[英]How to build trie tree to solve this parse algorithm
我正在尝试使用特里树来解决这个问题:
Symbol string generator consists of two parts, a set of the start symbol and a set of rules of generation.
For example:
Start symbol: ['S'], Rules of generation: ["S → abc", "S → aA", "A → b", "A → c"]
Then, symbolic string abc can be generated because S → abc.
Symbolic string ab can be generated because S → aA → ab.
Symbolic string abc can be generated because S → aA → ac.
Now, give you a symbolic string generator and a symbolic string, and you need to return True if the symbolic string can be generated, False otherwise
Example
Given generator = ["S -> abcd", "S -> Ad", "A -> ab", "A -> c"], startSymbol = S, symbolString = “abd”, return True.
explanation:
S → Ad → abd
Given generator = ["S → abc", "S → aA", "A → b", "A → c"], startSymbol = S, symbolString = “a”, return False
我发现此问题的关键是建立特里树。 我试图写:
def build_trie(values): #value is like ['abc', 'Ad'...]
root = {}
for word in values:
current = root
is_end = False
for c in word:
if 'A' <= c <= 'Z':
vals = m[c] #m is a mapping of {'S': ['abc', 'Ad'], ...}
rs = build_trie(vals)
for k in rs:
if k not in current:
current[k] = rs[k]
else:
# stuck here...
pass
# temp = collections.defaultdict(dict)
# for d in (current[k], rs[k]):
# for k, v in d.items():
# if k in temp and k != '__end__':
# temp[k].update(v)
# else:
# temp[k] = v
# # current[k].update(rs[k])
# current[k] = temp[k]
is_end = True
else:
current = current.setdefault(c, {})
is_end = False
if not is_end:
current['__end__'] = '__end__'
return root
但是卡在其他部分上...还没有弄清楚如何编写这棵特里树。 有什么线索吗?
您可能要使用python中的多个解析器库。 我使用了LARK解析器 。 他们给出了各种python解析器的比较。
在大学期间,我在C语言中实现了LALR(1)解析器。我想它的用处会更少。 我发现了一个Python有用的实现在这里 ,如果你想重新写入整个分析器。 我还没有测试该代码的工作原理。
对于给定的语法,我使用LARK编写了一个验证器,如下所示。
from lark import Lark
import sys
grammar = """
start: "abcd"
| A "d"
A: "ab"
| "c"
"""
parser = Lark(grammar)
def check_grammer(word):
try:
parser.parse(word)
return True
except Exception as exception:
print exception
return False
word = sys.argv[1]
print check_grammer(word)
希望能帮助到你!
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.