2 @file alchemy_extractor.py 3 @author Alexey <developers.hce@gmail.com> 4 @link http://hierarchical-cluster-engine.com/ 5 @copyright Copyright © 2013 IOIX Ukraine 6 @license http://hierarchical-cluster-engine.com/license/ 7 @package HCE project node API 24 def __init__(self, config, templ=None, domain=None, processorProperties=None):
25 BaseExtractor.__init__(self, config, templ, domain, processorProperties)
26 self.
name = CONSTS.EXTRACTOR_NAME_ALCHEMY
27 self.
data[
"extractor"] = CONSTS.EXTRACTOR_NAME_ALCHEMY
36 logger.info(
"AAAAAAA")
38 logger.info(
"BBBBBBB")
39 text = parser.text(
"html", resource.raw_html)
40 logger.info(
"CCCCCCC")
41 logger.info(
"Article's corpus: %s", text)
42 self.
addTag(result=reslt, \
43 tag_name=CONSTS.TAG_CONTENT_UTF8_ENCODED, \
45 logger.info(
"DDDDDDD")
46 except Exception, err:
def varDump(obj, stringify=True, strTypeMaxLen=256, strTypeCutSuffix='...', stringifyType=1, ignoreErrors=False, objectsHash=None, depth=0, indent=2, ensure_ascii=False, maxDepth=10)