From 56a2d0e023d6751c60d497abd0f791ce15a91351 Mon Sep 17 00:00:00 2001
From: dejavudwh <494319000@qq.com>
Date: Fri, 20 Sep 2019 16:11:22 +0800
Subject: [PATCH] Update the factor and and operations

---
 nfa/construction.py | 72 ++++++++++++++++++++++++++++++++++++++++++---
 nfa/nfa.py          | 17 +++++++----
 test.py             | 19 ++++++------
 3 files changed, 89 insertions(+), 19 deletions(-)

diff --git a/nfa/construction.py b/nfa/construction.py
index 2239052..5738a44 100644
--- a/nfa/construction.py
+++ b/nfa/construction.py
@@ -1,7 +1,10 @@
 from lex.token import Token
 from lex.lexer import scanner
 from lex.lexer import Lexer
-from nfa.nfa import Nfa
+from nfa.nfa import (
+    Nfa,
+    NfaPair,
+)
 from nfa.nfa import EPSILON
 from nfa.nfa import CCL
 from nfa.nfa import EMPTY
@@ -13,12 +16,13 @@
 
 
 # 对 . a (单个字符) [] 进行匹配
+# term -> a | [] | .
 def term(pair_out):
     if lexer.match(Token.L):
         nfa_single_char(pair_out)
     elif lexer.match(Token.ANY):
         nfa_dot_char(pair_out)
-    elif lexer.match(Token.CCL_START):    
+    elif lexer.match(Token.CCL_START):
         nfa_set_char(pair_out)
 
 
@@ -77,6 +81,47 @@ def dodash(input_set):
         lexer.advance()
 
 
+# factor connect
+# factor -> factor factor
+def factor_conn(pair_out):
+    if is_conn(lexer.current_token):
+        factor(pair_out)
+    
+    while is_conn(lexer.current_token):
+        pair = NfaPair()
+        factor(pair)
+        pair_out.end_node.next_1 = pair.start_node
+        pair_out.end_node = pair.end_node
+
+    return True
+
+
+def is_conn(token):
+    nc = [
+        Token.CLOSE_PAREN,
+        Token.AT_EOL,
+        Token.EOS,
+        Token.CLOSURE,
+        Token.PLUS_CLOSE,
+        Token.CCL_END,
+        Token.AT_BOL,
+    ]
+    print(lexer.current_token, token not in nc)
+    return token not in nc
+
+
+# factor * + ? closure
+# factor -> term* | term+ | term?
+def factor(pair_out):
+    term(pair_out)
+    if lexer.match(Token.CLOSURE):
+        nfa_star_closure(pair_out)
+    elif lexer.match(Token.PLUS_CLOSE):
+        nfa_plus_closure(pair_out)
+    elif lexer.match(Token.OPTIONAL):
+        nfa_option_closure(pair_out)
+
+
 # * 闭包操作
 def nfa_star_closure(pair_out):
     if not lexer.match(Token.CLOSURE):
@@ -105,10 +150,29 @@ def nfa_plus_closure(pair_out):
     start.next_1 = pair_out.start_node
 
     pair_out.end_node.next_1 = pair_out.start_node
-    pair_out.end_node.next_2 = end    
+    pair_out.end_node.next_2 = end
 
     pair_out.start_node = start
     pair_out.end_node = end
 
     lexer.advance()
-    return True
\ No newline at end of file
+    return True
+
+
+# ?
+def nfa_option_closure(pair_out):
+    if not lexer.match(Token.OPTIONAL):
+        return False
+    start = Nfa()
+    end = Nfa()
+
+    start.next_1 = pair_out.start_node
+    start.next_2 = end
+    pair_out.end_node.next_1 = end
+
+    pair_out.start_node = start
+    pair_out.end_node = end
+    print(lexer.current_token, '****')
+    lexer.advance()
+    print(lexer.current_token, '****')
+    return True
diff --git a/nfa/nfa.py b/nfa/nfa.py
index 7efea15..05dcf36 100644
--- a/nfa/nfa.py
+++ b/nfa/nfa.py
@@ -36,13 +36,18 @@ def __init__(self):
 
 
 def log_nfa(start_node):
-    if start_node is None or (start_node.next_1 is None and start_node.next_2 is None) or start_node.visited:
-        return
-    log('from: ', start_node.status_num, 'to: ',
-        start_node.next_1.status_num, 'in: ', start_node.edge)
+    log('NFA: ******** ')
+    if start_node.next_1 is not None:
+        log('from: ', start_node.status_num)
+        log('to: ', start_node.next_1.status_num)
+        log('in: ', start_node.edge)
+    else:
+        log('accept: ', start_node.status_num)    
     start_node.visited = True
     if hasattr(start_node, 'input_set'):
         log('input set: ', start_node.input_set)
 
-    log_nfa(start_node.next_1)
-    log_nfa(start_node.next_2)
\ No newline at end of file
+    if start_node.next_1 is not None and not start_node.next_1.visited:
+        log_nfa(start_node.next_1)
+    if start_node.next_2 is not None and not start_node.next_2.visited:
+        log_nfa(start_node.next_2)
diff --git a/test.py b/test.py
index a19d142..b4c0e0e 100644
--- a/test.py
+++ b/test.py
@@ -1,10 +1,11 @@
 from nfa.construction import (
-    term,
-    nfa_star_closure,
-    nfa_plus_closure,
+    factor,
+    factor_conn,
 )
 from nfa.nfa import NfaPair
-from nfa.nfa import log_nfa
+from nfa.nfa import (
+    log_nfa,
+)
 
 
 nfa_pair = NfaPair()
@@ -16,9 +17,9 @@
 # node2.next_2 = Nfa() # 5
 # term(nfa_pair)
 # nfa_star_closure(nfa_pair)
-term(nfa_pair)
-nfa_plus_closure(nfa_pair)
+# term(nfa_pair)
+# nfa_option_closure(nfa_pair)
 # debugNfa(nfa_pair.start_node)
-log_nfa(nfa_pair.start_node)
-# print(nfa_pair.start_node.next_1)
-
+# factor(nfa_pair)
+factor_conn(nfa_pair)
+log_nfa(nfa_pair.start_node)
\ No newline at end of file