README.md
1<!--
2Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
3
4SPDX-License-Identifier: curl
5-->
6
7# The curl Test Suite
8
9# Running
10
11## Requires to run
12
13 - perl (and a unix-style shell)
14 - python (and a unix-style shell, for SMB and TELNET tests)
15 - python-impacket (for SMB tests)
16 - diff (when a test fails, a diff is shown)
17 - stunnel (for HTTPS and FTPS tests)
18 - OpenSSH or SunSSH (for SCP, SFTP and SOCKS4/5 tests)
19 - nghttpx (for HTTP/2 and HTTP/3 tests)
20 - nroff (for --manual tests)
21 - An available `en_US.UTF-8` locale
22
23### Installation of python-impacket
24
25 The Python-based test servers support both recent Python 2 and 3.
26 You can figure out your default Python interpreter with python -V
27
28 Please install python-impacket in the correct Python environment.
29 You can use pip or your OS' package manager to install 'impacket'.
30
31 On Debian/Ubuntu the package names are:
32
33 - Python 2: 'python-impacket'
34 - Python 3: 'python3-impacket'
35
36 On FreeBSD the package names are:
37
38 - Python 2: 'py27-impacket'
39 - Python 3: 'py37-impacket'
40
41 On any system where pip is available:
42
43 - Python 2: 'pip2 install impacket'
44 - Python 3: 'pip3 install impacket'
45
46 You may also need to manually install the Python package 'six'
47 as that may be a missing requirement for impacket on Python 3.
48
49### Port numbers used by test servers
50
51 All test servers run on "random" port numbers. All tests should be written
52 to use suitable variables instead of fixed port numbers so that test cases
53 continue to work independent on what port numbers the test servers actually
54 use.
55
56 See [`FILEFORMAT`](FILEFORMAT.md) for the port number variables.
57
58### Test servers
59
60 The test suite runs stand-alone servers on random ports to which it makes
61 requests. For SSL tests, it runs stunnel to handle encryption to the regular
62 servers. For SSH, it runs a standard OpenSSH server. For SOCKS4/5 tests SSH
63 is used to perform the SOCKS functionality and requires a SSH client and
64 server.
65
66 The listen port numbers for the test servers are picked randomly to allow
67 users to run multiple test cases concurrently and to not collide with other
68 existing services that might listen to ports on the machine.
69
70 The HTTP server supports listening on a Unix domain socket, the default
71 location is 'http.sock'.
72
73 For HTTP/2 and HTTP/3 testing an installed `nghttpx` is used. HTTP/3
74 tests check if nghttpx supports the protocol. To override the nghttpx
75 used, set the environment variable `NGHTTPX`. The default can also be
76 changed by specifying `--with-test-nghttpx=<path>` as argument to `configure`.
77
78### Run
79
80 `./configure && make && make test`. This builds the test suite support code
81 and invokes the 'runtests.pl' perl script to run all the tests. Edit the top
82 variables of that script in case you have some specific needs, or run the
83 script manually (after the support code has been built).
84
85 The script breaks on the first test that doesn't do OK. Use `-a` to prevent
86 the script from aborting on the first error. Run the script with `-v` for
87 more verbose output. Use `-d` to run the test servers with debug output
88 enabled as well. Specifying `-k` keeps all the log files generated by the
89 test intact.
90
91 Use `-s` for shorter output, or pass test numbers to run specific tests only
92 (like `./runtests.pl 3 4` to test 3 and 4 only). It also supports test case
93 ranges with 'to', as in `./runtests.pl 3 to 9` which runs the seven tests
94 from 3 to 9. Any test numbers starting with ! are disabled, as are any test
95 numbers found in the files `data/DISABLED` or `data/DISABLED.local` (one per
96 line). The latter is meant for local temporary disables and will be ignored
97 by git.
98
99 Test cases mentioned in `DISABLED` can still be run if `-f` is provided.
100
101 When `-s` is not present, each successful test will display on one line the
102 test number and description and on the next line a set of flags, the test
103 result, current test sequence, total number of tests to be run and an
104 estimated amount of time to complete the test run. The flags consist of
105 these letters describing what is checked in this test:
106
107 s stdout
108 d data
109 u upload
110 p protocol
111 o output
112 e exit code
113 m memory
114 v valgrind
115
116### Shell startup scripts
117
118 Tests which use the ssh test server, SCP/SFTP/SOCKS tests, might be badly
119 influenced by the output of system wide or user specific shell startup
120 scripts, .bashrc, .profile, /etc/csh.cshrc, .login, /etc/bashrc, etc. which
121 output text messages or escape sequences on user login. When these shell
122 startup messages or escape sequences are output they might corrupt the
123 expected stream of data which flows to the sftp-server or from the ssh
124 client which can result in bad test behavior or even prevent the test server
125 from running.
126
127 If the test suite ssh or sftp server fails to start up and logs the message
128 'Received message too long' then you are certainly suffering the unwanted
129 output of a shell startup script. Locate, cleanup or adjust the shell
130 script.
131
132### Memory test
133
134 The test script will check that all allocated memory is freed properly IF
135 curl has been built with the `CURLDEBUG` define set. The script will
136 automatically detect if that is the case, and it will use the
137 `memanalyze.pl` script to analyze the memory debugging output.
138
139 Also, if you run tests on a machine where valgrind is found, the script will
140 use valgrind to run the test with (unless you use `-n`) to further verify
141 correctness.
142
143 The `runtests.pl` `-t` option enables torture testing mode. It runs each
144 test many times and makes each different memory allocation fail on each
145 successive run. This tests the out of memory error handling code to ensure
146 that memory leaks do not occur even in those situations. It can help to
147 compile curl with `CPPFLAGS=-DMEMDEBUG_LOG_SYNC` when using this option, to
148 ensure that the memory log file is properly written even if curl crashes.
149
150### Debug
151
152 If a test case fails, you can conveniently get the script to invoke the
153 debugger (gdb) for you with the server running and the same command line
154 parameters that failed. Just invoke `runtests.pl <test number> -g` and then
155 just type 'run' in the debugger to perform the command through the debugger.
156
157### Logs
158
159 All logs are generated in the log/ subdirectory (it is emptied first in the
160 runtests.pl script). They remain in there after a test run.
161
162### Log Verbosity
163
164 A curl build with `--enable-debug` offers more verbose output in the logs.
165 This applies not only for test cases, but also when running it standalone
166 with `curl -v`. While a curl debug built is
167 ***not suitable for production***, it is often helpful in tracking down
168 problems.
169
170 Sometimes, one needs detailed logging of operations, but does not want
171 to drown in output. The newly introduced *connection filters* allows one to
172 dynamically increase log verbosity for a particular *filter type*. Example:
173
174 CURL_DEBUG=ssl curl -v https://curl.se
175
176 will make the `ssl` connection filter log more details. One may do that for
177 every filter type and also use a combination of names, separated by `,` or
178 space.
179
180 CURL_DEBUG=ssl,http/2 curl -v https://curl.se
181
182 The order of filter type names is not relevant. Names used here are
183 case insensitive. Note that these names are implementation internals and
184 subject to change.
185
186 Some, likely stable names are `tcp`, `ssl`, `http/2`. For a current list,
187 one may search the sources for `struct Curl_cftype` definitions and find
188 the names there. Also, some filters are only available with certain build
189 options, of course.
190
191### Test input files
192
193 All test cases are put in the `data/` subdirectory. Each test is stored in
194 the file named according to the test number.
195
196 See [`FILEFORMAT`](FILEFORMAT.md) for a description of the test case file
197 format.
198
199### Code coverage
200
201 gcc provides a tool that can determine the code coverage figures for the
202 test suite. To use it, configure curl with `CFLAGS='-fprofile-arcs
203 -ftest-coverage -g -O0'`. Make sure you run the normal and torture tests to
204 get more full coverage, i.e. do:
205
206 make test
207 make test-torture
208
209 The graphical tool `ggcov` can be used to browse the source and create
210 coverage reports on \*nix hosts:
211
212 ggcov -r lib src
213
214 The text mode tool `gcov` may also be used, but it doesn't handle object
215 files in more than one directory correctly.
216
217### Remote testing
218
219 The runtests.pl script provides some hooks to allow curl to be tested on a
220 machine where perl can not be run. The test framework in this case runs on
221 a workstation where perl is available, while curl itself is run on a remote
222 system using ssh or some other remote execution method. See the comments at
223 the beginning of runtests.pl for details.
224
225## Test case numbering
226
227 Test cases used to be numbered by category ranges, but the ranges filled
228 up. Subsets of tests can now be selected by passing keywords to the
229 runtests.pl script via the make `TFLAGS` variable.
230
231 New tests are added by finding a free number in `tests/data/Makefile.inc`.
232
233## Write tests
234
235 Here's a quick description on writing test cases. We basically have three
236 kinds of tests: the ones that test the curl tool, the ones that build small
237 applications and test libcurl directly and the unit tests that test
238 individual (possibly internal) functions.
239
240### test data
241
242 Each test has a master file that controls all the test data. What to read,
243 what the protocol exchange should look like, what exit code to expect and
244 what command line arguments to use etc.
245
246 These files are `tests/data/test[num]` where `[num]` is just a unique
247 identifier described above, and the XML-like file format of them is
248 described in the separate [`FILEFORMAT`](FILEFORMAT.md) document.
249
250### curl tests
251
252 A test case that runs the curl tool and verifies that it gets the correct
253 data, it sends the correct data, it uses the correct protocol primitives
254 etc.
255
256### libcurl tests
257
258 The libcurl tests are identical to the curl ones, except that they use a
259 specific and dedicated custom-built program to run instead of "curl". This
260 tool is built from source code placed in `tests/libtest` and if you want to
261 make a new libcurl test that is where you add your code.
262
263### unit tests
264
265 Unit tests are placed in `tests/unit`. There's a tests/unit/README
266 describing the specific set of checks and macros that may be used when
267 writing tests that verify behaviors of specific individual functions.
268
269 The unit tests depend on curl being built with debug enabled.
270