Rsa rework

pyformlang/rsa/box.py

@@ -5,6 +5,7 @@
 from pyformlang.finite_automaton.epsilon_nfa import EpsilonNFA
 from pyformlang.finite_automaton.finite_automaton import to_symbol
 from pyformlang.finite_automaton.symbol import Symbol
+import networkx as nx
 
 
 class Box:
@@ -16,56 +17,57 @@ class Box:
     ----------
     enfa : :class:`~pyformlang.finite_automaton.EpsilonNFA`
         A epsilon nfa
-    label : :class:`~pyformlang.finite_automaton.Symbol`
-        A label for epsilon nfa
+    nonterminal : :class:`~pyformlang.finite_automaton.Symbol`
+        A nonterminal for epsilon nfa
 
     """
 
-    def __init__(self, enfa: EpsilonNFA = None, label: Symbol = None):
-        if enfa is not None:
-            enfa = enfa.minimize()
-        self._dfa = enfa or EpsilonNFA()
-
-        if label is not None:
-            label = to_symbol(label)
-        self._label = label or Symbol("")
-
-    def change_label(self, label: Symbol):
-        """ Set a new label
-
-        Parameters
-        -----------
-        label : :class:`~pyformlang.finite_automaton.Symbol`
-            The new label for automaton
-
-        """
-        self._label = to_symbol(label)
-
-    def change_dfa(self, enfa: EpsilonNFA):
-        """ Set an epsilon finite automaton
-
-        Parameters
-        -----------
-        enfa : :class:`~pyformlang.finite_automaton.EpsilonNFA`
-            The new epsilon finite automaton
-
-        """
-        enfa = enfa.minimize()
+    def __init__(self, enfa: EpsilonNFA, nonterminal: Symbol | str):
         self._dfa = enfa
 
+        nonterminal = to_symbol(nonterminal)
+        self._nonterminal = nonterminal
+
+    def to_subgraph_dot(self):
+        graph = self._dfa.to_networkx()
+        strange_nodes = []
+        dot_string = (f'subgraph cluster_{self._nonterminal}\n{{ label="{self._nonterminal}"\n'
+                      f'fontname="Helvetica,Arial,sans-serif"\n'
+                      f'node [fontname="Helvetica,Arial,sans-serif"]\n'
+                      f'edge [fontname="Helvetica,Arial,sans-serif"]\nrankdir=LR;\n'
+                      f'node [shape = circle style=filled fillcolor=white]')
+        for node, data in graph.nodes(data=True):
+            if 'is_start' not in data.keys() or 'is_final' not in data.keys():
+                strange_nodes.append(node)
+                continue
+            node = node.replace(";", "")
+            if data['is_start']:
+                dot_string += f'\n{node} [fillcolor = green];'
+            if data['is_final']:
+                dot_string += f'\n{node} [shape = doublecircle];'
+        for strange_node in strange_nodes:
+            graph.remove_node(strange_node)
+        for node_from, node_to, data in graph.edges(data=True):
+            node_from = node_from.replace(";", "")
+            node_to = node_to.replace(";", "")
+            label = data['label']
+            dot_string += f'\n{node_from} -> {node_to} [label = "{label}"];'
+        dot_string += "\n}"
+        return dot_string
+
     @property
     def dfa(self):
         """ Box's dfa """
         return self._dfa
 
     @property
-    def label(self):
-        """ Box's label """
-        return self._label
+    def nonterminal(self):
+        """ Box's nonterminal """
+        return self._nonterminal
 
     @property
-    def start_state(self):
-        """ The start state """
+    def start_states(self):
+        """ The start states """
         return self._dfa.start_states
 
     @property
@@ -90,14 +92,10 @@ def is_equivalent_to(self, other):
         if not isinstance(other, Box):
             return False
 
-        if self._dfa.is_equivalent_to(other.dfa) and \
-                self._label == other.label:
-            return True
-
-        return False
+        return self._dfa.is_equivalent_to(other.dfa) and self.nonterminal == other.nonterminal
 
     def __eq__(self, other):
         return self.is_equivalent_to(other)
 
     def __hash__(self):
-        return self._label.__hash__()
+        return self._nonterminal.__hash__()

pyformlang/rsa/recursive_automaton.py

@@ -4,6 +4,8 @@
 
 from typing import AbstractSet
 
+import pydot
+
 from pyformlang.finite_automaton.finite_automaton import to_symbol
 from pyformlang.finite_automaton.symbol import Symbol
 from pyformlang.regular_expression import Regex
@@ -19,89 +21,49 @@ class RecursiveAutomaton:
 
     Parameters
     ----------
-    labels : set of :class:`~pyformlang.finite_automaton.Symbol`, optional
-        A finite set of labels for boxes
-    initial_label : :class:`~pyformlang.finite_automaton.Symbol`, optional
-        A start label for automaton
-    boxes : set of :class:`~pyformlang.rsa.Box`, optional
+    start_box : :class:`~pyformlang.rsa.Box`
+        Start box
+    boxes : set of :class:`~pyformlang.rsa.Box`
         A finite set of boxes
 
     """
 
     def __init__(self,
-                 labels: AbstractSet[Symbol] = None,
-                 initial_label: Symbol = None,
-                 boxes: AbstractSet[Box] = None):
-
-        if labels is not None:
-            labels = {to_symbol(x) for x in labels}
-        self._labels = labels or set()
-
-        if initial_label is not None:
-            initial_label = to_symbol(initial_label)
-            if initial_label not in self._labels:
-                self._labels.add(initial_label)
-        self._initial_label = initial_label or Symbol("")
-
+                 start_box: Box,
+                 boxes: AbstractSet[Box]):
         self._boxes = {}
-        if boxes is not None:
-            for box in boxes:
-                self._boxes.update({to_symbol(box.label): box})
-                self._labels.add(box.label)
+        if start_box not in boxes:
+            self._boxes.update({to_symbol(start_box.nonterminal): start_box})
+        self._start_nonterminal = to_symbol(start_box.nonterminal)
+        for box in boxes:
+            self._boxes.update({to_symbol(box.nonterminal): box})
 
-        for label in self._labels:
-            box = self.get_box(label)
-            if box is None:
-                raise ValueError(
-                    "RSA must have the same number of labels and DFAs")
+    def get_box_by_nonterminal(self, nonterminal: Symbol | str):
+        """ Box by nonterminal """
 
-    def get_box(self, label: Symbol):
-        """ Box by label """
-
-        label = to_symbol(label)
-        if label in self._boxes:
-            return self._boxes[label]
+        nonterminal = to_symbol(nonterminal)
+        if nonterminal in self._boxes:
+            return self._boxes[nonterminal]
 
         return None
 
-    def add_box(self, new_box: Box):
-        """ Set a box
-
-        Parameters
-        -----------
-        new_box : :class:`~pyformlang.rsa.Box`
-            The new box
-
-        """
-
-        self._boxes.update({new_box.label: new_box})
-        self._labels.add(to_symbol(new_box.label))
-
     def get_number_of_boxes(self):
         """ Size of set of boxes """
 
         return len(self._boxes)
 
-    def change_initial_label(self, new_initial_label: Symbol):
-        """ Set an initial label
-
-        Parameters
-        -----------
-        new_initial_label : :class:`~pyformlang.finite_automaton.Symbol`
-            The new initial label
-
-        """
-
-        new_initial_label = to_symbol(new_initial_label)
-        if new_initial_label not in self._labels:
-            raise ValueError(
-                "New initial label not in set of labels for boxes")
+    def to_dot(self):
+        dot_string = f'digraph ""{{'
+        for box in self._boxes.values():
+            dot_string += f'\n{box.to_subgraph_dot()}'
+        dot_string += "\n}"
+        return dot_string
 
     @property
-    def labels(self) -> set:
-        """ The set of labels """
+    def nonterminals(self) -> set:
+        """ The set of nonterminals """
 
-        return self._labels
+        return set(self._boxes.keys())
 
     @property
     def boxes(self) -> dict:
@@ -110,52 +72,57 @@ def boxes(self) -> dict:
         return self._boxes
 
     @property
-    def initial_label(self) -> Symbol:
-        """ The initial label """
+    def start_nonterminal(self) -> Symbol:
+        """ The start nonterminal """
+
+        return self._start_nonterminal
+
+    @property
+    def start_box(self):
+        """ The start box """
 
-        return self._initial_label
+        return self.boxes[self.start_nonterminal]
 
     @classmethod
-    def from_regex(cls, regex: Regex, initial_label: Symbol):
+    def from_regex(cls, regex: Regex, start_nonterminal: Symbol | str):
         """ Create a recursive automaton from regular expression
 
         Parameters
         -----------
         regex : :class:`~pyformlang.regular_expression.Regex`
             The regular expression
-        initial_label : :class:`~pyformlang.finite_automaton.Symbol`
-            The initial label for the recursive automaton
+        start_nonterminal : :class:`~pyformlang.finite_automaton.Symbol` | str
+            The start nonterminal for the recursive automaton
 
         Returns
         -----------
         rsa : :class:`~pyformlang.rsa.RecursiveAutomaton`
             The new recursive automaton built from regular expression
         """
-
-        initial_label = to_symbol(initial_label)
-        box = Box(regex.to_epsilon_nfa().minimize(), initial_label)
-        return RecursiveAutomaton({initial_label}, initial_label, {box})
+        start_nonterminal = to_symbol(start_nonterminal)
+        box = Box(regex.to_epsilon_nfa().minimize(), start_nonterminal)
+        return RecursiveAutomaton(box, {box})
 
     @classmethod
-    def from_text(cls, text, start_symbol: Symbol = Symbol("S")):
-        """ Create a recursive automaton from text
+    def from_ebnf(cls, text, start_nonterminal: Symbol | str = Symbol("S")):
+        """ Create a recursive automaton from ebnf
 
         Parameters
         -----------
         text : str
             The text of transform
-        start_symbol : str, optional
-            The start symbol, S by default
+        start_nonterminal : :class:`~pyformlang.finite_automaton.Symbol` | str, optional
+            The start nonterminal, S by default
 
         Returns
         -----------
         rsa : :class:`~pyformlang.rsa.RecursiveAutomaton`
             The new recursive automaton built from context-free grammar
         """
-
+        start_nonterminal = to_symbol(start_nonterminal)
         productions = {}
         boxes = set()
-        labels = set()
+        nonterminals = set()
         for production in text.splitlines():
             production = production.strip()
             if "->" not in production:
@@ -164,7 +131,7 @@ def from_text(cls, text, start_symbol: Symbol = Symbol("S")):
             head, body = production.split("->")
             head = head.strip()
             body = body.strip()
-            labels.add(to_symbol(head))
+            nonterminals.add(to_symbol(head))
 
             if body == "":
                 body = Epsilon().to_text()
@@ -177,11 +144,12 @@ def from_text(cls, text, start_symbol: Symbol = Symbol("S")):
         for head, body in productions.items():
             boxes.add(Box(Regex(body).to_epsilon_nfa().minimize(),
                           to_symbol(head)))
+        start_box = Box(Regex(productions[start_nonterminal.value]).to_epsilon_nfa().minimize(), start_nonterminal)
+        return RecursiveAutomaton(start_box, boxes)
 
-        return RecursiveAutomaton(labels, start_symbol, boxes)
-
-    def is_equivalent_to(self, other):
-        """ Check whether two recursive automata are equivalent
+# equivalency not in terms of formal languages theory. Just mapping boxes.
+    def is_equals_to(self, other):
+        """ Check whether two recursive automata are equals by boxes
 
         Parameters
         ----------
@@ -191,23 +159,11 @@ def is_equivalent_to(self, other):
         Returns
         ----------
         are_equivalent : bool
-            Whether the two recursive automata are equivalent or not
+            Whether the two recursive automata are equals or not
         """
-
         if not isinstance(other, RecursiveAutomaton):
             return False
-
-        if self._labels != other._labels:
-            return False
-
-        for label in self._labels:
-            box_1 = self.get_box(label)
-            box_2 = other.get_box(label)
-
-            if box_1 != box_2:
-                return False
-
-        return True
+        return self.boxes == other.boxes
 
     def __eq__(self, other):
-        return self.is_equivalent_to(other)
+        return self.is_equals_to(other)