Simple fulltex search in-memory

simplefulltextsearch.py

# -*- coding: utf-8 -*-
import re
import shlex

def search(query):
    pieces = shlex.split(query.encode('utf-8'))
    include, or_include, exclude = [], [], []
    for piece in pieces:
        p = piece.decode('utf-8')
        if p.startswith('-'):
            exclude.append(re.compile(ur'\b%s\b' % p[1:], re.U))
        elif p.startswith('|'):
            or_include.append(re.compile(ur'\b%s\b'  % p[1:], re.U))
        else:
            include.append(re.compile(ur'\b%s\b' % p, re.U))
            
    def validator(s):
        return (
        	(
        		all(p.search(s) for p in include) and
        		not any(p.search(s) for p in exclude)
        	)
            or any(p.search(s) for p in or_include)
        )
    return validator

query = u'"Con mèo trèo"'
match = search(query)
print match(u"Con mèo trèo lên cây cau. Hỏi thăm chú chuột đi đâu vắng nhà. Chú chuột đi chợ đường xa. Mua mắm mua muối giỗ cha chú mèo")