.. currentmodule:: sardana.macroserver.macro .. _sardana-macros-scanframework: ============== Scan Framework ============== In general terms, we call *scan* to a macro that moves one or more motors and acquires data along the path of the motor(s). See the :ref:`introduction to the concept of scan in Sardana `. While a scan macro could be written from scratch, Sardana provides a higher- level API (the *scan framework*) that greatly simplifies the development of scan macros by taking care of the details about synchronization of motors and of acquisitions. The scan framework is implemented in the :mod:`~sardana.macroserver.scan` module, which provides the :class:`~sardana.macroserver.scan.GScan` base class and its specialized derived classes :class:`~sardana.macroserver.scan.SScan` and :class:`~sardana.macroserver.scan.CScan` for step and continuous scans, respectively. Creating a scan macro consists in writing a generic macro (see :ref:`the generic macro writing instructions `) in which an instance of :class:`~sardana.macroserver.scan.GScan` is created (typically in the :meth:`~Macro.prepare` method) which is then invoked in the :meth:`~Macro.run` method. Central to the scan framework is the :attr:`~sardana.macroserver.scan.GScan.generator` function, which must be passed to the GScan constructor. This generator is a function that allows to construct the path of the scan (see :class:`~sardana.macroserver.scan.GScan` for detailed information on the generator). A basic example on writing a step scan -------------------------------------- Step scans are built using an instance of the :class:`~sardana.macroserver.scan.SScan` class, which requires a step generator that defines the path for the motion. Since in a step scan the data is acquired at each step, the generator controls both the motion and the acquisition. Note that in general, the generator does not need to generate a determinate (or even finite) number of steps. Also note that it is possible to write generators that vary their current step based on the acquired values (e.g., changing step sizes as a function of some counter reading). The ``ascan_demo`` macro illustrates the most basic features of a step scan:: class ascan_demo(Macro): """ This is a basic reimplementation of the ascan` macro for demonstration purposes of the Generic Scan framework. The "real" implementation of :class:`sardana.macroserver.macros.ascan` derives from :class:`sardana.macroserver.macros.aNscan` and provides some extra features. """ hints = { 'scan' : 'ascan_demo'} #this is used to indicate other codes that the macro is a scan env = ('ActiveMntGrp',) #this hints that the macro requires the ActiveMntGrp environment variable to be set param_def = [ ['motor', Type.Moveable, None, 'Motor to move'], ['start_pos', Type.Float, None, 'Scan start position'], ['final_pos', Type.Float, None, 'Scan final position'], ['nb_interv', Type.Integer, None, 'Number of scan intervals'], ['integ_time', Type.Float, None, 'Integration time'] ] def prepare(self, motor, start_pos, final_pos, nb_interv, integ_time, **opts): #parse the user parameters self.start = numpy.array([start_pos], dtype='d') self.final = numpy.array([final_pos], dtype='d') self.integ_time = integ_time self.nb_points = nb_interv+1 self.interv_size = ( self.final - self.start) / nb_interv self.name='ascan_demo' env = opts.get('env',{}) #the "env" dictionary may be passed as an option #create an instance of GScan (in this case, of its child, SScan self._gScan=SScan(self, generator=self._generator, moveables=[motor], env=env) def _generator(self): step = {} step["integ_time"] = self.integ_time #integ_time is the same for all steps for point_no in xrange(self.nb_points): step["positions"] = self.start + point_no * self.interv_size #note that this is a numpy array step["point_id"] = point_no yield step def run(self,*args): for step in self._gScan.step_scan(): #just go through the steps yield step @property def data(self): return self._gScan.data #the GScan provides scan data The :class:`~sardana.macroserver.macros.examples.ascan_demo` shows only basic features of the scan framework, but it already shows that writing a step scan macro is mostly just a matter of writing a generator function. It also shows that the :class:`~sardana.macroserver.scan.GScan`'s :attr:`~sardana.macroserver.scan.GScan.data` property can be used to provide the needed return value of the :class:`~Macro`'s :attr:`~Macro.data`. A basic example on writing a continuous scan -------------------------------------------- Continuous scans are built using an instance of the :class:`~sardana.macroserver.scan.CScan` class. Since in the continuous scans the acquisition and motion are decoupled, CScan requires two independent generators: * a *waypoint generator*: which defines the path for the motion in a very similar way as the step generator does for a step scan. The steps generated by this generator are also called "waypoints". * a *period generator* which controls the data acquisition steps. Essentially, :class:`~sardana.macroserver.scan.CScan` implements the continuous scan as an acquisition loop (controlled by the period generator) nested within a motion loop (controlled by the waypoint generator). Note that each loop is run on an independent thread, and only limited communication occurs between the two (basically the acquisition starts at the beginning of each movement and ends when a waypoint is reached). The :class:`~sardana.macroserver.macros.examples.ascanc_demo` macro illustrates the most basic features of a continuous scan:: :: class ascanc_demo(Macro): """ This is a basic reimplementation of the ascanc` macro for demonstration purposes of the Generic Scan framework. The "real" implementation of :class:`sardana.macroserver.macros.ascanc` derives from :class:`sardana.macroserver.macros.aNscan` and provides some extra features. """ hints = { 'scan' : 'ascanc_demo'} #this is used to indicate other codes that the macro is a scan env = ('ActiveMntGrp',) #this hints that the macro requires the ActiveMntGrp environment variable to be set param_def = [ ['motor', Type.Moveable, None, 'Motor to move'], ['start_pos', Type.Float, None, 'Scan start position'], ['final_pos', Type.Float, None, 'Scan final position'], ['integ_time', Type.Float, None, 'Integration time'] ] def prepare(self, motor, start_pos, final_pos, integ_time, **opts): self.name='ascanc_demo' #parse the user parameters self.start = numpy.array([start_pos], dtype='d') self.final = numpy.array([final_pos], dtype='d') self.integ_time = integ_time env = opts.get('env',{}) #the "env" dictionary may be passed as an option #create an instance of GScan (in this case, of its child, CScan self._gScan = CScan(self, waypointGenerator=self._waypoint_generator, periodGenerator=self._period_generator, moveables=[motor], env=env) def _waypoint_generator(self): #a very simple waypoint generator! only start and stop points! yield {"positions":self.start, "waypoint_id": 0} yield {"positions":self.final, "waypoint_id": 1} def _period_generator(self): step = {} step["integ_time"] = self.integ_time point_no = 0 while(True): #infinite generator. The acquisition loop is started/stopped at begin and end of each waypoint point_no += 1 step["point_id"] = point_no yield step def run(self,*args): for step in self._gScan.step_scan(): yield step .. seealso:: for another example of a continuous scan implementation (with more elaborated waypoint generator), see the code of :class:`~sardana.macroserver.macros.scan.meshc` .. _sardana-macros-scanframework-determscan: Deterministic scans ------------------- By *deterministic scans* we call these scans which know *a priori* the number of points and the integration time. In some cases the experimental channel controllers may take profit of this information and prepare for the whole scan measurement up-front instead of preparing before each scan point. When writing this type of scan macros it is enough to set two attributes: ``nb_points`` and ``integ_time`` in your macro. When these attributes are present in your macro, the *generic scan framework* will take care of preparing the experimental channels for the whole scan measurement upfront. .. warning:: Since ``nb_points`` and ``integ_time`` attributes identify deterministic scan macros as deterministic and you must not use these attribute names for any other needs. .. seealso:: More information about deterministic scans and measurement preparation can be found in SEP18_. Hooks support in scans ---------------------- In general, the Hooks API provided by the :class:`~sardana.macroserver.macro.Hookable` base class allows a macro to run other code (the hook callable) at certain points of its execution. When registering, hooks are identified by *hook places* so the macro knows how/when they should be executed. The *hook places* are just arbitrary strings and are not fixed by the API, being up to each macro to identify, use and/or ignore them. See the code of the ``hooked_scan`` macro in :ref:`sardana-devel-macro-hooks-examples` that demonstrates how to use the Hooks API in scans. Other examples of the same chapter can be illustrative. Also, note that the Sardana-Taurus widget Sequencer widget allows the user to dynamically add hooks to existing macros before execution. In the case of the scan macros, the hooks can be registered not only via the Hooks API but also passed as ``key:value`` pairs, where key is a concatenation of a *hook place* and the **-hooks** string e.g. pre-scan-hooks and value is a list of callable hooks, of the "step" dictionary returned by the scan :attr:`~sardana.macroserver.scan.GScan.generator` (see :class:`~sardana.macroserver.scan.GScan` for more details and :class:`~sardana.macroserver.macros.scan.aNscan` as a use case). The following is a list of the supported keys: - 'pre-scan-hooks' : before starting the scan. - 'pre-move-hooks' : for steps: before starting to move. - 'post-move-hooks': for steps: after finishing the move. - 'pre-acq-hooks' : for steps: before starting to acquire. - 'post-acq-hooks' : for steps: after finishing acquisition but before recording the step. - 'post-step-hooks' : for steps: after finishing recording the step. - 'post-scan-hooks' : after finishing the scan. More examples ------------- Other macros in :ref:`sardana-devel-macro-scans-examples` illustrate more features of the scan framework. See also the code of the standard scan macros in the :mod:`~sardana.macroserver.macros.scan` module. Finally, the documentation and code of :mod:`~sardana.macroserver.scan.GScan`, :class:`~sardana.macroserver.scan.SScan` and :class:`~sardana.macroserver.scan.CScan` may be helpful. .. _SEP18: http://www.sardana-controls.org/sep/?SEP18.md