Source code for pyibisami.ami.parser

"""IBIS-AMI parameter parsing and configuration utilities.

Original author: David Banas <capn.freako@gmail.com>

Original date:   December 17, 2016

Copyright (c) 2019 David Banas; all rights reserved World wide.
"""

import re

from parsec import ParseError, generate, many, regex, string
from traits.api import Bool, Enum, HasTraits, Range, Trait
from traitsui.api import Group, Item, View
from traitsui.menu import ModalButtons

from pyibisami.ami.parameter import AMIParamError, AMIParameter

#####
# AMI parameter configurator.
#####


[docs] class AMIParamConfigurator(HasTraits): """Customizable IBIS-AMI model parameter configurator. This class can be configured to present a customized GUI to the user for configuring a particular IBIS-AMI model. The intended use model is as follows: 1. Instantiate this class only once per IBIS-AMI model invocation. When instantiating, provide the unprocessed contents of the AMI file, as a single string. This class will take care of getting that string parsed properly, and report any errors or warnings it encounters, in its ``ami_parsing_errors`` property. 2. When you want to let the user change the AMI parameter configuration, call the ``open_gui`` member function. (Or, just call the instance as if it were executable.) The instance will then present a GUI to the user, allowing him to modify the values of any *In* or *InOut* parameters. The resultant AMI parameter dictionary, suitable for passing into the ``ami_params`` parameter of the ``AMIModelInitializer`` constructor, can be accessed, via the instance's ``input_ami_params`` property. The latest user selections will be remembered, as long as the instance remains in scope. The entire AMI parameter definition dictionary, which should *not* be passed to the ``AMIModelInitializer`` constructor, is available in the instance's ``ami_param_defs`` property. Any errors or warnings encountered while parsing are available, in the ``ami_parsing_errors`` property. """ def __init__(self, ami_file_contents_str): """ Args: ami_file_contents_str (str): The unprocessed contents of the AMI file, as a single string. """ # Super-class initialization is ABSOLUTELY NECESSARY, in order # to get all the Traits/UI machinery setup correctly. super().__init__() # Parse the AMI file contents, storing any errors or warnings, # and customize the view accordingly. err_str, param_dict = parse_ami_param_defs(ami_file_contents_str) if not param_dict: print("Empty dictionary returned by parse_ami_param_defs()!") print(f"Error message:\n{err_str}") raise KeyError("Failed to parse AMI file; see console for more detail.") top_branch = list(param_dict.items())[0] param_dict = top_branch[1] if "Reserved_Parameters" not in param_dict: print(f"Error: {err_str}\nParameters: {param_dict}") raise KeyError("Unable to get 'Reserved_Parameters' from the parameter set.") if "Model_Specific" not in param_dict: print(f"Error: {err_str}\nParameters: {param_dict}") raise KeyError("Unable to get 'Model_Specific' from the parameter set.") pdict = param_dict["Reserved_Parameters"].copy() pdict.update(param_dict["Model_Specific"]) gui_items, new_traits = make_gui_items("Model In/InOut Parameters", pdict, first_call=True) trait_names = [] for trait in new_traits: self.add_trait(trait[0], trait[1]) trait_names.append(trait[0]) self._content = gui_items self._param_trait_names = trait_names self._root_name = top_branch[0] self._ami_parsing_errors = err_str self._content = gui_items self._param_dict = param_dict try: self._info_dict = {name: p.pvalue for (name, p) in list(param_dict["Reserved_Parameters"].items())} except Exception as err: print(f"{err}") print(f"param_dict['Reserved_Parameters']: {param_dict['Reserved_Parameters']}") raise def __call__(self): self.open_gui()
[docs] def open_gui(self): """Present a customized GUI to the user, for parameter customization.""" self.configure_traits()
def default_traits_view(self): "Default Traits/UI view definition." view = View( resizable=False, buttons=ModalButtons, title="PyBERT AMI Parameter Configurator", id="pybert.pybert_ami.param_config", ) view.set_content(self._content) return view
[docs] def fetch_param(self, branch_names): """Returns the parameter found by traversing 'branch_names' or None if not found. Note: 'branch_names' should *not* begin with 'root_name'. """ param_dict = self.ami_param_defs while branch_names: branch_name = branch_names.pop(0) if branch_name in param_dict: param_dict = param_dict[branch_name] else: return None if isinstance(param_dict, AMIParameter): return param_dict return None
[docs] def fetch_param_val(self, branch_names): """Returns the value of the parameter found by traversing 'branch_names' or None if not found. Note: 'branch_names' should *not* begin with 'root_name'. """ _param = self.fetch_param(branch_names) if _param: return _param.pvalue return None
[docs] def set_param_val(self, branch_names, new_val): """Sets the value of the parameter found by traversing 'branch_names' or raises an exception if not found. Note: 'branch_names' should *not* begin with 'root_name'. Note: Be careful! There is no checking done here! """ param_dict = self.ami_param_defs while branch_names: branch_name = branch_names.pop(0) if branch_name in param_dict: param_dict = param_dict[branch_name] else: raise ValueError( f"Failed parameter tree search looking for: {branch_name}; available keys: {param_dict.keys()}" ) if isinstance(param_dict, AMIParameter): param_dict.pvalue = new_val try: eval(f"self.set({branch_name}_={new_val})") # pylint: disable=eval-used except Exception: # pylint: disable=broad-exception-caught eval(f"self.set({branch_name}={new_val})") # pylint: disable=eval-used else: raise TypeError(f"{param_dict} is not of type: AMIParameter!")
@property def ami_parsing_errors(self): """Any errors or warnings encountered, while parsing the AMI parameter definition file contents.""" return self._ami_parsing_errors @property def ami_param_defs(self): """The entire AMI parameter definition dictionary. Should *not* be passed to ``AMIModelInitializer`` constructor! """ return self._param_dict @property def input_ami_params(self): """The dictionary of *Model Specific* AMI parameters of type 'In' or 'InOut', along with their user selected values. Should be passed to ``AMIModelInitializer`` constructor. """ res = {} res["root_name"] = self._root_name params = self.ami_param_defs["Model_Specific"] for pname in params: res.update(self.input_ami_param(params, pname)) return res
[docs] def input_ami_param(self, params, pname): """Retrieve one AMI parameter, or dictionary of subparameters.""" res = {} param = params[pname] if isinstance(param, AMIParameter): if pname in self._param_trait_names: # If model specific and In or InOut... # See the docs on the *HasTraits* class, if this is confusing. try: # Querry for a mapped trait, first, by trying to get '<trait_name>_'. (Note the underscore.) res[pname] = self.get(pname + "_")[pname + "_"] except ( Exception # pylint: disable=broad-exception-caught ): # If we get an exception, we have an ordinary (i.e. - not mapped) trait. res[pname] = self.get(pname)[pname] elif isinstance(param, dict): # We received a dictionary of subparameters, in 'param'. subs = {} for sname in param.keys(): subs.update(self.input_ami_param(param, sname)) res[pname] = subs return res
@property def info_ami_params(self): "Dictionary of *Reserved* AMI parameter values." return self._info_dict
##### # AMI file parser. ##### # ignore cases. whitespace = regex(r"\s+", re.MULTILINE) comment = regex(r"\|.*") ignore = many(whitespace | comment)
[docs] def lexeme(p): """Lexer for words.""" return p << ignore # skip all ignored characters.
[docs] def int2tap(x): """Convert integer to tap position.""" x = x.strip() if x[0] == "-": res = "pre" + x[1:] else: res = "post" + x return res
lparen = lexeme(string("(")) rparen = lexeme(string(")")) number = lexeme(regex(r"[-+]?[0-9]*\.?[0-9]+([eE][-+]?[0-9]+)?")) integ = lexeme(regex(r"[-+]?[0-9]+")) nat = lexeme(regex(r"[0-9]+")) tap_ix = (integ << whitespace).parsecmap(int2tap) symbol = lexeme(regex(r"[0-9a-zA-Z_][^\s()]*")) true = lexeme(string("True")).result(True) false = lexeme(string("False")).result(False) ami_string = lexeme(regex(r'"[^"]*"')) atom = number | symbol | ami_string | (true | false) node_name = tap_ix ^ symbol # `tap_ix` is new and gives the tap position; negative positions are allowed. @generate("AMI node") def node(): "Parse AMI node." yield lparen label = yield node_name values = yield many(expr) yield rparen return (label, values) expr = atom | node ami_defs = ignore >> node
[docs] def proc_branch(branch): """Process a branch in a AMI parameter definition tree. That is, build a dictionary from a pair containing: - a parameter name, and - a list of either: - parameter definition tags, or - subparameters. We distinguish between the two possible kinds of payloads, by peaking at the names of the first two items in the list and noting whether they are keys of 'AMIParameter._param_def_tag_procs'. We have to do this twice, due to the dual use of the 'Description' tag and the fact that we have no guarantee of any particular ordering of subparameter branch items. Args: p (str, list): A pair, as described above. Returns: (str, dict): A pair containing: err_str: String containing any errors or warnings encountered, while building the parameter dictionary. param_dict: Resultant parameter dictionary. """ results = ("", {}) # Empty Results if len(branch) != 2: if not branch: err_str = "ERROR: Empty branch provided to proc_branch()!\n" else: err_str = f"ERROR: Malformed item: {branch[0]}\n" results = (err_str, {}) param_name = branch[0] param_tags = branch[1] if not param_tags: err_str = f"ERROR: No tags/subparameters provided for parameter, '{param_name}'\n" results = (err_str, {}) try: if ( (len(param_tags) > 1) and ( # noqa: W503 param_tags[0][0] in AMIParameter._param_def_tag_procs # pylint: disable=protected-access # noqa: W503 ) and ( # noqa: W503 param_tags[1][0] in AMIParameter._param_def_tag_procs # pylint: disable=protected-access # noqa: W503 ) ): try: results = ("", {param_name: AMIParameter(param_name, param_tags)}) except AMIParamError as err: results = (str(err), {}) elif param_name == "Description": results = ("", {"description": param_tags[0].strip('"')}) else: err_str = "" param_dict = {} param_dict[param_name] = {} for param_tag in param_tags: temp_str, temp_dict = proc_branch(param_tag) param_dict[param_name].update(temp_dict) if temp_str: err_str = ( f"Error returned by recursive call, while processing parameter, '{param_name}':\n{temp_str}" ) results = (err_str, param_dict) results = (err_str, param_dict) except Exception: # pylint: disable=broad-exception-caught print(f"Error processing branch:\n{param_tags}") return results
[docs] def parse_ami_param_defs(param_str): # pylint: disable=too-many-branches """Parse the contents of a IBIS-AMI parameter definition file. Args: param_str (str): The contents of the file, as a single string. Example: :: with open(<ami_file_name>) as ami_file: param_str = ami_file.read() (err_str, param_dict) = parse_ami_param_defs(param_str) Returns: (str, dict): A pair containing: err_str: - None, if parser succeeds. - Helpful message, if it fails. param_dict: Dictionary containing parameter definitions. (Empty, on failure.) It has a single key, at the top level, which is the model root name. This key indexes the actual parameter dictionary, which has the following structure:: { 'description' : <optional model description string> 'Reserved_Parameters' : <dictionary of reserved parameter defintions> 'Model_Specific' : <dictionary of model specific parameter definitions> } The keys of the 'Reserved_Parameters' dictionary are limited to those called out in the IBIS-AMI specification. The keys of the 'Model_Specific' dictionary can be anything. The values of both are either: - instances of class *AMIParameter*, or - sub-dictionaries following the same pattern. """ try: res = ami_defs.parse(param_str) except ParseError as pe: err_str = f"Expected {pe.expected} at {pe.loc()} in:\n{pe.text[pe.index:]}" return err_str, {} err_str, param_dict = proc_branch(res) if err_str: return (err_str, {"res": res, "dict": param_dict}) reserved_found = False init_returns_impulse_found = False getwave_exists_found = False model_spec_found = False params = list(param_dict.items())[0][1] for label in list(params.keys()): if label == "Reserved_Parameters": reserved_found = True tmp_params = params[label] for param_name in list(tmp_params.keys()): if param_name not in AMIParameter.RESERVED_PARAM_NAMES: err_str += f"WARNING: Unrecognized reserved parameter name, '{param_name}', found in parameter definition string!\n" continue param = tmp_params[param_name] if param.pname == "AMI_Version": if param.pusage != "Info" or param.ptype != "String": err_str += "WARNING: Malformed 'AMI_Version' parameter.\n" elif param.pname == "Init_Returns_Impulse": init_returns_impulse_found = True elif param.pname == "GetWave_Exists": getwave_exists_found = True elif label == "Model_Specific": model_spec_found = True elif label == "description": pass else: err_str += f"WARNING: Unrecognized group with label, '{label}', found in parameter definition string!\n" if not reserved_found: err_str += "ERROR: Reserved parameters section not found! It is required." if not init_returns_impulse_found: err_str += "ERROR: Reserved parameter, 'Init_Returns_Impulse', not found! It is required." if not getwave_exists_found: err_str += "ERROR: Reserved parameter, 'GetWave_Exists', not found! It is required." if not model_spec_found: err_str += "WARNING: Model specific parameters section not found!" return (err_str, param_dict)
[docs] def make_gui_items( pname, param, first_call=False ): # pylint: disable=too-many-locals,too-many-branches,too-many-statements """Builds list of GUI items from AMI parameter dictionary.""" gui_items = [] new_traits = [] if isinstance(param, AMIParameter): # pylint: disable=too-many-nested-blocks pusage = param.pusage if pusage in ("In", "InOut"): if param.ptype == "Boolean": new_traits.append((pname, Bool(param.pvalue))) gui_items.append(Item(pname, tooltip=param.pdescription)) else: pformat = param.pformat if pformat == "Range": new_traits.append((pname, Range(param.pmin, param.pmax, param.pvalue))) gui_items.append(Item(pname, tooltip=param.pdescription)) elif pformat == "List": list_tips = param.plist_tip default = param.pdefault if list_tips: tmp_dict = {} tmp_dict.update(list(zip(list_tips, param.pvalue))) val = list(tmp_dict.keys())[0] if default: for tip in tmp_dict.items(): if tip == default: val = tip break new_traits.append((pname, Trait(val, tmp_dict))) else: val = param.pvalue[0] if default: val = default new_traits.append((pname, Enum([val] + param.pvalue))) gui_items.append(Item(pname, tooltip=param.pdescription)) else: # Value new_traits.append((pname, param.pvalue)) gui_items.append(Item(pname, tooltip=param.pdescription)) else: # subparameter branch subparam_names = list(param.keys()) subparam_names.sort() sub_items = [] group_desc = "" # Build GUI items for this branch. for subparam_name in subparam_names: if subparam_name == "description": group_desc = param[subparam_name] else: tmp_items, tmp_traits = make_gui_items(subparam_name, param[subparam_name]) sub_items.extend(tmp_items) new_traits.extend(tmp_traits) # Put all top-level ungrouped parameters in a single VGroup. top_lvl_params = [] sub_params = [] for item in sub_items: if isinstance(item, Item): top_lvl_params.append(item) else: sub_params.append(item) sub_items = [Group(top_lvl_params)] + sub_params # Make the top-level group an HGroup; all others VGroups (default). if first_call: gui_items.append( Group([Item(label=group_desc)] + sub_items, label=pname, show_border=True, orientation="horizontal") ) else: gui_items.append(Group([Item(label=group_desc)] + sub_items, label=pname, show_border=True)) return gui_items, new_traits