1 | /* Copyright (C) 2006 MySQL AB |
2 | |
3 | This program is free software; you can redistribute it and/or modify |
4 | it under the terms of the GNU General Public License as published by |
5 | the Free Software Foundation; version 2 of the License. |
6 | |
7 | This program is distributed in the hope that it will be useful, |
8 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
9 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
10 | GNU General Public License for more details. |
11 | |
12 | You should have received a copy of the GNU General Public License |
13 | along with this program; if not, write to the Free Software |
14 | Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, |
15 | MA 02111-1301, USA |
16 | |
17 | Library for providing TAP support for testing C and C++ was written |
18 | by Mats Kindahl <mats@mysql.com>. |
19 | */ |
20 | |
21 | #include "tap.h" |
22 | |
23 | #include "ma_global.h" |
24 | |
25 | #include <stdlib.h> |
26 | #include <stdarg.h> |
27 | #include <stdio.h> |
28 | #include <string.h> |
29 | #include <signal.h> |
30 | |
31 | /* |
32 | Visual Studio 2003 does not know vsnprintf but knows _vsnprintf. |
33 | We don't put this #define in config-win.h because we prefer |
34 | ma_vsnprintf everywhere instead, except when linking with libmysys |
35 | is not desirable - the case here. |
36 | */ |
37 | #if defined(_MSC_VER) && ( _MSC_VER == 1310 ) |
38 | #define vsnprintf _vsnprintf |
39 | #endif |
40 | |
41 | /** |
42 | @defgroup MyTAP_Internal MyTAP Internals |
43 | |
44 | Internal functions and data structures for the MyTAP implementation. |
45 | */ |
46 | |
47 | /** |
48 | Test data structure. |
49 | |
50 | Data structure containing all information about the test suite. |
51 | |
52 | @ingroup MyTAP_Internal |
53 | */ |
54 | static TEST_DATA g_test = { 0, 0, 0, "" }; |
55 | |
56 | /** |
57 | Output stream for test report message. |
58 | |
59 | The macro is just a temporary solution. |
60 | |
61 | @ingroup MyTAP_Internal |
62 | */ |
63 | #define tapout stdout |
64 | |
65 | /** |
66 | Emit the beginning of a test line, that is: "(not) ok", test number, |
67 | and description. |
68 | |
69 | To emit the directive, use the emit_dir() function |
70 | |
71 | @ingroup MyTAP_Internal |
72 | |
73 | @see emit_dir |
74 | |
75 | @param pass 'true' if test passed, 'false' otherwise |
76 | @param fmt Description of test in printf() format. |
77 | @param ap Vararg list for the description string above. |
78 | */ |
79 | static void |
80 | vemit_tap(int pass, char const *fmt, va_list ap) |
81 | { |
82 | fprintf(tapout, "%sok %d%s" , |
83 | pass ? "" : "not " , |
84 | ++g_test.last, |
85 | (fmt && *fmt) ? " - " : "" ); |
86 | if (fmt && *fmt) |
87 | vfprintf(tapout, fmt, ap); |
88 | } |
89 | |
90 | |
91 | /** |
92 | Emit a TAP directive. |
93 | |
94 | TAP directives are comments after that have the form: |
95 | |
96 | @code |
97 | ok 1 # skip reason for skipping |
98 | not ok 2 # todo some text explaining what remains |
99 | @endcode |
100 | |
101 | @ingroup MyTAP_Internal |
102 | |
103 | @param dir Directive as a string |
104 | @param why Explanation string |
105 | */ |
106 | static void |
107 | emit_dir(const char *dir, const char *why) |
108 | { |
109 | fprintf(tapout, " # %s %s" , dir, why); |
110 | } |
111 | |
112 | |
113 | /** |
114 | Emit a newline to the TAP output stream. |
115 | |
116 | @ingroup MyTAP_Internal |
117 | */ |
118 | static void |
119 | emit_endl() |
120 | { |
121 | fprintf(tapout, "\n" ); |
122 | } |
123 | |
124 | static void |
125 | handle_core_signal(int signo) |
126 | { |
127 | BAIL_OUT("Signal %d thrown" , signo); |
128 | } |
129 | |
130 | void |
131 | BAIL_OUT(char const *fmt, ...) |
132 | { |
133 | va_list ap; |
134 | va_start(ap, fmt); |
135 | fprintf(tapout, "Bail out! " ); |
136 | vfprintf(tapout, fmt, ap); |
137 | emit_endl(); |
138 | va_end(ap); |
139 | exit(255); |
140 | } |
141 | |
142 | |
143 | void |
144 | diag(char const *fmt, ...) |
145 | { |
146 | va_list ap; |
147 | va_start(ap, fmt); |
148 | fprintf(tapout, "# " ); |
149 | vfprintf(tapout, fmt, ap); |
150 | emit_endl(); |
151 | va_end(ap); |
152 | } |
153 | |
154 | typedef struct signal_entry { |
155 | int signo; |
156 | void (*handler)(int); |
157 | } signal_entry; |
158 | |
159 | static signal_entry install_signal[]= { |
160 | #ifdef SIGQUIT |
161 | { SIGQUIT, handle_core_signal }, |
162 | #endif |
163 | { SIGILL, handle_core_signal }, |
164 | { SIGABRT, handle_core_signal }, |
165 | { SIGFPE, handle_core_signal }, |
166 | { SIGSEGV, handle_core_signal } |
167 | #ifdef SIGBUS |
168 | , { SIGBUS, handle_core_signal } |
169 | #endif |
170 | #ifdef SIGXCPU |
171 | , { SIGXCPU, handle_core_signal } |
172 | #endif |
173 | #ifdef SIGXCPU |
174 | , { SIGXFSZ, handle_core_signal } |
175 | #endif |
176 | #ifdef SIGXCPU |
177 | , { SIGSYS, handle_core_signal } |
178 | #endif |
179 | #ifdef SIGXCPU |
180 | , { SIGTRAP, handle_core_signal } |
181 | #endif |
182 | }; |
183 | |
184 | int skip_big_tests= 1; |
185 | |
186 | void |
187 | plan(int const count) |
188 | { |
189 | char *config= getenv("MYTAP_CONFIG" ); |
190 | size_t i; |
191 | |
192 | if (config) |
193 | skip_big_tests= strcmp(config, "big" ); |
194 | |
195 | setvbuf(tapout, 0, _IONBF, 0); /* provide output at once */ |
196 | /* |
197 | Install signal handler |
198 | */ |
199 | |
200 | for (i= 0; i < sizeof(install_signal)/sizeof(*install_signal); ++i) |
201 | signal(install_signal[i].signo, install_signal[i].handler); |
202 | |
203 | g_test.plan= count; |
204 | switch (count) |
205 | { |
206 | case NO_PLAN: |
207 | break; |
208 | default: |
209 | if (count > 0) |
210 | fprintf(tapout, "1..%d\n" , count); |
211 | break; |
212 | } |
213 | } |
214 | |
215 | |
216 | void |
217 | skip_all(char const *reason, ...) |
218 | { |
219 | va_list ap; |
220 | va_start(ap, reason); |
221 | fprintf(tapout, "1..0 # skip " ); |
222 | vfprintf(tapout, reason, ap); |
223 | va_end(ap); |
224 | exit(0); |
225 | } |
226 | |
227 | void |
228 | ok(int const pass, char const *fmt, ...) |
229 | { |
230 | va_list ap; |
231 | va_start(ap, fmt); |
232 | |
233 | if (!pass && *g_test.todo == '\0') |
234 | ++g_test.failed; |
235 | |
236 | vemit_tap(pass, fmt, ap); |
237 | va_end(ap); |
238 | if (*g_test.todo != '\0') |
239 | emit_dir("todo" , g_test.todo); |
240 | emit_endl(); |
241 | } |
242 | |
243 | |
244 | void |
245 | skip(int how_many, char const *const fmt, ...) |
246 | { |
247 | char reason[80]; |
248 | if (fmt && *fmt) |
249 | { |
250 | va_list ap; |
251 | va_start(ap, fmt); |
252 | vsnprintf(reason, sizeof(reason), fmt, ap); |
253 | va_end(ap); |
254 | } |
255 | else |
256 | reason[0] = '\0'; |
257 | |
258 | while (how_many-- > 0) |
259 | { |
260 | va_list ap; |
261 | memset((char*) &ap, 0, sizeof(ap)); /* Keep compiler happy */ |
262 | vemit_tap(1, NULL, ap); |
263 | emit_dir("skip" , reason); |
264 | emit_endl(); |
265 | } |
266 | } |
267 | |
268 | void |
269 | todo_start(char const *message, ...) |
270 | { |
271 | va_list ap; |
272 | va_start(ap, message); |
273 | vsnprintf(g_test.todo, sizeof(g_test.todo), message, ap); |
274 | va_end(ap); |
275 | } |
276 | |
277 | void |
278 | todo_end() |
279 | { |
280 | *g_test.todo = '\0'; |
281 | } |
282 | |
283 | int exit_status() { |
284 | /* |
285 | If there were no plan, we write one last instead. |
286 | */ |
287 | if (g_test.plan == NO_PLAN) |
288 | plan(g_test.last); |
289 | |
290 | if (g_test.plan != g_test.last) |
291 | { |
292 | diag("%d tests planned but%s %d executed" , |
293 | g_test.plan, (g_test.plan > g_test.last ? " only" : "" ), g_test.last); |
294 | return EXIT_FAILURE; |
295 | } |
296 | |
297 | if (g_test.failed > 0) |
298 | { |
299 | diag("Failed %d tests!" , g_test.failed); |
300 | return EXIT_FAILURE; |
301 | } |
302 | |
303 | return EXIT_SUCCESS; |
304 | } |
305 | |
306 | /** |
307 | @mainpage Testing C and C++ using MyTAP |
308 | |
309 | @section IntroSec Introduction |
310 | |
311 | Unit tests are used to test individual components of a system. In |
312 | contrast, functional tests usually test the entire system. The |
313 | rationale is that each component should be correct if the system is |
314 | to be correct. Unit tests are usually small pieces of code that |
315 | tests an individual function, class, a module, or other unit of the |
316 | code. |
317 | |
318 | Observe that a correctly functioning system can be built from |
319 | "faulty" components. The problem with this approach is that as the |
320 | system evolves, the bugs surface in unexpected ways, making |
321 | maintenance harder. |
322 | |
323 | The advantages of using unit tests to test components of the system |
324 | are several: |
325 | |
326 | - The unit tests can make a more thorough testing than the |
327 | functional tests by testing correctness even for pathological use |
328 | (which shouldn't be present in the system). This increases the |
329 | overall robustness of the system and makes maintenance easier. |
330 | |
331 | - It is easier and faster to find problems with a malfunctioning |
332 | component than to find problems in a malfunctioning system. This |
333 | shortens the compile-run-edit cycle and therefore improves the |
334 | overall performance of development. |
335 | |
336 | - The component has to support at least two uses: in the system and |
337 | in a unit test. This leads to more generic and stable interfaces |
338 | and in addition promotes the development of reusable components. |
339 | |
340 | For example, the following are typical functional tests: |
341 | - Does transactions work according to specifications? |
342 | - Can we connect a client to the server and execute statements? |
343 | |
344 | In contrast, the following are typical unit tests: |
345 | |
346 | - Can the 'String' class handle a specified list of character sets? |
347 | - Does all operations for 'my_bitmap' produce the correct result? |
348 | - Does all the NIST test vectors for the AES implementation encrypt |
349 | correctly? |
350 | |
351 | |
352 | @section UnitTest Writing unit tests |
353 | |
354 | The purpose of writing unit tests is to use them to drive component |
355 | development towards a solution that passes the tests. This means that the |
356 | unit tests has to be as complete as possible, testing at least: |
357 | |
358 | - Normal input |
359 | - Borderline cases |
360 | - Faulty input |
361 | - Error handling |
362 | - Bad environment |
363 | |
364 | @subsection NormalSubSec Normal input |
365 | |
366 | This is to test that the component have the expected behaviour. |
367 | This is just plain simple: test that it works. For example, test |
368 | that you can unpack what you packed, adding gives the sum, pincing |
369 | the duck makes it quack. |
370 | |
371 | This is what everybody does when they write tests. |
372 | |
373 | |
374 | @subsection BorderlineTests Borderline cases |
375 | |
376 | If you have a size anywhere for your component, does it work for |
377 | size 1? Size 0? Sizes close to <code>UINT_MAX</code>? |
378 | |
379 | It might not be sensible to have a size 0, so in this case it is |
380 | not a borderline case, but rather a faulty input (see @ref |
381 | FaultyInputTests). |
382 | |
383 | |
384 | @subsection FaultyInputTests Faulty input |
385 | |
386 | Does your bitmap handle 0 bits size? Well, it might not be designed |
387 | for it, but is should <em>not</em> crash the application, but |
388 | rather produce an error. This is called defensive programming. |
389 | |
390 | Unfortunately, adding checks for values that should just not be |
391 | entered at all is not always practical: the checks cost cycles and |
392 | might cost more than it's worth. For example, some functions are |
393 | designed so that you may not give it a null pointer. In those |
394 | cases it's not sensible to pass it <code>NULL</code> just to see it |
395 | crash. |
396 | |
397 | Since every experienced programmer add an <code>assert()</code> to |
398 | ensure that you get a proper failure for the debug builds when a |
399 | null pointer passed (you add asserts too, right?), you will in this |
400 | case instead have a controlled (early) crash in the debug build. |
401 | |
402 | |
403 | @subsection ErrorHandlingTests Error handling |
404 | |
405 | This is testing that the errors your component is designed to give |
406 | actually are produced. For example, testing that trying to open a |
407 | non-existing file produces a sensible error code. |
408 | |
409 | |
410 | @subsection BadEnvironmentTests Environment |
411 | |
412 | Sometimes, modules has to behave well even when the environment |
413 | fails to work correctly. Typical examples are when the computer is |
414 | out of dynamic memory or when the disk is full. You can emulate |
415 | this by replacing, e.g., <code>malloc()</code> with your own |
416 | version that will work for a while, but then fail. Some things are |
417 | worth to keep in mind here: |
418 | |
419 | - Make sure to make the function fail deterministically, so that |
420 | you really can repeat the test. |
421 | |
422 | - Make sure that it doesn't just fail immediately. The unit might |
423 | have checks for the first case, but might actually fail some time |
424 | in the near future. |
425 | |
426 | |
427 | @section UnitTest How to structure a unit test |
428 | |
429 | In this section we will give some advice on how to structure the |
430 | unit tests to make the development run smoothly. The basic |
431 | structure of a test is: |
432 | |
433 | - Plan |
434 | - Test |
435 | - Report |
436 | |
437 | |
438 | @subsection TestPlanning Plan the test |
439 | |
440 | Planning the test means telling how many tests there are. In the |
441 | event that one of the tests causes a crash, it is then possible to |
442 | see that there are fewer tests than expected, and print a proper |
443 | error message. |
444 | |
445 | To plan a test, use the @c plan() function in the following manner: |
446 | |
447 | @code |
448 | int main(int argc, char *argv[]) |
449 | { |
450 | plan(5); |
451 | . |
452 | . |
453 | . |
454 | } |
455 | @endcode |
456 | |
457 | If you don't call the @c plan() function, the number of tests |
458 | executed will be printed at the end. This is intended to be used |
459 | while developing the unit and you are constantly adding tests. It |
460 | is not indented to be used after the unit has been released. |
461 | |
462 | |
463 | @subsection TestRunning Execute the test |
464 | |
465 | To report the status of a test, the @c ok() function is used in the |
466 | following manner: |
467 | |
468 | @code |
469 | int main(int argc, char *argv[]) |
470 | { |
471 | plan(5); |
472 | ok(ducks == paddling_ducks, |
473 | "%d ducks did not paddle", ducks - paddling_ducks); |
474 | . |
475 | . |
476 | . |
477 | } |
478 | @endcode |
479 | |
480 | This will print a test result line on the standard output in TAP |
481 | format, which allows TAP handling frameworks (like Test::Harness) |
482 | to parse the status of the test. |
483 | |
484 | @subsection TestReport Report the result of the test |
485 | |
486 | At the end, a complete test report should be written, with some |
487 | statistics. If the test returns EXIT_SUCCESS, all tests were |
488 | successful, otherwise at least one test failed. |
489 | |
490 | To get a TAP compliant output and exit status, report the exit |
491 | status in the following manner: |
492 | |
493 | @code |
494 | int main(int argc, char *argv[]) |
495 | { |
496 | plan(5); |
497 | ok(ducks == paddling_ducks, |
498 | "%d ducks did not paddle", ducks - paddling_ducks); |
499 | . |
500 | . |
501 | . |
502 | return exit_status(); |
503 | } |
504 | @endcode |
505 | |
506 | @section DontDoThis Ways to not do unit testing |
507 | |
508 | In this section, we'll go through some quite common ways to write |
509 | tests that are <em>not</em> a good idea. |
510 | |
511 | @subsection BreadthFirstTests Doing breadth-first testing |
512 | |
513 | If you're writing a library with several functions, don't test all |
514 | functions using size 1, then all functions using size 2, etc. If a |
515 | test for size 42 fails, you have no easy way of tracking down why |
516 | it failed. |
517 | |
518 | It is better to concentrate on getting one function to work at a |
519 | time, which means that you test each function for all sizes that |
520 | you think is reasonable. Then you continue with the next function, |
521 | doing the same. This is usually also the way that a library is |
522 | developed (one function at a time) so stick to testing that is |
523 | appropriate for now the unit is developed. |
524 | |
525 | @subsection JustToBeSafeTest Writing unnecessarily large tests |
526 | |
527 | Don't write tests that use parameters in the range 1-1024 unless |
528 | you have a very good reason to believe that the component will |
529 | succeed for 562 but fail for 564 (the numbers picked are just |
530 | examples). |
531 | |
532 | It is very common to write extensive tests "just to be safe." |
533 | Having a test suite with a lot of values might give you a warm |
534 | fuzzy feeling, but it doesn't really help you find the bugs. Good |
535 | tests fail; seriously, if you write a test that you expect to |
536 | succeed, you don't need to write it. If you think that it |
537 | <em>might</em> fail, <em>then</em> you should write it. |
538 | |
539 | Don't take this as an excuse to avoid writing any tests at all |
540 | "since I make no mistakes" (when it comes to this, there are two |
541 | kinds of people: those who admit they make mistakes, and those who |
542 | don't); rather, this means that there is no reason to test that |
543 | using a buffer with size 100 works when you have a test for buffer |
544 | size 96. |
545 | |
546 | The drawback is that the test suite takes longer to run, for little |
547 | or no benefit. It is acceptable to do a exhaustive test if it |
548 | doesn't take too long to run and it is quite common to do an |
549 | exhaustive test of a function for a small set of values. |
550 | Use your judgment to decide what is excessive: your milage may |
551 | vary. |
552 | */ |
553 | |
554 | /** |
555 | @example simple.t.c |
556 | |
557 | This is an simple example of how to write a test using the |
558 | library. The output of this program is: |
559 | |
560 | @code |
561 | 1..1 |
562 | # Testing basic functions |
563 | ok 1 - Testing gcs() |
564 | @endcode |
565 | |
566 | The basic structure is: plan the number of test points using the |
567 | plan() function, perform the test and write out the result of each |
568 | test point using the ok() function, print out a diagnostics message |
569 | using diag(), and report the result of the test by calling the |
570 | exit_status() function. Observe that this test does excessive |
571 | testing (see @ref JustToBeSafeTest), but the test point doesn't |
572 | take very long time. |
573 | */ |
574 | |
575 | /** |
576 | @example todo.t.c |
577 | |
578 | This example demonstrates how to use the <code>todo_start()</code> |
579 | and <code>todo_end()</code> function to mark a sequence of tests to |
580 | be done. Observe that the tests are assumed to fail: if any test |
581 | succeeds, it is considered a "bonus". |
582 | */ |
583 | |
584 | /** |
585 | @example skip.t.c |
586 | |
587 | This is an example of how the <code>SKIP_BLOCK_IF</code> can be |
588 | used to skip a predetermined number of tests. Observe that the |
589 | macro actually skips the following statement, but it's not sensible |
590 | to use anything than a block. |
591 | */ |
592 | |
593 | /** |
594 | @example skip_all.t.c |
595 | |
596 | Sometimes, you skip an entire test because it's testing a feature |
597 | that doesn't exist on the system that you're testing. To skip an |
598 | entire test, use the <code>skip_all()</code> function according to |
599 | this example. |
600 | */ |
601 | |