cobweb-launcher 1.1.5__tar.gz → 1.1.9__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {cobweb-launcher-1.1.5/cobweb_launcher.egg-info → cobweb-launcher-1.1.9}/PKG-INFO +1 -1
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/crawlers/file_crawler.py +3 -1
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/launchers/launcher.py +10 -6
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/launchers/launcher_pro.py +19 -18
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9/cobweb_launcher.egg-info}/PKG-INFO +1 -1
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/setup.py +2 -2
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/LICENSE +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/README.md +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/common_queue.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/decorators.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/item.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/log.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/request.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/response.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/base/seed.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/constant.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/crawlers/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/crawlers/base_crawler.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/db/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/db/redis_db.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/exceptions/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/exceptions/oss_db_exception.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/launchers/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/pipelines/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/pipelines/base_pipeline.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/pipelines/loghub_pipeline.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/setting.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/utils/__init__.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/utils/oss.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb/utils/tools.py +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb_launcher.egg-info/SOURCES.txt +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb_launcher.egg-info/dependency_links.txt +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb_launcher.egg-info/requires.txt +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/cobweb_launcher.egg-info/top_level.txt +0 -0
- {cobweb-launcher-1.1.5 → cobweb-launcher-1.1.9}/setup.cfg +0 -0
|
@@ -123,12 +123,14 @@ class FileCrawlerPro(Crawler):
|
|
|
123
123
|
item.seed, response, filter=True, msg=f"response content type is {content_type}",
|
|
124
124
|
bucket_name=bucket_name, data_size=content_length, **seed_dict
|
|
125
125
|
)
|
|
126
|
+
response.close()
|
|
126
127
|
elif position == 1 and min_upload_size >= content_length > 0:
|
|
127
128
|
"""过小文件标识返回"""
|
|
128
129
|
yield Response(
|
|
129
130
|
item.seed, response, filter=True, msg="file size is too small",
|
|
130
131
|
bucket_name=bucket_name, data_size=content_length, **seed_dict
|
|
131
132
|
)
|
|
133
|
+
response.close()
|
|
132
134
|
elif position == 1 and chunk_size > content_length > min_upload_size:
|
|
133
135
|
"""小文件直接下载"""
|
|
134
136
|
for part_data in response.iter_content(chunk_size):
|
|
@@ -153,12 +155,12 @@ class FileCrawlerPro(Crawler):
|
|
|
153
155
|
seed_dict['position'] = position
|
|
154
156
|
seed_dict['start'] = upload_content_length
|
|
155
157
|
|
|
156
|
-
response.close()
|
|
157
158
|
if content:
|
|
158
159
|
oss_util.put_part(key, upload_id, position, content)
|
|
159
160
|
content_length += len(content)
|
|
160
161
|
oss_util.merge(key, upload_id)
|
|
161
162
|
yield Response(item.seed, response, bucket_name=bucket_name, data_size=content_length, **seed_dict)
|
|
163
|
+
response.close()
|
|
162
164
|
|
|
163
165
|
elif item.seed.params.identifier == "merge":
|
|
164
166
|
oss_util.merge(key, seed_dict["upload_id"])
|
|
@@ -34,7 +34,7 @@ class Launcher(threading.Thread):
|
|
|
34
34
|
"_delete",
|
|
35
35
|
]
|
|
36
36
|
|
|
37
|
-
def __init__(self, task, project, custom_setting=None):
|
|
37
|
+
def __init__(self, task, project, custom_setting=None, **kwargs):
|
|
38
38
|
super().__init__()
|
|
39
39
|
self.task = task
|
|
40
40
|
self.project = project
|
|
@@ -42,10 +42,11 @@ class Launcher(threading.Thread):
|
|
|
42
42
|
self._stop = threading.Event() # 结束事件
|
|
43
43
|
self._pause = threading.Event() # 暂停事件
|
|
44
44
|
|
|
45
|
+
_setting = dict()
|
|
46
|
+
|
|
45
47
|
if custom_setting:
|
|
46
|
-
setting_ = dict()
|
|
47
48
|
if isinstance(custom_setting, dict):
|
|
48
|
-
|
|
49
|
+
_setting = custom_setting
|
|
49
50
|
else:
|
|
50
51
|
if isinstance(custom_setting, str):
|
|
51
52
|
custom_setting = importlib.import_module(custom_setting)
|
|
@@ -53,9 +54,12 @@ class Launcher(threading.Thread):
|
|
|
53
54
|
raise Exception
|
|
54
55
|
for k, v in custom_setting.__dict__.items():
|
|
55
56
|
if not k.startswith("__") and not inspect.ismodule(v):
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
57
|
+
_setting[k] = v
|
|
58
|
+
|
|
59
|
+
_setting.update(**kwargs)
|
|
60
|
+
|
|
61
|
+
for k, v in _setting.items():
|
|
62
|
+
setattr(setting, k, v)
|
|
59
63
|
|
|
60
64
|
self._Crawler = dynamic_load_class(setting.CRAWLER)
|
|
61
65
|
self._Pipeline = dynamic_load_class(setting.PIPELINE)
|
|
@@ -9,8 +9,8 @@ from cobweb.constant import DealModel, LogTemplate
|
|
|
9
9
|
|
|
10
10
|
class LauncherPro(Launcher):
|
|
11
11
|
|
|
12
|
-
def __init__(self, task, project, custom_setting=None):
|
|
13
|
-
super().__init__(task, project, custom_setting)
|
|
12
|
+
def __init__(self, task, project, custom_setting=None, **kwargs):
|
|
13
|
+
super().__init__(task, project, custom_setting, **kwargs)
|
|
14
14
|
self._todo = "{%s:%s}:todo" % (project, task)
|
|
15
15
|
self._done = "{%s:%s}:done" % (project, task)
|
|
16
16
|
self._fail = "{%s:%s}:fail" % (project, task)
|
|
@@ -140,22 +140,23 @@ class LauncherPro(Launcher):
|
|
|
140
140
|
while not self._stop.is_set():
|
|
141
141
|
queue_not_empty_count = 0
|
|
142
142
|
pooling_wait_seconds = 30
|
|
143
|
-
|
|
144
|
-
|
|
145
|
-
|
|
146
|
-
|
|
147
|
-
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
|
|
151
|
-
|
|
152
|
-
|
|
153
|
-
|
|
154
|
-
|
|
155
|
-
|
|
156
|
-
|
|
157
|
-
|
|
158
|
-
|
|
143
|
+
for q in self.__LAUNCHER_QUEUE__.values():
|
|
144
|
+
if q.length != 0:
|
|
145
|
+
queue_not_empty_count += 1
|
|
146
|
+
if self._pause.is_set():
|
|
147
|
+
self._pause.clear()
|
|
148
|
+
self._execute()
|
|
149
|
+
elif queue_not_empty_count == 0:
|
|
150
|
+
pooling_wait_seconds = 5
|
|
151
|
+
check_emtpy_times += 1
|
|
152
|
+
else:
|
|
153
|
+
check_emtpy_times = 0
|
|
154
|
+
|
|
155
|
+
if not self._db.zcount(self._todo, _min=0, _max="(1000") and check_emtpy_times > 2:
|
|
156
|
+
check_emtpy_times = 0
|
|
157
|
+
self.__DOING__ = {}
|
|
158
|
+
self._pause.set()
|
|
159
|
+
|
|
159
160
|
if not self._pause.is_set():
|
|
160
161
|
logger.info(LogTemplate.launcher_pro_polling.format(
|
|
161
162
|
task=self.task,
|
|
@@ -5,7 +5,7 @@ with open("README.md", "r", encoding="utf-8") as fh:
|
|
|
5
5
|
|
|
6
6
|
setup(
|
|
7
7
|
name="cobweb-launcher",
|
|
8
|
-
version="1.1.
|
|
8
|
+
version="1.1.9",
|
|
9
9
|
packages=find_packages(),
|
|
10
10
|
url="https://github.com/Juannie-PP/cobweb",
|
|
11
11
|
license="MIT",
|
|
@@ -22,4 +22,4 @@ setup(
|
|
|
22
22
|
"cobweb-launcher, cobweb",
|
|
23
23
|
],
|
|
24
24
|
python_requires=">=3.7",
|
|
25
|
-
)
|
|
25
|
+
)
|
|
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-1.1.5 → cobweb-launcher-1.1.9}/cobweb_launcher.egg-info/dependency_links.txt
RENAMED
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|