From 840dd928db5f4e477dd136a1cb7573cb6aaaa9be Mon Sep 17 00:00:00 2001 From: FongMi Date: Tue, 23 Jul 2024 11:31:08 +0800 Subject: [PATCH] Update spider.py --- chaquo/src/main/python/base/spider.py | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/chaquo/src/main/python/base/spider.py b/chaquo/src/main/python/base/spider.py index db8613304..2560db238 100644 --- a/chaquo/src/main/python/base/spider.py +++ b/chaquo/src/main/python/base/spider.py @@ -80,6 +80,13 @@ class Spider(metaclass=ABCMeta): path = os.path.join(os.path.join(cache_dir, 'py'), f'{name}.py') return SourceFileLoader(name, path).load_module() + def regStr(self, src, reg, group=1): + m = re.search(reg, src) + src = '' + if m: + src = m.group(group) + return src + def removeHtmlTags(self, src): clean = re.compile('<.*?>') return re.sub(clean, '', src)