Complete the interpretation of the NFA to parse the regular expressio…

…n, and fix a few fucking bugs

nfa/construction.py

@@ -21,7 +21,7 @@ def pattern(pattern_string):
     lexer = Lexer(pattern_string)
     lexer.advance()
     nfa_pair = NfaPair()
-    expr(nfa_pair)
+    group(nfa_pair)
     log_nfa(nfa_pair.start_node)
 
     return nfa_pair.start_node
@@ -139,6 +139,7 @@ def factor_conn(pair_out):
 
 def is_conn(token):
     nc = [
+        Token.OPEN_PAREN,
         Token.CLOSE_PAREN,
         Token.AT_EOL,
         Token.EOS,
@@ -237,4 +238,34 @@ def expr(pair_out):
         pair_out.end_node.next_2 = end
         pair_out.end_node = end
 
-    return True
+    return True
+
+
+def group(pair_out):
+    if lexer.match(Token.OPEN_PAREN):
+        lexer.advance()
+        expr(pair_out)
+        if lexer.match(Token.CLOSE_PAREN):
+            lexer.advance()
+    elif lexer.match(Token.EOS):
+        return False
+    else: 
+        expr(pair_out)
+
+    while True:
+        pair = NfaPair()
+        if lexer.match(Token.OPEN_PAREN):
+            lexer.advance()
+            expr(pair)
+            pair_out.end_node.next_1 = pair.start_node
+            pair_out.end_node = pair.end_node
+            if lexer.match(Token.CLOSE_PAREN):
+                lexer.advance()
+        elif lexer.match(Token.EOS):
+            return False
+        else: 
+            expr(pair)
+            pair_out.end_node.next_1 = pair.start_node
+            pair_out.end_node = pair.end_node
+
+

parse/parse.py

@@ -2,9 +2,12 @@
     EPSILON,
     CCL,
 )
+from nfa.construction import pattern
 
 
-def compiler(input_string, start_node):
+def match(input_string, pattern_string):
+    start_node = pattern(pattern_string)
+
     current_nfa_set = [start_node]
     next_nfa_set = closure(current_nfa_set)
 
@@ -56,7 +59,7 @@ def closure(input_set):
 def move(input_set, ch):
     out_set = []
     for nfa in input_set:
-        print('debug move 1****', ch, nfa.status_num, nfa.edge, nfa.input_set)
+        print('debug move no 1****', ch, nfa.status_num, nfa.edge, nfa.input_set)
         if nfa.edge == ch or (nfa.edge == CCL and ch in nfa.input_set):
             print('debug in move 1****', ch, nfa.status_num, nfa.edge, nfa.input_set)
             out_set.append(nfa.next_1)
@@ -68,4 +71,5 @@ def move(input_set, ch):
 def has_accepted_state(nfa_set):
     for nfa in nfa_set:
         if nfa.next_1 is None and nfa.next_2 is None:
+            print('fuck******* ', nfa.status_num)
             return True

regex.py

@@ -1,4 +1,16 @@
-from parse.parse import compiler
+from parse.parse import match
 
-def regex(pattern, input_string):
-    pass
+
+class Regex(object):
+    def __init__(self, input_string, pattern_string):
+        self.input_string = input_string
+        self.pattern_string = pattern_string
+
+    def match(self):
+        return match(self.input_string, self.pattern_string)
+
+    def replace():
+        pass
+
+    def search():
+        pass