion_point_p, |
160 | enum location_aspect aspect) | |
447924ef JM |
161 | { |
162 | expanded_location xloc; | |
0e50b624 | 163 | const line_map_ordinary *map; |
c4ca1a09 | 164 | enum location_resolution_kind lrk = LRK_MACRO_EXPANSION_POINT; |
5368224f DC |
165 | tree block = NULL; |
166 | ||
167 | if (IS_ADHOC_LOC (loc)) | |
168 | { | |
169 | block = LOCATION_BLOCK (loc); | |
170 | loc = LOCATION_LOCUS (loc); | |
171 | } | |
c4ca1a09 DS |
172 | |
173 | memset (&xloc, 0, sizeof (xloc)); | |
84756fd4 | 174 | |
c4ca1a09 DS |
175 | if (loc >= RESERVED_LOCATION_COUNT) |
176 | { | |
177 | if (!expansion_point_p) | |
178 | { | |
179 | /* We want to resolve LOC to its spelling location. | |
180 | ||
181 | But if that spelling location is a reserved location that | |
182 | appears in the context of a macro expansion (like for a | |
183 | location for a built-in token), let's consider the first | |
184 | location (toward the expansion point) that is not reserved; | |
185 | that is, the first location that is in real source code. */ | |
186 | loc = linemap_unwind_to_first_non_reserved_loc (line_table, | |
0e50b624 | 187 | loc, NULL); |
c4ca1a09 DS |
188 | lrk = LRK_SPELLING_LOCATION; |
189 | } | |
c471c6ed DM |
190 | loc = linemap_resolve_location (line_table, loc, lrk, &map); |
191 | ||
192 | /* loc is now either in an ordinary map, or is a reserved location. | |
193 | If it is a compound location, the caret is in a spelling location, | |
194 | but the start/finish might still be a virtual location. | |
195 | Depending of what the caller asked for, we may need to recurse | |
196 | one level in order to resolve any virtual locations in the | |
197 | end-points. */ | |
198 | switch (aspect) | |
199 | { | |
200 | default: | |
201 | gcc_unreachable (); | |
202 | /* Fall through. */ | |
203 | case LOCATION_ASPECT_CARET: | |
204 | break; | |
205 | case LOCATION_ASPECT_START: | |
206 | { | |
620e594b | 207 | location_t start = get_start (loc); |
c471c6ed DM |
208 | if (start != loc) |
209 | return expand_location_1 (start, expansion_point_p, aspect); | |
210 | } | |
211 | break; | |
212 | case LOCATION_ASPECT_FINISH: | |
213 | { | |
620e594b | 214 | location_t finish = get_finish (loc); |
c471c6ed DM |
215 | if (finish != loc) |
216 | return expand_location_1 (finish, expansion_point_p, aspect); | |
217 | } | |
218 | break; | |
219 | } | |
c4ca1a09 DS |
220 | xloc = linemap_expand_location (line_table, map, loc); |
221 | } | |
84756fd4 | 222 | |
5368224f | 223 | xloc.data = block; |
447924ef | 224 | if (loc <= BUILTINS_LOCATION) |
84756fd4 DS |
225 | xloc.file = loc == UNKNOWN_LOCATION ? NULL : _("<built-in>"); |
226 | ||
447924ef JM |
227 | return xloc; |
228 | } | |
64a1a422 | 229 | |
7ecc3eb9 DS |
230 | /* Initialize the set of cache used for files accessed by caret |
231 | diagnostic. */ | |
232 | ||
233 | static void | |
234 | diagnostic_file_cache_init (void) | |
235 | { | |
236 | if (fcache_tab == NULL) | |
237 | fcache_tab = new fcache[fcache_tab_size]; | |
238 | } | |
239 | ||
592f32fa | 240 | /* Free the resources used by the set of cache used for files accessed |
7ecc3eb9 DS |
241 | by caret diagnostic. */ |
242 | ||
243 | void | |
244 | diagnostic_file_cache_fini (void) | |
245 | { | |
246 | if (fcache_tab) | |
247 | { | |
248 | delete [] (fcache_tab); | |
249 | fcache_tab = NULL; | |
250 | } | |
251 | } | |
252 | ||
253 | /* Return the total lines number that have been read so far by the | |
254 | line map (in the preprocessor) so far. For languages like C++ that | |
255 | entirely preprocess the input file before starting to parse, this | |
256 | equals the actual number of lines of the file. */ | |
257 | ||
258 | static size_t | |
259 | total_lines_num (const char *file_path) | |
260 | { | |
261 | size_t r = 0; | |
620e594b | 262 | location_t l = 0; |
7ecc3eb9 DS |
263 | if (linemap_get_file_highest_location (line_table, file_path, &l)) |
264 | { | |
265 | gcc_assert (l >= RESERVED_LOCATION_COUNT); | |
266 | expanded_location xloc = expand_location (l); | |
267 | r = xloc.line; | |
268 | } | |
269 | return r; | |
270 | } | |
271 | ||
272 | /* Lookup the cache used for the content of a given file accessed by | |
273 | caret diagnostic. Return the found cached file, or NULL if no | |
274 | cached file was found. */ | |
275 | ||
276 | static fcache* | |
277 | lookup_file_in_cache_tab (const char *file_path) | |
278 | { | |
279 | if (file_path == NULL) | |
280 | return NULL; | |
281 | ||
282 | diagnostic_file_cache_init (); | |
283 | ||
284 | /* This will contain the found cached file. */ | |
285 | fcache *r = NULL; | |
286 | for (unsigned i = 0; i < fcache_tab_size; ++i) | |
287 | { | |
288 | fcache *c = &fcache_tab[i]; | |
289 | if (c->file_path && !strcmp (c->file_path, file_path)) | |
290 | { | |
291 | ++c->use_count; | |
292 | r = c; | |
293 | } | |
294 | } | |
295 | ||
296 | if (r) | |
297 | ++r->use_count; | |
298 | ||
299 | return r; | |
300 | } | |
301 | ||
f89b03b6 DM |
302 | /* Purge any mention of FILENAME from the cache of files used for |
303 | printing source code. For use in selftests when working | |
304 | with tempfiles. */ | |
305 | ||
306 | void | |
307 | diagnostics_file_cache_forcibly_evict_file (const char *file_path) | |
308 | { | |
309 | gcc_assert (file_path); | |
310 | ||
311 | fcache *r = lookup_file_in_cache_tab (file_path); | |
312 | if (!r) | |
313 | /* Not found. */ | |
314 | return; | |
315 | ||
316 | r->file_path = NULL; | |
317 | if (r->fp) | |
318 | fclose (r->fp); | |
319 | r->fp = NULL; | |
320 | r->nb_read = 0; | |
321 | r->line_start_idx = 0; | |
322 | r->line_num = 0; | |
323 | r->line_record.truncate (0); | |
324 | r->use_count = 0; | |
325 | r->total_lines = 0; | |
c65236d6 | 326 | r->missing_trailing_newline = true; |
f89b03b6 DM |
327 | } |
328 | ||
7ecc3eb9 DS |
329 | /* Return the file cache that has been less used, recently, or the |
330 | first empty one. If HIGHEST_USE_COUNT is non-null, | |
331 | *HIGHEST_USE_COUNT is set to the highest use count of the entries | |
332 | in the cache table. */ | |
333 | ||
334 | static fcache* | |
335 | evicted_cache_tab_entry (unsigned *highest_use_count) | |
336 | { | |
337 | diagnostic_file_cache_init (); | |
338 | ||
339 | fcache *to_evict = &fcache_tab[0]; | |
340 | unsigned huc = to_evict->use_count; | |
341 | for (unsigned i = 1; i < fcache_tab_size; ++i) | |
342 | { | |
343 | fcache *c = &fcache_tab[i]; | |
344 | bool c_is_empty = (c->file_path == NULL); | |
345 | ||
346 | if (c->use_count < to_evict->use_count | |
347 | || (to_evict->file_path && c_is_empty)) | |
348 | /* We evict C because it's either an entry with a lower use | |
349 | count or one that is empty. */ | |
350 | to_evict = c; | |
351 | ||
352 | if (huc < c->use_count) | |
353 | huc = c->use_count; | |
354 | ||
355 | if (c_is_empty) | |
356 | /* We've reached the end of the cache; subsequent elements are | |
357 | all empty. */ | |
358 | break; | |
359 | } | |
360 | ||
361 | if (highest_use_count) | |
362 | *highest_use_count = huc; | |
363 | ||
364 | return to_evict; | |
365 | } | |
366 | ||
367 | /* Create the cache used for the content of a given file to be | |
368 | accessed by caret diagnostic. This cache is added to an array of | |
369 | cache and can be retrieved by lookup_file_in_cache_tab. This | |
370 | function returns the created cache. Note that only the last | |
371 | fcache_tab_size files are cached. */ | |
372 | ||
373 | static fcache* | |
374 | add_file_to_cache_tab (const char *file_path) | |
375 | { | |
376 | ||
377 | FILE *fp = fopen (file_path, "r"); | |
317363b4 DS |
378 | if (fp == NULL) |
379 | return NULL; | |
7ecc3eb9 DS |
380 | |
381 | unsigned highest_use_count = 0; | |
382 | fcache *r = evicted_cache_tab_entry (&highest_use_count); | |
383 | r->file_path = file_path; | |
384 | if (r->fp) | |
385 | fclose (r->fp); | |
386 | r->fp = fp; | |
387 | r->nb_read = 0; | |
388 | r->line_start_idx = 0; | |
389 | r->line_num = 0; | |
390 | r->line_record.truncate (0); | |
391 | /* Ensure that this cache entry doesn't get evicted next time | |
392 | add_file_to_cache_tab is called. */ | |
393 | r->use_count = ++highest_use_count; | |
394 | r->total_lines = total_lines_num (file_path); | |
c65236d6 | 395 | r->missing_trailing_newline = true; |
7ecc3eb9 DS |
396 | |
397 | return r; | |
398 | } | |
399 | ||
400 | /* Lookup the cache used for the content of a given file accessed by | |
401 | caret diagnostic. If no cached file was found, create a new cache | |
402 | for this file, add it to the array of cached file and return | |
403 | it. */ | |
404 | ||
405 | static fcache* | |
406 | lookup_or_add_file_to_cache_tab (const char *file_path) | |
407 | { | |
408 | fcache *r = lookup_file_in_cache_tab (file_path); | |
409 | if (r == NULL) | |
410 | r = add_file_to_cache_tab (file_path); | |
411 | return r; | |
412 | } | |
413 | ||
414 | /* Default constructor for a cache of file used by caret | |
415 | diagnostic. */ | |
416 | ||
417 | fcache::fcache () | |
418 | : use_count (0), file_path (NULL), fp (NULL), data (0), | |
419 | size (0), nb_read (0), line_start_idx (0), line_num (0), | |
c65236d6 | 420 | total_lines (0), missing_trailing_newline (true) |
7ecc3eb9 DS |
421 | { |
422 | line_record.create (0); | |
423 | } | |
424 | ||
425 | /* Destructor for a cache of file used by caret diagnostic. */ | |
426 | ||
427 | fcache::~fcache () | |
428 | { | |
429 | if (fp) | |
430 | { | |
431 | fclose (fp); | |
432 | fp = NULL; | |
433 | } | |
434 | if (data) | |
435 | { | |
436 | XDELETEVEC (data); | |
437 | data = 0; | |
438 | } | |
439 | line_record.release (); | |
440 | } | |
441 | ||
442 | /* Returns TRUE iff the cache would need to be filled with data coming | |
443 | from the file. That is, either the cache is empty or full or the | |
444 | current line is empty. Note that if the cache is full, it would | |
445 | need to be extended and filled again. */ | |
446 | ||
447 | static bool | |
448 | needs_read (fcache *c) | |
449 | { | |
450 | return (c->nb_read == 0 | |
451 | || c->nb_read == c->size | |
452 | || (c->line_start_idx >= c->nb_read - 1)); | |
453 | } | |
454 | ||
455 | /* Return TRUE iff the cache is full and thus needs to be | |
456 | extended. */ | |
457 | ||
458 | static bool | |
459 | needs_grow (fcache *c) | |
460 | { | |
461 | return c->nb_read == c->size; | |
462 | } | |
463 | ||
464 | /* Grow the cache if it needs to be extended. */ | |
465 | ||
466 | static void | |
467 | maybe_grow (fcache *c) | |
9fec0042 | 468 | { |
7ecc3eb9 DS |
469 | if (!needs_grow (c)) |
470 | return; | |
471 | ||
472 | size_t size = c->size == 0 ? fcache_buffer_size : c->size * 2; | |
1adae327 | 473 | c->data = XRESIZEVEC (char, c->data, size); |
7ecc3eb9 DS |
474 | c->size = size; |
475 | } | |
9fec0042 | 476 | |
7ecc3eb9 DS |
477 | /* Read more data into the cache. Extends the cache if need be. |
478 | Returns TRUE iff new data could be read. */ | |
479 | ||
480 | static bool | |
481 | read_data (fcache *c) | |
482 | { | |
483 | if (feof (c->fp) || ferror (c->fp)) | |
484 | return false; | |
485 | ||
486 | maybe_grow (c); | |
487 | ||
488 | char * from = c->data + c->nb_read; | |
489 | size_t to_read = c->size - c->nb_read; | |
490 | size_t nb_read = fread (from, 1, to_read, c->fp); | |
491 | ||
492 | if (ferror (c->fp)) | |
493 | return false; | |
494 | ||
495 | c->nb_read += nb_read; | |
496 | return !!nb_read; | |
497 | } | |
498 | ||
499 | /* Read new data iff the cache needs to be filled with more data | |
500 | coming from the file FP. Return TRUE iff the cache was filled with | |
501 | mode data. */ | |
502 | ||
503 | static bool | |
504 | maybe_read_data (fcache *c) | |
505 | { | |
506 | if (!needs_read (c)) | |
507 | return false; | |
508 | return read_data (c); | |
509 | } | |
510 | ||
511 | /* Read a new line from file FP, using C as a cache for the data | |
512 | coming from the file. Upon successful completion, *LINE is set to | |
1adae327 BE |
513 | the beginning of the line found. *LINE points directly in the |
514 | line cache and is only valid until the next call of get_next_line. | |
7ecc3eb9 DS |
515 | *LINE_LEN is set to the length of the line. Note that the line |
516 | does not contain any terminal delimiter. This function returns | |
517 | true if some data was read or process from the cache, false | |
1adae327 BE |
518 | otherwise. Note that subsequent calls to get_next_line might |
519 | make the content of *LINE invalid. */ | |
7ecc3eb9 DS |
520 | |
521 | static bool | |
522 | get_next_line (fcache *c, char **line, ssize_t *line_len) | |
523 | { | |
524 | /* Fill the cache with data to process. */ | |
525 | maybe_read_data (c); | |
526 | ||
527 | size_t remaining_size = c->nb_read - c->line_start_idx; | |
528 | if (remaining_size == 0) | |
529 | /* There is no more data to process. */ | |
530 | return false; | |
531 | ||
532 | char *line_start = c->data + c->line_start_idx; | |
533 | ||
534 | char *next_line_start = NULL; | |
535 | size_t len = 0; | |
536 | char *line_end = (char *) memchr (line_start, '\n', remaining_size); | |
537 | if (line_end == NULL) | |
9fec0042 | 538 | { |
7ecc3eb9 DS |
539 | /* We haven't found the end-of-line delimiter in the cache. |
540 | Fill the cache with more data from the file and look for the | |
541 | '\n'. */ | |
542 | while (maybe_read_data (c)) | |
543 | { | |
544 | line_start = c->data + c->line_start_idx; | |
545 | remaining_size = c->nb_read - c->line_start_idx; | |
546 | line_end = (char *) memchr (line_start, '\n', remaining_size); | |
547 | if (line_end != NULL) | |
548 | { | |
549 | next_line_start = line_end + 1; | |
550 | break; | |
551 | } | |
552 | } | |
553 | if (line_end == NULL) | |
c65236d6 DM |
554 | { |
555 | /* We've loadded all the file into the cache and still no | |
556 | '\n'. Let's say the line ends up at one byte passed the | |
557 | end of the file. This is to stay consistent with the case | |
558 | of when the line ends up with a '\n' and line_end points to | |
559 | that terminal '\n'. That consistency is useful below in | |
560 | the len calculation. */ | |
561 | line_end = c->data + c->nb_read ; | |
562 | c->missing_trailing_newline = true; | |
563 | } | |
564 | else | |
565 | c->missing_trailing_newline = false; | |
9fec0042 | 566 | } |
7ecc3eb9 | 567 | else |
c65236d6 DM |
568 | { |
569 | next_line_start = line_end + 1; | |
570 | c->missing_trailing_newline = false; | |
571 | } | |
7ecc3eb9 DS |
572 | |
573 | if (ferror (c->fp)) | |
1adae327 | 574 | return false; |
7ecc3eb9 DS |
575 | |
576 | /* At this point, we've found the end of the of line. It either | |
577 | points to the '\n' or to one byte after the last byte of the | |
578 | file. */ | |
579 | gcc_assert (line_end != NULL); | |
9fec0042 | 580 | |
7ecc3eb9 DS |
581 | len = line_end - line_start; |
582 | ||
583 | if (c->line_start_idx < c->nb_read) | |
584 | *line = line_start; | |
585 | ||
586 | ++c->line_num; | |
587 | ||
588 | /* Before we update our line record, make sure the hint about the | |
589 | total number of lines of the file is correct. If it's not, then | |
590 | we give up recording line boundaries from now on. */ | |
591 | bool update_line_record = true; | |
592 | if (c->line_num > c->total_lines) | |
593 | update_line_record = false; | |
594 | ||
595 | /* Now update our line record so that re-reading lines from the | |
596 | before c->line_start_idx is faster. */ | |
597 | if (update_line_record | |
598 | && c->line_record.length () < fcache_line_record_size) | |
599 | { | |
600 | /* If the file lines fits in the line record, we just record all | |
601 | its lines ...*/ | |
602 | if (c->total_lines <= fcache_line_record_size | |
603 | && c->line_num > c->line_record.length ()) | |
604 | c->line_record.safe_push (fcache::line_info (c->line_num, | |
605 | c->line_start_idx, | |
606 | line_end - c->data)); | |
607 | else if (c->total_lines > fcache_line_record_size) | |
608 | { | |
609 | /* ... otherwise, we just scale total_lines down to | |
610 | (fcache_line_record_size lines. */ | |
611 | size_t n = (c->line_num * fcache_line_record_size) / c->total_lines; | |
612 | if (c->line_record.length () == 0 | |
613 | || n >= c->line_record.length ()) | |
614 | c->line_record.safe_push (fcache::line_info (c->line_num, | |
615 | c->line_start_idx, | |
616 | line_end - c->data)); | |
617 | } | |
618 | } | |
619 | ||
620 | /* Update c->line_start_idx so that it points to the next line to be | |
621 | read. */ | |
622 | if (next_line_start) | |
623 | c->line_start_idx = next_line_start - c->data; | |
624 | else | |
625 | /* We didn't find any terminal '\n'. Let's consider that the end | |
626 | of line is the end of the data in the cache. The next | |
627 | invocation of get_next_line will either read more data from the | |
628 | underlying file or return false early because we've reached the | |
629 | end of the file. */ | |
630 | c->line_start_idx = c->nb_read; | |
631 | ||
632 | *line_len = len; | |
633 | ||
634 | return true; | |
635 | } | |
636 | ||
7ecc3eb9 DS |
637 | /* Consume the next bytes coming from the cache (or from its |
638 | underlying file if there are remaining unread bytes in the file) | |
639 | until we reach the next end-of-line (or end-of-file). There is no | |
640 | copying from the cache involved. Return TRUE upon successful | |
641 | completion. */ | |
642 | ||
643 | static bool | |
644 | goto_next_line (fcache *cache) | |
645 | { | |
646 | char *l; | |
647 | ssize_t len; | |
648 | ||
649 | return get_next_line (cache, &l, &len); | |
650 | } | |
651 | ||
652 | /* Read an arbitrary line number LINE_NUM from the file cached in C. | |
1adae327 BE |
653 | If the line was read successfully, *LINE points to the beginning |
654 | of the line in the file cache and *LINE_LEN is the length of the | |
655 | line. *LINE is not nul-terminated, but may contain zero bytes. | |
656 | *LINE is only valid until the next call of read_line_num. | |
7ecc3eb9 DS |
657 | This function returns bool if a line was read. */ |
658 | ||
659 | static bool | |
660 | read_line_num (fcache *c, size_t line_num, | |
1adae327 | 661 | char **line, ssize_t *line_len) |
7ecc3eb9 DS |
662 | { |
663 | gcc_assert (line_num > 0); | |
664 | ||
665 | if (line_num <= c->line_num) | |
9789a912 | 666 | { |
7ecc3eb9 DS |
667 | /* We've been asked to read lines that are before c->line_num. |
668 | So lets use our line record (if it's not empty) to try to | |
669 | avoid re-reading the file from the beginning again. */ | |
7f4d640c | 670 | |
7ecc3eb9 | 671 | if (c->line_record.is_empty ()) |
9fec0042 | 672 | { |
7ecc3eb9 DS |
673 | c->line_start_idx = 0; |
674 | c->line_num = 0; | |
675 | } | |
676 | else | |
677 | { | |
678 | fcache::line_info *i = NULL; | |
679 | if (c->total_lines <= fcache_line_record_size) | |
680 | { | |
681 | /* In languages where the input file is not totally | |
682 | preprocessed up front, the c->total_lines hint | |
683 | can be smaller than the number of lines of the | |
684 | file. In that case, only the first | |
685 | c->total_lines have been recorded. | |
686 | ||
687 | Otherwise, the first c->total_lines we've read have | |
688 | their start/end recorded here. */ | |
689 | i = (line_num <= c->total_lines) | |
690 | ? &c->line_record[line_num - 1] | |
691 | : &c->line_record[c->total_lines - 1]; | |
692 | gcc_assert (i->line_num <= line_num); | |
693 | } | |
694 | else | |
695 | { | |
696 | /* So the file had more lines than our line record | |
697 | size. Thus the number of lines we've recorded has | |
698 | been scaled down to fcache_line_reacord_size. Let's | |
699 | pick the start/end of the recorded line that is | |
700 | closest to line_num. */ | |
701 | size_t n = (line_num <= c->total_lines) | |
702 | ? line_num * fcache_line_record_size / c->total_lines | |
703 | : c ->line_record.length () - 1; | |
704 | if (n < c->line_record.length ()) | |
705 | { | |
706 | i = &c->line_record[n]; | |
707 | gcc_assert (i->line_num <= line_num); | |
708 | } | |
709 | } | |
710 | ||
711 | if (i && i->line_num == line_num) | |
712 | { | |
1adae327 BE |
713 | /* We have the start/end of the line. */ |
714 | *line = c->data + i->start_pos; | |
715 | *line_len = i->end_pos - i->start_pos; | |
7ecc3eb9 DS |
716 | return true; |
717 | } | |
718 | ||
719 | if (i) | |
720 | { | |
721 | c->line_start_idx = i->start_pos; | |
722 | c->line_num = i->line_num - 1; | |
723 | } | |
724 | else | |
725 | { | |
726 | c->line_start_idx = 0; | |
727 | c->line_num = 0; | |
728 | } | |
9fec0042 | 729 | } |
9fec0042 | 730 | } |
7ecc3eb9 DS |
731 | |
732 | /* Let's walk from line c->line_num up to line_num - 1, without | |
733 | copying any line. */ | |
734 | while (c->line_num < line_num - 1) | |
735 | if (!goto_next_line (c)) | |
736 | return false; | |
737 | ||
738 | /* The line we want is the next one. Let's read and copy it back to | |
739 | the caller. */ | |
1adae327 | 740 | return get_next_line (c, line, line_len); |
9fec0042 MLI |
741 | } |
742 | ||
1adae327 BE |
743 | /* Return the physical source line that corresponds to FILE_PATH/LINE. |
744 | The line is not nul-terminated. The returned pointer is only | |
745 | valid until the next call of location_get_source_line. | |
746 | Note that the line can contain several null characters, | |
7761dfbe DM |
747 | so the returned value's length has the actual length of the line. |
748 | If the function fails, a NULL char_span is returned. */ | |
9fec0042 | 749 | |
7761dfbe DM |
750 | char_span |
751 | location_get_source_line (const char *file_path, int line) | |
9fec0042 | 752 | { |
ac2a97db | 753 | char *buffer = NULL; |
1adae327 | 754 | ssize_t len; |
7ecc3eb9 | 755 | |
31bdd08a | 756 | if (line == 0) |
7761dfbe | 757 | return char_span (NULL, 0); |
367c8286 | 758 | |
31bdd08a | 759 | fcache *c = lookup_or_add_file_to_cache_tab (file_path); |
367c8286 | 760 | if (c == NULL) |
7761dfbe | 761 | return char_span (NULL, 0); |
367c8286 | 762 | |
31bdd08a | 763 | bool read = read_line_num (c, line, &buffer, &len); |
7761dfbe DM |
764 | if (!read) |
765 | return char_span (NULL, 0); | |
9fec0042 | 766 | |
7761dfbe | 767 | return char_span (buffer, len); |
9fec0042 MLI |
768 | } |
769 | ||
c65236d6 DM |
770 | /* Determine if FILE_PATH missing a trailing newline on its final line. |
771 | Only valid to call once all of the file has been loaded, by | |
772 | requesting a line number beyond the end of the file. */ | |
773 | ||
774 | bool | |
775 | location_missing_trailing_newline (const char *file_path) | |
776 | { | |
777 | fcache *c = lookup_or_add_file_to_cache_tab (file_path); | |
778 | if (c == NULL) | |
779 | return false; | |
780 | ||
781 | return c->missing_trailing_newline; | |
782 | } | |
783 | ||
c468587a DS |
784 | /* Test if the location originates from the spelling location of a |
785 | builtin-tokens. That is, return TRUE if LOC is a (possibly | |
786 | virtual) location of a built-in token that appears in the expansion | |
787 | list of a macro. Please note that this function also works on | |
788 | tokens that result from built-in tokens. For instance, the | |
789 | function would return true if passed a token "4" that is the result | |
790 | of the expansion of the built-in __LINE__ macro. */ | |
791 | bool | |
620e594b | 792 | is_location_from_builtin_token (location_t loc) |
c468587a | 793 | { |
0e50b624 | 794 | const line_map_ordinary *map = NULL; |
c468587a DS |
795 | loc = linemap_resolve_location (line_table, loc, |
796 | LRK_SPELLING_LOCATION, &map); | |
797 | return loc == BUILTINS_LOCATION; | |
798 | } | |
799 | ||
7eb918cc DS |
800 | /* Expand the source location LOC into a human readable location. If |
801 | LOC is virtual, it resolves to the expansion point of the involved | |
802 | macro. If LOC resolves to a builtin location, the file name of the | |
803 | readable location is set to the string "<built-in>". */ | |
804 | ||
805 | expanded_location | |
620e594b | 806 | expand_location (location_t loc) |
7eb918cc | 807 | { |
c471c6ed DM |
808 | return expand_location_1 (loc, /*expansion_point_p=*/true, |
809 | LOCATION_ASPECT_CARET); | |
7eb918cc DS |
810 | } |
811 | ||
812 | /* Expand the source location LOC into a human readable location. If | |
813 | LOC is virtual, it resolves to the expansion location of the | |
814 | relevant macro. If LOC resolves to a builtin location, the file | |
815 | name of the readable location is set to the string | |
816 | "<built-in>". */ | |
817 | ||
818 | expanded_location | |
620e594b | 819 | expand_location_to_spelling_point (location_t loc, |
0d48e877 | 820 | enum location_aspect aspect) |
7eb918cc | 821 | { |
0d48e877 | 822 | return expand_location_1 (loc, /*expansion_point_p=*/false, aspect); |
7eb918cc DS |
823 | } |
824 | ||
8a645150 | 825 | /* The rich_location class within libcpp requires a way to expand |
620e594b | 826 | location_t instances, and relies on the client code |
8a645150 DM |
827 | providing a symbol named |
828 | linemap_client_expand_location_to_spelling_point | |
829 | to do this. | |
830 | ||
831 | This is the implementation for libcommon.a (all host binaries), | |
c471c6ed | 832 | which simply calls into expand_location_1. */ |
8a645150 DM |
833 | |
834 | expanded_location | |
620e594b | 835 | linemap_client_expand_location_to_spelling_point (location_t loc, |
c471c6ed | 836 | enum location_aspect aspect) |
8a645150 | 837 | { |
c471c6ed | 838 | return expand_location_1 (loc, /*expansion_point_p=*/false, aspect); |
8a645150 DM |
839 | } |
840 | ||
841 | ||
e1f0c178 MLI |
842 | /* If LOCATION is in a system header and if it is a virtual location for |
843 | a token coming from the expansion of a macro, unwind it to the | |
844 | location of the expansion point of the macro. Otherwise, just return | |
70dc395a DS |
845 | LOCATION. |
846 | ||
847 | This is used for instance when we want to emit diagnostics about a | |
e1f0c178 MLI |
848 | token that may be located in a macro that is itself defined in a |
849 | system header, for example, for the NULL macro. In such a case, if | |
850 | LOCATION were passed directly to diagnostic functions such as | |
851 | warning_at, the diagnostic would be suppressed (unless | |
852 | -Wsystem-headers). */ | |
70dc395a | 853 | |
620e594b DM |
854 | location_t |
855 | expansion_point_location_if_in_system_header (location_t location) | |
70dc395a DS |
856 | { |
857 | if (in_system_header_at (location)) | |
858 | location = linemap_resolve_location (line_table, location, | |
859 | LRK_MACRO_EXPANSION_POINT, | |
860 | NULL); | |
861 | return location; | |
862 | } | |
7eb918cc | 863 | |
79ce98bc MP |
864 | /* If LOCATION is a virtual location for a token coming from the expansion |
865 | of a macro, unwind to the location of the expansion point of the macro. */ | |
866 | ||
620e594b DM |
867 | location_t |
868 | expansion_point_location (location_t location) | |
79ce98bc MP |
869 | { |
870 | return linemap_resolve_location (line_table, location, | |
871 | LRK_MACRO_EXPANSION_POINT, NULL); | |
872 | } | |
873 | ||
a01fc549 DM |
874 | /* Construct a location with caret at CARET, ranging from START to |
875 | finish e.g. | |
876 | ||
877 | 11111111112 | |
878 | 12345678901234567890 | |
879 | 522 | |
880 | 523 return foo + bar; | |
881 | ~~~~^~~~~ | |
882 | 524 | |
883 | ||
884 | The location's caret is at the "+", line 523 column 15, but starts | |
885 | earlier, at the "f" of "foo" at column 11. The finish is at the "r" | |
886 | of "bar" at column 19. */ | |
887 | ||
888 | location_t | |
889 | make_location (location_t caret, location_t start, location_t finish) | |
890 | { | |
891 | location_t pure_loc = get_pure_location (caret); | |
892 | source_range src_range; | |
9144eabb DM |
893 | src_range.m_start = get_start (start); |
894 | src_range.m_finish = get_finish (finish); | |
a01fc549 DM |
895 | location_t combined_loc = COMBINE_LOCATION_DATA (line_table, |
896 | pure_loc, | |
897 | src_range, | |
898 | NULL); | |
899 | return combined_loc; | |
900 | } | |
901 | ||
a32c8316 MP |
902 | /* Same as above, but taking a source range rather than two locations. */ |
903 | ||
904 | location_t | |
905 | make_location (location_t caret, source_range src_range) | |
906 | { | |
907 | location_t pure_loc = get_pure_location (caret); | |
908 | return COMBINE_LOCATION_DATA (line_table, pure_loc, src_range, NULL); | |
909 | } | |
910 | ||
ee925640 LH |
911 | /* An expanded_location stores the column in byte units. This function |
912 | converts that column to display units. That requires reading the associated | |
913 | source line in order to calculate the display width. If that cannot be done | |
914 | for any reason, then returns the byte column as a fallback. */ | |
915 | int | |
916 | location_compute_display_column (expanded_location exploc) | |
917 | { | |
918 | if (!(exploc.file && *exploc.file && exploc.line && exploc.column)) | |
919 | return exploc.column; | |
920 | char_span line = location_get_source_line (exploc.file, exploc.line); | |
921 | /* If line is NULL, this function returns exploc.column which is the | |
922 | desired fallback. */ | |
923 | return cpp_byte_column_to_display_column (line.get_buffer (), line.length (), | |
924 | exploc.column); | |
925 | } | |
926 | ||
64a1a422 TT |
927 | /* Dump statistics to stderr about the memory usage of the line_table |
928 | set of line maps. This also displays some statistics about macro | |
929 | expansion. */ | |
930 | ||
931 | void | |
932 | dump_line_table_statistics (void) | |
933 | { | |
934 | struct linemap_stats s; | |
d17687f6 | 935 | long total_used_map_size, |
64a1a422 TT |
936 | macro_maps_size, |
937 | total_allocated_map_size; | |
938 | ||
939 | memset (&s, 0, sizeof (s)); | |
940 | ||
941 | linemap_get_statistics (line_table, &s); | |
942 | ||
943 | macro_maps_size = s.macro_maps_used_size | |
944 | + s.macro_maps_locations_size; | |
945 | ||
946 | total_allocated_map_size = s.ordinary_maps_allocated_size | |
947 | + s.macro_maps_allocated_size | |
948 | + s.macro_maps_locations_size; | |
949 | ||
950 | total_used_map_size = s.ordinary_maps_used_size | |
951 | + s.macro_maps_used_size | |
952 | + s.macro_maps_locations_size; | |
953 | ||
d17687f6 | 954 | fprintf (stderr, "Number of expanded macros: %5ld\n", |
64a1a422 TT |
955 | s.num_expanded_macros); |
956 | if (s.num_expanded_macros != 0) | |
d17687f6 | 957 | fprintf (stderr, "Average number of tokens per macro expansion: %5ld\n", |
64a1a422 TT |
958 | s.num_macro_tokens / s.num_expanded_macros); |
959 | fprintf (stderr, | |
960 | "\nLine Table allocations during the " | |
40ce7fa6 | 961 | "compilation process\n"); |
a0b48080 | 962 | fprintf (stderr, "Number of ordinary maps used: " PRsa (5) "\n", |
40ce7fa6 | 963 | SIZE_AMOUNT (s.num_ordinary_maps_used)); |
a0b48080 | 964 | fprintf (stderr, "Ordinary map used size: " PRsa (5) "\n", |
40ce7fa6 | 965 | SIZE_AMOUNT (s.ordinary_maps_used_size)); |
a0b48080 | 966 | fprintf (stderr, "Number of ordinary maps allocated: " PRsa (5) "\n", |
40ce7fa6 | 967 | SIZE_AMOUNT (s.num_ordinary_maps_allocated)); |
a0b48080 | 968 | fprintf (stderr, "Ordinary maps allocated size: " PRsa (5) "\n", |
40ce7fa6 | 969 | SIZE_AMOUNT (s.ordinary_maps_allocated_size)); |
a0b48080 | 970 | fprintf (stderr, "Number of macro maps used: " PRsa (5) "\n", |
40ce7fa6 | 971 | SIZE_AMOUNT (s.num_macro_maps_used)); |
a0b48080 | 972 | fprintf (stderr, "Macro maps used size: " PRsa (5) "\n", |
40ce7fa6 | 973 | SIZE_AMOUNT (s.macro_maps_used_size)); |
a0b48080 | 974 | fprintf (stderr, "Macro maps locations size: " PRsa (5) "\n", |
40ce7fa6 | 975 | SIZE_AMOUNT (s.macro_maps_locations_size)); |
a0b48080 | 976 | fprintf (stderr, "Macro maps size: " PRsa (5) "\n", |
40ce7fa6 | 977 | SIZE_AMOUNT (macro_maps_size)); |
a0b48080 | 978 | fprintf (stderr, "Duplicated maps locations size: " PRsa (5) "\n", |
40ce7fa6 | 979 | SIZE_AMOUNT (s.duplicated_macro_maps_locations_size)); |
a0b48080 | 980 | fprintf (stderr, "Total allocated maps size: " PRsa (5) "\n", |
40ce7fa6 | 981 | SIZE_AMOUNT (total_allocated_map_size)); |
a0b48080 | 982 | fprintf (stderr, "Total used maps size: " PRsa (5) "\n", |
40ce7fa6 | 983 | SIZE_AMOUNT (total_used_map_size)); |
a0b48080 | 984 | fprintf (stderr, "Ad-hoc table size: " PRsa (5) "\n", |
40ce7fa6 | 985 | SIZE_AMOUNT (s.adhoc_table_size)); |
a0b48080 | 986 | fprintf (stderr, "Ad-hoc table entries used: " PRsa (5) "\n", |
40ce7fa6 | 987 | SIZE_AMOUNT (s.adhoc_table_entries_used)); |
a0b48080 | 988 | fprintf (stderr, "optimized_ranges: " PRsa (5) "\n", |
40ce7fa6 | 989 | SIZE_AMOUNT (line_table->num_optimized_ranges)); |
a0b48080 | 990 | fprintf (stderr, "unoptimized_ranges: " PRsa (5) "\n", |
40ce7fa6 | 991 | SIZE_AMOUNT (line_table->num_unoptimized_ranges)); |
ee015909 | 992 | |
64a1a422 TT |
993 | fprintf (stderr, "\n"); |
994 | } | |
ba4ad400 DM |
995 | |
996 | /* Get location one beyond the final location in ordinary map IDX. */ | |
997 | ||
620e594b | 998 | static location_t |
99b1c316 | 999 | get_end_location (class line_maps *set, unsigned int idx) |
ba4ad400 DM |
1000 | { |
1001 | if (idx == LINEMAPS_ORDINARY_USED (set) - 1) | |
1002 | return set->highest_location; | |
1003 | ||
1004 | struct line_map *next_map = LINEMAPS_ORDINARY_MAP_AT (set, idx + 1); | |
1005 | return MAP_START_LOCATION (next_map); | |
1006 | } | |
1007 | ||
1008 | /* Helper function for write_digit_row. */ | |
1009 | ||
1010 | static void | |
1011 | write_digit (FILE *stream, int digit) | |
1012 | { | |
1013 | fputc ('0' + (digit % 10), stream); | |
1014 | } | |
1015 | ||
1016 | /* Helper function for dump_location_info. | |
1017 | Write a row of numbers to STREAM, numbering a source line, | |
1018 | giving the units, tens, hundreds etc of the column number. */ | |
1019 | ||
1020 | static void | |
1021 | write_digit_row (FILE *stream, int indent, | |
ebedc9a3 | 1022 | const line_map_ordinary *map, |
620e594b | 1023 | location_t loc, int max_col, int divisor) |
ba4ad400 DM |
1024 | { |
1025 | fprintf (stream, "%*c", indent, ' '); | |
1026 | fprintf (stream, "|"); | |
1027 | for (int column = 1; column < max_col; column++) | |
1028 | { | |
620e594b | 1029 | location_t column_loc = loc + (column << map->m_range_bits); |
ba4ad400 DM |
1030 | write_digit (stream, column_loc / divisor); |
1031 | } | |
1032 | fprintf (stream, "\n"); | |
1033 | } | |
1034 | ||
1035 | /* Write a half-closed (START) / half-open (END) interval of | |
620e594b | 1036 | location_t to STREAM. */ |
ba4ad400 DM |
1037 | |
1038 | static void | |
1039 | dump_location_range (FILE *stream, | |
620e594b | 1040 | location_t start, location_t end) |
ba4ad400 DM |
1041 | { |
1042 | fprintf (stream, | |
620e594b | 1043 | " location_t interval: %u <= loc < %u\n", |
ba4ad400 DM |
1044 | start, end); |
1045 | } | |
1046 | ||
1047 | /* Write a labelled description of a half-closed (START) / half-open (END) | |
620e594b | 1048 | interval of location_t to STREAM. */ |
ba4ad400 DM |
1049 | |
1050 | static void | |
1051 | dump_labelled_location_range (FILE *stream, | |
1052 | const char *name, | |
620e594b | 1053 | location_t start, location_t end) |
ba4ad400 DM |
1054 | { |
1055 | fprintf (stream, "%s\n", name); | |
1056 | dump_location_range (stream, start, end); | |
1057 | fprintf (stream, "\n"); | |
1058 | } | |
1059 | ||
1060 | /* Write a visualization of the locations in the line_table to STREAM. */ | |
1061 | ||
1062 | void | |
1063 | dump_location_info (FILE *stream) | |
1064 | { | |
1065 | /* Visualize the reserved locations. */ | |
1066 | dump_labelled_location_range (stream, "RESERVED LOCATIONS", | |
1067 | 0, RESERVED_LOCATION_COUNT); | |
1068 | ||
1069 | /* Visualize the ordinary line_map instances, rendering the sources. */ | |
1070 | for (unsigned int idx = 0; idx < LINEMAPS_ORDINARY_USED (line_table); idx++) | |
1071 | { | |
620e594b | 1072 | location_t end_location = get_end_location (line_table, idx); |
ba4ad400 DM |
1073 | /* half-closed: doesn't include this one. */ |
1074 | ||
0e50b624 DM |
1075 | const line_map_ordinary *map |
1076 | = LINEMAPS_ORDINARY_MAP_AT (line_table, idx); | |
ba4ad400 DM |
1077 | fprintf (stream, "ORDINARY MAP: %i\n", idx); |
1078 | dump_location_range (stream, | |
1079 | MAP_START_LOCATION (map), end_location); | |
1080 | fprintf (stream, " file: %s\n", ORDINARY_MAP_FILE_NAME (map)); | |
1081 | fprintf (stream, " starting at line: %i\n", | |
1082 | ORDINARY_MAP_STARTING_LINE_NUMBER (map)); | |
ebedc9a3 DM |
1083 | fprintf (stream, " column and range bits: %i\n", |
1084 | map->m_column_and_range_bits); | |
ba4ad400 | 1085 | fprintf (stream, " column bits: %i\n", |
ebedc9a3 DM |
1086 | map->m_column_and_range_bits - map->m_range_bits); |
1087 | fprintf (stream, " range bits: %i\n", | |
1088 | map->m_range_bits); | |
bc65bad2 MG |
1089 | const char * reason; |
1090 | switch (map->reason) { | |
1091 | case LC_ENTER: | |
1092 | reason = "LC_ENTER"; | |
1093 | break; | |
1094 | case LC_LEAVE: | |
1095 | reason = "LC_LEAVE"; | |
1096 | break; | |
1097 | case LC_RENAME: | |
1098 | reason = "LC_RENAME"; | |
1099 | break; | |
1100 | case LC_RENAME_VERBATIM: | |
1101 | reason = "LC_RENAME_VERBATIM"; | |
1102 | break; | |
1103 | case LC_ENTER_MACRO: | |
1104 | reason = "LC_RENAME_MACRO"; | |
1105 | break; | |
1106 | default: | |
1107 | reason = "Unknown"; | |
1108 | } | |
1109 | fprintf (stream, " reason: %d (%s)\n", map->reason, reason); | |
1110 | ||
1111 | const line_map_ordinary *includer_map | |
1112 | = linemap_included_from_linemap (line_table, map); | |
1113 | fprintf (stream, " included from location: %d", | |
1114 | linemap_included_from (map)); | |
1115 | if (includer_map) { | |
1116 | fprintf (stream, " (in ordinary map %d)", | |
1117 | int (includer_map - line_table->info_ordinary.maps)); | |
1118 | } | |
1119 | fprintf (stream, "\n"); | |
ba4ad400 DM |
1120 | |
1121 | /* Render the span of source lines that this "map" covers. */ | |
620e594b | 1122 | for (location_t loc = MAP_START_LOCATION (map); |
ba4ad400 | 1123 | loc < end_location; |
ebedc9a3 | 1124 | loc += (1 << map->m_range_bits) ) |
ba4ad400 | 1125 | { |
ebedc9a3 DM |
1126 | gcc_assert (pure_location_p (line_table, loc) ); |
1127 | ||
ba4ad400 DM |
1128 | expanded_location exploc |
1129 | = linemap_expand_location (line_table, map, loc); | |
1130 | ||
01512446 | 1131 | if (exploc.column == 0) |
ba4ad400 DM |
1132 | { |
1133 | /* Beginning of a new source line: draw the line. */ | |
1134 | ||
7761dfbe DM |
1135 | char_span line_text = location_get_source_line (exploc.file, |
1136 | exploc.line); | |
ba4ad400 DM |
1137 | if (!line_text) |
1138 | break; | |
1139 | fprintf (stream, | |
1140 | "%s:%3i|loc:%5i|%.*s\n", | |
1141 | exploc.file, exploc.line, | |
1142 | loc, | |
7761dfbe | 1143 | (int)line_text.length (), line_text.get_buffer ()); |
ba4ad400 DM |
1144 | |
1145 | /* "loc" is at column 0, which means "the whole line". | |
1146 | Render the locations *within* the line, by underlining | |
620e594b | 1147 | it, showing the location_t numeric values |
ba4ad400 | 1148 | at each column. */ |
7761dfbe DM |
1149 | size_t max_col = (1 << map->m_column_and_range_bits) - 1; |
1150 | if (max_col > line_text.length ()) | |
1151 | max_col = line_text.length () + 1; | |
ba4ad400 | 1152 | |
bc65bad2 MG |
1153 | int len_lnum = num_digits (exploc.line); |
1154 | if (len_lnum < 3) | |
1155 | len_lnum = 3; | |
1156 | int len_loc = num_digits (loc); | |
1157 | if (len_loc < 5) | |
1158 | len_loc = 5; | |
1159 | ||
1160 | int indent = 6 + strlen (exploc.file) + len_lnum + len_loc; | |
ba4ad400 DM |
1161 | |
1162 | /* Thousands. */ | |
1163 | if (end_location > 999) | |
ebedc9a3 | 1164 | write_digit_row (stream, indent, map, loc, max_col, 1000); |
ba4ad400 DM |
1165 | |
1166 | /* Hundreds. */ | |
1167 | if (end_location > 99) | |
ebedc9a3 | 1168 | write_digit_row (stream, indent, map, loc, max_col, 100); |
ba4ad400 DM |
1169 | |
1170 | /* Tens. */ | |
ebedc9a3 | 1171 | write_digit_row (stream, indent, map, loc, max_col, 10); |
ba4ad400 DM |
1172 | |
1173 | /* Units. */ | |
ebedc9a3 | 1174 | write_digit_row (stream, indent, map, loc, max_col, 1); |
ba4ad400 DM |
1175 | } |
1176 | } | |
1177 | fprintf (stream, "\n"); | |
1178 | } | |
1179 | ||
1180 | /* Visualize unallocated values. */ | |
1181 | dump_labelled_location_range (stream, "UNALLOCATED LOCATIONS", | |
1182 | line_table->highest_location, | |
1183 | LINEMAPS_MACRO_LOWEST_LOCATION (line_table)); | |
1184 | ||
1185 | /* Visualize the macro line_map instances, rendering the sources. */ | |
1186 | for (unsigned int i = 0; i < LINEMAPS_MACRO_USED (line_table); i++) | |
1187 | { | |
620e594b | 1188 | /* Each macro map that is allocated owns location_t values |
ba4ad400 DM |
1189 | that are *lower* that the one before them. |
1190 | Hence it's meaningful to view them either in order of ascending | |
1191 | source locations, or in order of ascending macro map index. */ | |
620e594b DM |
1192 | const bool ascending_location_ts = true; |
1193 | unsigned int idx = (ascending_location_ts | |
ba4ad400 DM |
1194 | ? (LINEMAPS_MACRO_USED (line_table) - (i + 1)) |
1195 | : i); | |
0e50b624 | 1196 | const line_map_macro *map = LINEMAPS_MACRO_MAP_AT (line_table, idx); |
ba4ad400 DM |
1197 | fprintf (stream, "MACRO %i: %s (%u tokens)\n", |
1198 | idx, | |
1199 | linemap_map_get_macro_name (map), | |
1200 | MACRO_MAP_NUM_MACRO_TOKENS (map)); | |
1201 | dump_location_range (stream, | |
1202 | map->start_location, | |
1203 | (map->start_location | |
1204 | + MACRO_MAP_NUM_MACRO_TOKENS (map))); | |
1205 | inform (MACRO_MAP_EXPANSION_POINT_LOCATION (map), | |
1206 | "expansion point is location %i", | |
1207 | MACRO_MAP_EXPANSION_POINT_LOCATION (map)); | |
1208 | fprintf (stream, " map->start_location: %u\n", | |
1209 | map->start_location); | |
1210 | ||
1211 | fprintf (stream, " macro_locations:\n"); | |
1212 | for (unsigned int i = 0; i < MACRO_MAP_NUM_MACRO_TOKENS (map); i++) | |
1213 | { | |
620e594b DM |
1214 | location_t x = MACRO_MAP_LOCATIONS (map)[2 * i]; |
1215 | location_t y = MACRO_MAP_LOCATIONS (map)[(2 * i) + 1]; | |
ba4ad400 DM |
1216 | |
1217 | /* linemap_add_macro_token encodes token numbers in an expansion | |
1218 | by putting them after MAP_START_LOCATION. */ | |
1219 | ||
1220 | /* I'm typically seeing 4 uninitialized entries at the end of | |
1221 | 0xafafafaf. | |
1222 | This appears to be due to macro.c:replace_args | |
1223 | adding 2 extra args for padding tokens; presumably there may | |
1224 | be a leading and/or trailing padding token injected, | |
1225 | each for 2 more location slots. | |
620e594b | 1226 | This would explain there being up to 4 location_ts slots |
ba4ad400 DM |
1227 | that may be uninitialized. */ |
1228 | ||
1229 | fprintf (stream, " %u: %u, %u\n", | |
1230 | i, | |
1231 | x, | |
1232 | y); | |
1233 | if (x == y) | |
1234 | { | |
1235 | if (x < MAP_START_LOCATION (map)) | |
a9c697b8 MS |
1236 | inform (x, "token %u has %<x-location == y-location == %u%>", |
1237 | i, x); | |
ba4ad400 DM |
1238 | else |
1239 | fprintf (stream, | |
1240 | "x-location == y-location == %u encodes token # %u\n", | |
1241 | x, x - MAP_START_LOCATION (map)); | |
1242 | } | |
1243 | else | |
1244 | { | |
a9c697b8 MS |
1245 | inform (x, "token %u has %<x-location == %u%>", i, x); |
1246 | inform (x, "token %u has %<y-location == %u%>", i, y); | |
ba4ad400 DM |
1247 | } |
1248 | } | |
1249 | fprintf (stream, "\n"); | |
1250 | } | |
1251 | ||
620e594b | 1252 | /* It appears that MAX_LOCATION_T itself is never assigned to a |
ba4ad400 DM |
1253 | macro map, presumably due to an off-by-one error somewhere |
1254 | between the logic in linemap_enter_macro and | |
1255 | LINEMAPS_MACRO_LOWEST_LOCATION. */ | |
620e594b DM |
1256 | dump_labelled_location_range (stream, "MAX_LOCATION_T", |
1257 | MAX_LOCATION_T, | |
1258 | MAX_LOCATION_T + 1); | |
ba4ad400 DM |
1259 | |
1260 | /* Visualize ad-hoc values. */ | |
1261 | dump_labelled_location_range (stream, "AD-HOC LOCATIONS", | |
620e594b | 1262 | MAX_LOCATION_T + 1, UINT_MAX); |
ba4ad400 | 1263 | } |
d9b950dd | 1264 | |
88fa5555 DM |
1265 | /* string_concat's constructor. */ |
1266 | ||
1267 | string_concat::string_concat (int num, location_t *locs) | |
1268 | : m_num (num) | |
1269 | { | |
1270 | m_locs = ggc_vec_alloc <location_t> (num); | |
1271 | for (int i = 0; i < num; i++) | |
1272 | m_locs[i] = locs[i]; | |
1273 | } | |
1274 | ||
1275 | /* string_concat_db's constructor. */ | |
1276 | ||
1277 | string_concat_db::string_concat_db () | |
1278 | { | |
1279 | m_table = hash_map <location_hash, string_concat *>::create_ggc (64); | |
1280 | } | |
1281 | ||
1282 | /* Record that a string concatenation occurred, covering NUM | |
1283 | string literal tokens. LOCS is an array of size NUM, containing the | |
1284 | locations of the tokens. A copy of LOCS is taken. */ | |
1285 | ||
1286 | void | |
1287 | string_concat_db::record_string_concatenation (int num, location_t *locs) | |
1288 | { | |
1289 | gcc_assert (num > 1); | |
1290 | gcc_assert (locs); | |
1291 | ||
1292 | location_t key_loc = get_key_loc (locs[0]); | |
1293 | ||
1294 | string_concat *concat | |
1295 | = new (ggc_alloc <string_concat> ()) string_concat (num, locs); | |
1296 | m_table->put (key_loc, concat); | |
1297 | } | |
1298 | ||
700d4cb0 | 1299 | /* Determine if LOC was the location of the initial token of a |
88fa5555 DM |
1300 | concatenation of string literal tokens. |
1301 | If so, *OUT_NUM is written to with the number of tokens, and | |
1302 | *OUT_LOCS with the location of an array of locations of the | |
1303 | tokens, and return true. *OUT_LOCS is a borrowed pointer to | |
1304 | storage owned by the string_concat_db. | |
1305 | Otherwise, return false. */ | |
1306 | ||
1307 | bool | |
1308 | string_concat_db::get_string_concatenation (location_t loc, | |
1309 | int *out_num, | |
1310 | location_t **out_locs) | |
1311 | { | |
1312 | gcc_assert (out_num); | |
1313 | gcc_assert (out_locs); | |
1314 | ||
1315 | location_t key_loc = get_key_loc (loc); | |
1316 | ||
1317 | string_concat **concat = m_table->get (key_loc); | |
1318 | if (!concat) | |
1319 | return false; | |
1320 | ||
1321 | *out_num = (*concat)->m_num; | |
1322 | *out_locs =(*concat)->m_locs; | |
1323 | return true; | |
1324 | } | |
1325 | ||
1326 | /* Internal function. Canonicalize LOC into a form suitable for | |
1327 | use as a key within the database, stripping away macro expansion, | |
1328 | ad-hoc information, and range information, using the location of | |
1329 | the start of LOC within an ordinary linemap. */ | |
1330 | ||
1331 | location_t | |
1332 | string_concat_db::get_key_loc (location_t loc) | |
1333 | { | |
1334 | loc = linemap_resolve_location (line_table, loc, LRK_SPELLING_LOCATION, | |
1335 | NULL); | |
1336 | ||
1337 | loc = get_range_from_loc (line_table, loc).m_start; | |
1338 | ||
1339 | return loc; | |
1340 | } | |
1341 | ||
1342 | /* Helper class for use within get_substring_ranges_for_loc. | |
1343 | An vec of cpp_string with responsibility for releasing all of the | |
1344 | str->text for each str in the vector. */ | |
1345 | ||
1346 | class auto_cpp_string_vec : public auto_vec <cpp_string> | |
1347 | { | |
1348 | public: | |
1349 | auto_cpp_string_vec (int alloc) | |
1350 | : auto_vec <cpp_string> (alloc) {} | |
1351 | ||
1352 | ~auto_cpp_string_vec () | |
1353 | { | |
1354 | /* Clean up the copies within this vec. */ | |
1355 | int i; | |
1356 | cpp_string *str; | |
1357 | FOR_EACH_VEC_ELT (*this, i, str) | |
1358 | free (const_cast <unsigned char *> (str->text)); | |
1359 | } | |
1360 | }; | |
1361 | ||
1362 | /* Attempt to populate RANGES with source location information on the | |
1363 | individual characters within the string literal found at STRLOC. | |
1364 | If CONCATS is non-NULL, then any string literals that the token at | |
1365 | STRLOC was concatenated with are also added to RANGES. | |
1366 | ||
1367 | Return NULL if successful, or an error message if any errors occurred (in | |
1368 | which case RANGES may be only partially populated and should not | |
1369 | be used). | |
1370 | ||
1371 | This is implemented by re-parsing the relevant source line(s). */ | |
1372 | ||
1373 | static const char * | |
1374 | get_substring_ranges_for_loc (cpp_reader *pfile, | |
1375 | string_concat_db *concats, | |
1376 | location_t strloc, | |
1377 | enum cpp_ttype type, | |
1378 | cpp_substring_ranges &ranges) | |
1379 | { | |
1380 | gcc_assert (pfile); | |
1381 | ||
1382 | if (strloc == UNKNOWN_LOCATION) | |
1383 | return "unknown location"; | |
1384 | ||
67b5d0b2 DM |
1385 | /* Reparsing the strings requires accurate location information. |
1386 | If -ftrack-macro-expansion has been overridden from its default | |
1387 | of 2, then we might have a location of a macro expansion point, | |
1388 | rather than the location of the literal itself. | |
1389 | Avoid this by requiring that we have full macro expansion tracking | |
1390 | for substring locations to be available. */ | |
1391 | if (cpp_get_options (pfile)->track_macro_expansion != 2) | |
1392 | return "track_macro_expansion != 2"; | |
1393 | ||
94f597df DM |
1394 | /* If #line or # 44 "file"-style directives are present, then there's |
1395 | no guarantee that the line numbers we have can be used to locate | |
1396 | the strings. For example, we might have a .i file with # directives | |
1397 | pointing back to lines within a .c file, but the .c file might | |
1398 | have been edited since the .i file was created. | |
1399 | In such a case, the safest course is to disable on-demand substring | |
1400 | locations. */ | |
1401 | if (line_table->seen_line_directive) | |
1402 | return "seen line directive"; | |
1403 | ||
88fa5555 DM |
1404 | /* If string concatenation has occurred at STRLOC, get the locations |
1405 | of all of the literal tokens making up the compound string. | |
1406 | Otherwise, just use STRLOC. */ | |
1407 | int num_locs = 1; | |
1408 | location_t *strlocs = &strloc; | |
1409 | if (concats) | |
1410 | concats->get_string_concatenation (strloc, &num_locs, &strlocs); | |
1411 | ||
1412 | auto_cpp_string_vec strs (num_locs); | |
1413 | auto_vec <cpp_string_location_reader> loc_readers (num_locs); | |
1414 | for (int i = 0; i < num_locs; i++) | |
1415 | { | |
1416 | /* Get range of strloc. We will use it to locate the start and finish | |
1417 | of the literal token within the line. */ | |
1418 | source_range src_range = get_range_from_loc (line_table, strlocs[i]); | |
1419 | ||
1420 | if (src_range.m_start >= LINEMAPS_MACRO_LOWEST_LOCATION (line_table)) | |
0d48e877 DM |
1421 | { |
1422 | /* If the string token was within a macro expansion, then we can | |
1423 | cope with it for the simple case where we have a single token. | |
1424 | Otherwise, bail out. */ | |
1425 | if (src_range.m_start != src_range.m_finish) | |
1426 | return "macro expansion"; | |
1427 | } | |
1428 | else | |
1429 | { | |
1430 | if (src_range.m_start >= LINE_MAP_MAX_LOCATION_WITH_COLS) | |
1431 | /* If so, we can't reliably determine where the token started within | |
1432 | its line. */ | |
1433 | return "range starts after LINE_MAP_MAX_LOCATION_WITH_COLS"; | |
1434 | ||
1435 | if (src_range.m_finish >= LINE_MAP_MAX_LOCATION_WITH_COLS) | |
1436 | /* If so, we can't reliably determine where the token finished | |
1437 | within its line. */ | |
1438 | return "range ends after LINE_MAP_MAX_LOCATION_WITH_COLS"; | |
1439 | } | |
88fa5555 DM |
1440 | |
1441 | expanded_location start | |
0d48e877 DM |
1442 | = expand_location_to_spelling_point (src_range.m_start, |
1443 | LOCATION_ASPECT_START); | |
88fa5555 | 1444 | expanded_location finish |
0d48e877 DM |
1445 | = expand_location_to_spelling_point (src_range.m_finish, |
1446 | LOCATION_ASPECT_FINISH); | |
88fa5555 DM |
1447 | if (start.file != finish.file) |
1448 | return "range endpoints are in different files"; | |
1449 | if (start.line != finish.line) | |
1450 | return "range endpoints are on different lines"; | |
1451 | if (start.column > finish.column) | |
1452 | return "range endpoints are reversed"; | |
1453 | ||
7761dfbe DM |
1454 | char_span line = location_get_source_line (start.file, start.line); |
1455 | if (!line) | |
88fa5555 DM |
1456 | return "unable to read source line"; |
1457 | ||
1458 | /* Determine the location of the literal (including quotes | |
1459 | and leading prefix chars, such as the 'u' in a u"" | |
1460 | token). */ | |
7761dfbe | 1461 | size_t literal_length = finish.column - start.column + 1; |
88fa5555 | 1462 | |
7cfa044d | 1463 | /* Ensure that we don't crash if we got the wrong location. */ |
7761dfbe | 1464 | if (line.length () < (start.column - 1 + literal_length)) |
7cfa044d DM |
1465 | return "line is not wide enough"; |
1466 | ||
7761dfbe DM |
1467 | char_span literal = line.subspan (start.column - 1, literal_length); |
1468 | ||
88fa5555 DM |
1469 | cpp_string from; |
1470 | from.len = literal_length; | |
1471 | /* Make a copy of the literal, to avoid having to rely on | |
1472 | the lifetime of the copy of the line within the cache. | |
1473 | This will be released by the auto_cpp_string_vec dtor. */ | |
7761dfbe | 1474 | from.text = (unsigned char *)literal.xstrdup (); |
88fa5555 DM |
1475 | strs.safe_push (from); |
1476 | ||
1477 | /* For very long lines, a new linemap could have started | |
1478 | halfway through the token. | |
1479 | Ensure that the loc_reader uses the linemap of the | |
1480 | *end* of the token for its start location. */ | |
05d57d65 DM |
1481 | const line_map_ordinary *start_ord_map; |
1482 | linemap_resolve_location (line_table, src_range.m_start, | |
1483 | LRK_SPELLING_LOCATION, &start_ord_map); | |
88fa5555 DM |
1484 | const line_map_ordinary *final_ord_map; |
1485 | linemap_resolve_location (line_table, src_range.m_finish, | |
05d57d65 | 1486 | LRK_SPELLING_LOCATION, &final_ord_map); |
3d0a5393 DM |
1487 | if (start_ord_map == NULL || final_ord_map == NULL) |
1488 | return "failed to get ordinary maps"; | |
05d57d65 DM |
1489 | /* Bulletproofing. We ought to only have different ordinary maps |
1490 | for start vs finish due to line-length jumps. */ | |
1491 | if (start_ord_map != final_ord_map | |
1492 | && start_ord_map->to_file != final_ord_map->to_file) | |
ef33afeb DM |
1493 | return "start and finish are spelled in different ordinary maps"; |
1494 | /* The file from linemap_resolve_location ought to match that from | |
1495 | expand_location_to_spelling_point. */ | |
1496 | if (start_ord_map->to_file != start.file) | |
1497 | return "mismatching file after resolving linemap"; | |
1498 | ||
88fa5555 DM |
1499 | location_t start_loc |
1500 | = linemap_position_for_line_and_column (line_table, final_ord_map, | |
1501 | start.line, start.column); | |
1502 | ||
1503 | cpp_string_location_reader loc_reader (start_loc, line_table); | |
1504 | loc_readers.safe_push (loc_reader); | |
1505 | } | |
1506 | ||
1507 | /* Rerun cpp_interpret_string, or rather, a modified version of it. */ | |
1508 | const char *err = cpp_interpret_string_ranges (pfile, strs.address (), | |
1509 | loc_readers.address (), | |
1510 | num_locs, &ranges, type); | |
1511 | if (err) | |
1512 | return err; | |
1513 | ||
1514 | /* Success: "ranges" should now contain information on the string. */ | |
1515 | return NULL; | |
1516 | } | |
1517 | ||
65e736c0 DM |
1518 | /* Attempt to populate *OUT_LOC with source location information on the |
1519 | given characters within the string literal found at STRLOC. | |
1520 | CARET_IDX, START_IDX, and END_IDX refer to offsets within the execution | |
1521 | character set. | |
1522 | ||
1523 | For example, given CARET_IDX = 4, START_IDX = 3, END_IDX = 7 | |
1524 | and string literal "012345\n789" | |
1525 | *OUT_LOC is written to with: | |
1526 | "012345\n789" | |
1527 | ~^~~~~ | |
1528 | ||
88fa5555 DM |
1529 | If CONCATS is non-NULL, then any string literals that the token at |
1530 | STRLOC was concatenated with are also considered. | |
1531 | ||
1532 | This is implemented by re-parsing the relevant source line(s). | |
1533 | ||
1534 | Return NULL if successful, or an error message if any errors occurred. | |
1535 | Error messages are intended for GCC developers (to help debugging) rather | |
1536 | than for end-users. */ | |
1537 | ||
1538 | const char * | |
620e594b DM |
1539 | get_location_within_string (cpp_reader *pfile, |
1540 | string_concat_db *concats, | |
1541 | location_t strloc, | |
1542 | enum cpp_ttype type, | |
1543 | int caret_idx, int start_idx, int end_idx, | |
1544 | location_t *out_loc) | |
65e736c0 DM |
1545 | { |
1546 | gcc_checking_assert (caret_idx >= 0); | |
88fa5555 DM |
1547 | gcc_checking_assert (start_idx >= 0); |
1548 | gcc_checking_assert (end_idx >= 0); | |
65e736c0 | 1549 | gcc_assert (out_loc); |
88fa5555 DM |
1550 | |
1551 | cpp_substring_ranges ranges; | |
1552 | const char *err | |
1553 | = get_substring_ranges_for_loc (pfile, concats, strloc, type, ranges); | |
1554 | if (err) | |
1555 | return err; | |
1556 | ||
65e736c0 DM |
1557 | if (caret_idx >= ranges.get_num_ranges ()) |
1558 | return "caret_idx out of range"; | |
88fa5555 DM |
1559 | if (start_idx >= ranges.get_num_ranges ()) |
1560 | return "start_idx out of range"; | |
1561 | if (end_idx >= ranges.get_num_ranges ()) | |
1562 | return "end_idx out of range"; | |
1563 | ||
65e736c0 DM |
1564 | *out_loc = make_location (ranges.get_range (caret_idx).m_start, |
1565 | ranges.get_range (start_idx).m_start, | |
1566 | ranges.get_range (end_idx).m_finish); | |
1567 | return NULL; | |
1568 | } | |
1569 | ||
0e06d2b3 DM |
1570 | #if CHECKING_P |
1571 | ||
1572 | namespace selftest { | |
1573 | ||
1574 | /* Selftests of location handling. */ | |
1575 | ||
65e736c0 DM |
1576 | /* Attempt to populate *OUT_RANGE with source location information on the |
1577 | given character within the string literal found at STRLOC. | |
1578 | CHAR_IDX refers to an offset within the execution character set. | |
1579 | If CONCATS is non-NULL, then any string literals that the token at | |
1580 | STRLOC was concatenated with are also considered. | |
1581 | ||
1582 | This is implemented by re-parsing the relevant source line(s). | |
1583 | ||
1584 | Return NULL if successful, or an error message if any errors occurred. | |
1585 | Error messages are intended for GCC developers (to help debugging) rather | |
1586 | than for end-users. */ | |
1587 | ||
1588 | static const char * | |
1589 | get_source_range_for_char (cpp_reader *pfile, | |
1590 | string_concat_db *concats, | |
1591 | location_t strloc, | |
1592 | enum cpp_ttype type, | |
1593 | int char_idx, | |
1594 | source_range *out_range) | |
1595 | { | |
1596 | gcc_checking_assert (char_idx >= 0); | |
1597 | gcc_assert (out_range); | |
1598 | ||
1599 | cpp_substring_ranges ranges; | |
1600 | const char *err | |
1601 | = get_substring_ranges_for_loc (pfile, concats, strloc, type, ranges); | |
1602 | if (err) | |
1603 | return err; | |
1604 | ||
1605 | if (char_idx >= ranges.get_num_ranges ()) | |
1606 | return "char_idx out of range"; | |
1607 | ||
1608 | *out_range = ranges.get_range (char_idx); | |
88fa5555 DM |
1609 | return NULL; |
1610 | } | |
1611 | ||
65e736c0 | 1612 | /* As get_source_range_for_char, but write to *OUT the number |
88fa5555 DM |
1613 | of ranges that are available. */ |
1614 | ||
0e06d2b3 | 1615 | static const char * |
88fa5555 DM |
1616 | get_num_source_ranges_for_substring (cpp_reader *pfile, |
1617 | string_concat_db *concats, | |
1618 | location_t strloc, | |
1619 | enum cpp_ttype type, | |
1620 | int *out) | |
1621 | { | |
1622 | gcc_assert (out); | |
1623 | ||
1624 | cpp_substring_ranges ranges; | |
1625 | const char *err | |
1626 | = get_substring_ranges_for_loc (pfile, concats, strloc, type, ranges); | |
1627 | ||
1628 | if (err) | |
1629 | return err; | |
1630 | ||
1631 | *out = ranges.get_num_ranges (); | |
1632 | return NULL; | |
1633 | } | |
1634 | ||
d9b950dd DM |
1635 | /* Selftests of location handling. */ |
1636 | ||
082284da DM |
1637 | /* Verify that compare() on linenum_type handles comparisons over the full |
1638 | range of the type. */ | |
1639 | ||
1640 | static void | |
1641 | test_linenum_comparisons () | |
1642 | { | |
1643 | linenum_type min_line (0); | |
1644 | linenum_type max_line (0xffffffff); | |
1645 | ASSERT_EQ (0, compare (min_line, min_line)); | |
1646 | ASSERT_EQ (0, compare (max_line, max_line)); | |
1647 | ||
1648 | ASSERT_GT (compare (max_line, min_line), 0); | |
1649 | ASSERT_LT (compare (min_line, max_line), 0); | |
1650 | } | |
1651 | ||
741d3be5 DM |
1652 | /* Helper function for verifying location data: when location_t |
1653 | values are > LINE_MAP_MAX_LOCATION_WITH_COLS, they are treated | |
1654 | as having column 0. */ | |
1655 | ||
1656 | static bool | |
1657 | should_have_column_data_p (location_t loc) | |
1658 | { | |
1659 | if (IS_ADHOC_LOC (loc)) | |
1660 | loc = get_location_from_adhoc_loc (line_table, loc); | |
1661 | if (loc > LINE_MAP_MAX_LOCATION_WITH_COLS) | |
1662 | return false; | |
1663 | return true; | |
1664 | } | |
1665 | ||
1666 | /* Selftest for should_have_column_data_p. */ | |
1667 | ||
1668 | static void | |
1669 | test_should_have_column_data_p () | |
1670 | { | |
1671 | ASSERT_TRUE (should_have_column_data_p (RESERVED_LOCATION_COUNT)); | |
1672 | ASSERT_TRUE | |
1673 | (should_have_column_data_p (LINE_MAP_MAX_LOCATION_WITH_COLS)); | |
1674 | ASSERT_FALSE | |
1675 | (should_have_column_data_p (LINE_MAP_MAX_LOCATION_WITH_COLS + 1)); | |
1676 | } | |
1677 | ||
d9b950dd DM |
1678 | /* Verify the result of LOCATION_FILE/LOCATION_LINE/LOCATION_COLUMN |
1679 | on LOC. */ | |
1680 | ||
1681 | static void | |
1682 | assert_loceq (const char *exp_filename, int exp_linenum, int exp_colnum, | |
1683 | location_t loc) | |
1684 | { | |
1685 | ASSERT_STREQ (exp_filename, LOCATION_FILE (loc)); | |
1686 | ASSERT_EQ (exp_linenum, LOCATION_LINE (loc)); | |
741d3be5 DM |
1687 | /* If location_t values are sufficiently high, then column numbers |
1688 | will be unavailable and LOCATION_COLUMN (loc) will be 0. | |
1689 | When close to the threshold, column numbers *may* be present: if | |
1690 | the final linemap before the threshold contains a line that straddles | |
1691 | the threshold, locations in that line have column information. */ | |
1692 | if (should_have_column_data_p (loc)) | |
1693 | ASSERT_EQ (exp_colnum, LOCATION_COLUMN (loc)); | |
1694 | } | |
1695 | ||
f87e22c5 DM |
1696 | /* Various selftests involve constructing a line table and one or more |
1697 | line maps within it. | |
741d3be5 DM |
1698 | |
1699 | For maximum test coverage we want to run these tests with a variety | |
1700 | of situations: | |
1701 | - line_table->default_range_bits: some frontends use a non-zero value | |
1702 | and others use zero | |
1703 | - the fallback modes within line-map.c: there are various threshold | |
620e594b | 1704 | values for location_t beyond line-map.c changes |
741d3be5 DM |
1705 | behavior (disabling of the range-packing optimization, disabling |
1706 | of column-tracking). We can exercise these by starting the line_table | |
1707 | at interesting values at or near these thresholds. | |
1708 | ||
1709 | The following struct describes a particular case within our test | |
1710 | matrix. */ | |
1711 | ||
6c1dae73 | 1712 | class line_table_case |
741d3be5 | 1713 | { |
6c1dae73 | 1714 | public: |
741d3be5 DM |
1715 | line_table_case (int default_range_bits, int base_location) |
1716 | : m_default_range_bits (default_range_bits), | |
1717 | m_base_location (base_location) | |
1718 | {} | |
1719 | ||
1720 | int m_default_range_bits; | |
1721 | int m_base_location; | |
1722 | }; | |
1723 | ||
f87e22c5 DM |
1724 | /* Constructor. Store the old value of line_table, and create a new |
1725 | one, using sane defaults. */ | |
741d3be5 | 1726 | |
f87e22c5 | 1727 | line_table_test::line_table_test () |
741d3be5 | 1728 | { |
f87e22c5 DM |
1729 | gcc_assert (saved_line_table == NULL); |
1730 | saved_line_table = line_table; | |
1731 | line_table = ggc_alloc<line_maps> (); | |
1732 | linemap_init (line_table, BUILTINS_LOCATION); | |
1733 | gcc_assert (saved_line_table->reallocator); | |
1734 | line_table->reallocator = saved_line_table->reallocator; | |
1735 | gcc_assert (saved_line_table->round_alloc_size); | |
1736 | line_table->round_alloc_size = saved_line_table->round_alloc_size; | |
1737 | line_table->default_range_bits = 0; | |
1738 | } | |
741d3be5 DM |
1739 | |
1740 | /* Constructor. Store the old value of line_table, and create a new | |
1741 | one, using the sitation described in CASE_. */ | |
1742 | ||
f87e22c5 | 1743 | line_table_test::line_table_test (const line_table_case &case_) |
741d3be5 | 1744 | { |
f87e22c5 DM |
1745 | gcc_assert (saved_line_table == NULL); |
1746 | saved_line_table = line_table; | |
741d3be5 DM |
1747 | line_table = ggc_alloc<line_maps> (); |
1748 | linemap_init (line_table, BUILTINS_LOCATION); | |
f87e22c5 DM |
1749 | gcc_assert (saved_line_table->reallocator); |
1750 | line_table->reallocator = saved_line_table->reallocator; | |
1751 | gcc_assert (saved_line_table->round_alloc_size); | |
1752 | line_table->round_alloc_size = saved_line_table->round_alloc_size; | |
741d3be5 DM |
1753 | line_table->default_range_bits = case_.m_default_range_bits; |
1754 | if (case_.m_base_location) | |
1755 | { | |
1756 | line_table->highest_location = case_.m_base_location; | |
1757 | line_table->highest_line = case_.m_base_location; | |
1758 | } | |
1759 | } | |
1760 | ||
1761 | /* Destructor. Restore the old value of line_table. */ | |
1762 | ||
f87e22c5 | 1763 | line_table_test::~line_table_test () |
741d3be5 | 1764 | { |
f87e22c5 DM |
1765 | gcc_assert (saved_line_table != NULL); |
1766 | line_table = saved_line_table; | |
1767 | saved_line_table = NULL; | |
d9b950dd DM |
1768 | } |
1769 | ||
1770 | /* Verify basic operation of ordinary linemaps. */ | |
1771 | ||
1772 | static void | |
741d3be5 | 1773 | test_accessing_ordinary_linemaps (const line_table_case &case_) |
d9b950dd | 1774 | { |
f87e22c5 | 1775 | line_table_test ltt (case_); |
741d3be5 | 1776 | |
d9b950dd DM |
1777 | /* Build a simple linemap describing some locations. */ |
1778 | linemap_add (line_table, LC_ENTER, false, "foo.c", 0); | |
1779 | ||
1780 | linemap_line_start (line_table, 1, 100); | |
1781 | location_t loc_a = linemap_position_for_column (line_table, 1); | |
1782 | location_t loc_b = linemap_position_for_column (line_table, 23); | |
1783 | ||
1784 | linemap_line_start (line_table, 2, 100); | |
1785 | location_t loc_c = linemap_position_for_column (line_table, 1); | |
1786 | location_t loc_d = linemap_position_for_column (line_table, 17); | |
1787 | ||
1788 | /* Example of a very long line. */ | |
1789 | linemap_line_start (line_table, 3, 2000); | |
1790 | location_t loc_e = linemap_position_for_column (line_table, 700); | |
1791 | ||
5ccf1d8d DM |
1792 | /* Transitioning back to a short line. */ |
1793 | linemap_line_start (line_table, 4, 0); | |
1794 | location_t loc_back_to_short = linemap_position_for_column (line_table, 100); | |
1795 | ||
1796 | if (should_have_column_data_p (loc_back_to_short)) | |
1797 | { | |
1798 | /* Verify that we switched to short lines in the linemap. */ | |
1799 | line_map_ordinary *map = LINEMAPS_LAST_ORDINARY_MAP (line_table); | |
1800 | ASSERT_EQ (7, map->m_column_and_range_bits - map->m_range_bits); | |
1801 | } | |
1802 | ||
b9f4757f DM |
1803 | /* Example of a line that will eventually be seen to be longer |
1804 | than LINE_MAP_MAX_COLUMN_NUMBER; the initially seen width is | |
1805 | below that. */ | |
1806 | linemap_line_start (line_table, 5, 2000); | |
1807 | ||
1808 | location_t loc_start_of_very_long_line | |
1809 | = linemap_position_for_column (line_table, 2000); | |
1810 | location_t loc_too_wide | |
1811 | = linemap_position_for_column (line_table, 4097); | |
1812 | location_t loc_too_wide_2 | |
1813 | = linemap_position_for_column (line_table, 4098); | |
1814 | ||
1815 | /* ...and back to a sane line length. */ | |
1816 | linemap_line_start (line_table, 6, 100); | |
1817 | location_t loc_sane_again = linemap_position_for_column (line_table, 10); | |
1818 | ||
d9b950dd DM |
1819 | linemap_add (line_table, LC_LEAVE, false, NULL, 0); |
1820 | ||
1821 | /* Multiple files. */ | |
1822 | linemap_add (line_table, LC_ENTER, false, "bar.c", 0); | |
1823 | linemap_line_start (line_table, 1, 200); | |
1824 | location_t loc_f = linemap_position_for_column (line_table, 150); | |
1825 | linemap_add (line_table, LC_LEAVE, false, NULL, 0); | |
1826 | ||
1827 | /* Verify that we can recover the location info. */ | |
1828 | assert_loceq ("foo.c", 1, 1, loc_a); | |
1829 | assert_loceq ("foo.c", 1, 23, loc_b); | |
1830 | assert_loceq ("foo.c", 2, 1, loc_c); | |
1831 | assert_loceq ("foo.c", 2, 17, loc_d); | |
1832 | assert_loceq ("foo.c", 3, 700, loc_e); | |
5ccf1d8d | 1833 | assert_loceq ("foo.c", 4, 100, loc_back_to_short); |
b9f4757f DM |
1834 | |
1835 | /* In the very wide line, the initial location should be fully tracked. */ | |
1836 | assert_loceq ("foo.c", 5, 2000, loc_start_of_very_long_line); | |
1837 | /* ...but once we exceed LINE_MAP_MAX_COLUMN_NUMBER column-tracking should | |
1838 | be disabled. */ | |
1839 | assert_loceq ("foo.c", 5, 0, loc_too_wide); | |
1840 | assert_loceq ("foo.c", 5, 0, loc_too_wide_2); | |
1841 | /*...and column-tracking should be re-enabled for subsequent lines. */ | |
1842 | assert_loceq ("foo.c", 6, 10, loc_sane_again); | |
1843 | ||
d9b950dd DM |
1844 | assert_loceq ("bar.c", 1, 150, loc_f); |
1845 | ||
1846 | ASSERT_FALSE (is_location_from_builtin_token (loc_a)); | |
a01fc549 DM |
1847 | ASSERT_TRUE (pure_location_p (line_table, loc_a)); |
1848 | ||
1849 | /* Verify using make_location to build a range, and extracting data | |
1850 | back from it. */ | |
1851 | location_t range_c_b_d = make_location (loc_c, loc_b, loc_d); | |
1852 | ASSERT_FALSE (pure_location_p (line_table, range_c_b_d)); | |
1853 | ASSERT_EQ (loc_c, get_location_from_adhoc_loc (line_table, range_c_b_d)); | |
1854 | source_range src_range = get_range_from_loc (line_table, range_c_b_d); | |
1855 | ASSERT_EQ (loc_b, src_range.m_start); | |
1856 | ASSERT_EQ (loc_d, src_range.m_finish); | |
d9b950dd DM |
1857 | } |
1858 | ||
1859 | /* Verify various properties of UNKNOWN_LOCATION. */ | |
1860 | ||
1861 | static void | |
1862 | test_unknown_location () | |
1863 | { | |
1864 | ASSERT_EQ (NULL, LOCATION_FILE (UNKNOWN_LOCATION)); | |
1865 | ASSERT_EQ (0, LOCATION_LINE (UNKNOWN_LOCATION)); | |
1866 | ASSERT_EQ (0, LOCATION_COLUMN (UNKNOWN_LOCATION)); | |
1867 | } | |
1868 | ||
1869 | /* Verify various properties of BUILTINS_LOCATION. */ | |
1870 | ||
1871 | static void | |
1872 | test_builtins () | |
1873 | { | |
10d2fc23 | 1874 | assert_loceq (_("<built-in>"), 0, 0, BUILTINS_LOCATION); |
d9b950dd DM |
1875 | ASSERT_PRED1 (is_location_from_builtin_token, BUILTINS_LOCATION); |
1876 | } | |
1877 | ||
9144eabb | 1878 | /* Regression test for make_location. |
cfa435e1 DM |
1879 | Ensure that we use pure locations for the start/finish of the range, |
1880 | rather than storing a packed or ad-hoc range as the start/finish. */ | |
9144eabb DM |
1881 | |
1882 | static void | |
1883 | test_make_location_nonpure_range_endpoints (const line_table_case &case_) | |
1884 | { | |
1885 | /* Issue seen with testsuite/c-c++-common/Wlogical-not-parentheses-2.c | |
1886 | with C++ frontend. | |
1887 | ....................0000000001111111111222. | |
1888 | ....................1234567890123456789012. */ | |
1889 | const char *content = " r += !aaa == bbb;\n"; | |
1890 | temp_source_file tmp (SELFTEST_LOCATION, ".C", content); | |
1891 | line_table_test ltt (case_); | |
1892 | linemap_add (line_table, LC_ENTER, false, tmp.get_filename (), 1); | |
1893 | ||
1894 | const location_t c11 = linemap_position_for_column (line_table, 11); | |
1895 | const location_t c12 = linemap_position_for_column (line_table, 12); | |
1896 | const location_t c13 = linemap_position_for_column (line_table, 13); | |
1897 | const location_t c14 = linemap_position_for_column (line_table, 14); | |
1898 | const location_t c21 = linemap_position_for_column (line_table, 21); | |
1899 | ||
1900 | if (c21 > LINE_MAP_MAX_LOCATION_WITH_COLS) | |
1901 | return; | |
1902 | ||
1903 | /* Use column 13 for the caret location, arbitrarily, to verify that we | |
1904 | handle start != caret. */ | |
1905 | const location_t aaa = make_location (c13, c12, c14); | |
1906 | ASSERT_EQ (c13, get_pure_location (aaa)); | |
1907 | ASSERT_EQ (c12, get_start (aaa)); | |
1908 | ASSERT_FALSE (IS_ADHOC_LOC (get_start (aaa))); | |
1909 | ASSERT_EQ (c14, get_finish (aaa)); | |
1910 | ASSERT_FALSE (IS_ADHOC_LOC (get_finish (aaa))); | |
1911 | ||
1912 | /* Make a location using a location with a range as the start-point. */ | |
1913 | const location_t not_aaa = make_location (c11, aaa, c14); | |
1914 | ASSERT_EQ (c11, get_pure_location (not_aaa)); | |
1915 | /* It should use the start location of the range, not store the range | |
1916 | itself. */ | |
1917 | ASSERT_EQ (c12, get_start (not_aaa)); | |
1918 | ASSERT_FALSE (IS_ADHOC_LOC (get_start (not_aaa))); | |
1919 | ASSERT_EQ (c14, get_finish (not_aaa)); | |
1920 | ASSERT_FALSE (IS_ADHOC_LOC (get_finish (not_aaa))); | |
1921 | ||
1922 | /* Similarly, make a location with a range as the end-point. */ | |
1923 | const location_t aaa_eq_bbb = make_location (c12, c12, c21); | |
1924 | ASSERT_EQ (c12, get_pure_location (aaa_eq_bbb)); | |
1925 | ASSERT_EQ (c12, get_start (aaa_eq_bbb)); | |
1926 | ASSERT_FALSE (IS_ADHOC_LOC (get_start (aaa_eq_bbb))); | |
1927 | ASSERT_EQ (c21, get_finish (aaa_eq_bbb)); | |
1928 | ASSERT_FALSE (IS_ADHOC_LOC (get_finish (aaa_eq_bbb))); | |
1929 | const location_t not_aaa_eq_bbb = make_location (c11, c12, aaa_eq_bbb); | |
1930 | /* It should use the finish location of the range, not store the range | |
1931 | itself. */ | |
1932 | ASSERT_EQ (c11, get_pure_location (not_aaa_eq_bbb)); | |
1933 | ASSERT_EQ (c12, get_start (not_aaa_eq_bbb)); | |
1934 | ASSERT_FALSE (IS_ADHOC_LOC (get_start (not_aaa_eq_bbb))); | |
1935 | ASSERT_EQ (c21, get_finish (not_aaa_eq_bbb)); | |
1936 | ASSERT_FALSE (IS_ADHOC_LOC (get_finish (not_aaa_eq_bbb))); | |
1937 | } | |
1938 | ||
d9b950dd DM |
1939 | /* Verify reading of input files (e.g. for caret-based diagnostics). */ |
1940 | ||
1941 | static void | |
1942 | test_reading_source_line () | |
1943 | { | |
85ecd05c | 1944 | /* Create a tempfile and write some text to it. */ |
741d3be5 DM |
1945 | temp_source_file tmp (SELFTEST_LOCATION, ".txt", |
1946 | "01234567890123456789\n" | |
1947 | "This is the test text\n" | |
1adae327 | 1948 | "This is the 3rd line"); |
85ecd05c DM |
1949 | |
1950 | /* Read back a specific line from the tempfile. */ | |
7761dfbe DM |
1951 | char_span source_line = location_get_source_line (tmp.get_filename (), 3); |
1952 | ASSERT_TRUE (source_line); | |
1953 | ASSERT_TRUE (source_line.get_buffer () != NULL); | |
1954 | ASSERT_EQ (20, source_line.length ()); | |
1adae327 | 1955 | ASSERT_TRUE (!strncmp ("This is the 3rd line", |
7761dfbe | 1956 | source_line.get_buffer (), source_line.length ())); |
1adae327 | 1957 | |
7761dfbe DM |
1958 | source_line = location_get_source_line (tmp.get_filename (), 2); |
1959 | ASSERT_TRUE (source_line); | |
1960 | ASSERT_TRUE (source_line.get_buffer () != NULL); | |
1961 | ASSERT_EQ (21, source_line.length ()); | |
1adae327 | 1962 | ASSERT_TRUE (!strncmp ("This is the test text", |
7761dfbe | 1963 | source_line.get_buffer (), source_line.length ())); |
85ecd05c | 1964 | |
7761dfbe DM |
1965 | source_line = location_get_source_line (tmp.get_filename (), 4); |
1966 | ASSERT_FALSE (source_line); | |
1967 | ASSERT_TRUE (source_line.get_buffer () == NULL); | |
d9b950dd DM |
1968 | } |
1969 | ||
741d3be5 DM |
1970 | /* Tests of lexing. */ |
1971 | ||
1972 | /* Verify that token TOK from PARSER has cpp_token_as_text | |
1973 | equal to EXPECTED_TEXT. */ | |
1974 | ||
1975 | #define ASSERT_TOKEN_AS_TEXT_EQ(PARSER, TOK, EXPECTED_TEXT) \ | |
1976 | SELFTEST_BEGIN_STMT \ | |
1977 | unsigned char *actual_txt = cpp_token_as_text ((PARSER), (TOK)); \ | |
1978 | ASSERT_STREQ ((EXPECTED_TEXT), (const char *)actual_txt); \ | |
1979 | SELFTEST_END_STMT | |
1980 | ||
1981 | /* Verify that TOK's src_loc is within EXP_FILENAME at EXP_LINENUM, | |
1982 | and ranges from EXP_START_COL to EXP_FINISH_COL. | |
1983 | Use LOC as the effective location of the selftest. */ | |
1984 | ||
1985 | static void | |
1986 | assert_token_loc_eq (const location &loc, | |
1987 | const cpp_token *tok, | |
1988 | const char *exp_filename, int exp_linenum, | |
1989 | int exp_start_col, int exp_finish_col) | |
1990 | { | |
1991 | location_t tok_loc = tok->src_loc; | |
1992 | ASSERT_STREQ_AT (loc, exp_filename, LOCATION_FILE (tok_loc)); | |
1993 | ASSERT_EQ_AT (loc, exp_linenum, LOCATION_LINE (tok_loc)); | |
1994 | ||
1995 | /* If location_t values are sufficiently high, then column numbers | |
1996 | will be unavailable. */ | |
1997 | if (!should_have_column_data_p (tok_loc)) | |
1998 | return; | |
1999 | ||
2000 | ASSERT_EQ_AT (loc, exp_start_col, LOCATION_COLUMN (tok_loc)); | |
2001 | source_range tok_range = get_range_from_loc (line_table, tok_loc); | |
2002 | ASSERT_EQ_AT (loc, exp_start_col, LOCATION_COLUMN (tok_range.m_start)); | |
2003 | ASSERT_EQ_AT (loc, exp_finish_col, LOCATION_COLUMN (tok_range.m_finish)); | |
2004 | } | |
2005 | ||
2006 | /* Use assert_token_loc_eq to verify the TOK->src_loc, using | |
2007 | SELFTEST_LOCATION as the effective location of the selftest. */ | |
2008 | ||
2009 | #define ASSERT_TOKEN_LOC_EQ(TOK, EXP_FILENAME, EXP_LINENUM, \ | |
2010 | EXP_START_COL, EXP_FINISH_COL) \ | |
2011 | assert_token_loc_eq (SELFTEST_LOCATION, (TOK), (EXP_FILENAME), \ | |
2012 | (EXP_LINENUM), (EXP_START_COL), (EXP_FINISH_COL)) | |
2013 | ||
2014 | /* Test of lexing a file using libcpp, verifying tokens and their | |
2015 | location information. */ | |
2016 | ||
2017 | static void | |
2018 | test_lexer (const line_table_case &case_) | |
2019 | { | |
2020 | /* Create a tempfile and write some text to it. */ | |
2021 | const char *content = | |
2022 | /*00000000011111111112222222222333333.3333444444444.455555555556 | |
2023 | 12345678901234567890123456789012345.6789012345678.901234567890. */ | |
2024 | ("test_name /* c-style comment */\n" | |
2025 | " \"test literal\"\n" | |
2026 | " // test c++-style comment\n" | |
2027 | " 42\n"); | |
2028 | temp_source_file tmp (SELFTEST_LOCATION, ".txt", content); | |
2029 | ||
f87e22c5 | 2030 | line_table_test ltt (case_); |
741d3be5 DM |
2031 | |
2032 | cpp_reader *parser = cpp_create_reader (CLK_GNUC89, NULL, line_table); | |
2033 | ||
2034 | const char *fname = cpp_read_main_file (parser, tmp.get_filename ()); | |
2035 | ASSERT_NE (fname, NULL); | |
2036 | ||
2037 | /* Verify that we get the expected tokens back, with the correct | |
2038 | location information. */ | |
2039 | ||
2040 | location_t loc; | |
2041 | const cpp_token *tok; | |
2042 | tok = cpp_get_token_with_location (parser, &loc); | |
2043 | ASSERT_NE (tok, NULL); | |
2044 | ASSERT_EQ (tok->type, CPP_NAME); | |
2045 | ASSERT_TOKEN_AS_TEXT_EQ (parser, tok, "test_name"); | |
2046 | ASSERT_TOKEN_LOC_EQ (tok, tmp.get_filename (), 1, 1, 9); | |
2047 | ||
2048 | tok = cpp_get_token_with_location (parser, &loc); | |
2049 | ASSERT_NE (tok, NULL); | |
2050 | ASSERT_EQ (tok->type, CPP_STRING); | |
2051 | ASSERT_TOKEN_AS_TEXT_EQ (parser, tok, "\"test literal\""); | |
2052 | ASSERT_TOKEN_LOC_EQ (tok, tmp.get_filename (), 2, 35, 48); | |
2053 | ||
2054 | tok = cpp_get_token_with_location (parser, &loc); | |
2055 | ASSERT_NE (tok, NULL); | |
2056 | ASSERT_EQ (tok->type, CPP_NUMBER); | |
2057 | ASSERT_TOKEN_AS_TEXT_EQ (parser, tok, "42"); | |
2058 | ASSERT_TOKEN_LOC_EQ (tok, tmp.get_filename (), 4, 4, 5); | |
2059 | ||
2060 | tok = cpp_get_token_with_location (parser, &loc); | |
2061 | ASSERT_NE (tok, NULL); | |
2062 | ASSERT_EQ (tok->type, CPP_EOF); | |
2063 | ||
2064 | cpp_finish (parser, NULL); | |
2065 | cpp_destroy (parser); | |
2066 | } | |
2067 | ||
88fa5555 DM |
2068 | /* Forward decls. */ |
2069 | ||
99b1c316 | 2070 | class lexer_test; |
88fa5555 DM |
2071 | class lexer_test_options; |
2072 | ||
2073 | /* A class for specifying options of a lexer_test. | |
2074 | The "apply" vfunc is called during the lexer_test constructor. */ | |
2075 | ||
2076 | class lexer_test_options | |
2077 | { | |
2078 | public: | |
2079 | virtual void apply (lexer_test &) = 0; | |
2080 | }; | |
2081 | ||
f5ea989d DM |
2082 | /* Wrapper around an cpp_reader *, which calls cpp_finish and cpp_destroy |
2083 | in its dtor. | |
2084 | ||
2085 | This is needed by struct lexer_test to ensure that the cleanup of the | |
2086 | cpp_reader happens *after* the cleanup of the temp_source_file. */ | |
2087 | ||
2088 | class cpp_reader_ptr | |
2089 | { | |
2090 | public: | |
2091 | cpp_reader_ptr (cpp_reader *ptr) : m_ptr (ptr) {} | |
2092 | ||
2093 | ~cpp_reader_ptr () | |
2094 | { | |
2095 | cpp_finish (m_ptr, NULL); | |
2096 | cpp_destroy (m_ptr); | |
2097 | } | |
2098 | ||
2099 | operator cpp_reader * () const { return m_ptr; } | |
2100 | ||
2101 | private: | |
2102 | cpp_reader *m_ptr; | |
2103 | }; | |
2104 | ||
88fa5555 DM |
2105 | /* A struct for writing lexer tests. */ |
2106 | ||
6c1dae73 | 2107 | class lexer_test |
88fa5555 | 2108 | { |
6c1dae73 | 2109 | public: |
88fa5555 DM |
2110 | lexer_test (const line_table_case &case_, const char *content, |
2111 | lexer_test_options *options); | |
2112 | ~lexer_test (); | |
2113 | ||
2114 | const cpp_token *get_token (); | |
2115 | ||
f5ea989d DM |
2116 | /* The ordering of these fields matters. |
2117 | The line_table_test must be first, since the cpp_reader_ptr | |
2118 | uses it. | |
2119 | The cpp_reader must be cleaned up *after* the temp_source_file | |
2120 | since the filenames in input.c's input cache are owned by the | |
2121 | cpp_reader; in particular, when ~temp_source_file evicts the | |
2122 | filename the filenames must still be alive. */ | |
f87e22c5 | 2123 | line_table_test m_ltt; |
f5ea989d DM |
2124 | cpp_reader_ptr m_parser; |
2125 | temp_source_file m_tempfile; | |
88fa5555 | 2126 | string_concat_db m_concats; |
a3998c2f | 2127 | bool m_implicitly_expect_EOF; |
88fa5555 DM |
2128 | }; |
2129 | ||
2130 | /* Use an EBCDIC encoding for the execution charset, specifically | |
2131 | IBM1047-encoded (aka "EBCDIC 1047", or "Code page 1047"). | |
2132 | ||
2133 | This exercises iconv integration within libcpp. | |
2134 | Not every build of iconv supports the given charset, | |
2135 | so we need to flag this error and handle it gracefully. */ | |
2136 | ||
2137 | class ebcdic_execution_charset : public lexer_test_options | |
2138 | { | |
2139 | public: | |
2140 | ebcdic_execution_charset () : m_num_iconv_errors (0) | |
2141 | { | |
2142 | gcc_assert (s_singleton == NULL); | |
2143 | s_singleton = this; | |
2144 | } | |
2145 | ~ebcdic_execution_charset () | |
2146 | { | |
2147 | gcc_assert (s_singleton == this); | |
2148 | s_singleton = NULL; | |
2149 | } | |
2150 | ||
2151 | void apply (lexer_test &test) FINAL OVERRIDE | |
2152 | { | |
2153 | cpp_options *cpp_opts = cpp_get_options (test.m_parser); | |
2154 | cpp_opts->narrow_charset = "IBM1047"; | |
2155 | ||
2156 | cpp_callbacks *callbacks = cpp_get_callbacks (test.m_parser); | |
c24300ba | 2157 | callbacks->diagnostic = on_diagnostic; |
88fa5555 DM |
2158 | } |
2159 | ||
c24300ba DM |
2160 | static bool on_diagnostic (cpp_reader *pfile ATTRIBUTE_UNUSED, |
2161 | enum cpp_diagnostic_level level ATTRIBUTE_UNUSED, | |
2162 | enum cpp_warning_reason reason ATTRIBUTE_UNUSED, | |
2163 | rich_location *richloc ATTRIBUTE_UNUSED, | |
2164 | const char *msgid, va_list *ap ATTRIBUTE_UNUSED) | |
88fa5555 DM |
2165 | ATTRIBUTE_FPTR_PRINTF(5,0) |
2166 | { | |
2167 | gcc_assert (s_singleton); | |
a7085816 JJ |
2168 | /* Avoid exgettext from picking this up, it is translated in libcpp. */ |
2169 | const char *msg = "conversion from %s to %s not supported by iconv"; | |
2170 | #ifdef ENABLE_NLS | |
2171 | msg = dgettext ("cpplib", msg); | |
2172 | #endif | |
88fa5555 DM |
2173 | /* Detect and record errors emitted by libcpp/charset.c:init_iconv_desc |
2174 | when the local iconv build doesn't support the conversion. */ | |
a7085816 | 2175 | if (strcmp (msgid, msg) == 0) |
88fa5555 DM |
2176 | { |
2177 | s_singleton->m_num_iconv_errors++; | |
2178 | return true; | |
2179 | } | |
2180 | ||
2181 | /* Otherwise, we have an unexpected error. */ | |
2182 | abort (); | |
2183 | } | |
2184 | ||
2185 | bool iconv_errors_occurred_p () const { return m_num_iconv_errors > 0; } | |
2186 | ||
2187 | private: | |
2188 | static ebcdic_execution_charset *s_singleton; | |
2189 | int m_num_iconv_errors; | |
2190 | }; | |
2191 | ||
2192 | ebcdic_execution_charset *ebcdic_execution_charset::s_singleton; | |
2193 | ||
c24300ba | 2194 | /* A lexer_test_options subclass that records a list of diagnostic |
a3998c2f DM |
2195 | messages emitted by the lexer. */ |
2196 | ||
c24300ba | 2197 | class lexer_diagnostic_sink : public lexer_test_options |
a3998c2f DM |
2198 | { |
2199 | public: | |
c24300ba | 2200 | lexer_diagnostic_sink () |
a3998c2f DM |
2201 | { |
2202 | gcc_assert (s_singleton == NULL); | |
2203 | s_singleton = this; | |
2204 | } | |
c24300ba | 2205 | ~lexer_diagnostic_sink () |
a3998c2f DM |
2206 | { |
2207 | gcc_assert (s_singleton == this); | |
2208 | s_singleton = NULL; | |
2209 | ||
2210 | int i; | |
2211 | char *str; | |
c24300ba | 2212 | FOR_EACH_VEC_ELT (m_diagnostics, i, str) |
a3998c2f DM |
2213 | free (str); |
2214 | } | |
2215 | ||
2216 | void apply (lexer_test &test) FINAL OVERRIDE | |
2217 | { | |
2218 | cpp_callbacks *callbacks = cpp_get_callbacks (test.m_parser); | |
c24300ba | 2219 | callbacks->diagnostic = on_diagnostic; |
a3998c2f DM |
2220 | } |
2221 | ||
c24300ba DM |
2222 | static bool on_diagnostic (cpp_reader *pfile ATTRIBUTE_UNUSED, |
2223 | enum cpp_diagnostic_level level ATTRIBUTE_UNUSED, | |
2224 | enum cpp_warning_reason reason ATTRIBUTE_UNUSED, | |
2225 | rich_location *richloc ATTRIBUTE_UNUSED, | |
2226 | const char *msgid, va_list *ap) | |
a3998c2f DM |
2227 | ATTRIBUTE_FPTR_PRINTF(5,0) |
2228 | { | |
2229 | char *msg = xvasprintf (msgid, *ap); | |
c24300ba | 2230 | s_singleton->m_diagnostics.safe_push (msg); |
a3998c2f DM |
2231 | return true; |
2232 | } | |
2233 | ||
c24300ba | 2234 | auto_vec<char *> m_diagnostics; |
a3998c2f DM |
2235 | |
2236 | private: | |
c24300ba | 2237 | static lexer_diagnostic_sink *s_singleton; |
a3998c2f DM |
2238 | }; |
2239 | ||
c24300ba | 2240 | lexer_diagnostic_sink *lexer_diagnostic_sink::s_singleton; |
a3998c2f | 2241 | |
88fa5555 DM |
2242 | /* Constructor. Override line_table with a new instance based on CASE_, |
2243 | and write CONTENT to a tempfile. Create a cpp_reader, and use it to | |
2244 | start parsing the tempfile. */ | |
2245 | ||
2246 | lexer_test::lexer_test (const line_table_case &case_, const char *content, | |
f5ea989d DM |
2247 | lexer_test_options *options) |
2248 | : m_ltt (case_), | |
2249 | m_parser (cpp_create_reader (CLK_GNUC99, NULL, line_table)), | |
88fa5555 DM |
2250 | /* Create a tempfile and write the text to it. */ |
2251 | m_tempfile (SELFTEST_LOCATION, ".c", content), | |
a3998c2f DM |
2252 | m_concats (), |
2253 | m_implicitly_expect_EOF (true) | |
88fa5555 DM |
2254 | { |
2255 | if (options) | |
2256 | options->apply (*this); | |
2257 | ||
2258 | cpp_init_iconv (m_parser); | |
2259 | ||
2260 | /* Parse the file. */ | |
2261 | const char *fname = cpp_read_main_file (m_parser, | |
2262 | m_tempfile.get_filename ()); | |
2263 | ASSERT_NE (fname, NULL); | |
2264 | } | |
2265 | ||
a3998c2f | 2266 | /* Destructor. By default, verify that the next token in m_parser is EOF. */ |
88fa5555 DM |
2267 | |
2268 | lexer_test::~lexer_test () | |
2269 | { | |
2270 | location_t loc; | |
2271 | const cpp_token *tok; | |
2272 | ||
a3998c2f DM |
2273 | if (m_implicitly_expect_EOF) |
2274 | { | |
2275 | tok = cpp_get_token_with_location (m_parser, &loc); | |
2276 | ASSERT_NE (tok, NULL); | |
2277 | ASSERT_EQ (tok->type, CPP_EOF); | |
2278 | } | |
88fa5555 DM |
2279 | } |
2280 | ||
2281 | /* Get the next token from m_parser. */ | |
2282 | ||
2283 | const cpp_token * | |
2284 | lexer_test::get_token () | |
2285 | { | |
2286 | location_t loc; | |
2287 | const cpp_token *tok; | |
2288 | ||
2289 | tok = cpp_get_token_with_location (m_parser, &loc); | |
2290 | ASSERT_NE (tok, NULL); | |
2291 | return tok; | |
2292 | } | |
2293 | ||
2294 | /* Verify that locations within string literals are correctly handled. */ | |
2295 | ||
2296 | /* Verify get_source_range_for_substring for token(s) at STRLOC, | |
2297 | using the string concatenation database for TEST. | |
2298 | ||
2299 | Assert that the character at index IDX is on EXPECTED_LINE, | |
2300 | and that it begins at column EXPECTED_START_COL and ends at | |
2301 | EXPECTED_FINISH_COL (unless the locations are beyond | |
2302 | LINE_MAP_MAX_LOCATION_WITH_COLS, in which case don't check their | |
2303 | columns). */ | |
2304 | ||
2305 | static void | |
2306 | assert_char_at_range (const location &loc, | |
2307 | lexer_test& test, | |
2308 | location_t strloc, enum cpp_ttype type, int idx, | |
2309 | int expected_line, int expected_start_col, | |
2310 | int expected_finish_col) | |
2311 | { | |
2312 | cpp_reader *pfile = test.m_parser; | |
2313 | string_concat_db *concats = &test.m_concats; | |
2314 | ||
a954833d | 2315 | source_range actual_range = source_range(); |
88fa5555 | 2316 | const char *err |
65e736c0 DM |
2317 | = get_source_range_for_char (pfile, concats, strloc, type, idx, |
2318 | &actual_range); | |
88fa5555 DM |
2319 | if (should_have_column_data_p (strloc)) |
2320 | ASSERT_EQ_AT (loc, NULL, err); | |
2321 | else | |
2322 | { | |
2323 | ASSERT_STREQ_AT (loc, | |
2324 | "range starts after LINE_MAP_MAX_LOCATION_WITH_COLS", | |
2325 | err); | |
2326 | return; | |
2327 | } | |
2328 | ||
2329 | int actual_start_line = LOCATION_LINE (actual_range.m_start); | |
2330 | ASSERT_EQ_AT (loc, expected_line, actual_start_line); | |
2331 | int actual_finish_line = LOCATION_LINE (actual_range.m_finish); | |
2332 | ASSERT_EQ_AT (loc, expected_line, actual_finish_line); | |
2333 | ||
2334 | if (should_have_column_data_p (actual_range.m_start)) | |
2335 | { | |
2336 | int actual_start_col = LOCATION_COLUMN (actual_range.m_start); | |
2337 | ASSERT_EQ_AT (loc, expected_start_col, actual_start_col); | |
2338 | } | |
2339 | if (should_have_column_data_p (actual_range.m_finish)) | |
2340 | { | |
2341 | int actual_finish_col = LOCATION_COLUMN (actual_range.m_finish); | |
2342 | ASSERT_EQ_AT (loc, expected_finish_col, actual_finish_col); | |
2343 | } | |
2344 | } | |
2345 | ||
2346 | /* Macro for calling assert_char_at_range, supplying SELFTEST_LOCATION for | |
2347 | the effective location of any errors. */ | |
2348 | ||
2349 | #define ASSERT_CHAR_AT_RANGE(LEXER_TEST, STRLOC, TYPE, IDX, EXPECTED_LINE, \ | |
2350 | EXPECTED_START_COL, EXPECTED_FINISH_COL) \ | |
2351 | assert_char_at_range (SELFTEST_LOCATION, (LEXER_TEST), (STRLOC), (TYPE), \ | |
2352 | (IDX), (EXPECTED_LINE), (EXPECTED_START_COL), \ | |
2353 | (EXPECTED_FINISH_COL)) | |
2354 | ||
2355 | /* Verify get_num_source_ranges_for_substring for token(s) at STRLOC, | |
2356 | using the string concatenation database for TEST. | |
2357 | ||
2358 | Assert that the token(s) at STRLOC contain EXPECTED_NUM_RANGES. */ | |
2359 | ||
2360 | static void | |
2361 | assert_num_substring_ranges (const location &loc, | |
2362 | lexer_test& test, | |
2363 | location_t strloc, | |
2364 | enum cpp_ttype type, | |
2365 | int expected_num_ranges) | |
2366 | { | |
2367 | cpp_reader *pfile = test.m_parser; | |
2368 | string_concat_db *concats = &test.m_concats; | |
2369 | ||
0e06d2b3 | 2370 | int actual_num_ranges = -1; |
88fa5555 DM |
2371 | const char *err |
2372 | = get_num_source_ranges_for_substring (pfile, concats, strloc, type, | |
2373 | &actual_num_ranges); | |
2374 | if (should_have_column_data_p (strloc)) | |
2375 | ASSERT_EQ_AT (loc, NULL, err); | |
2376 | else | |
2377 | { | |
2378 | ASSERT_STREQ_AT (loc, | |
2379 | "range starts after LINE_MAP_MAX_LOCATION_WITH_COLS", | |
2380 | err); | |
2381 | return; | |
2382 | } | |
2383 | ASSERT_EQ_AT (loc, expected_num_ranges, actual_num_ranges); | |
2384 | } | |
2385 | ||
2386 | /* Macro for calling assert_num_substring_ranges, supplying | |
2387 | SELFTEST_LOCATION for the effective location of any errors. */ | |
2388 | ||
2389 | #define ASSERT_NUM_SUBSTRING_RANGES(LEXER_TEST, STRLOC, TYPE, \ | |
2390 | EXPECTED_NUM_RANGES) \ | |
2391 | assert_num_substring_ranges (SELFTEST_LOCATION, (LEXER_TEST), (STRLOC), \ | |
2392 | (TYPE), (EXPECTED_NUM_RANGES)) | |
2393 | ||
2394 | ||
2395 | /* Verify that get_num_source_ranges_for_substring for token(s) at STRLOC | |
2396 | returns an error (using the string concatenation database for TEST). */ | |
2397 | ||
2398 | static void | |
2399 | assert_has_no_substring_ranges (const location &loc, | |
2400 | lexer_test& test, | |
2401 | location_t strloc, | |
2402 | enum cpp_ttype type, | |
2403 | const char *expected_err) | |
2404 | { | |
2405 | cpp_reader *pfile = test.m_parser; | |
2406 | string_concat_db *concats = &test.m_concats; | |
2407 | cpp_substring_ranges ranges; | |
2408 | const char *actual_err | |
2409 | = get_substring_ranges_for_loc (pfile, concats, strloc, | |
2410 | type, ranges); | |
2411 | if (should_have_column_data_p (strloc)) | |
2412 | ASSERT_STREQ_AT (loc, expected_err, actual_err); | |
2413 | else | |
2414 | ASSERT_STREQ_AT (loc, | |
2415 | "range starts after LINE_MAP_MAX_LOCATION_WITH_COLS", | |
2416 | actual_err); | |
2417 | } | |
2418 | ||
2419 | #define ASSERT_HAS_NO_SUBSTRING_RANGES(LEXER_TEST, STRLOC, TYPE, ERR) \ | |
2420 | assert_has_no_substring_ranges (SELFTEST_LOCATION, (LEXER_TEST), \ | |
2421 | (STRLOC), (TYPE), (ERR)) | |
2422 | ||
2423 | /* Lex a simple string literal. Verify the substring location data, before | |
2424 | and after running cpp_interpret_string on it. */ | |
2425 | ||
2426 | static void | |
2427 | test_lexer_string_locations_simple (const line_table_case &case_) | |
2428 | { | |
2429 | /* Digits 0-9 (with 0 at column 10), the simple way. | |
2430 | ....................000000000.11111111112.2222222223333333333 | |
2431 | ....................123456789.01234567890.1234567890123456789 | |
2432 | We add a trailing comment to ensure that we correctly locate | |
2433 | the end of the string literal token. */ | |
2434 | const char *content = " \"0123456789\" /* not a string */\n"; | |
2435 | lexer_test test (case_, content, NULL); | |
2436 | ||
2437 | /* Verify that we get the expected token back, with the correct | |
2438 | location information. */ | |
2439 | const cpp_token *tok = test.get_token (); | |
2440 | ASSERT_EQ (tok->type, CPP_STRING); | |
2441 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"0123456789\""); | |
2442 | ASSERT_TOKEN_LOC_EQ (tok, test.m_tempfile.get_filename (), 1, 9, 20); | |
2443 | ||
2444 | /* At this point in lexing, the quote characters are treated as part of | |
2445 | the string (they are stripped off by cpp_interpret_string). */ | |
2446 | ||
2447 | ASSERT_EQ (tok->val.str.len, 12); | |
2448 | ||
2449 | /* Verify that cpp_interpret_string works. */ | |
2450 | cpp_string dst_string; | |
2451 | const enum cpp_ttype type = CPP_STRING; | |
2452 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2453 | &dst_string, type); | |
2454 | ASSERT_TRUE (result); | |
2455 | ASSERT_STREQ ("0123456789", (const char *)dst_string.text); | |
2456 | free (const_cast <unsigned char *> (dst_string.text)); | |
2457 | ||
2458 | /* Verify ranges of individual characters. This no longer includes the | |
bbd6fcf3 DM |
2459 | opening quote, but does include the closing quote. */ |
2460 | for (int i = 0; i <= 10; i++) | |
88fa5555 DM |
2461 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, |
2462 | 10 + i, 10 + i); | |
2463 | ||
bbd6fcf3 | 2464 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, type, 11); |
88fa5555 DM |
2465 | } |
2466 | ||
2467 | /* As test_lexer_string_locations_simple, but use an EBCDIC execution | |
2468 | encoding. */ | |
2469 | ||
2470 | static void | |
2471 | test_lexer_string_locations_ebcdic (const line_table_case &case_) | |
2472 | { | |
2473 | /* EBCDIC support requires iconv. */ | |
2474 | if (!HAVE_ICONV) | |
2475 | return; | |
2476 | ||
2477 | /* Digits 0-9 (with 0 at column 10), the simple way. | |
2478 | ....................000000000.11111111112.2222222223333333333 | |
2479 | ....................123456789.01234567890.1234567890123456789 | |
2480 | We add a trailing comment to ensure that we correctly locate | |
2481 | the end of the string literal token. */ | |
2482 | const char *content = " \"0123456789\" /* not a string */\n"; | |
2483 | ebcdic_execution_charset use_ebcdic; | |
2484 | lexer_test test (case_, content, &use_ebcdic); | |
2485 | ||
2486 | /* Verify that we get the expected token back, with the correct | |
2487 | location information. */ | |
2488 | const cpp_token *tok = test.get_token (); | |
2489 | ASSERT_EQ (tok->type, CPP_STRING); | |
2490 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"0123456789\""); | |
2491 | ASSERT_TOKEN_LOC_EQ (tok, test.m_tempfile.get_filename (), 1, 9, 20); | |
2492 | ||
2493 | /* At this point in lexing, the quote characters are treated as part of | |
2494 | the string (they are stripped off by cpp_interpret_string). */ | |
2495 | ||
2496 | ASSERT_EQ (tok->val.str.len, 12); | |
2497 | ||
2498 | /* The remainder of the test requires an iconv implementation that | |
2499 | can convert from UTF-8 to the EBCDIC encoding requested above. */ | |
2500 | if (use_ebcdic.iconv_errors_occurred_p ()) | |
2501 | return; | |
2502 | ||
2503 | /* Verify that cpp_interpret_string works. */ | |
2504 | cpp_string dst_string; | |
2505 | const enum cpp_ttype type = CPP_STRING; | |
2506 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2507 | &dst_string, type); | |
2508 | ASSERT_TRUE (result); | |
2509 | /* We should now have EBCDIC-encoded text, specifically | |
2510 | IBM1047-encoded (aka "EBCDIC 1047", or "Code page 1047"). | |
2511 | The digits 0-9 are encoded as 240-249 i.e. 0xf0-0xf9. */ | |
2512 | ASSERT_STREQ ("\xf0\xf1\xf2\xf3\xf4\xf5\xf6\xf7\xf8\xf9", | |
2513 | (const char *)dst_string.text); | |
2514 | free (const_cast <unsigned char *> (dst_string.text)); | |
2515 | ||
2516 | /* Verify that we don't attempt to record substring location information | |
2517 | for such cases. */ | |
2518 | ASSERT_HAS_NO_SUBSTRING_RANGES | |
2519 | (test, tok->src_loc, type, | |
2520 | "execution character set != source character set"); | |
2521 | } | |
2522 | ||
2523 | /* Lex a string literal containing a hex-escaped character. | |
2524 | Verify the substring location data, before and after running | |
2525 | cpp_interpret_string on it. */ | |
2526 | ||
2527 | static void | |
2528 | test_lexer_string_locations_hex (const line_table_case &case_) | |
2529 | { | |
2530 | /* Digits 0-9, expressing digit 5 in ASCII as "\x35" | |
2531 | and with a space in place of digit 6, to terminate the escaped | |
2532 | hex code. | |
2533 | ....................000000000.111111.11112222. | |
2534 | ....................123456789.012345.67890123. */ | |
2535 | const char *content = " \"01234\\x35 789\"\n"; | |
2536 | lexer_test test (case_, content, NULL); | |
2537 | ||
2538 | /* Verify that we get the expected token back, with the correct | |
2539 | location information. */ | |
2540 | const cpp_token *tok = test.get_token (); | |
2541 | ASSERT_EQ (tok->type, CPP_STRING); | |
2542 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"01234\\x35 789\""); | |
2543 | ASSERT_TOKEN_LOC_EQ (tok, test.m_tempfile.get_filename (), 1, 9, 23); | |
2544 | ||
2545 | /* At this point in lexing, the quote characters are treated as part of | |
2546 | the string (they are stripped off by cpp_interpret_string). */ | |
2547 | ASSERT_EQ (tok->val.str.len, 15); | |
2548 | ||
2549 | /* Verify that cpp_interpret_string works. */ | |
2550 | cpp_string dst_string; | |
2551 | const enum cpp_ttype type = CPP_STRING; | |
2552 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2553 | &dst_string, type); | |
2554 | ASSERT_TRUE (result); | |
2555 | ASSERT_STREQ ("012345 789", (const char *)dst_string.text); | |
2556 | free (const_cast <unsigned char *> (dst_string.text)); | |
2557 | ||
2558 | /* Verify ranges of individual characters. This no longer includes the | |
bbd6fcf3 | 2559 | opening quote, but does include the closing quote. */ |
88fa5555 DM |
2560 | for (int i = 0; i <= 4; i++) |
2561 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 10 + i, 10 + i); | |
2562 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, 5, 1, 15, 18); | |
bbd6fcf3 | 2563 | for (int i = 6; i <= 10; i++) |
88fa5555 DM |
2564 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 13 + i, 13 + i); |
2565 | ||
bbd6fcf3 | 2566 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, type, 11); |
88fa5555 DM |
2567 | } |
2568 | ||
2569 | /* Lex a string literal containing an octal-escaped character. | |
2570 | Verify the substring location data after running cpp_interpret_string | |
2571 | on it. */ | |
2572 | ||
2573 | static void | |
2574 | test_lexer_string_locations_oct (const line_table_case &case_) | |
2575 | { | |
2576 | /* Digits 0-9, expressing digit 5 in ASCII as "\065" | |
2577 | and with a space in place of digit 6, to terminate the escaped | |
2578 | octal code. | |
2579 | ....................000000000.111111.11112222.2222223333333333444 | |
2580 | ....................123456789.012345.67890123.4567890123456789012 */ | |
2581 | const char *content = " \"01234\\065 789\" /* not a string */\n"; | |
2582 | lexer_test test (case_, content, NULL); | |
2583 | ||
2584 | /* Verify that we get the expected token back, with the correct | |
2585 | location information. */ | |
2586 | const cpp_token *tok = test.get_token (); | |
2587 | ASSERT_EQ (tok->type, CPP_STRING); | |
2588 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"01234\\065 789\""); | |
2589 | ||
2590 | /* Verify that cpp_interpret_string works. */ | |
2591 | cpp_string dst_string; | |
2592 | const enum cpp_ttype type = CPP_STRING; | |
2593 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2594 | &dst_string, type); | |
2595 | ASSERT_TRUE (result); | |
2596 | ASSERT_STREQ ("012345 789", (const char *)dst_string.text); | |
2597 | free (const_cast <unsigned char *> (dst_string.text)); | |
2598 | ||
2599 | /* Verify ranges of individual characters. This no longer includes the | |
bbd6fcf3 | 2600 | opening quote, but does include the closing quote. */ |
88fa5555 DM |
2601 | for (int i = 0; i < 5; i++) |
2602 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 10 + i, 10 + i); | |
2603 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, 5, 1, 15, 18); | |
bbd6fcf3 | 2604 | for (int i = 6; i <= 10; i++) |
88fa5555 DM |
2605 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 13 + i, 13 + i); |
2606 | ||
bbd6fcf3 | 2607 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, type, 11); |
88fa5555 DM |
2608 | } |
2609 | ||
2610 | /* Test of string literal containing letter escapes. */ | |
2611 | ||
2612 | static void | |
2613 | test_lexer_string_locations_letter_escape_1 (const line_table_case &case_) | |
2614 | { | |
2615 | /* The string "\tfoo\\\nbar" i.e. tab, "foo", backslash, newline, bar. | |
2616 | .....................000000000.1.11111.1.1.11222.22222223333333 | |
2617 | .....................123456789.0.12345.6.7.89012.34567890123456. */ | |
2618 | const char *content = (" \"\\tfoo\\\\\\nbar\" /* non-str */\n"); | |
2619 | lexer_test test (case_, content, NULL); | |
2620 | ||
2621 | /* Verify that we get the expected tokens back. */ | |
2622 | const cpp_token *tok = test.get_token (); | |
2623 | ASSERT_EQ (tok->type, CPP_STRING); | |
2624 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"\\tfoo\\\\\\nbar\""); | |
2625 | ||
2626 | /* Verify ranges of individual characters. */ | |
2627 | /* "\t". */ | |
2628 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
2629 | 0, 1, 10, 11); | |
2630 | /* "foo". */ | |
2631 | for (int i = 1; i <= 3; i++) | |
2632 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
2633 | i, 1, 11 + i, 11 + i); | |
2634 | /* "\\" and "\n". */ | |
2635 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
2636 | 4, 1, 15, 16); | |
2637 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
2638 | 5, 1, 17, 18); | |
2639 | ||
bbd6fcf3 DM |
2640 | /* "bar" and closing quote for nul-terminator. */ |
2641 | for (int i = 6; i <= 9; i++) | |
88fa5555 DM |
2642 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, |
2643 | i, 1, 13 + i, 13 + i); | |
2644 | ||
bbd6fcf3 | 2645 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, CPP_STRING, 10); |
88fa5555 DM |
2646 | } |
2647 | ||
2648 | /* Another test of a string literal containing a letter escape. | |
2649 | Based on string seen in | |
2650 | printf ("%-%\n"); | |
2651 | in gcc.dg/format/c90-printf-1.c. */ | |
2652 | ||
2653 | static void | |
2654 | test_lexer_string_locations_letter_escape_2 (const line_table_case &case_) | |
2655 | { | |
2656 | /* .....................000000000.1111.11.1111.22222222223. | |
2657 | .....................123456789.0123.45.6789.01234567890. */ | |
2658 | const char *content = (" \"%-%\\n\" /* non-str */\n"); | |
2659 | lexer_test test (case_, content, NULL); | |
2660 | ||
2661 | /* Verify that we get the expected tokens back. */ | |
2662 | const cpp_token *tok = test.get_token (); | |
2663 | ASSERT_EQ (tok->type, CPP_STRING); | |
2664 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"%-%\\n\""); | |
2665 | ||
2666 | /* Verify ranges of individual characters. */ | |
2667 | /* "%-%". */ | |
2668 | for (int i = 0; i < 3; i++) | |
2669 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
2670 | i, 1, 10 + i, 10 + i); | |
2671 | /* "\n". */ | |
2672 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
2673 | 3, 1, 13, 14); | |
2674 | ||
bbd6fcf3 DM |
2675 | /* Closing quote for nul-terminator. */ |
2676 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
2677 | 4, 1, 15, 15); | |
2678 | ||
2679 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, CPP_STRING, 5); | |
88fa5555 DM |
2680 | } |
2681 | ||
2682 | /* Lex a string literal containing UCN 4 characters. | |
2683 | Verify the substring location data after running cpp_interpret_string | |
2684 | on it. */ | |
2685 | ||
2686 | static void | |
2687 | test_lexer_string_locations_ucn4 (const line_table_case &case_) | |
2688 | { | |
2689 | /* Digits 0-9, expressing digits 5 and 6 as Roman numerals expressed | |
2690 | as UCN 4. | |
2691 | ....................000000000.111111.111122.222222223.33333333344444 | |
2692 | ....................123456789.012345.678901.234567890.12345678901234 */ | |
2693 | const char *content = " \"01234\\u2174\\u2175789\" /* non-str */\n"; | |
2694 | lexer_test test (case_, content, NULL); | |
2695 | ||
2696 | /* Verify that we get the expected token back, with the correct | |
2697 | location information. */ | |
2698 | const cpp_token *tok = test.get_token (); | |
2699 | ASSERT_EQ (tok->type, CPP_STRING); | |
2700 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"01234\\u2174\\u2175789\""); | |
2701 | ||
2702 | /* Verify that cpp_interpret_string works. | |
2703 | The string should be encoded in the execution character | |
700d4cb0 | 2704 | set. Assuming that is UTF-8, we should have the following: |
88fa5555 DM |
2705 | ----------- ---- ----- ------- ---------------- |
2706 | Byte offset Byte Octal Unicode Source Column(s) | |
2707 | ----------- ---- ----- ------- ---------------- | |
2708 | 0 0x30 '0' 10 | |
2709 | 1 0x31 '1' 11 | |
2710 | 2 0x32 '2' 12 | |
2711 | 3 0x33 '3' 13 | |
2712 | 4 0x34 '4' 14 | |
2713 | 5 0xE2 \342 U+2174 15-20 | |
2714 | 6 0x85 \205 (cont) 15-20 | |
2715 | 7 0xB4 \264 (cont) 15-20 | |
2716 | 8 0xE2 \342 U+2175 21-26 | |
2717 | 9 0x85 \205 (cont) 21-26 | |
2718 | 10 0xB5 \265 (cont) 21-26 | |
2719 | 11 0x37 '7' 27 | |
2720 | 12 0x38 '8' 28 | |
2721 | 13 0x39 '9' 29 | |
bbd6fcf3 | 2722 | 14 0x00 30 (closing quote) |
88fa5555 DM |
2723 | ----------- ---- ----- ------- ---------------. */ |
2724 | ||
2725 | cpp_string dst_string; | |
2726 | const enum cpp_ttype type = CPP_STRING; | |
2727 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2728 | &dst_string, type); | |
2729 | ASSERT_TRUE (result); | |
2730 | ASSERT_STREQ ("01234\342\205\264\342\205\265789", | |
2731 | (const char *)dst_string.text); | |
2732 | free (const_cast <unsigned char *> (dst_string.text)); | |
2733 | ||
2734 | /* Verify ranges of individual characters. This no longer includes the | |
bbd6fcf3 | 2735 | opening quote, but does include the closing quote. |
88fa5555 DM |
2736 | '01234'. */ |
2737 | for (int i = 0; i <= 4; i++) | |
2738 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 10 + i, 10 + i); | |
2739 | /* U+2174. */ | |
2740 | for (int i = 5; i <= 7; i++) | |
2741 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 15, 20); | |
2742 | /* U+2175. */ | |
2743 | for (int i = 8; i <= 10; i++) | |
2744 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 21, 26); | |
bbd6fcf3 DM |
2745 | /* '789' and nul terminator */ |
2746 | for (int i = 11; i <= 14; i++) | |
88fa5555 DM |
2747 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 16 + i, 16 + i); |
2748 | ||
bbd6fcf3 | 2749 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, type, 15); |
88fa5555 DM |
2750 | } |
2751 | ||
2752 | /* Lex a string literal containing UCN 8 characters. | |
2753 | Verify the substring location data after running cpp_interpret_string | |
2754 | on it. */ | |
2755 | ||
2756 | static void | |
2757 | test_lexer_string_locations_ucn8 (const line_table_case &case_) | |
2758 | { | |
2759 | /* Digits 0-9, expressing digits 5 and 6 as Roman numerals as UCN 8. | |
2760 | ....................000000000.111111.1111222222.2222333333333.344444 | |
2761 | ....................123456789.012345.6789012345.6789012345678.901234 */ | |
2762 | const char *content = " \"01234\\U00002174\\U00002175789\" /* */\n"; | |
2763 | lexer_test test (case_, content, NULL); | |
2764 | ||
2765 | /* Verify that we get the expected token back, with the correct | |
2766 | location information. */ | |
2767 | const cpp_token *tok = test.get_token (); | |
2768 | ASSERT_EQ (tok->type, CPP_STRING); | |
2769 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, | |
2770 | "\"01234\\U00002174\\U00002175789\""); | |
2771 | ||
2772 | /* Verify that cpp_interpret_string works. | |
2773 | The UTF-8 encoding of the string is identical to that from | |
2774 | the ucn4 testcase above; the only difference is the column | |
2775 | locations. */ | |
2776 | cpp_string dst_string; | |
2777 | const enum cpp_ttype type = CPP_STRING; | |
2778 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2779 | &dst_string, type); | |
2780 | ASSERT_TRUE (result); | |
2781 | ASSERT_STREQ ("01234\342\205\264\342\205\265789", | |
2782 | (const char *)dst_string.text); | |
2783 | free (const_cast <unsigned char *> (dst_string.text)); | |
2784 | ||
2785 | /* Verify ranges of individual characters. This no longer includes the | |
bbd6fcf3 | 2786 | opening quote, but does include the closing quote. |
88fa5555 DM |
2787 | '01234'. */ |
2788 | for (int i = 0; i <= 4; i++) | |
2789 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 10 + i, 10 + i); | |
2790 | /* U+2174. */ | |
2791 | for (int i = 5; i <= 7; i++) | |
2792 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 15, 24); | |
2793 | /* U+2175. */ | |
2794 | for (int i = 8; i <= 10; i++) | |
2795 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 25, 34); | |
2796 | /* '789' at columns 35-37 */ | |
2797 | for (int i = 11; i <= 13; i++) | |
2798 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 24 + i, 24 + i); | |
bbd6fcf3 DM |
2799 | /* Closing quote/nul-terminator at column 38. */ |
2800 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, 14, 1, 38, 38); | |
88fa5555 | 2801 | |
bbd6fcf3 | 2802 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, type, 15); |
88fa5555 DM |
2803 | } |
2804 | ||
2805 | /* Fetch a big-endian 32-bit value and convert to host endianness. */ | |
2806 | ||
2807 | static uint32_t | |
2808 | uint32_from_big_endian (const uint32_t *ptr_be_value) | |
2809 | { | |
2810 | const unsigned char *buf = (const unsigned char *)ptr_be_value; | |
2811 | return (((uint32_t) buf[0] << 24) | |
2812 | | ((uint32_t) buf[1] << 16) | |
2813 | | ((uint32_t) buf[2] << 8) | |
2814 | | (uint32_t) buf[3]); | |
2815 | } | |
2816 | ||
2817 | /* Lex a wide string literal and verify that attempts to read substring | |
2818 | location data from it fail gracefully. */ | |
2819 | ||
2820 | static void | |
2821 | test_lexer_string_locations_wide_string (const line_table_case &case_) | |
2822 | { | |
2823 | /* Digits 0-9. | |
2824 | ....................000000000.11111111112.22222222233333 | |
2825 | ....................123456789.01234567890.12345678901234 */ | |
2826 | const char *content = " L\"0123456789\" /* non-str */\n"; | |
2827 | lexer_test test (case_, content, NULL); | |
2828 | ||
2829 | /* Verify that we get the expected token back, with the correct | |
2830 | location information. */ | |
2831 | const cpp_token *tok = test.get_token (); | |
2832 | ASSERT_EQ (tok->type, CPP_WSTRING); | |
2833 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "L\"0123456789\""); | |
2834 | ||
2835 | /* Verify that cpp_interpret_string works, using CPP_WSTRING. */ | |
2836 | cpp_string dst_string; | |
2837 | const enum cpp_ttype type = CPP_WSTRING; | |
2838 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2839 | &dst_string, type); | |
2840 | ASSERT_TRUE (result); | |
2841 | /* The cpp_reader defaults to big-endian with | |
2842 | CHAR_BIT * sizeof (int) for the wchar_precision, so dst_string should | |
2843 | now be encoded as UTF-32BE. */ | |
2844 | const uint32_t *be32_chars = (const uint32_t *)dst_string.text; | |
2845 | ASSERT_EQ ('0', uint32_from_big_endian (&be32_chars[0])); | |
2846 | ASSERT_EQ ('5', uint32_from_big_endian (&be32_chars[5])); | |
2847 | ASSERT_EQ ('9', uint32_from_big_endian (&be32_chars[9])); | |
2848 | ASSERT_EQ (0, uint32_from_big_endian (&be32_chars[10])); | |
2849 | free (const_cast <unsigned char *> (dst_string.text)); | |
2850 | ||
2851 | /* We don't yet support generating substring location information | |
2852 | for L"" strings. */ | |
2853 | ASSERT_HAS_NO_SUBSTRING_RANGES | |
2854 | (test, tok->src_loc, type, | |
2855 | "execution character set != source character set"); | |
2856 | } | |
2857 | ||
2858 | /* Fetch a big-endian 16-bit value and convert to host endianness. */ | |
2859 | ||
2860 | static uint16_t | |
2861 | uint16_from_big_endian (const uint16_t *ptr_be_value) | |
2862 | { | |
2863 | const unsigned char *buf = (const unsigned char *)ptr_be_value; | |
2864 | return ((uint16_t) buf[0] << 8) | (uint16_t) buf[1]; | |
2865 | } | |
2866 | ||
2867 | /* Lex a u"" string literal and verify that attempts to read substring | |
2868 | location data from it fail gracefully. */ | |
2869 | ||
2870 | static void | |
2871 | test_lexer_string_locations_string16 (const line_table_case &case_) | |
2872 | { | |
2873 | /* Digits 0-9. | |
2874 | ....................000000000.11111111112.22222222233333 | |
2875 | ....................123456789.01234567890.12345678901234 */ | |
2876 | const char *content = " u\"0123456789\" /* non-str */\n"; | |
2877 | lexer_test test (case_, content, NULL); | |
2878 | ||
2879 | /* Verify that we get the expected token back, with the correct | |
2880 | location information. */ | |
2881 | const cpp_token *tok = test.get_token (); | |
2882 | ASSERT_EQ (tok->type, CPP_STRING16); | |
2883 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "u\"0123456789\""); | |
2884 | ||
2885 | /* Verify that cpp_interpret_string works, using CPP_STRING16. */ | |
2886 | cpp_string dst_string; | |
2887 | const enum cpp_ttype type = CPP_STRING16; | |
2888 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2889 | &dst_string, type); | |
2890 | ASSERT_TRUE (result); | |
2891 | ||
2892 | /* The cpp_reader defaults to big-endian, so dst_string should | |
2893 | now be encoded as UTF-16BE. */ | |
2894 | const uint16_t *be16_chars = (const uint16_t *)dst_string.text; | |
2895 | ASSERT_EQ ('0', uint16_from_big_endian (&be16_chars[0])); | |
2896 | ASSERT_EQ ('5', uint16_from_big_endian (&be16_chars[5])); | |
2897 | ASSERT_EQ ('9', uint16_from_big_endian (&be16_chars[9])); | |
2898 | ASSERT_EQ (0, uint16_from_big_endian (&be16_chars[10])); | |
2899 | free (const_cast <unsigned char *> (dst_string.text)); | |
2900 | ||
2901 | /* We don't yet support generating substring location information | |
2902 | for L"" strings. */ | |
2903 | ASSERT_HAS_NO_SUBSTRING_RANGES | |
2904 | (test, tok->src_loc, type, | |
2905 | "execution character set != source character set"); | |
2906 | } | |
2907 | ||
2908 | /* Lex a U"" string literal and verify that attempts to read substring | |
2909 | location data from it fail gracefully. */ | |
2910 | ||
2911 | static void | |
2912 | test_lexer_string_locations_string32 (const line_table_case &case_) | |
2913 | { | |
2914 | /* Digits 0-9. | |
2915 | ....................000000000.11111111112.22222222233333 | |
2916 | ....................123456789.01234567890.12345678901234 */ | |
2917 | const char *content = " U\"0123456789\" /* non-str */\n"; | |
2918 | lexer_test test (case_, content, NULL); | |
2919 | ||
2920 | /* Verify that we get the expected token back, with the correct | |
2921 | location information. */ | |
2922 | const cpp_token *tok = test.get_token (); | |
2923 | ASSERT_EQ (tok->type, CPP_STRING32); | |
2924 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "U\"0123456789\""); | |
2925 | ||
2926 | /* Verify that cpp_interpret_string works, using CPP_STRING32. */ | |
2927 | cpp_string dst_string; | |
2928 | const enum cpp_ttype type = CPP_STRING32; | |
2929 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2930 | &dst_string, type); | |
2931 | ASSERT_TRUE (result); | |
2932 | ||
2933 | /* The cpp_reader defaults to big-endian, so dst_string should | |
2934 | now be encoded as UTF-32BE. */ | |
2935 | const uint32_t *be32_chars = (const uint32_t *)dst_string.text; | |
2936 | ASSERT_EQ ('0', uint32_from_big_endian (&be32_chars[0])); | |
2937 | ASSERT_EQ ('5', uint32_from_big_endian (&be32_chars[5])); | |
2938 | ASSERT_EQ ('9', uint32_from_big_endian (&be32_chars[9])); | |
2939 | ASSERT_EQ (0, uint32_from_big_endian (&be32_chars[10])); | |
2940 | free (const_cast <unsigned char *> (dst_string.text)); | |
2941 | ||
2942 | /* We don't yet support generating substring location information | |
2943 | for L"" strings. */ | |
2944 | ASSERT_HAS_NO_SUBSTRING_RANGES | |
2945 | (test, tok->src_loc, type, | |
2946 | "execution character set != source character set"); | |
2947 | } | |
2948 | ||
2949 | /* Lex a u8-string literal. | |
2950 | Verify the substring location data after running cpp_interpret_string | |
2951 | on it. */ | |
2952 | ||
2953 | static void | |
2954 | test_lexer_string_locations_u8 (const line_table_case &case_) | |
2955 | { | |
2956 | /* Digits 0-9. | |
2957 | ....................000000000.11111111112.22222222233333 | |
2958 | ....................123456789.01234567890.12345678901234 */ | |
2959 | const char *content = " u8\"0123456789\" /* non-str */\n"; | |
2960 | lexer_test test (case_, content, NULL); | |
2961 | ||
2962 | /* Verify that we get the expected token back, with the correct | |
2963 | location information. */ | |
2964 | const cpp_token *tok = test.get_token (); | |
2965 | ASSERT_EQ (tok->type, CPP_UTF8STRING); | |
2966 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "u8\"0123456789\""); | |
2967 | ||
2968 | /* Verify that cpp_interpret_string works. */ | |
2969 | cpp_string dst_string; | |
2970 | const enum cpp_ttype type = CPP_STRING; | |
2971 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
2972 | &dst_string, type); | |
2973 | ASSERT_TRUE (result); | |
2974 | ASSERT_STREQ ("0123456789", (const char *)dst_string.text); | |
2975 | free (const_cast <unsigned char *> (dst_string.text)); | |
2976 | ||
2977 | /* Verify ranges of individual characters. This no longer includes the | |
bbd6fcf3 DM |
2978 | opening quote, but does include the closing quote. */ |
2979 | for (int i = 0; i <= 10; i++) | |
88fa5555 DM |
2980 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 10 + i, 10 + i); |
2981 | } | |
2982 | ||
2983 | /* Lex a string literal containing UTF-8 source characters. | |
2984 | Verify the substring location data after running cpp_interpret_string | |
2985 | on it. */ | |
2986 | ||
2987 | static void | |
2988 | test_lexer_string_locations_utf8_source (const line_table_case &case_) | |
2989 | { | |
2990 | /* This string literal is written out to the source file as UTF-8, | |
2991 | and is of the form "before mojibake after", where "mojibake" | |
2992 | is written as the following four unicode code points: | |
2993 | U+6587 CJK UNIFIED IDEOGRAPH-6587 | |
2994 | U+5B57 CJK UNIFIED IDEOGRAPH-5B57 | |
2995 | U+5316 CJK UNIFIED IDEOGRAPH-5316 | |
2996 | U+3051 HIRAGANA LETTER KE. | |
2997 | Each of these is 3 bytes wide when encoded in UTF-8, whereas the | |
2998 | "before" and "after" are 1 byte per unicode character. | |
2999 | ||
3000 | The numbering shown are "columns", which are *byte* numbers within | |
3001 | the line, rather than unicode character numbers. | |
3002 | ||
3003 | .................... 000000000.1111111. | |
3004 | .................... 123456789.0123456. */ | |
3005 | const char *content = (" \"before " | |
3006 | /* U+6587 CJK UNIFIED IDEOGRAPH-6587 | |
3007 | UTF-8: 0xE6 0x96 0x87 | |
3008 | C octal escaped UTF-8: \346\226\207 | |
3009 | "column" numbers: 17-19. */ | |
3010 | "\346\226\207" | |
3011 | ||
3012 | /* U+5B57 CJK UNIFIED IDEOGRAPH-5B57 | |
3013 | UTF-8: 0xE5 0xAD 0x97 | |
3014 | C octal escaped UTF-8: \345\255\227 | |
3015 | "column" numbers: 20-22. */ | |
3016 | "\345\255\227" | |
3017 | ||
3018 | /* U+5316 CJK UNIFIED IDEOGRAPH-5316 | |
3019 | UTF-8: 0xE5 0x8C 0x96 | |
3020 | C octal escaped UTF-8: \345\214\226 | |
3021 | "column" numbers: 23-25. */ | |
3022 | "\345\214\226" | |
3023 | ||
3024 | /* U+3051 HIRAGANA LETTER KE | |
3025 | UTF-8: 0xE3 0x81 0x91 | |
3026 | C octal escaped UTF-8: \343\201\221 | |
3027 | "column" numbers: 26-28. */ | |
3028 | "\343\201\221" | |
3029 | ||
3030 | /* column numbers 29 onwards | |
3031 | 2333333.33334444444444 | |
3032 | 9012345.67890123456789. */ | |
3033 | " after\" /* non-str */\n"); | |
3034 | lexer_test test (case_, content, NULL); | |
3035 | ||
3036 | /* Verify that we get the expected token back, with the correct | |
3037 | location information. */ | |
3038 | const cpp_token *tok = test.get_token (); | |
3039 | ASSERT_EQ (tok->type, CPP_STRING); | |
3040 | ASSERT_TOKEN_AS_TEXT_EQ | |
3041 | (test.m_parser, tok, | |
3042 | "\"before \346\226\207\345\255\227\345\214\226\343\201\221 after\""); | |
3043 | ||
3044 | /* Verify that cpp_interpret_string works. */ | |
3045 | cpp_string dst_string; | |
3046 | const enum cpp_ttype type = CPP_STRING; | |
3047 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
3048 | &dst_string, type); | |
3049 | ASSERT_TRUE (result); | |
3050 | ASSERT_STREQ | |
3051 | ("before \346\226\207\345\255\227\345\214\226\343\201\221 after", | |
3052 | (const char *)dst_string.text); | |
3053 | free (const_cast <unsigned char *> (dst_string.text)); | |
3054 | ||
3055 | /* Verify ranges of individual characters. This no longer includes the | |
bbd6fcf3 | 3056 | opening quote, but does include the closing quote. |
88fa5555 | 3057 | Assuming that both source and execution encodings are UTF-8, we have |
bbd6fcf3 | 3058 | a run of 25 octets in each, plus the NUL terminator. */ |
88fa5555 DM |
3059 | for (int i = 0; i < 25; i++) |
3060 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, i, 1, 10 + i, 10 + i); | |
bbd6fcf3 DM |
3061 | /* NUL-terminator should use the closing quote at column 35. */ |
3062 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, type, 25, 1, 35, 35); | |
88fa5555 | 3063 | |
bbd6fcf3 | 3064 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, type, 26); |
88fa5555 DM |
3065 | } |
3066 | ||
3067 | /* Test of string literal concatenation. */ | |
3068 | ||
3069 | static void | |
3070 | test_lexer_string_locations_concatenation_1 (const line_table_case &case_) | |
3071 | { | |
3072 | /* Digits 0-9. | |
3073 | .....................000000000.111111.11112222222222 | |
3074 | .....................123456789.012345.67890123456789. */ | |
3075 | const char *content = (" \"01234\" /* non-str */\n" | |
3076 | " \"56789\" /* non-str */\n"); | |
3077 | lexer_test test (case_, content, NULL); | |
3078 | ||
3079 | location_t input_locs[2]; | |
3080 | ||
3081 | /* Verify that we get the expected tokens back. */ | |
3082 | auto_vec <cpp_string> input_strings; | |
3083 | const cpp_token *tok_a = test.get_token (); | |
3084 | ASSERT_EQ (tok_a->type, CPP_STRING); | |
3085 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok_a, "\"01234\""); | |
3086 | input_strings.safe_push (tok_a->val.str); | |
3087 | input_locs[0] = tok_a->src_loc; | |
3088 | ||
3089 | const cpp_token *tok_b = test.get_token (); | |
3090 | ASSERT_EQ (tok_b->type, CPP_STRING); | |
3091 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok_b, "\"56789\""); | |
3092 | input_strings.safe_push (tok_b->val.str); | |
3093 | input_locs[1] = tok_b->src_loc; | |
3094 | ||
3095 | /* Verify that cpp_interpret_string works. */ | |
3096 | cpp_string dst_string; | |
3097 | const enum cpp_ttype type = CPP_STRING; | |
3098 | bool result = cpp_interpret_string (test.m_parser, | |
3099 | input_strings.address (), 2, | |
3100 | &dst_string, type); | |
3101 | ASSERT_TRUE (result); | |
3102 | ASSERT_STREQ ("0123456789", (const char *)dst_string.text); | |
3103 | free (const_cast <unsigned char *> (dst_string.text)); | |
3104 | ||
3105 | /* Simulate c-lex.c's lex_string in order to record concatenation. */ | |
3106 | test.m_concats.record_string_concatenation (2, input_locs); | |
3107 | ||
3108 | location_t initial_loc = input_locs[0]; | |
3109 | ||
bbd6fcf3 | 3110 | /* "01234" on line 1. */ |
88fa5555 DM |
3111 | for (int i = 0; i <= 4; i++) |
3112 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, i, 1, 10 + i, 10 + i); | |
bbd6fcf3 DM |
3113 | /* "56789" in line 2, plus its closing quote for the nul terminator. */ |
3114 | for (int i = 5; i <= 10; i++) | |
88fa5555 DM |
3115 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, i, 2, 5 + i, 5 + i); |
3116 | ||
bbd6fcf3 | 3117 | ASSERT_NUM_SUBSTRING_RANGES (test, initial_loc, type, 11); |
88fa5555 DM |
3118 | } |
3119 | ||
3120 | /* Another test of string literal concatenation. */ | |
3121 | ||
3122 | static void | |
3123 | test_lexer_string_locations_concatenation_2 (const line_table_case &case_) | |
3124 | { | |
3125 | /* Digits 0-9. | |
3126 | .....................000000000.111.11111112222222 | |
3127 | .....................123456789.012.34567890123456. */ | |
3128 | const char *content = (" \"01\" /* non-str */\n" | |
3129 | " \"23\" /* non-str */\n" | |
3130 | " \"45\" /* non-str */\n" | |
3131 | " \"67\" /* non-str */\n" | |
3132 | " \"89\" /* non-str */\n"); | |
3133 | lexer_test test (case_, content, NULL); | |
3134 | ||
3135 | auto_vec <cpp_string> input_strings; | |
3136 | location_t input_locs[5]; | |
3137 | ||
3138 | /* Verify that we get the expected tokens back. */ | |
3139 | for (int i = 0; i < 5; i++) | |
3140 | { | |
3141 | const cpp_token *tok = test.get_token (); | |
3142 | ASSERT_EQ (tok->type, CPP_STRING); | |
3143 | input_strings.safe_push (tok->val.str); | |
3144 | input_locs[i] = tok->src_loc; | |
3145 | } | |
3146 | ||
3147 | /* Verify that cpp_interpret_string works. */ | |
3148 | cpp_string dst_string; | |
3149 | const enum cpp_ttype type = CPP_STRING; | |
3150 | bool result = cpp_interpret_string (test.m_parser, | |
3151 | input_strings.address (), 5, | |
3152 | &dst_string, type); | |
3153 | ASSERT_TRUE (result); | |
3154 | ASSERT_STREQ ("0123456789", (const char *)dst_string.text); | |
3155 | free (const_cast <unsigned char *> (dst_string.text)); | |
3156 | ||
3157 | /* Simulate c-lex.c's lex_string in order to record concatenation. */ | |
3158 | test.m_concats.record_string_concatenation (5, input_locs); | |
3159 | ||
3160 | location_t initial_loc = input_locs[0]; | |
3161 | ||
3162 | /* Within ASSERT_CHAR_AT_RANGE (actually assert_char_at_range), we can | |
3163 | detect if the initial loc is after LINE_MAP_MAX_LOCATION_WITH_COLS | |
3164 | and expect get_source_range_for_substring to fail. | |
3165 | However, for a string concatenation test, we can have a case | |
3166 | where the initial string is fully before LINE_MAP_MAX_LOCATION_WITH_COLS, | |
3167 | but subsequent strings can be after it. | |
3168 | Attempting to detect this within assert_char_at_range | |
3169 | would overcomplicate the logic for the common test cases, so | |
3170 | we detect it here. */ | |
3171 | if (should_have_column_data_p (input_locs[0]) | |
3172 | && !should_have_column_data_p (input_locs[4])) | |
3173 | { | |
3174 | /* Verify that get_source_range_for_substring gracefully rejects | |
3175 | this case. */ | |
3176 | source_range actual_range; | |
3177 | const char *err | |
65e736c0 DM |
3178 | = get_source_range_for_char (test.m_parser, &test.m_concats, |
3179 | initial_loc, type, 0, &actual_range); | |
88fa5555 DM |
3180 | ASSERT_STREQ ("range starts after LINE_MAP_MAX_LOCATION_WITH_COLS", err); |
3181 | return; | |
3182 | } | |
3183 | ||
3184 | for (int i = 0; i < 5; i++) | |
3185 | for (int j = 0; j < 2; j++) | |
3186 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, (i * 2) + j, | |
3187 | i + 1, 10 + j, 10 + j); | |
3188 | ||
bbd6fcf3 DM |
3189 | /* NUL-terminator should use the final closing quote at line 5 column 12. */ |
3190 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, 10, 5, 12, 12); | |
3191 | ||
3192 | ASSERT_NUM_SUBSTRING_RANGES (test, initial_loc, type, 11); | |
88fa5555 DM |
3193 | } |
3194 | ||
3195 | /* Another test of string literal concatenation, this time combined with | |
3196 | various kinds of escaped characters. */ | |
3197 | ||
3198 | static void | |
3199 | test_lexer_string_locations_concatenation_3 (const line_table_case &case_) | |
3200 | { | |
3201 | /* Digits 0-9, expressing digit 5 in ASCII as hex "\x35" | |
3202 | digit 6 in ASCII as octal "\066", concatenating multiple strings. */ | |
3203 | const char *content | |
3204 | /* .000000000.111111.111.1.2222.222.2.2233.333.3333.34444444444555 | |
3205 | .123456789.012345.678.9.0123.456.7.8901.234.5678.90123456789012. */ | |
3206 | = (" \"01234\" \"\\x35\" \"\\066\" \"789\" /* non-str */\n"); | |
3207 | lexer_test test (case_, content, NULL); | |
3208 | ||
3209 | auto_vec <cpp_string> input_strings; | |
3210 | location_t input_locs[4]; | |
3211 | ||
3212 | /* Verify that we get the expected tokens back. */ | |
3213 | for (int i = 0; i < 4; i++) | |
3214 | { | |
3215 | const cpp_token *tok = test.get_token (); | |
3216 | ASSERT_EQ (tok->type, CPP_STRING); | |
3217 | input_strings.safe_push (tok->val.str); | |
3218 | input_locs[i] = tok->src_loc; | |
3219 | } | |
3220 | ||
3221 | /* Verify that cpp_interpret_string works. */ | |
3222 | cpp_string dst_string; | |
3223 | const enum cpp_ttype type = CPP_STRING; | |
3224 | bool result = cpp_interpret_string (test.m_parser, | |
3225 | input_strings.address (), 4, | |
3226 | &dst_string, type); | |
3227 | ASSERT_TRUE (result); | |
3228 | ASSERT_STREQ ("0123456789", (const char *)dst_string.text); | |
3229 | free (const_cast <unsigned char *> (dst_string.text)); | |
3230 | ||
3231 | /* Simulate c-lex.c's lex_string in order to record concatenation. */ | |
3232 | test.m_concats.record_string_concatenation (4, input_locs); | |
3233 | ||
3234 | location_t initial_loc = input_locs[0]; | |
3235 | ||
3236 | for (int i = 0; i <= 4; i++) | |
3237 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, i, 1, 10 + i, 10 + i); | |
3238 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, 5, 1, 19, 22); | |
3239 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, 6, 1, 27, 30); | |
3240 | for (int i = 7; i <= 9; i++) | |
3241 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, i, 1, 28 + i, 28 + i); | |
3242 | ||
bbd6fcf3 DM |
3243 | /* NUL-terminator should use the location of the final closing quote. */ |
3244 | ASSERT_CHAR_AT_RANGE (test, initial_loc, type, 10, 1, 38, 38); | |
3245 | ||
3246 | ASSERT_NUM_SUBSTRING_RANGES (test, initial_loc, type, 11); | |
88fa5555 DM |
3247 | } |
3248 | ||
3249 | /* Test of string literal in a macro. */ | |
3250 | ||
3251 | static void | |
3252 | test_lexer_string_locations_macro (const line_table_case &case_) | |
3253 | { | |
3254 | /* Digits 0-9. | |
3255 | .....................0000000001111111111.22222222223. | |
3256 | .....................1234567890123456789.01234567890. */ | |
3257 | const char *content = ("#define MACRO \"0123456789\" /* non-str */\n" | |
3258 | " MACRO"); | |
3259 | lexer_test test (case_, content, NULL); | |
3260 | ||
3261 | /* Verify that we get the expected tokens back. */ | |
3262 | const cpp_token *tok = test.get_token (); | |
3263 | ASSERT_EQ (tok->type, CPP_PADDING); | |
3264 | ||
3265 | tok = test.get_token (); | |
3266 | ASSERT_EQ (tok->type, CPP_STRING); | |
3267 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"0123456789\""); | |
3268 | ||
3269 | /* Verify ranges of individual characters. We ought to | |
3270 | see columns within the macro definition. */ | |
bbd6fcf3 | 3271 | for (int i = 0; i <= 10; i++) |
88fa5555 DM |
3272 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, |
3273 | i, 1, 20 + i, 20 + i); | |
3274 | ||
bbd6fcf3 | 3275 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, CPP_STRING, 11); |
88fa5555 DM |
3276 | |
3277 | tok = test.get_token (); | |
3278 | ASSERT_EQ (tok->type, CPP_PADDING); | |
3279 | } | |
3280 | ||
3281 | /* Test of stringification of a macro argument. */ | |
3282 | ||
3283 | static void | |
3284 | test_lexer_string_locations_stringified_macro_argument | |
3285 | (const line_table_case &case_) | |
3286 | { | |
3287 | /* .....................000000000111111111122222222223. | |
3288 | .....................123456789012345678901234567890. */ | |
3289 | const char *content = ("#define MACRO(X) #X /* non-str */\n" | |
3290 | "MACRO(foo)\n"); | |
3291 | lexer_test test (case_, content, NULL); | |
3292 | ||
3293 | /* Verify that we get the expected token back. */ | |
3294 | const cpp_token *tok = test.get_token (); | |
3295 | ASSERT_EQ (tok->type, CPP_PADDING); | |
3296 | ||
3297 | tok = test.get_token (); | |
3298 | ASSERT_EQ (tok->type, CPP_STRING); | |
3299 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "\"foo\""); | |
3300 | ||
3301 | /* We don't support getting the location of a stringified macro | |
3302 | argument. Verify that it fails gracefully. */ | |
3303 | ASSERT_HAS_NO_SUBSTRING_RANGES (test, tok->src_loc, CPP_STRING, | |
3304 | "cpp_interpret_string_1 failed"); | |
3305 | ||
3306 | tok = test.get_token (); | |
3307 | ASSERT_EQ (tok->type, CPP_PADDING); | |
3308 | ||
3309 | tok = test.get_token (); | |
3310 | ASSERT_EQ (tok->type, CPP_PADDING); | |
3311 | } | |
3312 | ||
3313 | /* Ensure that we are fail gracefully if something attempts to pass | |
3314 | in a location that isn't a string literal token. Seen on this code: | |
3315 | ||
3316 | const char a[] = " %d "; | |
3317 | __builtin_printf (a, 0.5); | |
3318 | ^ | |
3319 | ||
3320 | when c-format.c erroneously used the indicated one-character | |
3321 | location as the format string location, leading to a read past the | |
3322 | end of a string buffer in cpp_interpret_string_1. */ | |
3323 | ||
3324 | static void | |
3325 | test_lexer_string_locations_non_string (const line_table_case &case_) | |
3326 | { | |
3327 | /* .....................000000000111111111122222222223. | |
3328 | .....................123456789012345678901234567890. */ | |
3329 | const char *content = (" a\n"); | |
3330 | lexer_test test (case_, content, NULL); | |
3331 | ||
3332 | /* Verify that we get the expected token back. */ | |
3333 | const cpp_token *tok = test.get_token (); | |
3334 | ASSERT_EQ (tok->type, CPP_NAME); | |
3335 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "a"); | |
3336 | ||
3337 | /* At this point, libcpp is attempting to interpret the name as a | |
3338 | string literal, despite it not starting with a quote. We don't detect | |
3339 | that, but we should at least fail gracefully. */ | |
3340 | ASSERT_HAS_NO_SUBSTRING_RANGES (test, tok->src_loc, CPP_STRING, | |
3341 | "cpp_interpret_string_1 failed"); | |
3342 | } | |
3343 | ||
3344 | /* Ensure that we can read substring information for a token which | |
3345 | starts in one linemap and ends in another . Adapted from | |
3346 | gcc.dg/cpp/pr69985.c. */ | |
3347 | ||
3348 | static void | |
3349 | test_lexer_string_locations_long_line (const line_table_case &case_) | |
3350 | { | |
3351 | /* .....................000000.000111111111 | |
3352 | .....................123456.789012346789. */ | |
3353 | const char *content = ("/* A very long line, so that we start a new line map. */\n" | |
3354 | " \"0123456789012345678901234567890123456789" | |
3355 | "0123456789012345678901234567890123456789" | |
3356 | "0123456789012345678901234567890123456789" | |
3357 | "0123456789\"\n"); | |
3358 | ||
3359 | lexer_test test (case_, content, NULL); | |
3360 | ||
3361 | /* Verify that we get the expected token back. */ | |
3362 | const cpp_token *tok = test.get_token (); | |
3363 | ASSERT_EQ (tok->type, CPP_STRING); | |
3364 | ||
3365 | if (!should_have_column_data_p (line_table->highest_location)) | |
3366 | return; | |
3367 | ||
3368 | /* Verify ranges of individual characters. */ | |
bbd6fcf3 DM |
3369 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, CPP_STRING, 131); |
3370 | for (int i = 0; i < 131; i++) | |
88fa5555 DM |
3371 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, |
3372 | i, 2, 7 + i, 7 + i); | |
3373 | } | |
3374 | ||
b8f56412 DM |
3375 | /* Test of locations within a raw string that doesn't contain a newline. */ |
3376 | ||
3377 | static void | |
3378 | test_lexer_string_locations_raw_string_one_line (const line_table_case &case_) | |
3379 | { | |
3380 | /* .....................00.0000000111111111122. | |
3381 | .....................12.3456789012345678901. */ | |
3382 | const char *content = ("R\"foo(0123456789)foo\"\n"); | |
3383 | lexer_test test (case_, content, NULL); | |
3384 | ||
3385 | /* Verify that we get the expected token back. */ | |
3386 | const cpp_token *tok = test.get_token (); | |
3387 | ASSERT_EQ (tok->type, CPP_STRING); | |
3388 | ||
3389 | /* Verify that cpp_interpret_string works. */ | |
3390 | cpp_string dst_string; | |
3391 | const enum cpp_ttype type = CPP_STRING; | |
3392 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
3393 | &dst_string, type); | |
3394 | ASSERT_TRUE (result); | |
3395 | ASSERT_STREQ ("0123456789", (const char *)dst_string.text); | |
3396 | free (const_cast <unsigned char *> (dst_string.text)); | |
3397 | ||
3398 | if (!should_have_column_data_p (line_table->highest_location)) | |
3399 | return; | |
3400 | ||
3401 | /* 0-9, plus the nil terminator. */ | |
3402 | ASSERT_NUM_SUBSTRING_RANGES (test, tok->src_loc, CPP_STRING, 11); | |
3403 | for (int i = 0; i < 11; i++) | |
3404 | ASSERT_CHAR_AT_RANGE (test, tok->src_loc, CPP_STRING, | |
3405 | i, 1, 7 + i, 7 + i); | |
3406 | } | |
3407 | ||
3408 | /* Test of locations within a raw string that contains a newline. */ | |
3409 | ||
3410 | static void | |
3411 | test_lexer_string_locations_raw_string_multiline (const line_table_case &case_) | |
3412 | { | |
3413 | /* .....................00.0000. | |
3414 | .....................12.3456. */ | |
3415 | const char *content = ("R\"foo(\n" | |
3416 | /* .....................00000. | |
3417 | .....................12345. */ | |
3418 | "hello\n" | |
3419 | "world\n" | |
3420 | /* .....................00000. | |
3421 | .....................12345. */ | |
3422 | ")foo\"\n"); | |
3423 | lexer_test test (case_, content, NULL); | |
3424 | ||
3425 | /* Verify that we get the expected token back. */ | |
3426 | const cpp_token *tok = test.get_token (); | |
3427 | ASSERT_EQ (tok->type, CPP_STRING); | |
3428 | ||
3429 | /* Verify that cpp_interpret_string works. */ | |
3430 | cpp_string dst_string; | |
3431 | const enum cpp_ttype type = CPP_STRING; | |
3432 | bool result = cpp_interpret_string (test.m_parser, &tok->val.str, 1, | |
3433 | &dst_string, type); | |
3434 | ASSERT_TRUE (result); | |
3435 | ASSERT_STREQ ("\nhello\nworld\n", (const char *)dst_string.text); | |
3436 | free (const_cast <unsigned char *> (dst_string.text)); | |
3437 | ||
3438 | if (!should_have_column_data_p (line_table->highest_location)) | |
3439 | return; | |
3440 | ||
3441 | /* Currently we don't support locations within raw strings that | |
3442 | contain newlines. */ | |
3443 | ASSERT_HAS_NO_SUBSTRING_RANGES (test, tok->src_loc, tok->type, | |
3444 | "range endpoints are on different lines"); | |
3445 | } | |
3446 | ||
a3998c2f DM |
3447 | /* Test of parsing an unterminated raw string. */ |
3448 | ||
3449 | static void | |
3450 | test_lexer_string_locations_raw_string_unterminated (const line_table_case &case_) | |
3451 | { | |
3452 | const char *content = "R\"ouch()ouCh\" /* etc */"; | |
3453 | ||
c24300ba DM |
3454 | lexer_diagnostic_sink diagnostics; |
3455 | lexer_test test (case_, content, &diagnostics); | |
a3998c2f DM |
3456 | test.m_implicitly_expect_EOF = false; |
3457 | ||
3458 | /* Attempt to parse the raw string. */ | |
3459 | const cpp_token *tok = test.get_token (); | |
3460 | ASSERT_EQ (tok->type, CPP_EOF); | |
3461 | ||
c24300ba | 3462 | ASSERT_EQ (1, diagnostics.m_diagnostics.length ()); |
a3998c2f DM |
3463 | /* We expect the message "unterminated raw string" |
3464 | in the "cpplib" translation domain. | |
3465 | It's not clear that dgettext is available on all supported hosts, | |
3466 | so this assertion is commented-out for now. | |
3467 | ASSERT_STREQ (dgettext ("cpplib", "unterminated raw string"), | |
c24300ba | 3468 | diagnostics.m_diagnostics[0]); |
a3998c2f DM |
3469 | */ |
3470 | } | |
3471 | ||
88fa5555 DM |
3472 | /* Test of lexing char constants. */ |
3473 | ||
3474 | static void | |
3475 | test_lexer_char_constants (const line_table_case &case_) | |
3476 | { | |
3477 | /* Various char constants. | |
3478 | .....................0000000001111111111.22222222223. | |
3479 | .....................1234567890123456789.01234567890. */ | |
3480 | const char *content = (" 'a'\n" | |
3481 | " u'a'\n" | |
3482 | " U'a'\n" | |
3483 | " L'a'\n" | |
3484 | " 'abc'\n"); | |
3485 | lexer_test test (case_, content, NULL); | |
3486 | ||
3487 | /* Verify that we get the expected tokens back. */ | |
3488 | /* 'a'. */ | |
3489 | const cpp_token *tok = test.get_token (); | |
3490 | ASSERT_EQ (tok->type, CPP_CHAR); | |
3491 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "'a'"); | |
3492 | ||
3493 | unsigned int chars_seen; | |
3494 | int unsignedp; | |
3495 | cppchar_t cc = cpp_interpret_charconst (test.m_parser, tok, | |
3496 | &chars_seen, &unsignedp); | |
3497 | ASSERT_EQ (cc, 'a'); | |
3498 | ASSERT_EQ (chars_seen, 1); | |
3499 | ||
3500 | /* u'a'. */ | |
3501 | tok = test.get_token (); | |
3502 | ASSERT_EQ (tok->type, CPP_CHAR16); | |
3503 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "u'a'"); | |
3504 | ||
3505 | /* U'a'. */ | |
3506 | tok = test.get_token (); | |
3507 | ASSERT_EQ (tok->type, CPP_CHAR32); | |
3508 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "U'a'"); | |
3509 | ||
3510 | /* L'a'. */ | |
3511 | tok = test.get_token (); | |
3512 | ASSERT_EQ (tok->type, CPP_WCHAR); | |
3513 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "L'a'"); | |
3514 | ||
3515 | /* 'abc' (c-char-sequence). */ | |
3516 | tok = test.get_token (); | |
3517 | ASSERT_EQ (tok->type, CPP_CHAR); | |
3518 | ASSERT_TOKEN_AS_TEXT_EQ (test.m_parser, tok, "'abc'"); | |
3519 | } | |
741d3be5 DM |
3520 | /* A table of interesting location_t values, giving one axis of our test |
3521 | matrix. */ | |
3522 | ||
3523 | static const location_t boundary_locations[] = { | |
3524 | /* Zero means "don't override the default values for a new line_table". */ | |
3525 | 0, | |
3526 | ||
3527 | /* An arbitrary non-zero value that isn't close to one of | |
3528 | the boundary values below. */ | |
3529 | 0x10000, | |
3530 | ||
3531 | /* Values near LINE_MAP_MAX_LOCATION_WITH_PACKED_RANGES. */ | |
3532 | LINE_MAP_MAX_LOCATION_WITH_PACKED_RANGES - 0x100, | |
3533 | LINE_MAP_MAX_LOCATION_WITH_PACKED_RANGES - 1, | |
3534 | LINE_MAP_MAX_LOCATION_WITH_PACKED_RANGES, | |
3535 | LINE_MAP_MAX_LOCATION_WITH_PACKED_RANGES + 1, | |
3536 | LINE_MAP_MAX_LOCATION_WITH_PACKED_RANGES + 0x100, | |
3537 | ||
3538 | /* Values near LINE_MAP_MAX_LOCATION_WITH_COLS. */ | |
3539 | LINE_MAP_MAX_LOCATION_WITH_COLS - 0x100, | |
3540 | LINE_MAP_MAX_LOCATION_WITH_COLS - 1, | |
3541 | LINE_MAP_MAX_LOCATION_WITH_COLS, | |
3542 | LINE_MAP_MAX_LOCATION_WITH_COLS + 1, | |
3543 | LINE_MAP_MAX_LOCATION_WITH_COLS + 0x100, | |
3544 | }; | |
3545 | ||
f87e22c5 | 3546 | /* Run TESTCASE multiple times, once for each case in our test matrix. */ |
d9b950dd DM |
3547 | |
3548 | void | |
f87e22c5 | 3549 | for_each_line_table_case (void (*testcase) (const line_table_case &)) |
d9b950dd | 3550 | { |
741d3be5 DM |
3551 | /* As noted above in the description of struct line_table_case, |
3552 | we want to explore a test matrix of interesting line_table | |
3553 | situations, running various selftests for each case within the | |
3554 | matrix. */ | |
3555 | ||
3556 | /* Run all tests with: | |
3557 | (a) line_table->default_range_bits == 0, and | |
3558 | (b) line_table->default_range_bits == 5. */ | |
3559 | int num_cases_tested = 0; | |
3560 | for (int default_range_bits = 0; default_range_bits <= 5; | |
3561 | default_range_bits += 5) | |
3562 | { | |
3563 | /* ...and use each of the "interesting" location values as | |
3564 | the starting location within line_table. */ | |
3565 | const int num_boundary_locations | |
3566 | = sizeof (boundary_locations) / sizeof (boundary_locations[0]); | |
3567 | for (int loc_idx = 0; loc_idx < num_boundary_locations; loc_idx++) | |
3568 | { | |
3569 | line_table_case c (default_range_bits, boundary_locations[loc_idx]); | |
3570 | ||
f87e22c5 | 3571 | testcase (c); |
741d3be5 DM |
3572 | |
3573 | num_cases_tested++; | |
3574 | } | |
3575 | } | |
3576 | ||
3577 | /* Verify that we fully covered the test matrix. */ | |
3578 | ASSERT_EQ (num_cases_tested, 2 * 12); | |
f87e22c5 DM |
3579 | } |
3580 | ||
a4553534 DM |
3581 | /* Verify that when presented with a consecutive pair of locations with |
3582 | a very large line offset, we don't attempt to consolidate them into | |
3583 | a single ordinary linemap where the line offsets within the line map | |
3584 | would lead to overflow (PR lto/88147). */ | |
3585 | ||
3586 | static void | |
3587 | test_line_offset_overflow () | |
3588 | { | |
3589 | line_table_test ltt (line_table_case (5, 0)); | |
3590 | ||
3591 | linemap_add (line_table, LC_ENTER, false, "foo.c", 0); | |
3592 | linemap_line_start (line_table, 1, 100); | |
3593 | location_t loc_a = linemap_line_start (line_table, 2578, 255); | |
3594 | assert_loceq ("foo.c", 2578, 0, loc_a); | |
3595 | ||
3596 | const line_map_ordinary *ordmap_a = LINEMAPS_LAST_ORDINARY_MAP (line_table); | |
3597 | ASSERT_EQ (ordmap_a->m_column_and_range_bits, 13); | |
3598 | ASSERT_EQ (ordmap_a->m_range_bits, 5); | |
3599 | ||
3600 | location_t loc_b = linemap_line_start (line_table, 404198, 512); | |
3601 | assert_loceq ("foo.c", 404198, 0, loc_b); | |
3602 | ||
3603 | /* We should have started a new linemap, rather than attempting to store | |
3604 | a very large line offset. */ | |
3605 | const line_map_ordinary *ordmap_b = LINEMAPS_LAST_ORDINARY_MAP (line_table); | |
3606 | ASSERT_NE (ordmap_a, ordmap_b); | |
3607 | } | |
3608 | ||
ee925640 LH |
3609 | void test_cpp_utf8 () |
3610 | { | |
3611 | /* Verify that wcwidth of invalid UTF-8 or control bytes is 1. */ | |
3612 | { | |
3613 | int w_bad = cpp_display_width ("\xf0!\x9f!\x98!\x82!", 8); | |
3614 | ASSERT_EQ (8, w_bad); | |
3615 | int w_ctrl = cpp_display_width ("\r\t\n\v\0\1", 6); | |
3616 | ASSERT_EQ (6, w_ctrl); | |
3617 | } | |
3618 | ||
3619 | /* Verify that wcwidth of valid UTF-8 is as expected. */ | |
3620 | { | |
3621 | const int w_pi = cpp_display_width ("\xcf\x80", 2); | |
3622 | ASSERT_EQ (1, w_pi); | |
3623 | const int w_emoji = cpp_display_width ("\xf0\x9f\x98\x82", 4); | |
3624 | ASSERT_EQ (2, w_emoji); | |
3625 | const int w_umlaut_precomposed = cpp_display_width ("\xc3\xbf", 2); | |
3626 | ASSERT_EQ (1, w_umlaut_precomposed); | |
3627 | const int w_umlaut_combining = cpp_display_width ("y\xcc\x88", 3); | |
3628 | ASSERT_EQ (1, w_umlaut_combining); | |
3629 | const int w_han = cpp_display_width ("\xe4\xb8\xba", 3); | |
3630 | ASSERT_EQ (2, w_han); | |
3631 | const int w_ascii = cpp_display_width ("GCC", 3); | |
3632 | ASSERT_EQ (3, w_ascii); | |
3633 | const int w_mixed = cpp_display_width ("\xcf\x80 = 3.14 \xf0\x9f\x98\x82" | |
3634 | "\x9f! \xe4\xb8\xba y\xcc\x88", 24); | |
3635 | ASSERT_EQ (18, w_mixed); | |
3636 | } | |
3637 | ||
3638 | /* Verify that cpp_byte_column_to_display_column can go past the end, | |
3639 | and similar edge cases. */ | |
3640 | { | |
3641 | const char *str | |
3642 | /* Display columns. | |
3643 | 111111112345 */ | |
3644 | = "\xcf\x80 abc"; | |
3645 | /* 111122223456 | |
3646 | Byte columns. */ | |
3647 | ||
3648 | ASSERT_EQ (5, cpp_display_width (str, 6)); | |
3649 | ASSERT_EQ (105, cpp_byte_column_to_display_column (str, 6, 106)); | |
3650 | ASSERT_EQ (10000, cpp_byte_column_to_display_column (NULL, 0, 10000)); | |
3651 | ASSERT_EQ (0, cpp_byte_column_to_display_column (NULL, 10000, 0)); | |
3652 | } | |
3653 | ||
3654 | /* Verify that cpp_display_column_to_byte_column can go past the end, | |
3655 | and similar edge cases, and check invertibility. */ | |
3656 | { | |
3657 | const char *str | |
3658 | /* Display columns. | |
3659 | 000000000000000000000000000000000000011 | |
3660 | 111111112222222234444444455555555678901 */ | |
3661 | = "\xf0\x9f\x98\x82 \xf0\x9f\x98\x82 hello"; | |
3662 | /* 000000000000000000000000000000000111111 | |
3663 | 111122223333444456666777788889999012345 | |
3664 | Byte columns. */ | |
3665 | ASSERT_EQ (4, cpp_display_column_to_byte_column (str, 15, 2)); | |
3666 | ASSERT_EQ (15, cpp_display_column_to_byte_column (str, 15, 11)); | |
3667 | ASSERT_EQ (115, cpp_display_column_to_byte_column (str, 15, 111)); | |
3668 | ASSERT_EQ (10000, cpp_display_column_to_byte_column (NULL, 0, 10000)); | |
3669 | ASSERT_EQ (0, cpp_display_column_to_byte_column (NULL, 10000, 0)); | |
3670 | ||
3671 | /* Verify that we do not interrupt a UTF-8 sequence. */ | |
3672 | ASSERT_EQ (4, cpp_display_column_to_byte_column (str, 15, 1)); | |
3673 | ||
3674 | for (int byte_col = 1; byte_col <= 15; ++byte_col) | |
3675 | { | |
3676 | const int disp_col = cpp_byte_column_to_display_column (str, 15, | |
3677 | byte_col); | |
3678 | const int byte_col2 = cpp_display_column_to_byte_column (str, 15, | |
3679 | disp_col); | |
3680 | ||
3681 | /* If we ask for the display column in the middle of a UTF-8 | |
3682 | sequence, it will return the length of the partial sequence, | |
3683 | matching the behavior of GCC before display column support. | |
3684 | Otherwise check the round trip was successful. */ | |
3685 | if (byte_col < 4) | |
3686 | ASSERT_EQ (byte_col, disp_col); | |
3687 | else if (byte_col >= 6 && byte_col < 9) | |
3688 | ASSERT_EQ (3 + (byte_col - 5), disp_col); | |
3689 | else | |
3690 | ASSERT_EQ (byte_col2, byte_col); | |
3691 | } | |
3692 | } | |
3693 | ||
3694 | } | |
3695 | ||
f87e22c5 DM |
3696 | /* Run all of the selftests within this file. */ |
3697 | ||
3698 | void | |
3699 | input_c_tests () | |
3700 | { | |
082284da | 3701 | test_linenum_comparisons (); |
f87e22c5 DM |
3702 | test_should_have_column_data_p (); |
3703 | test_unknown_location (); | |
3704 | test_builtins (); | |
9144eabb | 3705 | for_each_line_table_case (test_make_location_nonpure_range_endpoints); |
f87e22c5 DM |
3706 | |
3707 | for_each_line_table_case (test_accessing_ordinary_linemaps); | |
3708 | for_each_line_table_case (test_lexer); | |
3709 | for_each_line_table_case (test_lexer_string_locations_simple); | |
3710 | for_each_line_table_case (test_lexer_string_locations_ebcdic); | |
3711 | for_each_line_table_case (test_lexer_string_locations_hex); | |
3712 | for_each_line_table_case (test_lexer_string_locations_oct); | |
3713 | for_each_line_table_case (test_lexer_string_locations_letter_escape_1); | |
3714 | for_each_line_table_case (test_lexer_string_locations_letter_escape_2); | |
3715 | for_each_line_table_case (test_lexer_string_locations_ucn4); | |
3716 | for_each_line_table_case (test_lexer_string_locations_ucn8); | |
3717 | for_each_line_table_case (test_lexer_string_locations_wide_string); | |
3718 | for_each_line_table_case (test_lexer_string_locations_string16); | |
3719 | for_each_line_table_case (test_lexer_string_locations_string32); | |
3720 | for_each_line_table_case (test_lexer_string_locations_u8); | |
3721 | for_each_line_table_case (test_lexer_string_locations_utf8_source); | |
3722 | for_each_line_table_case (test_lexer_string_locations_concatenation_1); | |
3723 | for_each_line_table_case (test_lexer_string_locations_concatenation_2); | |
3724 | for_each_line_table_case (test_lexer_string_locations_concatenation_3); | |
3725 | for_each_line_table_case (test_lexer_string_locations_macro); | |
3726 | for_each_line_table_case (test_lexer_string_locations_stringified_macro_argument); | |
3727 | for_each_line_table_case (test_lexer_string_locations_non_string); | |
3728 | for_each_line_table_case (test_lexer_string_locations_long_line); | |
b8f56412 DM |
3729 | for_each_line_table_case (test_lexer_string_locations_raw_string_one_line); |
3730 | for_each_line_table_case (test_lexer_string_locations_raw_string_multiline); | |
a3998c2f | 3731 | for_each_line_table_case (test_lexer_string_locations_raw_string_unterminated); |
f87e22c5 | 3732 | for_each_line_table_case (test_lexer_char_constants); |
741d3be5 | 3733 | |
d9b950dd | 3734 | test_reading_source_line (); |
a4553534 DM |
3735 | |
3736 | test_line_offset_overflow (); | |
ee925640 LH |
3737 | |
3738 | test_cpp_utf8 (); | |
d9b950dd DM |
3739 | } |
3740 | ||
3741 | } // namespace selftest | |
3742 | ||
3743 | #endif /* CHECKING_P */ |