2007 2013 Ericsson AB, All Rights Reserved Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. The Initial Developer of the Original Code is Ericsson AB. The Test Server Controller Siri Hansen, Peter Andersson test_server_ctrl_ref.sgml
test_server_ctrl This module provides a low level interface to the Test Server.

The test_server_ctrl module provides a low level interface to the Test Server. This interface is normally not used directly by the tester, but through a framework built on top of test_server_ctrl.

Common Test is such a framework, well suited for automated black box testing of target systems of any kind (not necessarily implemented in Erlang). Common Test is also a very useful tool for white box testing Erlang programs and OTP applications. Please see the Common Test User's Guide and reference manual for more information.

If you want to write your own framework, some more information can be found in the chapter "Writing your own test server framework" in the Test Server User's Guide. Details about the interface provided by test_server_ctrl follows below.

start() -> Result Starts the test server. Result = ok | {error, {already_started, pid()}

This function starts the test server.

 stop() -> ok Stops the test server immediately.

This stops the test server and all its activity. The running test suite (if any) will be halted.

 add_dir(Name, Dir) -> ok add_dir(Name, Dir, Pattern) -> ok add_dir(Name, [Dir|Dirs]) -> ok add_dir(Name, [Dir|Dirs], Pattern) -> ok Add a directory to the job queue. Name = term() The jobname for this directory. Dir = term() The directory to scan for test suites. Dirs = [term()] List of directories to scan for test suites. Pattern = term() Suite match pattern. Directories will be scanned for Pattern_SUITE.erl files.

Puts a collection of suites matching (*_SUITE) in given directories into the job queue. Name is an arbitrary name for the job, it can be any erlang term. If Pattern is given, only modules matching Pattern* will be added.

 add_module(Mod) -> ok add_module(Name, [Mod|Mods]) -> ok Add a module to the job queue with or without a given name. Mod = atom() Mods = [atom()] The name(s) of the module(s) to add. Name = term() Name for the job.

This function adds a module or a list of modules, to the test servers job queue. Name may be any Erlang term. When Name is not given, the job gets the name of the module.

 add_case(Mod, Case) -> ok Adds one test case to the job queue. Mod = atom() Name of the module the test case is in. Case = atom() Function name of the test case to add.

This function will add one test case to the job queue. The job will be given the module's name.

 add_case(Name, Mod, Case) -> ok Equivalent to add_case/2, but with specified name. Name = string() Name to use for the test job.

Equivalent to add_case/2, but the test job will get the specified name.

 add_cases(Mod, Cases) -> ok Adds a list of test cases to the job queue. Mod = atom() Name of the module the test case is in. Cases = [Case] Case = atom() Function names of the test cases to add.

This function will add one or more test cases to the job queue. The job will be given the module's name.

 add_cases(Name, Mod, Cases) -> ok Equivalent to add_cases/2, but with specified name. Name = string() Name to use for the test job.

Equivalent to add_cases/2, but the test job will get the specified name.

 add_spec(TestSpecFile) -> ok | {error, nofile} Adds a test specification file to the job queue. TestSpecFile = string() Name of the test specification file

This function will add the content of the given test specification file to the job queue. The job will be given the name of the test specification file, e.g. if the file is called test.spec, the job will be called test.

See the reference manual for the test server application for details about the test specification file.

 add_dir_with_skip(Name, [Dir|Dirs], Skip) -> ok add_dir_with_skip(Name, [Dir|Dirs], Pattern, Skip) -> ok add_module_with_skip(Mod, Skip) -> ok add_module_with_skip(Name, [Mod|Mods], Skip) -> ok add_case_with_skip(Mod, Case, Skip) -> ok add_case_with_skip(Name, Mod, Case, Skip) -> ok add_cases_with_skip(Mod, Cases, Skip) -> ok add_cases_with_skip(Name, Mod, Cases, Skip) -> ok Same purpose as functions listed above, but with extra Skip argument. Skip = [SkipItem] List of items to be skipped from the test. SkipItem = {Mod,Comment} | {Mod,Case,Comment} | {Mod,Cases,Comment} Mod = atom() Test suite name. Comment = string() Reason why suite or case is being skipped. Cases = [Case] Case = atom() Name of test case function.

These functions add test jobs just like the add_dir, add_module, add_case and add_cases functions above, but carry an additional argument, Skip. Skip is a list of items that should be skipped in the current test run. Test job items that occur in the Skip list will be logged as SKIPPED with the associated Comment.

 add_tests_with_skip(Name, Tests, Skip) -> ok Adds different types of jobs to the run queue. Name = term() The jobname for this directory. Tests = [TestItem] List of jobs to add to the run queue. TestItem = {Dir,all,all} | {Dir,Mods,all} | {Dir,Mod,Cases} Dir = term() The directory to scan for test suites. Mods = [Mod] Mod = atom() Test suite name. Cases = [Case] Case = atom() Name of test case function. Skip = [SkipItem] List of items to be skipped from the test. SkipItem = {Mod,Comment} | {Mod,Case,Comment} | {Mod,Cases,Comment} Comment = string() Reason why suite or case is being skipped.

This function adds various test jobs to the test_server_ctrl job queue. These jobs can be of different type (all or specific suites in one directory, all or specific cases in one suite, etc). It is also possible to get particular items skipped by passing them along in the Skip list (see the add_*_with_skip functions above).

 abort_current_testcase(Reason) -> ok | {error,no_testcase_running} Aborts the test case currently executing. Reason = term() The reason for stopping the test case, which will be printed in the log.

When calling this function, the currently executing test case will be aborted. It is the user's responsibility to know for sure which test case is currently executing. The function is therefore only safe to call from a function which has been called (or synchronously invoked) by the test case.

 set_levels(Console, Major, Minor) -> ok Sets the levels of I/O. Console = integer() Level for I/O to be sent to console. Major = integer() Level for I/O to be sent to the major logfile. Minor = integer() Level for I/O to be sent to the minor logfile.

Determines where I/O from test suites/test server will go. All text output from test suites and the test server is tagged with a priority value which ranges from 0 to 100, 100 being the most detailed. (see the section about log files in the user's guide). Output from the test cases (using io:format/2) has a detail level of 50. Depending on the levels set by this function, this I/O may be sent to the console, the major log file (for the whole test suite) or to the minor logfile (separate for each test case).

All output with detail level:

Less than or equal to Console is displayed on the screen (default 1) Less than or equal to Major is logged in the major log file (default 19) Greater than or equal to Minor is logged in the minor log files (default 10)

To view the currently set thresholds, use the get_levels/0 function.

 get_levels() -> {Console, Major, Minor} Returns the current levels.

Returns the current levels. See set_levels/3 for types.

 jobs() -> JobQueue Returns the job queue. JobQueue = [{list(), pid()}]

This function will return all the jobs currently in the job queue.

 multiply_timetraps(N) -> ok All timetraps started after this will be multiplied by N. N = integer() | infinity

This function should be called before a test is started which requires extended timetraps, e.g. if extensive tracing is used. All timetraps started after this call will be multiplied by N.

 scale_timetraps(Bool) -> ok . Bool = true | false

This function should be called before a test is started. The parameter specifies if test_server should attempt to automatically scale the timetrap value in order to compensate for delays caused by e.g. the cover tool.

 get_timetrap_parameters() -> {N,Bool} Read the parameter values that affect timetraps. N = integer() | infinity Bool = true | false

This function may be called to read the values set by multiply_timetraps/1 and scale_timetraps/1.

 cover(Application,Analyse) -> ok cover(CoverFile,Analyse) -> ok cover(App,CoverFile,Analyse) -> ok Informs the test_server controller that next test shall run with code coverage analysis. Application = atom() OTP application to cover compile CoverFile = string() Name of file listing modules to exclude from or include in cover compilation. The filename must include full path to the file. Analyse = details | overview

This function informs the test_server controller that next test shall run with code coverage analysis. All timetraps will automatically be multiplied by 10 when cover i run.

Application and CoverFile indicates what to cover compile. If Application is given, the default is that all modules in the ebin directory of the application will be cover compiled. The ebin directory is found by adding ebin to code:lib_dir(Application).

A CoverFile can have the following entries:

{exclude, all | ExcludeModuleList}. {include, IncludeModuleList}. {cross, CrossCoverInfo}.

Note that each line must end with a full stop. ExcludeModuleList and IncludeModuleList are lists of atoms, where each atom is a module name.

CrossCoverInfo is used when collecting cover data over multiple tests. Modules listed here are compiled, but they will not be analysed when the test is finished. See cross_cover_analyse/2 for more information about the cross cover mechanism and the format of CrossCoverInfo.

If both an Application and a CoverFile is given, all modules in the application are cover compiled, except for the modules listed in ExcludeModuleList. The modules in IncludeModuleList are also cover compiled.

If a CoverFile is given, but no Application, only the modules in IncludeModuleList are cover compiled.

Analyse indicates the detail level of the cover analysis. If Analyse = details, each cover compiled module will be analysed with cover:analyse_to_file/1. If Analyse = overview an overview of all cover compiled modules is created, listing the number of covered and not covered lines for each module.

If the test following this call starts any slave or peer nodes with test_server:start_node/3, the same cover compiled code will be loaded on all nodes. If the loading fails, e.g. if the node runs an old version of OTP, the node will simply not be a part of the coverage analysis. Note that slave or peer nodes must be stopped with test_server:stop_node/1 for the node to be part of the coverage analysis, else the test server will not be able to fetch coverage data from the node.

When the test is finished, the coverage analysis is automatically completed, logs are created and the cover compiled modules are unloaded. If another test is to be run with coverage analysis, test_server_ctrl:cover/2/3 must be called again.

cross_cover_analyse(Level, Tests) -> ok Analyse cover data collected from multiple tests Level = details | overview Tests = [{Tag,LogDir}] Tag = atom() Test identifier. LogDir = string() Log directory for the test identified by Tag. This can either be the run.<timestamp> directory or the parent directory of this (in which case the latest run.<timestamp> directory is chosen.

Analyse cover data collected from multiple tests. The modules analysed are the ones listed in cross statements in the cover files. These are modules that are heavily used by other tests than the one where they belong or are explicitly tested. They should then be listed as cross modules in the cover file for the test where they are used but do not belong. Se example below.

This function should be run after all tests are completed, and the result will be stored in a file called cross_cover.html in the run.<timestamp> directory of the test the modules belong to.

Note that the function can be executed on any node, and it does not require test_server_ctrl to be started first.

The cross statement in the cover file must be like this:

{cross,[{Tag,Modules}]}.

where Tag is the same as Tag in the Tests parameter to this function and Modules is a list of module names (atoms).

Example:

If the module m1 belongs to system s1 but is heavily used also in the tests for another system s2, then the cover files for the two systems' tests could be like this:

s1.cover: {include,[m1]}. s2.cover: {include,[....]}. % modules belonging to system s2 {cross,[{s1,[m1]}]}.

When the tests for both s1 and s2 are completed, run

test_server_ctrl:cross_cover_analyse(Level,[{s1,S1LogDir},{s2,S2LogDir}])

and the accumulated cover data for m1 will be written to S1LogDir/[run.<timestamp>/]cross_cover.html.

Note that the m1 module will also be presented in the normal coverage log for s1 (due to the include statement in s1.cover), but that only includes the coverage achieved by the s1 test itself.

The Tag in the cross statement in the cover file has no other purpose than mapping the list of modules ([m1] in the example above) to the correct log directory where it should be included in the cross_cover.html file (S1LogDir in the example above). I.e. the value of Tag has no meaning, it could be foo as well as s1 above, as long as the same Tag is used in the cover file and in the call to this function.

 trc(TraceInfoFile) -> ok | {error, Reason} Starts call trace on target and slave nodes TraceInfoFile = atom() | string() Name of a file defining which functions to trace and how

This function starts call trace on target and on slave or peer nodes that are started or will be started by the test suites.

Timetraps are not extended automatically when tracing is used. Use multiply_timetraps/1 if necessary.

Note that the trace support in the test server is in a very early stage of the implementation, and thus not yet as powerful as one might wish for.

The trace information file specified by the TraceInfoFile argument is a text file containing one or more of the following elements:

{SetTP,Module,Pattern}. {SetTP,Module,Function,Pattern}. {SetTP,Module,Function,Arity,Pattern}. ClearTP. {ClearTP,Module}. {ClearTP,Module,Function}. {ClearTP,Module,Function,Arity}. SetTP = tp | tpl This is maps to the corresponding functions in the ttb module in the observer application. tp means set trace pattern on global function calls. tpl means set trace pattern on local and global function calls. ClearTP = ctp | ctpl | ctpg This is maps to the corresponding functions in the ttb module in the observer application. ctp means clear trace pattern (i.e. turn off) on global and local function calls. ctpl means clear trace pattern on local function calls only and ctpg means clear trace pattern on global function calls only. Module = atom() The module to trace Function = atom() The name of the function to trace Arity = integer() The arity of the function to trace Pattern = [] | match_spec() The trace pattern to set for the module or function. For a description of the match_spec() syntax, please turn to the User's guide for the runtime system (erts). The chapter "Match Specification in Erlang" explains the general match specification language.

The trace result will be logged in a (binary) file called NodeName-test_server in the current directory of the test server controller node. The log must be formatted using ttb:format/1/2.

stop_trace() -> ok | {error, not_tracing} Stops tracing on target and slave nodes.

This function stops tracing on target, and on slave or peer nodes that are currently running. New slave or peer nodes will no longer be traced after this.
FUNCTIONS INVOKED FROM COMMAND LINE

The following functions are supposed to be invoked from the command line using the -s option when starting the erlang node.

run_test(CommandLine) -> ok Runs the tests specified on the command line. CommandLine = FlagList

This function is supposed to be invoked from the commandline. It starts the test server, interprets the argument supplied from the commandline, runs the tests specified and when all tests are done, stops the test server and returns to the Erlang prompt.

The CommandLine argument is a list of command line flags, typically ['KEY1', Value1, 'KEY2', Value2, ...]. The valid command line flags are listed below.

Under a UNIX command prompt, this function can be invoked like this:

erl -noshell -s test_server_ctrl run_test KEY1 Value1 KEY2 Value2 ... -s erlang halt

Or make an alias (this is for unix/tcsh)

alias erl_test 'erl -noshell -s test_server_ctrl run_test \!* -s erlang halt'

And then use it like this

erl_test KEY1 Value1 KEY2 Value2 ...

The valid command line flags are

DIR dir Adds all test modules in the directory dir to the job queue. MODULE mod Adds the module mod to the job queue. CASE mod case Adds the case case in module mod to the job queue. SPEC spec Runs the test specification file spec. SKIPMOD mod Skips all test cases in the module mod SKIPCASE mod case Skips the test case case in module mod. NAME name Names the test suite to something else than the default name. This does not apply to SPEC which keeps its names. COVER app cover_file analyse Indicates that the test should be run with cover analysis. app, cover_file and analyse corresponds to the parameters to test_server_ctrl:cover/3. If no cover file is used, the atom none should be given. TRACE traceinfofile Specifies a trace information file. When this option is given, call tracing is started on the target node and all slave or peer nodes that are started. The trace information file specifies which modules and functions to trace. See the function trc/1 above for more information about the syntax of this file.
FRAMEWORK CALLBACK FUNCTIONS

A test server framework can be defined by setting the environment variable TEST_SERVER_FRAMEWORK to a module name. This module will then be framework callback module, and it must export the following function:

get_suite(Mod,Func) -> TestCaseList Get subcases. Mod = atom() Test suite name. Func = atom() Name of test case. TestCaseList = [SubCase] List of test cases. SubCase = atom() Name of a case.

This function is called before a test case is started. The purpose is to retrieve a list of subcases. The default behaviour of this function should be to call Mod:Func(suite) and return the result from this call.

 init_tc(Mod,Func,Args0) -> {ok,Args1} | {skip,ReasonToSkip} | {auto_skip,ReasonToSkip} | {fail,ReasonToFail} Preparation for a test case or configuration function. Mod = atom() Test suite name. Func = atom() Name of test case or configuration function. Args0 = Args1 = [tuple()] Normally Args = [Config] ReasonToSkip = term() Reason to skip the test case or configuration function. ReasonToFail = term() Reason to fail the test case or configuration function.

This function is called before a test case or configuration function starts. It is called on the process executing the function Mod:Func. Typical use of this function can be to alter the input parameters to the test case function (Args) or to set properties for the executing process.

By returning {skip,Reason}, Func gets skipped. Func also gets skipped if {auto_skip,Reason} is returned, but then gets an auto skipped status (rather than user skipped).

To fail Func immediately instead of executing it, return {fail,ReasonToFail}.

 end_tc(Mod,Func,Status) -> ok | {fail,ReasonToFail} Cleanup after a test case or configuration function. Mod = atom() Test suite name. Func = atom() Name of test case or configuration function. Status = {Result,Args} | {TCPid,Result,Args} The status of the test case or configuration function. ReasonToFail = term() Reason to fail the test case or configuration function. Result = ok | Skip | Fail The final result of the test case or configuration function. TCPid = pid() Pid of the process executing Func Skip = {skip,SkipReason} SkipReason = term() | {failed,{Mod,init_per_testcase,term()}} Reason why the function was skipped. Fail = {error,term()} | {'EXIT',term()} | {timetrap_timeout,integer()} | {testcase_aborted,term()} | testcase_aborted_or_killed | {failed,term()} | {failed,{Mod,end_per_testcase,term()}} Reason why the function failed. Args = [tuple()] Normally Args = [Config]

This function is called when a test case, or a configuration function, is finished. It is normally called on the process where the function Mod:Func has been executing, but if not, the pid of the test case process is passed with the Status argument.

Typical use of the end_tc/3 function can be to clean up after init_tc/3.

If Func is a test case, it is possible to analyse the value of Result to verify that init_per_testcase/2 and end_per_testcase/2 executed successfully.

It is possible with end_tc/3 to fail an otherwise successful test case, by returning {fail,ReasonToFail}. The test case Func will be logged as failed with the provided term as reason.

 report(What,Data) -> ok Progress report for test. What = atom() Data = term()

This function is called in order to keep the framework up-to-date with the progress of the test. This is useful e.g. if the framework implements a GUI where the progress information is constantly updated. The following can be reported:

What = tests_start, Data = {Name,NumCases}

What = loginfo, Data = [{topdir,TestRootDir},{rundir,CurrLogDir}]

What = tests_done, Data = {Ok,Failed,{UserSkipped,AutoSkipped}}

What = tc_start, Data = {{Mod,{Func,GroupName}},TCLogFile}

What = tc_done, Data = {Mod,{Func,GroupName},Result}

What = tc_user_skip, Data = {Mod,{Func,GroupName},Comment}

What = tc_auto_skip, Data = {Mod,{Func,GroupName},Comment}

What = framework_error, Data = {{FWMod,FWFunc},Error}

Note that for a test case function that doesn't belong to a group, GroupName has value undefined, otherwise the name of the test case group.

 error_notification(Mod, Func, Args, Error) -> ok Inform framework of crashing testcase or configuration function. Mod = atom() Test suite name. Func = atom() Name of test case or configuration function. Args = [tuple()] Normally Args = [Config] Error = {Reason,Location} Reason = term() Reason for termination. Location = unknown | [{Mod,Func,Line}] Last known position in Mod before termination. Line = integer() Line number in file Mod.erl.

This function is called as the result of function Mod:Func failing with Reason at Location. The function is intended mainly to aid specific logging or error handling in the framework application. Note that for Location to have relevant values (i.e. other than unknown), the line macro or test_server_line parse transform must be used. For details, please see the section about test suite line numbers in the test_server reference manual page.

 warn(What) -> boolean() Ask framework if test server should issue a warning for What. What = processes | nodes

The test server checks the number of processes and nodes before and after the test is executed. This function is a question to the framework if the test server should warn when the number of processes or nodes has changed during the test execution. If true is returned, a warning will be written in the test case minor log file.

 target_info() -> InfoStr Print info about the target system to the test case log. InfoStr = string() | ""

The test server will ask the framework for information about the test target system and print InfoStr in the test case log file below the host information.