コード例 #1
0
class SuffixTreeTest(IndexTest):
    def __init__(self, corpus_str, defer_build=True, force_unicode=False):
        if not corpus_str[-1] == '$':
            corpus_str += '$'

        self.force_unicode = force_unicode
        self.corpus_str = corpus_str if not self.force_unicode else unicode(
            corpus_str, 'utf-8', 'replace')

        if not defer_build:
            self.build()
        else:
            self.suffix_tree = None

    def build(self):
        self.suffix_tree = SuffixTree(self.corpus_str)

    def search(self, search_string):
        search_string = search_string if not self.force_unicode else unicode(
            search_string, 'utf-8', 'replace')
        # problem, only returns single index. how to return all?
        return [self.suffix_tree.find_substring(search_string)]

    def supports_unicode(self):
        return True
コード例 #2
0
class SuffixTreeTest(IndexTest):
	def __init__(self, corpus_str, defer_build=True, force_unicode=False):
		if not corpus_str[-1] == '$':
			corpus_str += '$'

		self.force_unicode = force_unicode
		self.corpus_str = corpus_str if not self.force_unicode else unicode(corpus_str, 'utf-8', 'replace')

		if not defer_build:
			self.build()
		else:
			self.suffix_tree = None

	def build(self):
		self.suffix_tree = SuffixTree(self.corpus_str)

	def search(self, search_string):
		search_string = search_string if not self.force_unicode else unicode(search_string, 'utf-8', 'replace')
		# problem, only returns single index. how to return all?
		return [self.suffix_tree.find_substring(search_string)]

	def supports_unicode(self):
		return True
コード例 #3
0
 def build(self):
     self.suffix_tree = SuffixTree(self.corpus_str)
コード例 #4
0
	def build(self):
		self.suffix_tree = SuffixTree(self.corpus_str)