cobweb-launcher 1.2.42__tar.gz → 1.2.43__tar.gz
Sign up to get free protection for your applications and to get access to all the features.
- {cobweb-launcher-1.2.42/cobweb_launcher.egg-info → cobweb-launcher-1.2.43}/PKG-INFO +1 -1
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/crawlers/crawler.py +15 -14
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43/cobweb_launcher.egg-info}/PKG-INFO +1 -1
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/setup.py +1 -1
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/LICENSE +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/README.md +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/common_queue.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/decorators.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/item.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/log.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/request.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/response.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/base/seed.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/constant.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/crawlers/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/crawlers/base_crawler.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/crawlers/file_crawler.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/db/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/db/api_db.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/db/redis_db.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/exceptions/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/exceptions/oss_db_exception.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/launchers/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/launchers/launcher.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/launchers/launcher_air.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/launchers/launcher_api.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/launchers/launcher_pro.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/pipelines/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/pipelines/pipeline.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/pipelines/pipeline_console.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/pipelines/pipeline_loghub.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/setting.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/utils/__init__.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/utils/bloom.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/utils/dotting.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/utils/oss.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb/utils/tools.py +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb_launcher.egg-info/SOURCES.txt +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb_launcher.egg-info/dependency_links.txt +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb_launcher.egg-info/requires.txt +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb_launcher.egg-info/top_level.txt +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/setup.cfg +0 -0
- {cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/test/test.py +0 -0
@@ -8,6 +8,7 @@ from urllib.parse import urlparse
|
|
8
8
|
|
9
9
|
import urllib3
|
10
10
|
from requests import HTTPError, Response as Res
|
11
|
+
from requests.exceptions import ChunkedEncodingError
|
11
12
|
|
12
13
|
from cobweb.constant import DealModel, LogTemplate
|
13
14
|
from cobweb.base import (
|
@@ -165,7 +166,7 @@ class Crawler(threading.Thread):
|
|
165
166
|
|
166
167
|
if not iterator_status:
|
167
168
|
raise ValueError("request/download/parse function yield value error!")
|
168
|
-
except
|
169
|
+
except Exception as e:
|
169
170
|
exception_msg = ''.join(traceback.format_exception(type(e), e, e.__traceback__))
|
170
171
|
url = seed.url
|
171
172
|
status = str(e)
|
@@ -189,19 +190,19 @@ class Crawler(threading.Thread):
|
|
189
190
|
seed.params.retry += 1
|
190
191
|
self._set_seed(seed)
|
191
192
|
# time.sleep(self.time_sleep * seed.params.retry)
|
192
|
-
except Exception as e:
|
193
|
-
|
194
|
-
|
195
|
-
|
196
|
-
|
197
|
-
|
198
|
-
|
199
|
-
|
200
|
-
|
201
|
-
|
202
|
-
|
203
|
-
|
204
|
-
|
193
|
+
# except Exception as e:
|
194
|
+
# logger.info(LogTemplate.download_exception.format(
|
195
|
+
# detail=seed_detail_log_info,
|
196
|
+
# retry=seed.params.retry,
|
197
|
+
# priority=seed.params.priority,
|
198
|
+
# seed_version=seed.params.seed_version,
|
199
|
+
# identifier=seed.identifier or "",
|
200
|
+
# exception=''.join(traceback.format_exception(type(e), e, e.__traceback__))
|
201
|
+
# ))
|
202
|
+
# seed.params.retry += 1
|
203
|
+
# # self._todo.push(seed)
|
204
|
+
# self._set_seed(seed)
|
205
|
+
# # time.sleep(self.time_sleep * seed.params.retry)
|
205
206
|
finally:
|
206
207
|
time.sleep(0.1)
|
207
208
|
logger.info("spider thread close")
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
{cobweb-launcher-1.2.42 → cobweb-launcher-1.2.43}/cobweb_launcher.egg-info/dependency_links.txt
RENAMED
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|