Ejemplo n.º 1
0
    def __init__(self,
                 timeout=5.0,
                 server_host='localhost',
                 server_port=4444,
                 jar_path=None,
                 run_on_failure='Capture Screenshot'):
        """SeleniumLibrary can be imported with optional arguments.

        `timeout` is the default timeout used to wait for page load actions.
        It can be later set with `Set Selenium Timeout`

        `server_host` and `server_port` are used to connect to Selenium Server.
        Browsers opened with this SeleniumLibrary instance will be attached to
        that server. Note that the Selenium Server must be running before `Open
        Browser` keyword can be used. Selenium Server can be started with
        keyword `Start Selenium Server`. Starting from SeleniumLibrary 2.6.1,
        it is possible to give `server_host` as a URL with a possible embedded
        port, for example `http://192.168.52.1:4444`. If `server_host` contains
        port, the value of `server_port` is ignored.

        `jar_path` is the absolute path to the selenium-server.jar file to be
        used by the library. If set, a custom, modified version can be started
        instead of the default one distributed with the library.

        `run_on_failure` specifies the name of a SeleniumLibrary keyword to
        execute when another SeleniumLibrary keyword fails. By default
        `Capture Screenshot` will be used to take a screenshot of the situation.
        Using any value that is not a keyword name will disable this feature
        altogether. See `Register Keyword To Run On Failure` keyword for more
        information about this functionality that was added in SeleniumLibrary
        2.5.

        Because there are many optional arguments, it is often a good idea to
        use the handy named-arguments syntax supported by Robot Framework 2.5
        and later. This is demonstrated by the last two examples below.

        Examples:
        | Library | SeleniumLibrary | 15 | | | # Sets default timeout |
        | Library | SeleniumLibrary | | | 4455 | # Use default timeout and host but specify different port. |
        | Library | SeleniumLibrary | server_host=http://192.168.52.1:4444 | | | # Host as URL. |
        | Library | SeleniumLibrary | run_on_failure=Nothing | | | # Do nothing on failure. |
        """
        self._cache = utils.ConnectionCache()
        self._selenium = NoBrowserOpen()
        self.set_selenium_timeout(timeout or 5.0)
        self._server_host, self._server_port \
                = self._parse_host_and_port(server_host or 'localhost',
                                            server_port or 4444)
        self._jar_path = jar_path
        self._set_run_on_failure(run_on_failure)
        self._selenium_log = None
        self._locator_parser = LocatorParser(self)
        self._namegen = _NameGenerator()
Ejemplo n.º 2
0
    def __init__(
        self, timeout=5.0, server_host="localhost", server_port=4444, jar_path=None, run_on_failure="Capture Screenshot"
    ):
        """SeleniumLibrary can be imported with optional arguments.

        `timeout` is the default timeout used to wait for page load actions.
        It can be later set with `Set Selenium Timeout`

        `server_host` and `server_port` are used to connect to Selenium Server.
        Browsers opened with this SeleniumLibrary instance will be attached to
        that server. Note that the Selenium Server must be running before `Open
        Browser` keyword can be used. Selenium Server can be started with
        keyword `Start Selenium Server`. Starting from SeleniumLibrary 2.6.1,
        it is possible to give `server_host` as a URL with a possible embedded
        port, for example `http://192.168.52.1:4444`. If `server_host` contains
        port, the value of `server_port` is ignored.

        `jar_path` is the absolute path to the selenium-server.jar file to be
        used by the library. If set, a custom, modified version can be started
        instead of the default one distributed with the library.

        `run_on_failure` specifies the name of a SeleniumLibrary keyword to
        execute when another SeleniumLibrary keyword fails. By default
        `Capture Screenshot` will be used to take a screenshot of the situation.
        Using any value that is not a keyword name will disable this feature
        altogether. See `Register Keyword To Run On Failure` keyword for more
        information about this functionality that was added in SeleniumLibrary
        2.5.

        Because there are many optional arguments, it is often a good idea to
        use the handy named-arguments syntax supported by Robot Framework 2.5
        and later. This is demonstrated by the last two examples below.

        Examples:
        | Library | SeleniumLibrary | 15 | | | # Sets default timeout |
        | Library | SeleniumLibrary | | | 4455 | # Use default timeout and host but specify different port. |
        | Library | SeleniumLibrary | server_host=http://192.168.52.1:4444 | | | # Host as URL. |
        | Library | SeleniumLibrary | run_on_failure=Nothing | | | # Do nothing on failure. |
        """
        self._cache = utils.ConnectionCache()
        self._selenium = NoBrowserOpen()
        self.set_selenium_timeout(timeout or 5.0)
        self._server_host, self._server_port = self._parse_host_and_port(
            server_host or "localhost", server_port or 4444
        )
        self._jar_path = jar_path
        self._set_run_on_failure(run_on_failure)
        self._selenium_log = None
        self._locator_parser = LocatorParser(self)
        self._namegen = _NameGenerator()
Ejemplo n.º 3
0
class SeleniumLibrary(Browser, Page, Button, Click, JavaScript, Mouse, Select,
                      Element, Screenshot, Table, TextField, Flex):
    """SeleniumLibrary is a *deprecated* web testing library for Robot Framework.

    It uses the Selenium Remote Control tool internally to control a web browser.
    See http://selenium-rc.openqa.org/ for more information on Selenium tool.

    SeleniumLibrary runs tests in a real browser instance. It should work in
    most modern browsers and can be used with both Python and Jython interpreters.

    = Before running the tests =

    Prior to running test cases using SeleniumLibrary, the Selenium Server must
    be started. This can be done using keyword `Start Selenium Server` or from
    the command line by using command: `java -jar
    /path/to/selenium-server.jar`. The Selenium Server is included in the
    SeleniumLibrary distribution and can be found under
    `[PythonLibs]/site-packages/SeleniumLibrary/lib`. Additionally, `Open
    Browser` keyword must be used in order to open browser in the desired
    location before any other keyword from the library may be used.

    = Locating elements =

    All keywords in SeleniumLibrary that need to find an element on the page
    take an argument, `locator`. In the most common case, `locator` is
    matched against the values of key attributes of the particular element type.
    For example, `id` and `name` are key attributes to all elements, and
    locating elements is easy using just the `id` as a `locator`.

    Asterisk character may be used as a wildcard in locators, but it only works
    as the last character of the expression. In the middle of the locator it
    is interpreted as literal '*'.

    It is also possible to give an arbitrary XPath or DOM expression as
    `locator`. In this case, the expression must be prefixed with either
    'xpath=' or 'dom='.

    Examples:
    | Click Link      | my link | # Matches if either link text or 'id', 'name' or 'href' of a link equals 'my link' |
    | Page Should Contain Link | Link id * | # Passes if the page contain any link starting with 'Link id' |
    | Select Checkbox | xpath=//table[0]/input[@name='my_checkbox'] | # Using XPath |
    | Click Image     | dom=document.images[56] | # Using a DOM expression |

    Table related keywords, such as `Table Should Contain`, allow identifying
    tables either by an id, by a CSS locator, or by an XPath expression.
    The XPath support was added in SeleniumLibrary 2.6.

    Examples:
    | Table Should Contain | tableID | text |
    | Table Should Contain | css=h2.someClass ~ table:last-child() | text |
    | Table Should Contain | xpath=//table/[@name="myTable"] | text |

    = Locating Flex elements =

    SeleniumLibary 2.6 and newer support testing Adobe Flex and Flash
    applications using Flex Pilot tool. For more information, including the
    required bootstrapping, see
    https://github.com/robotframework/SeleniumLibrary/wiki/FlexTesting

    By default Flex elements are located based on `id` they have in Flex source
    code. Other supported locators are `name`, `automationName`, `text`,
    `htmlText`, `label` and xpath-like `chain`. To use them, you need to prefix
    the value with the locator type like `name=example`. Locators also support
    `*` as a wildcard.

    Examples:
    | Click Flex Element | foo          | # Search element by id |
    | Click Flex Element | name=myName  | # Search element by name |
    | Click Flex Element | label=Hello! | # Search element by label text |
    | Click Flex Element | chain=id:someId/name:someName | # Search element first by id and then its child by name |
    | Click Flex Element | name=wild*   | # Name with wildcard |
    | Click Flex Element | chain=name:*llo/name:world | # Chain with wildcard |

    = Handling page load events =

    Some keywords that may cause a page to load take an additional argument
    `dont_wait` that is used to determine whether a new page is expected to
    load or not. By default, a page load is expected to happen whenever a link
    or image is clicked, or a form submitted. If a page load does not happen
    (if the link only executes some JavaScript, for example), a non-empty value
    must be given for the `dont_wait` argument. How much to wait is determined
    by a timeout discussed in the next section.

    There are also some keywords that may cause a page to load but by default
    we expect them not to. For these cases, the keywords have an optional `wait`
    argument, and providing a non-empty value for it will cause the keyword to
    wait. An other possibility is using `Wait Until Page Loaded` keyword
    which also accepts a custom timeout.

    Examples:
    | Click Link | link text    |            |          | # A page is expected to load. |
    | Click Link | another link | don't wait |          | # A page is not expected to load. |
    | Select Radio Button | group1 | value1  |          | # A page is not expected to load. |
    | Select Radio Button | group2 | value2  | and wait | # A page is expected to load. |

    = Timeouts =

    How much to wait when a new page is loaded is specified by a timeout
    that can be given in `importing` (default is 5 seconds) or dynamically
    with `Set Selenium Timeout` keyword.

    There are also several `Wait ...` keywords that take timeout as an
    argument. Starting from SeleniumLibrary 2.6 all these timeouts are
    optional and the same timeout used with page loads is used as a default.

    All timeouts can be given as numbers considered seconds (e.g. 0.5 or 42)
    or in Robot Framework's time syntax (e.g. '1.5 seconds' or '1 min 30 s').
    For more information about the time syntax see:
    http://robotframework.org/robotframework/latest/RobotFrameworkUserGuide.html#time-format.

    = Testing sites using https =

    Usually, https works out of the box. However, there may be trouble with
    self-signed certificates. We have a Wiki page describing how to test
    against these, using Firefox:
    https://github.com/robotframework/SeleniumLibrary/wiki/HandlingSelfSignedCertificates
    """

    ROBOT_LIBRARY_SCOPE = 'GLOBAL'
    ROBOT_LIBRARY_VERSION = __version__

    def __init__(self, timeout=5.0, server_host='localhost', server_port=4444,
                 jar_path=None, run_on_failure='Capture Screenshot'):
        """SeleniumLibrary can be imported with optional arguments.

        `timeout` is the default timeout used to wait for page load actions.
        It can be later set with `Set Selenium Timeout`

        `server_host` and `server_port` are used to connect to Selenium Server.
        Browsers opened with this SeleniumLibrary instance will be attached to
        that server. Note that the Selenium Server must be running before `Open
        Browser` keyword can be used. Selenium Server can be started with
        keyword `Start Selenium Server`. Starting from SeleniumLibrary 2.6.1,
        it is possible to give `server_host` as a URL with a possible embedded
        port, for example `http://192.168.52.1:4444`. If `server_host` contains
        port, the value of `server_port` is ignored.

        `jar_path` is the absolute path to the selenium-server.jar file to be
        used by the library. If set, a custom, modified version can be started
        instead of the default one distributed with the library.

        `run_on_failure` specifies the name of a SeleniumLibrary keyword to
        execute when another SeleniumLibrary keyword fails. By default
        `Capture Screenshot` will be used to take a screenshot of the situation.
        Using any value that is not a keyword name will disable this feature
        altogether. See `Register Keyword To Run On Failure` keyword for more
        information about this functionality that was added in SeleniumLibrary
        2.5.

        Because there are many optional arguments, it is often a good idea to
        use the handy named-arguments syntax supported by Robot Framework 2.5
        and later. This is demonstrated by the last two examples below.

        Examples:
        | Library | SeleniumLibrary | 15 | | | # Sets default timeout |
        | Library | SeleniumLibrary | | | 4455 | # Use default timeout and host but specify different port. |
        | Library | SeleniumLibrary | server_host=http://192.168.52.1:4444 | | | # Host as URL. |
        | Library | SeleniumLibrary | run_on_failure=Nothing | | | # Do nothing on failure. |
        """
        self._cache = utils.ConnectionCache()
        self._selenium = NoBrowserOpen()
        self.set_selenium_timeout(timeout or 5.0)
        self._server_host, self._server_port \
                = self._parse_host_and_port(server_host or 'localhost',
                                            server_port or 4444)
        self._jar_path = jar_path
        self._set_run_on_failure(run_on_failure)
        self._selenium_log = None
        self._locator_parser = LocatorParser(self)
        self._namegen = _NameGenerator()

    def _parse_host_and_port(self, host, port):
        if '://' in host:
            host = urlparse.urlparse(host).netloc
        if ':' in host:
            host, port = host.split(':', 1)
        return host, int(port)

    @property
    def selenium(self):
        """The selenium remote control instance of currently active browser."""
        return self._selenium

    def start_selenium_server(self, *params):
        """Starts the Selenium Server provided with SeleniumLibrary.

        `params` can contain additional command line options given to the
        Selenium Server. This keyword uses some command line options
        automatically:

        1) The port given in `importing` is added to `params` automatically
        using the `-port` option.

        2) A custom Firefox profile that is included with the library
        and contains automation friendly settings is enabled via the
        `-firefoxProfileTemplate` option. You can override this
        profile with your own custom profile by using the same argument
        in `params` yourself. To use the default profile on your machine,
        use this argument with `DEFAULT` value (case-sensitive).

        3) Starting from SeleniumLibrary 2.6, if there is `user-extensions.js`
        file in the same directory as Selenium Server jar, it is loaded using
        the `-userExtensions` option.  This is not done if the option is
        defined in `params`.  By default, such extension file providing Flex
        testing support is loaded automatically.

        Special syntax `JVM=some jvm opts` can be used to define options to
        the java command itself used to start the selenium server. This
        possibility was added in SeleniumLibrary 2.9.1.

        Examples:
        | Start Selenium Server | | | # Default settings. Uses the Firefox profile supplied with the library. |
        | Start Selenium Server | -firefoxProfileTemplate | C:\\\\the\\\\path | # Uses custom Firefox profile. |
        | Start Selenium Server | -firefoxProfileTemplate | DEFAULT | # Uses default Firefox profile on your machine. |
        | Start Selenium Server | -avoidProxy | -ensureCleanSession | # Uses various Selenium Server settings. |
        | Start Selenium Server | JVM=-DserverName=somehost | # Define JVM options. |

        All Selenium Server output is written into `selenium_server_log.txt`
        file in the same directory as the Robot Framework log file.

        If the test execution round starts and stops Selenium Server multiple
        times, it is best to open the server to different port each time.

        *NOTE:* This keyword requires `subprocess` module which is available
        on Python/Jython 2.5 or newer.
        """
        params = ('-port', str(self._server_port)) + params
        logpath = os.path.join(self._get_log_dir(), 'selenium_server_log.txt')
        self._selenium_log = open(logpath, 'w')
        start_selenium_server(self._selenium_log, self._jar_path, *params)
        self._html('Selenium server log is written to <a href="file://%s">%s</a>.'
                   % (logpath.replace('\\', '/'), logpath))

    def _get_log_dir(self):
        variables = BuiltIn().get_variables()
        logfile = variables['${LOG FILE}']
        if logfile != 'NONE':
            return os.path.dirname(logfile)
        return variables['${OUTPUTDIR}']

    def stop_selenium_server(self):
        """Stops the selenium server (and closes all browsers)."""
        shut_down_selenium_server(self._server_host, self._server_port)
        self._selenium = NoBrowserOpen()
        if self._selenium_log:
            self._selenium_log.close()

    def set_selenium_timeout(self, seconds):
        """Sets the timeout used by various keywords.

        Keywords that expect a page load to happen will fail if the page
        is not loaded within the timeout specified with `seconds`.
        Starting from SeleniumLibrary 2.6, this timeout is also the default
        timeout with various `Wait ...` keywords. See `introduction` for
        more information about timeouts and handling page loads.

        The previous timeout value is returned by this keyword and can
        be used to set the old value back later. The default timeout
        is 5 seconds, but it can be altered in `importing`.

        Example:
        | ${orig timeout} = | Set Selenium Timeout | 15 seconds |
        | Open page that loads slowly |
        | Set Selenium Timeout | ${orig timeout} |
        """
        timeout = utils.timestr_to_secs(seconds)
        old = getattr(self, '_timeout', timeout)
        self._timeout = timeout
        self._selenium.set_timeout(timeout * 1000)
        return utils.secs_to_timestr(old)

    def set_selenium_speed(self, seconds):
        """Sets the delay that is waited after each Selenium command.

        This is useful mainly in slowing down the test execution to be able to
        view the execution.  `seconds` may be given in Robot Framework time
        format.  Returns the previous speed value.

        Example:
        | Set Selenium Speed | 2 seconds |
        """
        old = self._selenium.get_speed()
        seconds = str(int(utils.timestr_to_secs(seconds)*1000))
        self._selenium.set_speed(seconds)
        return utils.secs_to_timestr(float(old)/1000)

    def call_selenium_api(self, method_name, *args):
        """Calls a method in the Selenium remote control API directly.

        This keyword can be used if some functionality provided by
        Selenium is not yet exposed as a keyword.

        `method_name` is the name of the method to call in the Selenium API and
        `args` specify the arguments it expects.

        The keyword first tries to find a method in Selenium's Python API
        provided by the `selenium.py` file. If no matching method is found, the
        keyword calls the Selenium Server's Remote Controller API directly via
        the `do_command` method in the Python API [1]. In both cases the keyword
        returns the return value of the call directly without any modifications
        or verifications.

        Examples:
        | ${ret} = | Call Selenium API | is_element_present | # Python API |
        | Call Selenium API | double_click | element_id | # Python API |
        | Call Selenium API | doubleClick  | element_id | # RC API |

        [1] http://release.seleniumhq.org/selenium-remote-control/1.0-beta-2/doc/python/
        """
        try:
            method = getattr(self._selenium, method_name)
        except AttributeError:
            method = lambda *args: self._selenium.do_command(method_name, args)
        return method(*args)

    def add_location_strategy(self, strategy_name, function_definition):
        """Adds a custom location strategy.

        `strategy_name` is the name of the strategy; a prefix used when
        addressing an element.

        `function_definition` is the JavaScript that will be called. It must
        return a DOM reference, an array with DOM references, or null.

        Together with the modified selenium-server.jar it can provide a new
        method of locating elements on the page. For example, a jQuery
        strategy can be added to locate elements given jQuery selector syntax.

        For jQuery selector setup see:
        https://github.com/robotframework/SeleniumLibrary/wiki/jQueryElementSelectors

        Example:
        | Add Location Strategy | jquery | return Selenium.prototype.locateElementByJQuerySelector(locator, inDocument, inWindow); |
        | Page Should Contain Element | jquery=div.#data-table |
        """
        self._locator_parser.add_strategy(strategy_name)
        self._selenium.add_location_strategy(strategy_name, function_definition)

    def _log(self, message, level='INFO'):
        if level != 'NONE':
            print '*%s* %s' % (level, message)

    def _info(self, message):
        self._log(message)

    def _debug(self, message):
        self._log(message, 'DEBUG')

    def _warn(self, message):
        self._log(message,  "WARN")

    def _html(self, message):
        self._log(message, 'HTML')

    def _parse_locator(self, locator, tag=None):
        parsed_locator = self._locator_parser.locator_for(locator, tag)
        self._debug("Parsed locator '%s' to search expression '%s'"
                    % (locator, parsed_locator))
        return parsed_locator

    def _get_error_message(self, exception):
        # Cannot use unicode(exception) because it fails on Python 2.5 and
        # earlier if the message contains non-ASCII chars.
        # See for details: http://bugs.jython.org/issue1585
        return unicode(exception.args and exception.args[0] or '')

    def _error_contains(self, exception, message):
        return message in self._get_error_message(exception)

    def _wait_until(self, timeout, error, function, *args):
        timeout = self._get_timeout(timeout)
        error = error.replace('<TIMEOUT>', utils.secs_to_timestr(timeout))
        maxtime = time.time() + timeout
        while not function(*args):
            if time.time() > maxtime:
                raise AssertionError(error)
            time.sleep(0.2)

    def _get_timeout(self, timeout=None):
        if timeout:
            return utils.timestr_to_secs(timeout)
        return self._timeout

    def _log_list(self, items, what='item'):
        msg = ['Altogether %d %s%s.' % (len(items), what, ['s',''][len(items)==1])]
        for index, item in enumerate(items):
            msg.append('%d: %s' % (index+1, item))
        self._info('\n'.join(msg))
        return items
Ejemplo n.º 4
0
class SeleniumLibrary(Browser, Page, Button, Click, JavaScript, Mouse, Select,
                      Element, Screenshot, Table, TextField, Flex):
    """SeleniumLibrary is a web testing library for Robot Test Automation Framework.

    It uses the Selenium Remote Control tool internally to control a web browser.
    See http://selenium-rc.openqa.org/ for more information on Selenium tool.

    SeleniumLibrary runs tests in a real browser instance. It should work in
    most modern browsers and can be used with both Python and Jython interpreters.

    *Before running the tests*

    Prior to running test cases using SeleniumLibrary, the Selenium Server must
    be started. This can be done using keyword `Start Selenium Server` or from
    the command line by using command: `java -jar
    /path/to/selenium-server.jar`. The Selenium Server is included in the
    SeleniumLibrary distribution and can be found under
    `[PythonLibs]/site-packages/SeleniumLibrary/lib`. Additionally, `Open
    Browser` keyword must be used in order to open browser in the desired
    location before any other keyword from the library may be used.

    *Locating elements*

    All keywords in SeleniumLibrary that need to find an element on the page
    take an argument, `locator`. In the most common case, `locator` is
    matched against the values of key attributes of the particular element type.
    For example, `id` and `name` are key attributes to all elements, and
    locating elements is easy using just the `id` as a `locator`.

    Asterisk character may be used as a wildcard in locators, but it only works
    as the last character of the expression. In the middle of the locator it
    is interpreted as literal '*'.

    It is also possible to give an arbitrary XPath or DOM expression as
    `locator`. In this case, the expression must be prefixed with either
    'xpath=' or 'dom='.

    Examples:
    | Click Link      | my link | # Matches if either link text or 'id', 'name' or 'href' of a link equals 'my link' |
    | Page Should Contain Link | Link id * | # Passes if the page contain any link starting with 'Link id' |
    | Select Checkbox | xpath=//table[0]/input[@name='my_checkbox'] | # Using XPath |
    | Click Image     | dom=document.images[56] | # Using a DOM expression |

    Table related keywords, such as `Table Should Contain`, allow identifying
    tables either by an id, by a CSS locator, or by an XPath expression.
    The XPath support was added in SeleniumLibrary 2.6.

    Examples:
    | Table Should Contain | tableID | text |
    | Table Should Contain | css=h2.someClass ~ table:last-child() | text |
    | Table Should Contain | xpath=//table/[@name="myTable"] | text |

    *Locating Flex elements*

    SeleniumLibary 2.6 and newer support testing Adobe Flex and Flash
    applications using Flex Pilot tool. For more information, including the
    required bootstrapping, see
    http://code.google.com/p/robotframework-seleniumlibrary/wiki/FlexTesting

    By default Flex elements are located based on `id` they have in Flex source
    code. Other supported locators are `name`, `automationName`, `text`,
    `htmlText`, `label` and xpath-like `chain`. To use them, you need to prefix
    the value with the locator type like `name=example`. Locators also support
    `*` as a wildcard.

    Examples:
    | Click Flex Element | foo          | # Search element by id |
    | Click Flex Element | name=myName  | # Search element by name |
    | Click Flex Element | label=Hello! | # Search element by label text |
    | Click Flex Element | chain=id:someId/name:someName | # Search element first by id and then its child by name |
    | Click Flex Element | name=wild*   | # Name with wildcard |
    | Click Flex Element | chain=name:*llo/name:world | # Chain with wildcard |

    *Handling page load events*

    Some keywords that may cause a page to load take an additional argument
    `dont_wait` that is used to determine whether a new page is expected to
    load or not. By default, a page load is expected to happen whenever a link
    or image is clicked, or a form submitted. If a page load does not happen
    (if the link only executes some JavaScript, for example), a non-empty value
    must be given for the `dont_wait` argument. How much to wait is determined
    by a timeout discussed in the next section.

    There are also some keywords that may cause a page to load but by default
    we expect them not to. For these cases, the keywords have an optional `wait`
    argument, and providing a non-empty value for it will cause the keyword to
    wait. An other possibility is using `Wait Until Page Loaded` keyword
    which also accepts a custom timeout.

    Examples:
    | Click Link | link text    |            |          | # A page is expected to load. |
    | Click Link | another link | don't wait |          | # A page is not expected to load. |
    | Select Radio Button | group1 | value1  |          | # A page is not expected to load. |
    | Select Radio Button | group2 | value2  | and wait | # A page is expected to load. |

    *Timeouts*

    How much to wait when a new page is loaded is specified by a timeout
    that can be given in `importing` (default is 5 seconds) or dynamically
    with `Set Selenium Timeout` keyword.

    There are also several `Wait ...` keywords that take timeout as an
    argument. Starting from SeleniumLibrary 2.6 all these timeouts are
    optional and the same timeout used with page loads is used as a default.

    All timeouts can be given as numbers considered seconds (e.g. 0.5 or 42)
    or in Robot Framework's time syntax (e.g. '1.5 seconds' or '1 min 30 s').
    For more information about the time syntax see:
    http://robotframework.googlecode.com/svn/trunk/doc/userguide/RobotFrameworkUserGuide.html#time-format.

    *Testing sites using https*

    Usually, https works out of the box. However, there may be trouble with
    self-signed certificates. We have a Wiki page describing how to test
    against these, using Firefox:
    http://code.google.com/p/robotframework-seleniumlibrary/wiki/HandlingSelfSignedCertificates
    """

    ROBOT_LIBRARY_SCOPE = 'GLOBAL'
    ROBOT_LIBRARY_VERSION = __version__

    def __init__(self, timeout=5.0, server_host='localhost', server_port=4444,
                 jar_path=None, run_on_failure='Capture Screenshot'):
        """SeleniumLibrary can be imported with optional arguments.

        `timeout` is the default timeout used to wait for page load actions.
        It can be later set with `Set Selenium Timeout`

        `host` and `port` are used to connect to Selenium Server. Browsers
        opened with this SeleniumLibrary instance will be attached to that
        server. Note that the Selenium Server must be running before `Open
        Browser` keyword can be used. Selenium Server can be started with
        keyword `Start Selenium Server`.

        `jar_path` is the absolute path to the selenium-server.jar file to be
        used by the library. If set, a custom, modified version can be started
        instead of the default one distributed with the library.

        `run_on_failure` specifies the name of a SeleniumLibrary keyword to
        execute when another SeleniumLibrary keyword fails. By default
        `Capture Screenshot` will be used to take a screenshot of the situation.
        Using any value that is not a keyword name will disable this feature
        altogether. See `Register Keyword To Run On Failure` keyword for more
        information about this functionality that was added in SeleniumLibrary
        2.5.

        Because there are many optional arguments, it is often a good idea to
        use the handy named-arguments syntax supported by Robot Framework 2.5
        and later. This is demonstrated by the last example below.

        Examples:
        | Library | SeleniumLibrary | 15 | | | # Sets default timeout |
        | Library | SeleniumLibrary | | | 4455 | # Use default timeout and host but specify different port. |
        | Library | SeleniumLibrary | run_on_failure=Nothing | # Do nothing on failure. |
        """
        self._cache = utils.ConnectionCache()
        self._selenium = _NoBrowser()
        self.set_selenium_timeout(timeout or 5.0)
        self._server_host = server_host or 'localhost'
        self._server_port = int(server_port or 4444)
        self._jar_path = jar_path
        self._set_run_on_failure(run_on_failure)
        self._selenium_log = None
        self._locator_parser = LocatorParser(self)
        self._namegen = _NameGenerator()

    def start_selenium_server(self, *params):
        """Starts the Selenium Server provided with SeleniumLibrary.

        `params` can contain additional command line options given to the
        Selenium Server. This keyword uses some command line options
        automatically:

        1) The port given in `importing` is added to `params` automatically
        using the `-port` option.

        2) A custom Firefox profile that is included with the library
        and contains automation friendly settings is enabled via the
        `-firefoxProfileTemplate` option. You can override this
        profile with your own custom profile by using the same argument
        in `params` yourself. To use the default profile on your machine,
        use this argument with `DEFAULT` value (case-sensitive). Using a
        custom Firefox profile automatically is a new feature in
        SeleniumLibrary 2.5. For more information see
        http://code.google.com/p/robotframework-seleniumlibrary/wiki/CustomFirefoxProfile

        3) Starting from SeleniumLibrary 2.6, if there is `user-extensions.js`
        file in the same directory as Selenium Server jar, it is loaded using
        the `-userExtensions` option.  This is not done if the option is
        defined in `params`.  By default, such extension file providing Flex
        testing support is loaded automatically.

        Examples:
        | Start Selenium Server | | | # Default settings. Uses the Firefox profile supplied with the library. |
        | Start Selenium Server | -firefoxProfileTemplate | C:\\\\the\\\\path | # Uses custom Firefox profile. |
        | Start Selenium Server | -firefoxProfileTemplate | DEFAULT | # Uses default Firefox profile on your machine. |
        | Start Selenium Server | -avoidProxy | -ensureCleanSession | # Uses various Selenium Server settings. |

        All Selenium Server output is written into `selenium_server_log.txt`
        file in the same directory as the Robot Framework log file.

        If the test execution round starts and stops Selenium Server multiple
        times, it is best to open the server to different port each time.

        *NOTE:* This keyword requires `subprocess` module which is available
        on Python/Jython 2.5 or newer.
        """
        params = ('-port', str(self._server_port)) + params
        logpath = os.path.join(self._get_log_dir(), 'selenium_server_log.txt')
        self._selenium_log = open(logpath, 'w')
        start_selenium_server(self._selenium_log, self._jar_path, *params)
        self._html('Selenium server log is written to <a href="file://%s">%s</a>.'
                   % (logpath.replace('\\', '/'), logpath))

    def _get_log_dir(self):
        logfile = GLOBAL_VARIABLES['${LOG FILE}']
        if logfile != 'NONE':
            return os.path.dirname(logfile)
        return GLOBAL_VARIABLES['${OUTPUTDIR}']

    def stop_selenium_server(self):
        """Stops the selenium server (and closes all browsers)."""
        shut_down_selenium_server(self._server_host, self._server_port)
        self._selenium = _NoBrowser()
        if self._selenium_log:
            self._selenium_log.close()

    def open_browser(self, url, browser='firefox', alias=None):
        """Opens a new browser instance to given URL.

        Possible already opened connections are cached.

        Returns the index of this browser instance which can be used later to
        switch back to it. Index starts from 1 and is reset back to it when
        `Close All Browsers` keyword is used. See `Switch Browser` for
        example.

        Optional alias is a alias for the browser instance and it can be used
        for switching between browsers similarly as the index. See `Switch
        Browser` for more details about that.

        Possible values for `browser` are all the values supported by Selenium
        and some aliases that are defined for convenience. The table below
        lists the aliases for most common supported browsers.

        | firefox          | FireFox   |
        | ff               | FireFox   |
        | ie               | Internet Explorer |
        | internetexplorer | Internet Explorer |
        | safari           | Safari |
        | googlechrome     | Google Chrome |
        | opera            | Opera |

        Additionally, a string like `*custom /path/to/browser-executable` can
        be used to specify the browser directly. In this case, the path needs to
        point to an executable, not a script, otherwise the library may not be
        able to shut down the browser properly.

        Note, that you will encounter strange behavior, if you open
        multiple Internet Explorer browser instances. That is also why
        `Switch Browser` only works with one IE browser at most.
        For more information see:
        http://selenium-grid.seleniumhq.org/faq.html#i_get_some_strange_errors_when_i_run_multiple_internet_explorer_instances_on_the_same_machine
        """
        self._info("Opening browser '%s' to base url '%s'" % (browser, url))
        browser = self._get_browser(browser)
        self._selenium = selenium(self._server_host, self._server_port, browser,
                                  url)
        self._connect_to_selenium_server()
        self._selenium.set_timeout(self._timeout * 1000)
        self._selenium.open(url, ignoreResponseCode=True)
        self._debug('Opened browser with Selenium session id %s'
                    % self._selenium.sessionId)
        return self._cache.register(self._selenium, alias)

    def _get_browser(self, browser):
        return BROWSER_ALIASES.get(browser.lower().replace(' ', ''), browser)

    def _connect_to_selenium_server(self):
        timeout = time.time() + SELENIUM_CONNECTION_TIMEOUT
        while time.time() < timeout:
            try:
                self._selenium.start()
            # AssertionError occurs on Jython: http://bugs.jython.org/issue1697
            except (socket.error, AssertionError):
                time.sleep(2)
            else:
                return
        self._selenium = _NoBrowser()
        raise RuntimeError("Could not connect to Selenium Server in %d seconds. "
                           "Please make sure Selenium Server is running."
                           % SELENIUM_CONNECTION_TIMEOUT)

    def close_browser(self):
        """Closes the current browser."""
        if self._selenium:
            self._debug('Closing browser with Selenium session id %s'
                        % self._selenium.sessionId)
            self._selenium.stop()
            self._cache.current = None
            self._selenium = _NoBrowser()

    def close_all_browsers(self):
        """Closes all open browsers and empties the connection cache.

        After this keyword new indexes get from Open Browser keyword are reset
        to 1.

        This keyword should be used in test or suite teardown to make sure
        all browsers are closed.
        """
        # ConnectionCache's connections attribute was renamed to _connections
        # in RF 2.0.2 (which was actually a stupid decision)
        try:
            connections = self._cache._connections
        except AttributeError:
            connections = self._cache.connections
        for sel in connections:
            if sel is not None and sel.sessionId is not None:
                self._debug('Closing browser with Selenium session id %s'
                            % sel.sessionId)
                sel.stop()
        self._selenium = _NoBrowser()
        self._cache.empty_cache()

    def switch_browser(self, index_or_alias):
        """Switches between active browsers using index or alias.

        Index is got from `Open Browser` and alias can be given to it.

        Examples:
        | Open Browser        | http://google.com | ff       |
        | Location Should Be  | http://google.com |          |
        | Open Browser        | http://yahoo.com  | ie       | 2nd conn |
        | Location Should Be  | http://yahoo.com  |          |
        | Switch Browser      | 1                 | # index  |
        | Page Should Contain | I'm feeling lucky |          |
        | Switch Browser      | 2nd conn          | # alias  |
        | Page Should Contain | More Yahoo!       |          |
        | Close All Browsers  |                   |          |

        Above example expects that there was no other open browsers when
        opening the first one because it used index '1' when switching to it
        later. If you aren't sure about that you can store the index into
        a variable as below.

        | ${id} =            | Open Browser  | http://google.com | *firefox |
        | # Do something ... |
        | Switch Browser     | ${id}         |                   |          |
        """
        try:
            self._selenium = self._cache.switch(index_or_alias)
            self._debug('Switched to browser with Selenium session id %s'
                         % self._selenium.sessionId)
        except DataError:
            raise RuntimeError("No browser with index or alias '%s' found."
                               % index_or_alias)

    def set_selenium_timeout(self, seconds):
        """Sets the timeout used by various keywords.

        Keywords that expect a page load to happen will fail if the page
        is not loaded within the timeout specified with `seconds`.
        Starting from SeleniumLibrary 2.6, this timeout is also the default
        timeout with various `Wait ...` keywords. See `introduction` for
        more information about timeouts and handling page loads.

        The previous timeout value is returned by this keyword and can
        be used to set the old value back later. The default timeout
        is 5 seconds, but it can be altered in `importing`.

        Example:
        | ${orig timeout} = | Set Selenium Timeout | 15 seconds |
        | Open page that loads slowly |
        | Set Selenium Timeout | ${orig timeout} |
        """
        timeout = utils.timestr_to_secs(seconds)
        old = getattr(self, '_timeout', timeout)
        self._timeout = timeout
        self._selenium.set_timeout(timeout * 1000)
        return utils.secs_to_timestr(old)

    def set_selenium_speed(self, seconds):
        """Sets the delay that is waited after each Selenium command.

        This is useful mainly in slowing down the test execution to be able to
        view the execution.  `seconds` may be given in Robot Framework time
        format.  Returns the previous speed value.

        Example:
        | Set Selenium Speed | 2 seconds |
        """
        old = self._selenium.get_speed()
        seconds = str(int(utils.timestr_to_secs(seconds)*1000))
        self._selenium.set_speed(seconds)
        return utils.secs_to_timestr(float(old)/1000)

    def call_selenium_api(self, method_name, *args):
        """Calls a method in the Selenium remote control API directly.

        This keyword can be used if some functionality provided by
        Selenium is not yet exposed as a keyword.

        `method_name` is the name of the method to call in the Selenium API and
        `args` specify the arguments it expects.

        The keyword first tries to find a method in Selenium's Python API
        provided by the `selenium.py` file. If no matching method is found, the
        keyword calls the Selenium Server's Remote Controller API directly via
        the `do_command` method in the Python API [1]. In both cases the keyword
        returns the return value of the call directly without any modifications
        or verifications.

        Examples:
        | ${ret} = | Call Selenium API | is_element_present | # Python API |
        | Call Selenium API | double_click | element_id | # Python API |
        | Call Selenium API | doubleClick  | element_id | # RC API |

        [1] http://release.seleniumhq.org/selenium-remote-control/1.0-beta-2/doc/python/
        """
        try:
            method = getattr(self._selenium, method_name)
        except AttributeError:
            method = lambda *args: self._selenium.do_command(method_name, args)
        return method(*args)

    def add_location_strategy(self, strategy_name, function_definition):
        """Adds a custom location strategy.

        `strategy_name` is the name of the strategy; a prefix used when
        addressing an element.

        `function_definition` is the JavaScript that will be called. It must
        return a DOM reference, an array with DOM references, or null.

        Together with the modified selenium-server.jar it can provide a new
        method of locating elements on the page. For example, a jQuery
        strategy can be added to locate elements given jQuery selector syntax.

        For jQuery selector setup see:
        http://code.google.com/p/robotframework-seleniumlibrary/wiki/jQueryElementSelectors

        Example:
        | Add Location Strategy | jquery | return Selenium.prototype.locateElementByJQuerySelector(locator, inDocument, inWindow); |
        | Page Should Contain Element | jquery=div.#data-table |
        """
        self._locator_parser.add_strategy(strategy_name)
        self._selenium.add_location_strategy(strategy_name, function_definition)

    def _log(self, message, level='INFO'):
        if level != 'NONE':
            print '*%s* %s' % (level, message)

    def _info(self, message):
        self._log(message)

    def _debug(self, message):
        self._log(message, 'DEBUG')

    def _warn(self, message):
        self._log(message,  "WARN")

    def _html(self, message):
        self._log(message, 'HTML')

    def _parse_locator(self, locator, tag=None):
        parsed_locator = self._locator_parser.locator_for(locator, tag)
        self._debug("Parsed locator '%s' to search expression '%s'"
                    % (locator, parsed_locator))
        return parsed_locator

    def _get_error_message(self, exception):
        # Cannot use unicode(exception) because it fails on Python 2.5 and
        # earlier if the message contains non-ASCII chars.
        # See for details: http://bugs.jython.org/issue1585
        return unicode(exception.args and exception.args[0] or '')

    def _error_contains(self, exception, message):
        return message in self._get_error_message(exception)

    def _wait_until(self, timeout, error, function, *args):
        timeout = self._get_timeout(timeout)
        error = error % {'TIMEOUT': utils.secs_to_timestr(timeout)}
        maxtime = time.time() + timeout
        while not function(*args):
            if time.time() > maxtime:
                raise AssertionError(error)
            time.sleep(0.2)

    def _get_timeout(self, timeout=None):
        if timeout:
            return utils.timestr_to_secs(timeout)
        return self._timeout

    def _log_list(self, items, what='item'):
        msg = ['Altogether %d %s%s.' % (len(items), what, ['s',''][len(items)==1])]
        for index, item in enumerate(items):
            msg.append('%d: %s' % (index+1, item))
        self._info('\n'.join(msg))
        return items
Ejemplo n.º 5
0
class SeleniumLibrary(Browser, Page, Button, Click, JavaScript, Mouse, Select,
                      Element, Screenshot, Table, TextField, Flex):
    """SeleniumLibrary is a *deprecated* web testing library for Robot Framework.

    It uses the Selenium Remote Control tool internally to control a web browser.
    See http://selenium-rc.openqa.org/ for more information on Selenium tool.

    SeleniumLibrary runs tests in a real browser instance. It should work in
    most modern browsers and can be used with both Python and Jython interpreters.

    = Before running the tests =

    Prior to running test cases using SeleniumLibrary, the Selenium Server must
    be started. This can be done using keyword `Start Selenium Server` or from
    the command line by using command: `java -jar
    /path/to/selenium-server.jar`. The Selenium Server is included in the
    SeleniumLibrary distribution and can be found under
    `[PythonLibs]/site-packages/SeleniumLibrary/lib`. Additionally, `Open
    Browser` keyword must be used in order to open browser in the desired
    location before any other keyword from the library may be used.

    = Locating elements =

    All keywords in SeleniumLibrary that need to find an element on the page
    take an argument, `locator`. In the most common case, `locator` is
    matched against the values of key attributes of the particular element type.
    For example, `id` and `name` are key attributes to all elements, and
    locating elements is easy using just the `id` as a `locator`.

    Asterisk character may be used as a wildcard in locators, but it only works
    as the last character of the expression. In the middle of the locator it
    is interpreted as literal '*'.

    It is also possible to give an arbitrary XPath or DOM expression as
    `locator`. In this case, the expression must be prefixed with either
    'xpath=' or 'dom='.

    Examples:
    | Click Link      | my link | # Matches if either link text or 'id', 'name' or 'href' of a link equals 'my link' |
    | Page Should Contain Link | Link id * | # Passes if the page contain any link starting with 'Link id' |
    | Select Checkbox | xpath=//table[0]/input[@name='my_checkbox'] | # Using XPath |
    | Click Image     | dom=document.images[56] | # Using a DOM expression |

    Table related keywords, such as `Table Should Contain`, allow identifying
    tables either by an id, by a CSS locator, or by an XPath expression.
    The XPath support was added in SeleniumLibrary 2.6.

    Examples:
    | Table Should Contain | tableID | text |
    | Table Should Contain | css=h2.someClass ~ table:last-child() | text |
    | Table Should Contain | xpath=//table/[@name="myTable"] | text |

    = Locating Flex elements =

    SeleniumLibary 2.6 and newer support testing Adobe Flex and Flash
    applications using Flex Pilot tool. For more information, including the
    required bootstrapping, see
    https://github.com/robotframework/SeleniumLibrary/wiki/FlexTesting

    By default Flex elements are located based on `id` they have in Flex source
    code. Other supported locators are `name`, `automationName`, `text`,
    `htmlText`, `label` and xpath-like `chain`. To use them, you need to prefix
    the value with the locator type like `name=example`. Locators also support
    `*` as a wildcard.

    Examples:
    | Click Flex Element | foo          | # Search element by id |
    | Click Flex Element | name=myName  | # Search element by name |
    | Click Flex Element | label=Hello! | # Search element by label text |
    | Click Flex Element | chain=id:someId/name:someName | # Search element first by id and then its child by name |
    | Click Flex Element | name=wild*   | # Name with wildcard |
    | Click Flex Element | chain=name:*llo/name:world | # Chain with wildcard |

    = Handling page load events =

    Some keywords that may cause a page to load take an additional argument
    `dont_wait` that is used to determine whether a new page is expected to
    load or not. By default, a page load is expected to happen whenever a link
    or image is clicked, or a form submitted. If a page load does not happen
    (if the link only executes some JavaScript, for example), a non-empty value
    must be given for the `dont_wait` argument. How much to wait is determined
    by a timeout discussed in the next section.

    There are also some keywords that may cause a page to load but by default
    we expect them not to. For these cases, the keywords have an optional `wait`
    argument, and providing a non-empty value for it will cause the keyword to
    wait. An other possibility is using `Wait Until Page Loaded` keyword
    which also accepts a custom timeout.

    Examples:
    | Click Link | link text    |            |          | # A page is expected to load. |
    | Click Link | another link | don't wait |          | # A page is not expected to load. |
    | Select Radio Button | group1 | value1  |          | # A page is not expected to load. |
    | Select Radio Button | group2 | value2  | and wait | # A page is expected to load. |

    = Timeouts =

    How much to wait when a new page is loaded is specified by a timeout
    that can be given in `importing` (default is 5 seconds) or dynamically
    with `Set Selenium Timeout` keyword.

    There are also several `Wait ...` keywords that take timeout as an
    argument. Starting from SeleniumLibrary 2.6 all these timeouts are
    optional and the same timeout used with page loads is used as a default.

    All timeouts can be given as numbers considered seconds (e.g. 0.5 or 42)
    or in Robot Framework's time syntax (e.g. '1.5 seconds' or '1 min 30 s').
    For more information about the time syntax see:
    http://robotframework.org/robotframework/latest/RobotFrameworkUserGuide.html#time-format.

    = Testing sites using https =

    Usually, https works out of the box. However, there may be trouble with
    self-signed certificates. We have a Wiki page describing how to test
    against these, using Firefox:
    https://github.com/robotframework/SeleniumLibrary/wiki/HandlingSelfSignedCertificates
    """

    ROBOT_LIBRARY_SCOPE = 'GLOBAL'
    ROBOT_LIBRARY_VERSION = __version__

    def __init__(self,
                 timeout=5.0,
                 server_host='localhost',
                 server_port=4444,
                 jar_path=None,
                 run_on_failure='Capture Screenshot'):
        """SeleniumLibrary can be imported with optional arguments.

        `timeout` is the default timeout used to wait for page load actions.
        It can be later set with `Set Selenium Timeout`

        `server_host` and `server_port` are used to connect to Selenium Server.
        Browsers opened with this SeleniumLibrary instance will be attached to
        that server. Note that the Selenium Server must be running before `Open
        Browser` keyword can be used. Selenium Server can be started with
        keyword `Start Selenium Server`. Starting from SeleniumLibrary 2.6.1,
        it is possible to give `server_host` as a URL with a possible embedded
        port, for example `http://192.168.52.1:4444`. If `server_host` contains
        port, the value of `server_port` is ignored.

        `jar_path` is the absolute path to the selenium-server.jar file to be
        used by the library. If set, a custom, modified version can be started
        instead of the default one distributed with the library.

        `run_on_failure` specifies the name of a SeleniumLibrary keyword to
        execute when another SeleniumLibrary keyword fails. By default
        `Capture Screenshot` will be used to take a screenshot of the situation.
        Using any value that is not a keyword name will disable this feature
        altogether. See `Register Keyword To Run On Failure` keyword for more
        information about this functionality that was added in SeleniumLibrary
        2.5.

        Because there are many optional arguments, it is often a good idea to
        use the handy named-arguments syntax supported by Robot Framework 2.5
        and later. This is demonstrated by the last two examples below.

        Examples:
        | Library | SeleniumLibrary | 15 | | | # Sets default timeout |
        | Library | SeleniumLibrary | | | 4455 | # Use default timeout and host but specify different port. |
        | Library | SeleniumLibrary | server_host=http://192.168.52.1:4444 | | | # Host as URL. |
        | Library | SeleniumLibrary | run_on_failure=Nothing | | | # Do nothing on failure. |
        """
        self._cache = utils.ConnectionCache()
        self._selenium = NoBrowserOpen()
        self.set_selenium_timeout(timeout or 5.0)
        self._server_host, self._server_port \
                = self._parse_host_and_port(server_host or 'localhost',
                                            server_port or 4444)
        self._jar_path = jar_path
        self._set_run_on_failure(run_on_failure)
        self._selenium_log = None
        self._locator_parser = LocatorParser(self)
        self._namegen = _NameGenerator()

    def _parse_host_and_port(self, host, port):
        if '://' in host:
            host = urlparse.urlparse(host).netloc
        if ':' in host:
            host, port = host.split(':', 1)
        return host, int(port)

    @property
    def selenium(self):
        """The selenium remote control instance of currently active browser."""
        return self._selenium

    def start_selenium_server(self, *params):
        """Starts the Selenium Server provided with SeleniumLibrary.

        `params` can contain additional command line options given to the
        Selenium Server. This keyword uses some command line options
        automatically:

        1) The port given in `importing` is added to `params` automatically
        using the `-port` option.

        2) A custom Firefox profile that is included with the library
        and contains automation friendly settings is enabled via the
        `-firefoxProfileTemplate` option. You can override this
        profile with your own custom profile by using the same argument
        in `params` yourself. To use the default profile on your machine,
        use this argument with `DEFAULT` value (case-sensitive).

        3) Starting from SeleniumLibrary 2.6, if there is `user-extensions.js`
        file in the same directory as Selenium Server jar, it is loaded using
        the `-userExtensions` option.  This is not done if the option is
        defined in `params`.  By default, such extension file providing Flex
        testing support is loaded automatically.

        Special syntax `JVM=some jvm opts` can be used to define options to
        the java command itself used to start the selenium server. This
        possibility was added in SeleniumLibrary 2.9.1.

        Examples:
        | Start Selenium Server | | | # Default settings. Uses the Firefox profile supplied with the library. |
        | Start Selenium Server | -firefoxProfileTemplate | C:\\\\the\\\\path | # Uses custom Firefox profile. |
        | Start Selenium Server | -firefoxProfileTemplate | DEFAULT | # Uses default Firefox profile on your machine. |
        | Start Selenium Server | -avoidProxy | -ensureCleanSession | # Uses various Selenium Server settings. |
        | Start Selenium Server | JVM=-DserverName=somehost | # Define JVM options. |

        All Selenium Server output is written into `selenium_server_log.txt`
        file in the same directory as the Robot Framework log file.

        If the test execution round starts and stops Selenium Server multiple
        times, it is best to open the server to different port each time.

        *NOTE:* This keyword requires `subprocess` module which is available
        on Python/Jython 2.5 or newer.
        """
        params = ('-port', str(self._server_port)) + params
        logpath = os.path.join(self._get_log_dir(), 'selenium_server_log.txt')
        self._selenium_log = open(logpath, 'w')
        start_selenium_server(self._selenium_log, self._jar_path, *params)
        self._html(
            'Selenium server log is written to <a href="file://%s">%s</a>.' %
            (logpath.replace('\\', '/'), logpath))

    def _get_log_dir(self):
        variables = BuiltIn().get_variables()
        logfile = variables['${LOG FILE}']
        if logfile != 'NONE':
            return os.path.dirname(logfile)
        return variables['${OUTPUTDIR}']

    def stop_selenium_server(self):
        """Stops the selenium server (and closes all browsers)."""
        shut_down_selenium_server(self._server_host, self._server_port)
        self._selenium = NoBrowserOpen()
        if self._selenium_log:
            self._selenium_log.close()

    def set_selenium_timeout(self, seconds):
        """Sets the timeout used by various keywords.

        Keywords that expect a page load to happen will fail if the page
        is not loaded within the timeout specified with `seconds`.
        Starting from SeleniumLibrary 2.6, this timeout is also the default
        timeout with various `Wait ...` keywords. See `introduction` for
        more information about timeouts and handling page loads.

        The previous timeout value is returned by this keyword and can
        be used to set the old value back later. The default timeout
        is 5 seconds, but it can be altered in `importing`.

        Example:
        | ${orig timeout} = | Set Selenium Timeout | 15 seconds |
        | Open page that loads slowly |
        | Set Selenium Timeout | ${orig timeout} |
        """
        timeout = utils.timestr_to_secs(seconds)
        old = getattr(self, '_timeout', timeout)
        self._timeout = timeout
        self._selenium.set_timeout(timeout * 1000)
        return utils.secs_to_timestr(old)

    def set_selenium_speed(self, seconds):
        """Sets the delay that is waited after each Selenium command.

        This is useful mainly in slowing down the test execution to be able to
        view the execution.  `seconds` may be given in Robot Framework time
        format.  Returns the previous speed value.

        Example:
        | Set Selenium Speed | 2 seconds |
        """
        old = self._selenium.get_speed()
        seconds = str(int(utils.timestr_to_secs(seconds) * 1000))
        self._selenium.set_speed(seconds)
        return utils.secs_to_timestr(float(old) / 1000)

    def call_selenium_api(self, method_name, *args):
        """Calls a method in the Selenium remote control API directly.

        This keyword can be used if some functionality provided by
        Selenium is not yet exposed as a keyword.

        `method_name` is the name of the method to call in the Selenium API and
        `args` specify the arguments it expects.

        The keyword first tries to find a method in Selenium's Python API
        provided by the `selenium.py` file. If no matching method is found, the
        keyword calls the Selenium Server's Remote Controller API directly via
        the `do_command` method in the Python API [1]. In both cases the keyword
        returns the return value of the call directly without any modifications
        or verifications.

        Examples:
        | ${ret} = | Call Selenium API | is_element_present | # Python API |
        | Call Selenium API | double_click | element_id | # Python API |
        | Call Selenium API | doubleClick  | element_id | # RC API |

        [1] http://release.seleniumhq.org/selenium-remote-control/1.0-beta-2/doc/python/
        """
        try:
            method = getattr(self._selenium, method_name)
        except AttributeError:
            method = lambda *args: self._selenium.do_command(method_name, args)
        return method(*args)

    def add_location_strategy(self, strategy_name, function_definition):
        """Adds a custom location strategy.

        `strategy_name` is the name of the strategy; a prefix used when
        addressing an element.

        `function_definition` is the JavaScript that will be called. It must
        return a DOM reference, an array with DOM references, or null.

        Together with the modified selenium-server.jar it can provide a new
        method of locating elements on the page. For example, a jQuery
        strategy can be added to locate elements given jQuery selector syntax.

        For jQuery selector setup see:
        https://github.com/robotframework/SeleniumLibrary/wiki/jQueryElementSelectors

        Example:
        | Add Location Strategy | jquery | return Selenium.prototype.locateElementByJQuerySelector(locator, inDocument, inWindow); |
        | Page Should Contain Element | jquery=div.#data-table |
        """
        self._locator_parser.add_strategy(strategy_name)
        self._selenium.add_location_strategy(strategy_name,
                                             function_definition)

    def _log(self, message, level='INFO'):
        if level != 'NONE':
            print '*%s* %s' % (level, message)

    def _info(self, message):
        self._log(message)

    def _debug(self, message):
        self._log(message, 'DEBUG')

    def _warn(self, message):
        self._log(message, "WARN")

    def _html(self, message):
        self._log(message, 'HTML')

    def _parse_locator(self, locator, tag=None):
        parsed_locator = self._locator_parser.locator_for(locator, tag)
        self._debug("Parsed locator '%s' to search expression '%s'" %
                    (locator, parsed_locator))
        return parsed_locator

    def _get_error_message(self, exception):
        # Cannot use unicode(exception) because it fails on Python 2.5 and
        # earlier if the message contains non-ASCII chars.
        # See for details: http://bugs.jython.org/issue1585
        return unicode(exception.args and exception.args[0] or '')

    def _error_contains(self, exception, message):
        return message in self._get_error_message(exception)

    def _wait_until(self, timeout, error, function, *args):
        timeout = self._get_timeout(timeout)
        error = error.replace('<TIMEOUT>', utils.secs_to_timestr(timeout))
        maxtime = time.time() + timeout
        while not function(*args):
            if time.time() > maxtime:
                raise AssertionError(error)
            time.sleep(0.2)

    def _get_timeout(self, timeout=None):
        if timeout:
            return utils.timestr_to_secs(timeout)
        return self._timeout

    def _log_list(self, items, what='item'):
        msg = [
            'Altogether %d %s%s.' %
            (len(items), what, ['s', ''][len(items) == 1])
        ]
        for index, item in enumerate(items):
            msg.append('%d: %s' % (index + 1, item))
        self._info('\n'.join(msg))
        return items