1Notmuch test suite
2==================
3This directory contains the test suite for notmuch.
4
5When fixing bugs or enhancing notmuch, you are strongly encouraged to
6add tests in this directory to cover what you are trying to fix or
7enhance.
8
9Prerequisites
10-------------
11The test system itself requires:
12
13 - bash(1) version 4.0 or newer
14
15Without bash 4.0+ the tests just refuse to run.
16
17Some tests require external dependencies to run. Without them, they
18will be skipped, or (rarely) marked failed. Please install these, so
19that you know if you break anything.
20
21 - GNU tar(1)
22 - dtach(1)
23 - emacs(1)
24 - emacsclient(1)
25 - gdb(1)
26 - gpg(1)
27 - python(1)
28 - xapian-metadata(1)
29
30If your system lacks these tools or have older, non-upgradable versions
31of these, please (possibly compile and) install these to some other
32path, for example /usr/local/bin or /opt/gnu/bin. Then prepend the
33chosen directory to your PATH before running the tests.
34
35e.g. env PATH=/opt/gnu/bin:$PATH make test
36
37For FreeBSD you need to install latest gdb from ports or packages and
38provide path to it in TEST_GDB environment variable before executing
39the tests, native FreeBSD gdb does not not work. If you install
40coreutils, which provides GNU versions of basic utils like 'date' and
41'base64' on FreeBSD, the test suite will use these instead of the
42native ones. This provides robustness against portability issues with
43these system tools. Most often the tests are written, reviewed and
44tested on Linux system so such portability issues arise from time to
45time.
46
47Running Tests
48-------------
49The easiest way to run tests is to say "make test", (or simply run the
50notmuch-test script). Either command will run all available tests.
51
52Alternately, you can run a specific subset of tests by simply invoking
53one of the executable scripts in this directory, (such as ./T*-search.sh,
54./T*-reply.sh, etc). Note that you will probably want "make test-binaries"
55before running individual tests.
56
57The following command-line options are available when running tests:
58
59--debug::
60 This may help the person who is developing a new test.
61 It causes the command defined with test_debug to run.
62
63--immediate::
64 This causes the test to immediately exit upon the first
65 failed test.
66
67--valgrind::
68 Execute notmuch with valgrind and exit with status
69 126 on errors (just like regular tests, this will only stop
70 the test script when running under -i). Valgrind errors
71 go to stderr, so you might want to pass the -v option, too.
72
73 Since it makes no sense to run the tests with --valgrind and
74 not see any output, this option implies --verbose. For
75 convenience, it also implies --tee.
76
77--tee::
78 In addition to printing the test output to the terminal,
79 write it to files named 't/test-results/$TEST_NAME.out'.
80 As the names depend on the tests' file names, it is safe to
81 run the tests with this option in parallel.
82
83When invoking the test suite via "make test" any of the above options
84can be specified as follows:
85
86 make test OPTIONS="--verbose"
87
88You can choose an emacs binary (and corresponding emacsclient) to run
89the tests in one of the following ways.
90
91 TEST_EMACS=my-emacs TEST_EMACSCLIENT=my-emacsclient make test
92 TEST_EMACS=my-emacs TEST_EMACSCLIENT=my-emacsclient ./T*-emacs.sh
93 make test TEST_EMACS=my-emacs TEST_EMACSCLIENT=my-emacsclient
94
95Some tests may require a c compiler. You can choose the name and flags similarly
96to with emacs, e.g.
97
98 make test TEST_CC=gcc TEST_CFLAGS="-g -O2"
99
100Parallel Execution
101------------------
102If either the moreutils or GNU "parallel" utility is available all
103tests will be run in parallel. If the NOTMUCH_TEST_SERIALIZE variable
104is non-null all tests will be executed sequentially.
105
106Quiet Execution
107---------------
108Normally, when new script starts and when test PASSes you get a message
109printed on screen. This printing can be disabled by setting the
110NOTMUCH_TEST_QUIET variable to a non-null value. Message on test
111failures and skips are still printed.
112
113Skipping Tests
114--------------
115If, for any reason, you need to skip one or more tests, you can do so
116by setting the NOTMUCH_SKIP_TESTS variable to the name of one or more
117sections of tests.
118
119For example:
120
121 $ NOTMUCH_SKIP_TESTS="search reply" make test
122
123Even more fine-grained skipping is possible by appending a test number
124(or glob pattern) after the section name. For example, the first
125search test and the second reply test could be skipped with:
126
127 $ NOTMUCH_SKIP_TESTS="search.1 reply.2" make test
128
129Note that some tests in the existing test suite rely on previous test
130items, so you cannot arbitrarily skip any test and expect the
131remaining tests to be unaffected.
132
133Currently we do not consider skipped tests as build failures. For
134maximum robustness, when setting up automated build processes, you
135should explicitly skip tests, rather than relying on notmuch's
136detection of missing prerequisites. In the future we may treat tests
137unable to run because of missing prerequisites, but not explicitly
138skipped by the user, as failures.
139
140Writing Tests
141-------------
142The test script is written as a shell script. It is to be named as
143Tddd-testname.sh where 'ddd' is three digits and 'testname' the "bare"
144name of your test. Tests will be run in order the 'ddd' part determines.
145
146The test script should start with the standard "#!/usr/bin/env bash"
147and an assignment to variable 'test_description', like this:
148
149 #!/usr/bin/env bash
150
151 test_description='xxx test (option --frotz)
152
153 This test exercises the "notmuch xxx" command when
154 given the option --frotz.'
155
156Source 'test-lib.sh'
157--------------------
158After assigning test_description, the test script should source
159test-lib.sh like this:
160
161 . ./test-lib.sh || exit 1
162
163This test harness library does the following things:
164
165 - If the script is invoked with command line argument --help
166 (or -h), it shows the test_description and exits.
167
168 - Creates a temporary directory with default notmuch-config and a
169 mail store with a corpus of mail, (initially, 50 early messages
170 sent to the notmuch list). This directory is
171 test/tmp.<test-basename>. The path to notmuch-config is exported in
172 NOTMUCH_CONFIG environment variable and mail store path is stored
173 in MAIL_DIR variable.
174
175 - Defines standard test helper functions for your scripts to
176 use. These functions are designed to make all scripts behave
177 consistently when command line arguments --verbose (or -v),
178 --debug (or -d), and --immediate (or -i) is given.
179
180End with test_done
181------------------
182Your script will be a sequence of tests, using helper functions
183from the test harness library. At the end of the script, call
184'test_done'.
185
186Test harness library
187--------------------
188There are a handful helper functions defined in the test harness
189library for your script to use.
190
191 test_begin_subtest <message>
192
193 Set the test description message for a subsequent test_expect_*
194 invocation (see below).
195
196 test_expect_success <script>
197
198 This takes a string as parameter, and evaluates the
199 <script>. If it yields success, test is considered
200 successful.
201
202 test_expect_code <code> <script>
203
204 This takes two strings as parameter, and evaluates the <script>.
205 If it yields <code> exit status, test is considered successful.
206
207 test_subtest_known_broken
208
209 Mark the current test as broken. Such tests are expected to fail.
210 Unlike the normal tests, which say "PASS" on success and "FAIL" on
211 failure, these will say "FIXED" on success and "BROKEN" on failure.
212 Failures from these tests won't cause -i (immediate) to stop. A
213 test must call this before any test_expect_* function.
214
215 test_expect_equal <output> <expected>
216
217 This is an often-used convenience function built on top of
218 test_expect_success. It uses the message from the last
219 test_begin_subtest call, so call before calling
220 test_expect_equal. This function generates a successful test if
221 both the <output> and <expected> strings are identical. If not, it
222 will generate a failure and print the difference of the two
223 strings.
224
225 test_expect_equal_file <file1> <file2>
226
227 Identical to test_expect_equal, except that <file1> and <file2>
228 are files instead of strings. This is a much more robust method to
229 compare formatted textual information, since it also notices
230 whitespace and closing newline differences.
231
232 test_expect_equal_json <output> <expected>
233
234 Identical to test_expect_equal, except that the two strings are
235 treated as JSON and canonicalized before equality testing. This is
236 useful to abstract away from whitespace differences in the expected
237 output and that generated by running a notmuch command.
238
239 test_debug <script>
240
241 This takes a single argument, <script>, and evaluates it only
242 when the test script is started with --debug command line
243 argument. This is primarily meant for use during the
244 development of a new test script.
245
246 test_emacs <emacs-lisp-expressions>
247
248 This function executes the provided emacs lisp script within
249 emacs. The script can be a sequence of emacs lisp expressions,
250 (that is, they will be evaluated within a progn form). Emacs
251 stdout and stderr is not available, the common way to get output
252 is to save it to a file. There are some auxiliary functions
253 useful in emacs tests provided in test-lib.el. Do not use `setq'
254 for setting variables in Emacs tests because it affects other
255 tests that may run in the same Emacs instance. Use `let' instead
256 so the scope of the changed variables is limited to a single test.
257
258 test_emacs_expect_t <emacs-lisp-expressions>
259
260 This function executes the provided emacs lisp script within
261 emacs in a manner similar to 'test_emacs'. The expressions should
262 return the value `t' to indicate that the test has passed. If the
263 test does not return `t' then it is considered failed and all data
264 returned by the test is reported to the tester.
265
266 test_done
267
268 Your test script must have test_done at the end. Its purpose
269 is to summarize successes and failures in the test script and
270 exit with an appropriate error code.
271
272There are also a number of notmuch-specific auxiliary functions and
273variables which are useful in writing tests:
274
275 generate_message
276
277 Generates a message with an optional template. Most tests will
278 actually prefer to call add_message. See below.
279
280 add_message
281
282 Generate a message and add it to the database (by calling "notmuch
283 new"). It is sufficient to simply call add_message with no
284 arguments if you don't care about the content of the message. If
285 more control is needed, arguments can be provide to specify many
286 different header values for the new message. See the documentation
287 within test-lib.sh or refer to many example calls within existing
288 tests.
289
290 add_email_corpus
291
292 This function should be called at the beginning of a test file
293 when a test needs to operate on a non-empty body of messages. It
294 will initialize the mail database to a known state of 50 sample
295 messages, (culled from the early history of the notmuch mailing
296 list).
297
298 notmuch_counter_reset
299 $notmuch_counter_command
300 notmuch_counter_value
301
302 These allow to count how many times notmuch binary is called.
303 notmuch_counter_reset() function generates a script that counts
304 how many times it is called and resets the counter to zero. The
305 function sets $notmuch_counter_command variable to the path to the
306 generated script that should be called instead of notmuch to do
307 the counting. The notmuch_counter_value() function prints the
308 current counter value.
309
310There are also functions which remove various environment-dependent
311values from notmuch output; these are useful to ensure that test
312results remain consistent across different machines.
313
314 notmuch_search_sanitize
315 notmuch_show_sanitize
316 notmuch_show_sanitize_all
317 notmuch_json_show_sanitize
318
319 All these functions should receive the text to be sanitized as the
320 input of a pipe, e.g.
321 output=`notmuch search "..." | notmuch_search_sanitize`
322