def feature_extraction(csvfile,
datapath,
resultcsv,
fe_path=config.path_fe_space):
fe = FeatureExtract(fe_path)
fe.print_featuremap()
# feature extraction
out = open(resultcsv, "w")
line = "%s,%s,%s" % ('id', 'label', fe.str_featuremap_line())
# print line
out.write(line + "\n")
out.flush()
with open(csvfile, 'rb') as csvfile:
reader = csv.reader(csvfile, delimiter=',')
header = True
for row in reader:
if not header:
item = UrlItem()
ct = open(datapath + "/" + row[0], "r").read()
item['url'] = row[1]
item['content'] = ct
item['title'] = item.get_short_title()
f = fe.extract_item(item)
line = "%s,%s,%s" % (row[0], row[2],
FeatureExtract.str_feature(f))
print line
out.write(line + "\n")
out.flush()
else:
header = False
out.close()
def feature_extraction(csvfile, datapath, resultcsv,fe_path=config.path_fe_space):
fe = FeatureExtract(fe_path)
fe.print_featuremap()
# feature extraction
out = open(resultcsv, "w")
line = "%s,%s,%s" % ('id', 'label', fe.str_featuremap_line())
# print line
out.write(line+"\n")
out.flush()
with open(csvfile, 'rb') as csvfile:
reader = csv.reader(csvfile, delimiter=',')
header = True
for row in reader:
if not header:
item = UrlItem()
ct = open(datapath + "/" + row[0], "r").read()
item['url'] = row[1]
item['content'] = ct
item['title'] = item.get_short_title()
f = fe.extract_item(item)
line = "%s,%s,%s" % (row[0], row[2], FeatureExtract.str_feature(f))
print line
out.write(line+"\n")
out.flush()
else:
header = False
out.close()
def parse_item(response):
# response are all updated or new
# db not changed
# is this url in URL_LIB
item = UrlItem.load_with_content(url=response.url, response=response)
logging.info("PC get page [%s]:- %s" % (item['id'], item['url']))
yield item
def __op_new(self, data_loaded, connection):
item_id = int(data_loaded['id'])
item = UrlItem.load_with_content(
id=item_id,file_path=config.path_judge_inbox)
feature = self.__fe.extract_item(item)
if 'decision' not in data_loaded.keys():
decision, confidence = self.__auto_judge(feature)
log.info("[%s]: [%s] # %s # %s%%" % (item_id, FeatureExtract.str_feature(feature), decision, confidence))
else:
decision, confidence = data_loaded['decision'],100
log.info("[%s]: back from Extractor # %s # %s%%" % (item_id, decision, confidence))
self.__relearn_clf(feature,decision)
if confidence > config.const_CONFIDENCE_THRESHOLD:
item['is_target'] = decision
item.save()
if int(item['is_target']) in [config.const_IS_TARGET_MULTIPLE, config.const_IS_TARGET_SIGNLE]:
self.__send_to_extractor(item)
else:
os.remove(config.path_judge_inbox + "/%s" % item.filename())
else:
item['is_target'] = config.const_IS_TARGET_UNKNOW
item.save()
self.__judge_queue[item_id] = {
"title": item['title'],
"url": item['url'],
"filename": item.filename(),
"confidence": round(confidence,2),
"decision": decision,
"feature": feature
}
pass
def __op_new(self, data_loaded, connection):
item_id = int(data_loaded['id'])
item = UrlItem.load_with_content(
id=item_id,file_path=config.path_extractor_inbox)
count, maps = db.get_url_with_same_layout_hash(item['layout_hash'])
log.info(str(maps))
log.info(count)
if len(maps) > 0:
import operator
tar_ext = max(maps.iteritems(), key=operator.itemgetter(1))
log.info(float(tar_ext[1]) / len(maps))
if tar_ext[1] > config.extractor_same_layout_number:
extractor = tool.str2extractor(tar_ext[0])
self.__extract(item,extractor)
return
extractor = config.const_RULE_UNKNOW
self.__ext_queue[item_id] = {
"title": item['title'],
"url": item['url'],
"filename": item.filename(),
"decision": item['is_target'],
"extractor": extractor
}
log.info("[%s]: # %s " % (item_id, extractor))
pass
def __op_new(self, data_loaded, connection):
item_id = int(data_loaded['id'])
item = UrlItem.load_with_content(id=item_id,
file_path=config.path_extractor_inbox)
count, maps = db.get_url_with_same_layout_hash(item['layout_hash'])
log.info(str(maps))
log.info(count)
if len(maps) > 0:
import operator
tar_ext = max(maps.iteritems(), key=operator.itemgetter(1))
log.info(float(tar_ext[1]) / len(maps))
if tar_ext[1] > config.extractor_same_layout_number:
extractor = tool.str2extractor(tar_ext[0])
self.__extract(item, extractor)
return
extractor = config.const_RULE_UNKNOW
self.__ext_queue[item_id] = {
"title": item['title'],
"url": item['url'],
"filename": item.filename(),
"decision": item['is_target'],
"extractor": extractor
}
log.info("[%s]: # %s " % (item_id, extractor))
pass
def parse_item(response):
# response are all updated or new
# db not changed
# is this url in URL_LIB
item = UrlItem.load_with_content(url=response.url, response=response)
logging.info("PC get page [%s]:- %s" % (item['id'], item['url']))
yield item
def __op_rejudge_done(self, data_loaded, connection):
item_id = int(data_loaded['id'])
decision = int(data_loaded['decision'])
item = UrlItem.load(id=item_id)
del self.__ext_queue[item_id]
self.__send_back_to_judge(item, decision)
tool.send_msg(connection, "0")
pass
def __op_rejudge_done(self, data_loaded, connection):
item_id = int(data_loaded['id'])
decision = int(data_loaded['decision'])
item = UrlItem.load(id=item_id)
del self.__ext_queue[item_id]
self.__send_back_to_judge(item, decision)
tool.send_msg(connection, "0")
pass
def __op_add_extractor(self, data_loaded, connection):
item_id = int(data_loaded['id'])
extractor = data_loaded['extractor']
item = UrlItem.load_with_content(item_id,
file_path=config.path_extractor_inbox)
self.__extract(item, extractor)
del self.__ext_queue[item['id']]
tool.send_msg(connection, "0")
pass
def __op_test_rule(self, data_loaded, connection):
item_id = int(data_loaded['id'])
rule = data_loaded['rule']
attrid = int(data_loaded['attrid'])
item = UrlItem.load_with_content(id=item_id,
file_path=config.path_extractor_inbox)
tool.send_msg(
connection,
self.__ie.extract_attr(item, rule_id_or_dict=rule, attr_id=attrid))
pass
def __op_test_rule(self, data_loaded, connection):
item_id = int(data_loaded['id'])
rule = data_loaded['rule']
attrid = int(data_loaded['attrid'])
item = UrlItem.load_with_content(
id=item_id, file_path=config.path_extractor_inbox)
tool.send_msg(
connection,
self.__ie.extract_attr(item, rule_id_or_dict=rule, attr_id=attrid)
)
pass
def __op_add_extractor(self, data_loaded, connection):
item_id = int(data_loaded['id'])
extractor = data_loaded['extractor']
item = UrlItem.load_with_content(
item_id, file_path=config.path_extractor_inbox)
self.__extract(item,extractor)
del self.__ext_queue[item['id']]
tool.send_msg(
connection,
"0"
)
pass
def __op_preview(self, data_loaded, connection):
log.info(data_loaded['extractor'])
if data_loaded['extractor'] == config.const_RULE_UNKNOW:
result = {x: "" for x in xrange(1, self.__ie.num_attr() + 1)}
else:
item_id = int(data_loaded['id'])
item = UrlItem.load_with_content(
item_id, file_path=config.path_extractor_inbox)
extractor = data_loaded['extractor']
result = self.__ie.extract(item, extractor)
preview = list()
for att, str in result.iteritems():
preview.insert(att, dict(name=self.__ie.name(att), value=str))
log.info(preview)
tool.send_msg(connection, pickle.dumps(preview, -1))
pass
def __op_preview(self, data_loaded, connection):
log.info(data_loaded['extractor'])
if data_loaded['extractor'] == config.const_RULE_UNKNOW:
result = {x: "" for x in xrange(1, self.__ie.num_attr() + 1)}
else:
item_id = int(data_loaded['id'])
item = UrlItem.load_with_content(
item_id, file_path=config.path_extractor_inbox)
extractor = data_loaded['extractor']
result = self.__ie.extract(item, extractor)
preview = list()
for att, str in result.iteritems():
preview.insert(att, dict(name=self.__ie.name(att), value=str))
log.info(preview)
tool.send_msg(connection, pickle.dumps(preview, -1))
pass
def __refresh_list(self):
delete_ids = []
for key, ent in self.__judge_queue.iteritems():
decision, confidence = self.__auto_judge(ent['feature'])
if confidence > config.const_CONFIDENCE_THRESHOLD:
item = UrlItem.load(id=key)
item['is_target'] = decision
item.save()
delete_ids.append(key)
if int(item['is_target']) in [config.const_IS_TARGET_MULTIPLE, config.const_IS_TARGET_SIGNLE]:
self.__send_to_extractor(item)
else:
os.remove(config.path_judge_inbox + "/%s" % ent['filename'])
else:
self.__judge_queue[key]['confidence'] = confidence
self.__judge_queue[key]['decision'] = decision
for ent_id in delete_ids:
del self.__judge_queue[ent_id]
def __op_done(self, data_loaded, connection):
item_id = int(data_loaded['id'])
decision = int(data_loaded['decision'])
item = UrlItem.load(id=item_id)
item['is_target'] = decision
item.save()
if int(item['is_target']) in [config.const_IS_TARGET_MULTIPLE, config.const_IS_TARGET_SIGNLE]:
self.__send_to_extractor(item)
else:
os.remove(config.path_judge_inbox + "/%s" % item.filename())
self.__relearn_clf(self.__judge_queue[item_id]['feature'],decision)
del self.__judge_queue[item_id]
tool.send_msg(connection, "0")
pass
def __op_new(self, data_loaded, connection):
item_id = int(data_loaded['id'])
item = UrlItem.load_with_content(id=item_id,
file_path=config.path_judge_inbox)
feature = self.__fe.extract_item(item)
if 'decision' not in data_loaded.keys():
decision, confidence = self.__auto_judge(feature)
log.info("[%s]: [%s] # %s # %s%%" %
(item_id, FeatureExtract.str_feature(feature), decision,
confidence))
else:
decision, confidence = data_loaded['decision'], 100
log.info("[%s]: back from Extractor # %s # %s%%" %
(item_id, decision, confidence))
self.__relearn_clf(feature, decision)
if confidence > config.const_CONFIDENCE_THRESHOLD:
# pretty sure, save to db, and pass to extract
item['is_target'] = decision
item.save()
if int(item['is_target']) in [
config.const_IS_TARGET_MULTIPLE,
config.const_IS_TARGET_SIGNLE
]:
# item is target
self.__send_to_extractor(item)
else:
# item is not target
os.remove(config.path_judge_inbox + "/%s" % item.filename())
else:
# not sure, put it in queue, involving human-being
item['is_target'] = config.const_IS_TARGET_UNKNOW
item.save()
self.__judge_queue[item_id] = {
"title": item['title'],
"url": item['url'],
"filename": item.filename(),
"confidence": round(confidence, 2),
"decision": decision,
"feature": feature
}
pass
def __op_refresh(self, data_loaded, connection):
delete_ids = []
for key, ent in self.__ext_queue.iteritems():
item_id = int(key)
item = UrlItem.load_with_content(
id=item_id, file_path=config.path_extractor_inbox)
count, maps = db.get_url_with_same_layout_hash(item['layout_hash'])
log.info(str(maps))
log.info(count)
if len(maps) > 0:
import operator
tar_ext = max(maps.iteritems(), key=operator.itemgetter(1))
log.info(float(tar_ext[1]) / len(maps))
if tar_ext[1] > config.extractor_same_layout_number:
extractor = tool.str2extractor(tar_ext[0])
self.__extract(item, extractor)
delete_ids.append(item_id)
# clear delete_ids
for ent_id in delete_ids:
del self.__ext_queue[ent_id]
def __op_done(self, data_loaded, connection):
item_id = int(data_loaded['id'])
decision = int(data_loaded['decision'])
item = UrlItem.load(id=item_id)
item['is_target'] = decision
item.save()
if int(item['is_target']) in [
config.const_IS_TARGET_MULTIPLE, config.const_IS_TARGET_SIGNLE
]:
# item is target
self.__send_to_extractor(item)
else:
# item is not target
os.remove(config.path_judge_inbox + "/%s" % item.filename())
self.__relearn_clf(self.__judge_queue[item_id]['feature'], decision)
del self.__judge_queue[item_id]
tool.send_msg(connection, "0")
pass
def __op_refresh(self,data_loaded,connection):
delete_ids = []
for key, ent in self.__ext_queue.iteritems():
item_id = int(key)
if not os.path.isfile(
config.path_extractor_inbox+"/"+str(item_id)+".html"):
delete_ids.append(item_id)
continue
item = UrlItem.load_with_content(
id=item_id,file_path=config.path_extractor_inbox)
count, maps = db.get_url_with_same_layout_hash(item['layout_hash'])
log.info(str(maps))
log.info(count)
if len(maps) > 0:
import operator
tar_ext = max(maps.iteritems(), key=operator.itemgetter(1))
log.info(float(tar_ext[1]) / len(maps))
if tar_ext[1] > config.extractor_same_layout_number:
extractor = tool.str2extractor(tar_ext[0])
self.__extract(item,extractor)
delete_ids.append(item_id)
for ent_id in delete_ids:
del self.__ext_queue[ent_id]
def __refresh_list(self):
delete_ids = []
for key, ent in self.__judge_queue.iteritems():
decision, confidence = self.__auto_judge(ent['feature'])
if confidence > config.const_CONFIDENCE_THRESHOLD:
# pretty sure, save to db, and pass to extract
item = UrlItem.load(id=key)
item['is_target'] = decision
item.save()
delete_ids.append(key)
if int(item['is_target']) in [
config.const_IS_TARGET_MULTIPLE,
config.const_IS_TARGET_SIGNLE
]:
self.__send_to_extractor(item)
else:
os.remove(config.path_judge_inbox +
"/%s" % ent['filename'])
else:
self.__judge_queue[key]['confidence'] = confidence
self.__judge_queue[key]['decision'] = decision
# clear delete_ids
for ent_id in delete_ids:
del self.__judge_queue[ent_id]
__author__ = 'LeoDong'
import requests
from util import config
from bs4 import BeautifulSoup
from extractor.InfoExtractor import InfoExtractor
from SAECrawlers.items import UrlItem
import json
ie = InfoExtractor(config.path_extract_onto+"/seminar.xml",config.path_extract_onto)
item = UrlItem.load_with_content(id=1,file_path=config.path_judge_inbox)
# print item.get_part('soup').prettify()
# print item.get_part('content')
rule={
"on": "content",
"scope": {
"sel":"section#visible-body .logo",
"target":"text"
},
"description": "url",
"actions": [
2
],
"substring": {
"after": "H",
"before": ""
},
}
print ie.extract_attr(item,rule_id_or_dict=rule)
__author__ = 'LeoDong'
import requests
from util import config
from bs4 import BeautifulSoup
from extractor.InfoExtractor import InfoExtractor
from SAECrawlers.items import UrlItem
import json
ie = InfoExtractor(config.path_extract_onto + "/seminar.xml",
config.path_extract_onto)
item = UrlItem.load_with_content(id=1, file_path=config.path_judge_inbox)
# print item.get_part('soup').prettify()
# print item.get_part('content')
rule = {
"on": "content",
"scope": {
"sel": "section#visible-body .logo",
"target": "text"
},
"description": "url",
"actions": [2],
"substring": {
"after": "H",
"before": ""
},
}
print ie.extract_attr(item, rule_id_or_dict=rule)
# print json.dumps(ie.map(1), indent=2)
def parse(self, response):
item = UrlItem.load_with_content(url=response.url, response=response)
logging.debug("Updater get page [%s]:- %s" % (item['id'], item['url']))
yield item
