Source code for mph.node

"""Provides the wrapper class for a model node."""
__license__ = 'MIT'


########################################
# Dependencies                         #
########################################
from numpy import array, ndarray       # numerical array
from jpype import JBoolean             # Java boolean
from jpype import JInt                 # Java integer
from jpype import JDouble              # Java float
from jpype import JString              # Java string
from jpype import JArray               # Java array
from pathlib import Path               # file-system path
from re import split                   # string splitting
from json import load as json_load     # JSON decoder
from difflib import get_close_matches  # fuzzy matching
from functools import lru_cache        # function cache
from logging import getLogger          # event logging


########################################
# Globals                              #
########################################
logger = getLogger(__package__)        # event logger


########################################
# Node                                #
########################################
[docs]class Node: """ Represents a model node. Nodes work similarly to `pathlib.Path` objects from Python's standard library. They support string concatenation to the right with the division operator in order to reference child nodes: ```python >>> node = model/'functions' >>> node Node('functions') >>> node/'step' Node('functions/step') ``` Note how the `model` object also supports the division operator in order to generate node references. As mere references, nodes must must not necessarily exist in the model tree: ```python >>> (node/'new function').exists() False ``` In rare cases, the node name itself might contain a forward slash, such as the dataset `sweep/solution` that happens to exist in the demo model from the Tutorial. These literal forward slashes can be escaped by doubling the character: ```python >>> node = model/'datasets/sweep//solution' >>> node.name() 'sweep//solution' >>> node.parent() Node('datasets') ``` This class allows inspecting a node, such as its properties and child nodes, as well as manipulating it to some extent, like toggling it on/off, creating child nodes, or "running" it. Not all actions made available by Comsol are exposed here. Those missing can however be triggered in the Java layer, which is accessible via the `.java` property. """ #################################### # Internal # #################################### def __init__(self, model, path=None): if path is None: self.path = ('',) elif isinstance(path, str): self.path = parse(path) elif isinstance(path, Node): self.path = path.path else: error = f'Node path {path!r} is not a string or Node instance.' logger.error(error) raise TypeError(error) self.alias = { 'parameter': 'parameters', 'function': 'functions', 'component': 'components', 'geometry': 'geometries', 'view': 'views', 'selection': 'selections', 'variable': 'variables', 'coupling': 'couplings', 'material': 'materials', 'mesh': 'meshes', 'study': 'studies', 'solution': 'solutions', 'batch': 'batches', 'dataset': 'datasets', 'evaluation': 'evaluations', 'table': 'tables', 'plot': 'plots', 'result': 'plots', 'results': 'plots', 'export': 'exports', } if self.path[0] in self.alias: self.path = (self.alias[self.path[0]],) + self.path[1:] self.groups = { 'parameters': 'self.model.java.param().group()', 'functions': 'self.model.java.func()', 'components': 'self.model.java.component()', 'geometries': 'self.model.java.geom()', 'views': 'self.model.java.view()', 'selections': 'self.model.java.selection()', 'coordinates': 'self.model.java.coordSystem()', 'variables': 'self.model.java.variable()', 'couplings': 'self.model.java.cpl()', 'physics': 'self.model.java.physics()', 'multiphysics': 'self.model.java.multiphysics()', 'materials': 'self.model.java.material()', 'meshes': 'self.model.java.mesh()', 'studies': 'self.model.java.study()', 'solutions': 'self.model.java.sol()', 'batches': 'self.model.java.batch()', 'datasets': 'self.model.java.result().dataset()', 'evaluations': 'self.model.java.result().numerical()', 'tables': 'self.model.java.result().table()', 'plots': 'self.model.java.result()', 'exports': 'self.model.java.result().export()', } self.model = model def __str__(self): return join(self.path) def __repr__(self): return f"{self.__class__.__name__}('{self}')" def __eq__(self, other): return (self.path == other.path and self.model == other.model) def __truediv__(self, other): if isinstance(other, str): other = other.lstrip('/') return Node(self.model, join(parse(f'{self}/{other}'))) return NotImplemented def __contains__(self, node): if isinstance(node, str): if (self/node).exists(): return True elif isinstance(node, Node): if node.parent() == self and node.exists(): return True return False def __iter__(self): yield from self.children() @property def java(self): """ Returns the Java object this node maps to. Note that this is a property, not an attribute. Internally, it is a function that performs a top-down search of the model tree in order to resolve the node reference. So it introduces a certain overhead every time it is accessed. """ if self.is_root(): return self.model.java name = self.name() if self.is_group(): return eval(self.groups.get(name)) parent = self.parent() java = parent.java if not java: return None container = java if parent.is_group() else java.feature() for tag in container.tags(): member = container.get(tag) if name == escape(member.name()): return member #################################### # Navigation # ####################################
[docs] def name(self): """Returns the node's name.""" return f'{self.model}' if self.is_root() else escape(self.path[-1])
[docs] def tag(self): """Returns the node's tag.""" java = self.java return str(java.tag()) if java else None
[docs] def type(self): """ Returns the node's feature type. This a something like `'Block'` for "a right-angled solid or surface block in 3D". Refer to the Comsol documentation for details. Feature types are displayed in the Comsol GUI at the top of the `Settings` tab. """ java = self.java return str(java.getType()) if hasattr(java, 'getType') else None
[docs] def parent(self): """Returns the parent node.""" if self.is_root(): return None else: return Node(self.model, join(self.path[:-1]))
[docs] def children(self): """Returns all child nodes.""" java = self.java if self.is_root(): return [Node(self.model, group) for group in self.groups] elif self.is_group(): return [self/escape(java.get(tag).name()) for tag in java.tags()] elif hasattr(java, 'feature'): return [self/escape(java.feature(tag).name()) for tag in java.feature().tags()] else: return []
[docs] def is_root(self): """Checks if the node is the model's root node.""" return bool(len(self.path) == 1 and not self.path[0])
[docs] def is_group(self): """Checks if the node refers to a built-in group.""" return bool(len(self.path) == 1 and self.path[0])
[docs] def exists(self): """Checks if the node exists in the model tree.""" return (self.java is not None)
#################################### # Interaction # ####################################
[docs] def rename(self, name): """Renames the node.""" if self.is_root(): error = 'Cannot rename the root node.' logger.error(error) raise PermissionError(error) if self.is_group(): error = 'Cannot rename a built-in group.' logger.error(error) raise PermissionError(error) java = self.java if java: java.name(name) self.path = self.path[:-1] + (name,)
[docs] def retag(self, tag): """Assigns a new tag to the node.""" if self.is_root(): error = 'Cannot change tag of root node.' logger.error(error) raise PermissionError(error) if self.is_group(): error = 'Cannot change tag of built-in group.' logger.error(error) raise PermissionError(error) java = self.java if not java: error = f'Node "{self}" does not exist in model tree.' logger.error(error) raise LookupError(error) java.tag(tag)
[docs] def property(self, name, value=None): """ Returns or changes the value of the named property. If no `value` is given, returns the value of property `name`. Otherwise sets the property to the given value. """ if value is None: return get(self.java, name) else: self.java.set(name, cast(value))
[docs] def properties(self): """Returns names and values of all node properties as a dictionary.""" java = self.java if not hasattr(java, 'properties'): return {} names = sorted(str(name) for name in java.properties()) return {name: get(java, name) for name in names}
[docs] def toggle(self, action='flip'): """ Enables or disables the node. If `action` is `'flip'` (the default), it enables the feature in the model tree if it is currently disabled or disables it if enabled. Pass `'enable'` or `'on'` to enable the feature regardless of its current state. Pass `'disable'` or `'off'` to disable it. """ java = self.java if not java: error = f'Node "{self}" does not exist in model tree.' logger.error(error) raise LookupError(error) if action == 'flip': java.active(not java.isActive()) elif action in ('enable', 'on', 'activate'): java.active(True) elif action in ('disable', 'off', 'deactivate'): java.active(False)
[docs] def run(self): """Performs the "run" action if the node implements it.""" java = self.java if not java: error = f'Node "{self}" does not exist in model tree.' logger.error(error) raise LookupError(error) if not hasattr(java, 'run'): error = f'Node "{self}" does not implement "run" operation.' logger.error(error) raise RuntimeError(error) java.run()
[docs] def create(self, *arguments, name=None): """ Creates a new child node. Refer to the Comsol documentation for the values of valid `arguments`. It is often just the feature type of the child node to be created, given as a string such as `'Block'`, but may also require different or more arguments. If `name` is not given, a unique name/label will be assigned automatically. """ if self.is_root(): error = 'Cannot create nodes at root of model tree.' logger.error(error) raise PermissionError(error) java = self.java container = None if self.is_group(): if not hasattr(java, 'uniquetag') and hasattr(java, 'feature'): container = java.feature() elif hasattr(java, 'uniquetag') and hasattr(java, 'create'): container = java elif hasattr(java, 'feature'): container = java.feature() if not hasattr(container, 'uniquetag'): error = f'Node {self} does not support feature creation.' logger.error(error) raise RuntimeError(error) for argument in arguments: if isinstance(argument, str): type = argument break else: type = '?' pattern = tag_pattern(feature_path(self) + [type]) if pattern.endswith('*'): tag = container.uniquetag(pattern[:-1]) elif pattern in container.tags(): tag = container.uniquetag(pattern) else: tag = pattern if not arguments: container.create(tag) else: container.create(tag, *[cast(argument) for argument in arguments]) if name: container.get(tag).name(unescape(name)) else: name = escape(container.get(tag).name()) child = self/name check = tag_pattern(feature_path(child)) if pattern != check: pattern = check if pattern.endswith('*'): tag = container.uniquetag(pattern[:-1]) elif pattern in container.tags(): tag = container.uniquetag(pattern) else: tag = pattern logger.debug(f'Retagging "{child}": "{child.tag()}" → "{tag}".') child.retag(tag) return child
[docs] def remove(self): """Removes the node from the model tree.""" if self.is_root(): error = 'Cannot remove the root node.' logger.error(error) raise PermissionError(error) if self.is_group(): error = 'Cannot remove a built-in group.' logger.error(error) raise PermissionError(error) if not self.exists(): error = f'Node "{self}" does not exist in model tree.' logger.error(error) raise LookupError(error) parent = self.parent() container = parent.java if parent.is_group() else parent.java.feature() container.remove(self.java.tag())
######################################## # Name parsing # ######################################## def parse(string): """Parses a node path given as string to a tuple.""" # Force-cast str subclasses to str, just like `pathlib` does. # See bugs.python.org/issue21127 for the rationale. string = str(string) # Remove all leading and trailing forward slashes. string = string.lstrip('/').rstrip('/') # Split at forward slashes, but not double forward slashes. path = tuple(unescape(name) for name in split(r'(?<!/)/(?!/)', string)) return path def join(path): """Joins a node path given as tuple into a string.""" return '/'.join(escape(name) for name in path) def escape(name): """Escapes forward slashes in a node name.""" # Also accept Java strings, but always return Python string. name = str(name) return name.replace('/', '//') def unescape(name): """Reverses escaping of forward slashes in a node name.""" return name.replace('//', '/') ######################################## # Tag patterns # ######################################## @lru_cache(maxsize=1) def load_patterns(): """Loads the look-up table for tag patterns indexed by feature path.""" file = Path(__file__).parent/'tags.json' with file.open(encoding='UTF-8-sig') as stream: patterns = json_load(stream) return patterns def feature_path(node): """Returns the feature path of a node.""" if node.is_group(): return [node.name()] type = node.type() if not type: type = '?' return feature_path(node.parent()) + [type] def tag_pattern(feature_path): """Looks up the tag pattern for the best match to given feature path.""" (group, type) = (feature_path[0], feature_path[-1]) patterns = load_patterns() selected = [key for key in patterns if key.startswith(group) and key.endswith(type)] matches = get_close_matches(' → '.join(feature_path), selected) if matches: return patterns[matches[0]] elif type != '?': return type.lower()[:3] + '*' else: return 'tag*' ######################################## # Type casting # ######################################## def cast(value): """Casts a value from its Python data type to a suitable Java data type.""" if isinstance(value, Node): return JString(value.tag()) elif isinstance(value, bool): return JBoolean(value) elif isinstance(value, int): return JInt(value) elif isinstance(value, float): return JDouble(value) elif isinstance(value, str): return JString(value) elif isinstance(value, Path): return JString(str(value)) elif isinstance(value, (list, tuple)): dimension = 0 item = value while isinstance(item, (list, tuple)): dimension += 1 if not len(item): datatype = JString value = [] break item = item[0] else: datatype = cast(item).__class__ value = [cast(item) for item in value] return JArray(datatype, dimension)(value) elif isinstance(value, ndarray): if value.dtype.kind == 'b': return JArray(JBoolean, value.ndim)(value) elif value.dtype.kind == 'f': return JArray(JDouble, value.ndim)(value) elif value.dtype.kind == 'i': return JArray(JInt, value.ndim)(value) elif value.dtype.kind == 'O': if value.ndim > 2: error = 'Cannot cast object arrays of dimension higher than 2.' logger.error(error) raise TypeError(error) if len(value) > 2: error = 'Will not cast object arrays with more than two rows.' logger.error(error) raise TypeError(error) rows = [row.astype(float) for row in value] return JArray(JDouble, 2)(rows) else: error = f'Cannot cast arrays of data type "{value.dtype}".' logger.error(error) raise TypeError(error) else: error = f'Cannot cast values of data type "{type(value).__name__}".' logger.error(error) raise TypeError(error) def get(java, name): """Returns the value of a Java node property as a Python data type.""" datatype = java.getValueType(name) if datatype == 'Boolean': return java.getBoolean(name) elif datatype == 'BooleanArray': return array(java.getBooleanArray(name)) elif datatype == 'BooleanMatrix': return array([line for line in java.getBooleanMatrix(name)]) elif datatype == 'Double': return java.getDouble(name) elif datatype == 'DoubleArray': return array(java.getDoubleArray(name)) elif datatype == 'DoubleMatrix': return array([line for line in java.getDoubleMatrix(name)]) elif datatype == 'DoubleRowMatrix': value = java.getDoubleMatrix(name) if len(value) == 0: rows = [] elif len(value) == 1: rows = [array(value[0])] elif len(value) == 2: rows = [array(value[0]), array(value[1])] else: error = 'Cannot convert double-row matrix with more than two rows.' logger.error(error) raise TypeError(error) return array(rows, dtype=object) elif datatype == 'File': return Path(str(java.getString(name))) elif datatype == 'Int': return int(java.getInt(name)) elif datatype == 'IntArray': return array(java.getIntArray(name)) elif datatype == 'IntMatrix': return array([line for line in java.getIntMatrix(name)]) elif datatype == 'None': return None elif datatype == 'Selection': return [str(string) for string in java.getEntryKeys(name)] elif datatype == 'String': value = java.getString(name) return str(value) if value else None elif datatype == 'StringArray': return [str(string) for string in java.getStringArray(name)] elif datatype == 'StringMatrix': value = java.getStringMatrix(name) if value: return [[str(string) for string in line] for line in value] else: return [[]] else: error = f'Cannot convert Java data type "{datatype}".' logger.error(error) raise TypeError(error) ######################################## # Inspection # ########################################
[docs]def inspect(java): """ Inspects a Java node object. This is basically a "pretty-fied" version of the output from the built-in `dir` command. It displays (prints to the console) the methods of a model node (as given by the Comsol API) as well as the node's properties (if any are defined). The node's name, tag, and documentation reference marker are listed first. These access methods and a few others, which are common to all objects, are suppressed in the method list further down, for the sake of clarity. """ # Also accept Node and Model instances. if hasattr(java, 'java'): java = java.java # Display general information about the feature. print(f'name: {java.name()}') print(f'tag: {java.tag()}') if hasattr(java, 'getType'): print(f'type: {java.getType()}') print(f'display: {java.getDisplayString()}') print(f'doc: {java.docMarker()}') # Display comments and notify if feature is deactivated or has warnings. comments = str(java.comments()) if comments: print(f'comment: {comments}') if not java.isActive(): print('This feature is currently deactivated.') if hasattr(java, 'hasWarning') and java.hasWarning(): print('This feature has warnings.') # Introspect the feature's attributes. attributes = [attribute for attribute in dir(java)] # Display properties if any are defined. if 'properties' in attributes: print('properties:') names = [str(name) for name in java.properties()] for name in names: value = get(java, name) print(f' {name}: {value}') # Define a list of common methods to be suppressed in the output. suppress = ['name', 'label', 'tag', 'getType', 'getDisplayString', 'docMarker', 'help', 'comments', 'toString', 'icon', 'properties', 'hasProperty', 'set', 'getEntryKeys', 'getEntryKeyIndex', 'getValueType', 'getInt', 'getIntArray', 'getIntMatrix', 'getBoolean', 'getBooleanArray', 'getBooleanMatrix', 'getDouble', 'getDoubleArray', 'getDoubleMatrix', 'getString', 'getStringArray', 'getStringMatrix', 'version', 'author', 'resetAuthor', 'lastModifiedBy', 'dateCreated', 'dateModified', 'timeCreated', 'timeModified', 'active', 'isActive', 'isactive', 'hasWarning', 'class_', 'getClass', 'hashCode', 'notify', 'notifyAll', 'wait'] # Display the feature's methods. print('methods:') for name in attributes: if name.startswith('_') or name in suppress: continue print(f' {name}')
[docs]def tree(node, levels=[], max_depth=None): """ Displays the model tree. This function displays a representation of the model tree in the console. `node` would typically be the model's root node. `levels` is used internally when traversing the model tree recursively. Specify `max_depth` to possibly limit the number of lower branches. Note that this function performs poorly in client–server mode, the default on Linux and macOS, especially for complex models. The client–server communication introduces inefficiencies that do not occur in stand-alone mode, the default on Windows, where the model tree, i.e. the hierarchy of related Java objects, can be traversed reasonably fast. """ if not isinstance(node, Node): # Support passing the model directly instead of a node. node = node/None if max_depth and len(levels) > max_depth: return markers = ''.join(' ' if last else '│ ' for last in levels[:-1]) markers += '' if not levels else '└─ ' if levels[-1] else '├─ ' print(f'{markers}{node.name()}') children = node.children() last = len(children) - 1 for (index, child) in enumerate(children): tree(child, levels + [index == last], max_depth)