cobweb-launcher 1.0.0__py3-none-any.whl → 1.0.2__py3-none-any.whl

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.

Potentially problematic release.


This version of cobweb-launcher might be problematic. Click here for more details.

Files changed (45) hide show
  1. cobweb/__init__.py +1 -1
  2. cobweb/launchers/launcher.py +1 -2
  3. {cobweb_launcher-1.0.0.dist-info → cobweb_launcher-1.0.2.dist-info}/METADATA +1 -1
  4. cobweb_launcher-1.0.2.dist-info/RECORD +32 -0
  5. cobweb/bbb.py +0 -191
  6. cobweb/config.py +0 -164
  7. cobweb/db/oss_db.py +0 -128
  8. cobweb/db/scheduler/__init__.py +0 -1
  9. cobweb/db/scheduler/default.py +0 -8
  10. cobweb/db/scheduler/textfile.py +0 -27
  11. cobweb/db/storer/__init__.py +0 -1
  12. cobweb/db/storer/console.py +0 -9
  13. cobweb/db/storer/loghub.py +0 -54
  14. cobweb/db/storer/redis.py +0 -15
  15. cobweb/db/storer/textfile.py +0 -15
  16. cobweb/decorators.py +0 -16
  17. cobweb/distributed/__init__.py +0 -0
  18. cobweb/distributed/launcher.py +0 -243
  19. cobweb/distributed/models.py +0 -143
  20. cobweb/equip/__init__.py +0 -8
  21. cobweb/equip/dev/__init__.py +0 -0
  22. cobweb/equip/dev/launcher.py +0 -202
  23. cobweb/equip/dev/models.py +0 -156
  24. cobweb/equip/distributed/__init__.py +0 -0
  25. cobweb/equip/distributed/launcher.py +0 -219
  26. cobweb/equip/distributed/models.py +0 -158
  27. cobweb/equip/download/__init__.py +0 -0
  28. cobweb/equip/download/launcher.py +0 -203
  29. cobweb/equip/download/models.py +0 -169
  30. cobweb/equip/single/__init__.py +0 -0
  31. cobweb/equip/single/launcher.py +0 -203
  32. cobweb/equip/single/models.py +0 -173
  33. cobweb/interface.py +0 -34
  34. cobweb/log.py +0 -96
  35. cobweb/new.py +0 -20
  36. cobweb/single/__init__.py +0 -0
  37. cobweb/single/launcher.py +0 -231
  38. cobweb/single/models.py +0 -134
  39. cobweb/single/nest.py +0 -153
  40. cobweb/task.py +0 -61
  41. cobweb/utils.py +0 -90
  42. cobweb_launcher-1.0.0.dist-info/RECORD +0 -69
  43. {cobweb_launcher-1.0.0.dist-info → cobweb_launcher-1.0.2.dist-info}/LICENSE +0 -0
  44. {cobweb_launcher-1.0.0.dist-info → cobweb_launcher-1.0.2.dist-info}/WHEEL +0 -0
  45. {cobweb_launcher-1.0.0.dist-info → cobweb_launcher-1.0.2.dist-info}/top_level.txt +0 -0
@@ -1,203 +0,0 @@
1
- import time
2
- import threading
3
-
4
- from .. import log, sqn, rtn, pim
5
- from .. import Queue, DBItem, RedisDB, Setting, OssDB
6
- from .models import Scheduler, Spider, Storer
7
-
8
-
9
- def check(stop, last, spider, scheduler, storer, ready_seed_length, spider_queue_length):
10
- log.info("run check thread after 30 seconds...")
11
- time.sleep(30)
12
- spider_info = """
13
- ------------------- check: {0} ------------------
14
- redis_spider_seed_length: {1}
15
- redis_ready_seed_length: {2}
16
- running_spider_thread_num: {3}
17
- memory_seed_queue_length: {4}
18
- storer_queue_length_info: {5}
19
- ----------------------- end -----------------------"""
20
- while True:
21
- status = "running"
22
- running_spider_thread_num = spider.spider_in_progress.length
23
- redis_ready_seed_length = ready_seed_length()
24
- redis_spider_seed_length = spider_queue_length()
25
- memory_seed_queue_length = scheduler.queue.length
26
- storer_upload_queue_length = storer.queue.length if storer else None
27
- if (
28
- scheduler.stop and
29
- not memory_seed_queue_length and
30
- not running_spider_thread_num
31
- ):
32
- if not Setting.LAUNCHER_MODEL:
33
- log.info("spider is done?")
34
- last.set()
35
- time.sleep(3)
36
- storer_queue_empty = True
37
- if storer and storer.queue.length:
38
- storer_queue_empty = False
39
- storer_upload_queue_length = storer.queue.length if storer else None
40
- if (
41
- storer_queue_empty and
42
- not redis_ready_seed_length and
43
- not redis_spider_seed_length
44
- ):
45
- if Setting.LAUNCHER_MODEL:
46
- log.info("waiting for push seeds...")
47
- status = "waiting"
48
- time.sleep(30)
49
- else:
50
- log.info("spider done!")
51
- break
52
-
53
- last.clear()
54
-
55
- log.info(spider_info.format(
56
- status,
57
- redis_spider_seed_length,
58
- redis_ready_seed_length,
59
- running_spider_thread_num,
60
- memory_seed_queue_length,
61
- storer_upload_queue_length
62
- ))
63
-
64
- time.sleep(3)
65
- stop.set()
66
-
67
-
68
- def launcher(task):
69
- """
70
- 任务启动装饰器
71
- :param task: 任务配置信息
72
- """
73
- def decorator(func):
74
- # 程序结束事件
75
- last = threading.Event()
76
- # 停止采集事件
77
- stop = threading.Event()
78
-
79
- # 初始化redis信息
80
- redis_db = RedisDB(task.project, task.task_name, task.redis_info)
81
-
82
- # new item
83
- item = type("Item", (object,), {"redis_client": redis_db.client})()
84
- if task.oss_config:
85
- item.oss = OssDB(**task.oss_config)
86
-
87
- log.info("初始化cobweb!")
88
-
89
- seed_queue = Queue()
90
-
91
- scheduler_info = task.scheduler_info or dict()
92
- # 调度器动态继承
93
- sql = scheduler_info.get("sql")
94
- table = scheduler_info.get("table")
95
- size = scheduler_info.get("size")
96
- scheduler_config = scheduler_info.get("config")
97
- scheduler_db = scheduler_info.get("db", "default")
98
- DB, class_name = pim(scheduler_db, "scheduler")
99
- # SchedulerDB, table, sql, length, size, config = task.scheduler_info
100
- SchedulerTmp = type(class_name, (Scheduler, DB), {})
101
- # 初始化调度器
102
- scheduler = SchedulerTmp(
103
- table=table, sql=sql, size=size, queue=seed_queue,
104
- length=task.scheduler_queue_length, config=scheduler_config
105
- )
106
-
107
- storer = None
108
- storer_info = task.storer_info or dict()
109
-
110
- if storer_info:
111
- storer_db = storer_info["db"]
112
- fields = storer_info["fields"]
113
- storer_table = storer_info.get("table", "console")
114
- storer_config = storer_info.get("config")
115
-
116
- StorerDB, class_name = pim(storer_db, "storer")
117
- StorerTmp = type(class_name, (Storer, StorerDB), {})
118
-
119
- db_name = class_name.lower()
120
- if not getattr(item, db_name, None):
121
- instance = type(db_name, (DBItem,), {})
122
- setattr(item, db_name, instance)
123
-
124
- storer_item_instance = getattr(item, db_name)
125
- storer_item_instance.init_item(storer_table, fields)
126
-
127
- storer_queue = sqn(db_name, storer_table)
128
- queue = getattr(storer_item_instance, storer_queue)
129
- # 初始话存储器
130
- table_name = rtn(table_name=storer_table)
131
- storer = StorerTmp(
132
- table=table_name, fields=fields,
133
- length=task.storer_queue_length,
134
- queue=queue, config=storer_config
135
- )
136
-
137
- # 初始化采集器
138
- spider = Spider(seed_queue, storer and True, task.max_retries)
139
-
140
- threading.Thread(target=redis_db.check_spider_queue, args=(stop, 0)).start()
141
- threading.Thread(target=redis_db.set_heartbeat, args=(stop,)).start()
142
-
143
- # 推送初始种子
144
- # seeds = start_seeds(task.start_seed)
145
- redis_db.add_seed(task.seeds)
146
- # 启动调度器, 调度至redis队列
147
- threading.Thread(
148
- # name="xxxx_schedule_seeds",
149
- target=scheduler.schedule_seed,
150
- args=(
151
- redis_db.ready_seed_length,
152
- redis_db.get_scheduler_lock,
153
- redis_db.add_seed
154
- )
155
- ).start()
156
-
157
- # 启动调度器, 调度任务队列
158
- threading.Thread(
159
- # name="xxxx_schedule_task",
160
- target=scheduler.schedule_task,
161
- args=(
162
- stop, redis_db.get_seed,
163
- redis_db.ready_seed_length
164
- )
165
- ).start()
166
-
167
- # 启动采集器
168
- for index in range(task.spider_num):
169
- threading.Thread(
170
- # name=f"xxxx_spider_task:{index}",
171
- target=spider.spider_task,
172
- args=(
173
- stop, func, item,
174
- redis_db.del_seed,
175
- redis_db.add_seed
176
- )
177
- ).start()
178
-
179
- # 启动存储器
180
- if storer:
181
- threading.Thread(
182
- # name=f"xxxx_store_task:{storer.table}",
183
- target=storer.store_task,
184
- args=(
185
- stop, last,
186
- redis_db.reset_seed,
187
- redis_db.del_seed
188
- )
189
- ).start()
190
-
191
- threading.Thread(
192
- # name="check_spider",
193
- target=check,
194
- args=(
195
- stop, last, spider,
196
- scheduler, storer,
197
- redis_db.ready_seed_length,
198
- redis_db.spider_queue_length,
199
- )
200
- ).start()
201
-
202
- return decorator
203
-
@@ -1,169 +0,0 @@
1
- import time
2
- from inspect import isgenerator
3
- # from pympler import asizeof
4
- from .. import log, ici
5
- from .. import DealModel, Queue, Seed, Setting
6
-
7
-
8
- class Scheduler:
9
-
10
- def schedule_seed(self, ready_seed_length, get_scheduler_lock, add_seed):
11
-
12
- inf_name = "SchedulerInterface"
13
- if not ici(self.__class__, inf_name):
14
- raise Exception("not have schedule function!")
15
-
16
- if self.__class__.__name__ == "Default":
17
- self.stop = True
18
- return None
19
-
20
- while not self.stop:
21
- length = ready_seed_length()
22
- if length > self.size:
23
- time.sleep(15)
24
-
25
- elif get_scheduler_lock():
26
- seeds = self.schedule()
27
- add_seed(seeds)
28
-
29
- log.info(f"close thread: schedule_seed")
30
-
31
- def schedule_task(self, stop, get_seed, ready_seed_length):
32
- time.sleep(3)
33
- while not stop.is_set():
34
-
35
- if not ready_seed_length():
36
- time.sleep(Setting.SCHEDULER_WAIT_TIME)
37
- continue
38
-
39
- if self.queue.length >= self.length:
40
- time.sleep(Setting.SCHEDULER_BLOCK_TIME)
41
- continue
42
-
43
- seeds = get_seed(self.length)
44
- self.queue.push(seeds)
45
- log.info(f"close thread: schedule_task")
46
-
47
-
48
- class Spider:
49
-
50
- def __init__(self, queue, storage, max_retries=5):
51
- self.spider_in_progress = Queue()
52
- self.max_retries = max_retries
53
- self.storage = storage
54
- self.queue = queue
55
-
56
- def spider_task(self, stop, func, item, del_seed, add_seed):
57
- while not stop.is_set():
58
-
59
- seed = self.queue.pop()
60
-
61
- if not seed:
62
- time.sleep(Setting.SPIDER_WAIT_TIME)
63
- continue
64
-
65
- elif seed._retry >= self.max_retries:
66
- del_seed(seed, spider_status=False)
67
- continue
68
-
69
- try:
70
- self.spider_in_progress.push(1, direct_insertion=True)
71
- log.info("spider seed: " + str(seed))
72
-
73
- store_queue = None
74
- store_data = list()
75
- add_seed_list = list()
76
- iterators = func(item, seed)
77
-
78
- if not isgenerator(iterators):
79
- if not self.storage:
80
- del_seed(seed, spider_status=True)
81
- continue
82
- raise TypeError(f"{func.__name__} isn't a generator")
83
-
84
- status = None
85
- for it in iterators:
86
- status = True
87
- if getattr(it, "table_name", None):
88
- if not store_queue:
89
- store_queue = it.queue()
90
- store_data.append(it.struct_data)
91
- elif isinstance(it, Seed):
92
- add_seed_list.append(it)
93
-
94
- elif isinstance(it, str) and it == DealModel.polling:
95
- self.queue.push(seed)
96
- break
97
- elif isinstance(it, str) and it == DealModel.success:
98
- del_seed(seed, spider_status=True)
99
- break
100
- elif isinstance(it, str) and it == DealModel.failure:
101
- del_seed(seed, spider_status=False)
102
- break
103
- else:
104
- raise TypeError("yield value type error!")
105
-
106
- if not status:
107
- raise Exception("yield value type error!")
108
-
109
- if store_queue and store_data:
110
- store_data.append(seed)
111
- store_queue.push(store_data)
112
-
113
- if add_seed_list:
114
- del_seed(seed, spider_status=True)
115
- add_seed(add_seed_list)
116
-
117
- except Exception as e:
118
- seed._retry += 1
119
- self.queue.push(seed)
120
- log.info(f"{str(seed)} -> {str(e)}")
121
- finally:
122
- self.spider_in_progress.pop()
123
- time.sleep(Setting.SPIDER_SLEEP_TIME)
124
- log.info(f"close thread: spider")
125
-
126
-
127
- class Storer:
128
-
129
- def store_task(self, stop, last, reset_seed, del_seed):
130
-
131
- inf_name = "StorerInterface"
132
- if not ici(self.__class__, inf_name):
133
- return None
134
-
135
- if not getattr(self, "store", None):
136
- raise Exception("not have store function!")
137
-
138
- storer_name = self.__class__.__name__ + self.table
139
-
140
- while not stop.is_set():
141
-
142
- storer_length = self.queue.length
143
- if not storer_length:
144
- time.sleep(5)
145
- continue
146
- elif not last.is_set() and storer_length < self.length:
147
- time.sleep(3)
148
- continue
149
-
150
- seeds, data_list = [], []
151
-
152
- while True:
153
- data = self.queue.pop()
154
- if not data:
155
- break
156
- if isinstance(data, Seed):
157
- seeds.append(data)
158
- if len(data_list) >= self.length:
159
- break
160
- continue
161
- data_list.append(data)
162
-
163
- if self.store(data_list):
164
- del_seed(seeds)
165
- else:
166
- reset_seed(seeds)
167
-
168
-
169
- log.info(f"close thread: {storer_name}")
File without changes
@@ -1,203 +0,0 @@
1
- import time
2
- import threading
3
-
4
- from .. import log, sqn, rtn, pim
5
- from .. import Queue, DBItem, RedisDB, Setting, OssDB
6
- from .models import Scheduler, Spider, Storer
7
-
8
-
9
- def check(stop, last, spider, scheduler, storer, ready_seed_length, spider_queue_length):
10
- log.info("run check thread after 30 seconds...")
11
- time.sleep(30)
12
- spider_info = """
13
- ------------------- check: {0} ------------------
14
- redis_spider_seed_length: {1}
15
- redis_ready_seed_length: {2}
16
- running_spider_thread_num: {3}
17
- memory_seed_queue_length: {4}
18
- storer_queue_length_info: {5}
19
- ----------------------- end -----------------------"""
20
- while True:
21
- status = "running"
22
- running_spider_thread_num = spider.spider_in_progress.length
23
- redis_ready_seed_length = ready_seed_length()
24
- redis_spider_seed_length = spider_queue_length()
25
- memory_seed_queue_length = scheduler.queue.length
26
- storer_upload_queue_length = storer.queue.length if storer else None
27
- if (
28
- scheduler.stop and
29
- not memory_seed_queue_length and
30
- not running_spider_thread_num
31
- ):
32
- if not Setting.LAUNCHER_MODEL:
33
- log.info("spider is done?")
34
- last.set()
35
- time.sleep(3)
36
- storer_queue_empty = True
37
- if storer and storer.queue.length:
38
- storer_queue_empty = False
39
- storer_upload_queue_length = storer.queue.length if storer else None
40
- if (
41
- storer_queue_empty and
42
- not redis_ready_seed_length and
43
- not redis_spider_seed_length
44
- ):
45
- if Setting.LAUNCHER_MODEL:
46
- log.info("waiting for push seeds...")
47
- status = "waiting"
48
- time.sleep(30)
49
- else:
50
- log.info("spider done!")
51
- break
52
-
53
- last.clear()
54
-
55
- log.info(spider_info.format(
56
- status,
57
- redis_spider_seed_length,
58
- redis_ready_seed_length,
59
- running_spider_thread_num,
60
- memory_seed_queue_length,
61
- storer_upload_queue_length
62
- ))
63
-
64
- time.sleep(3)
65
- stop.set()
66
-
67
-
68
- def launcher(task):
69
- """
70
- 任务启动装饰器
71
- :param task: 任务配置信息
72
- """
73
- def decorator(func):
74
- # 程序结束事件
75
- last = threading.Event()
76
- # 停止采集事件
77
- stop = threading.Event()
78
-
79
- # 初始化redis信息
80
- redis_db = RedisDB(task.project, task.task_name, task.redis_info)
81
-
82
- # new item
83
- item = type("Item", (object,), {"redis_client": redis_db.client})()
84
- if task.oss_config:
85
- item.oss = OssDB(**task.oss_config)
86
-
87
- log.info("初始化cobweb!")
88
-
89
- seed_queue = Queue()
90
-
91
- scheduler_info = task.scheduler_info or dict()
92
- # 调度器动态继承
93
- sql = scheduler_info.get("sql")
94
- table = scheduler_info.get("table")
95
- size = scheduler_info.get("size")
96
- scheduler_config = scheduler_info.get("config")
97
- scheduler_db = scheduler_info.get("db", "default")
98
- DB, class_name = pim(scheduler_db, "scheduler")
99
- # SchedulerDB, table, sql, length, size, config = task.scheduler_info
100
- SchedulerTmp = type(class_name, (Scheduler, DB), {})
101
- # 初始化调度器
102
- scheduler = SchedulerTmp(
103
- table=table, sql=sql, size=size, queue=seed_queue,
104
- length=task.scheduler_queue_length, config=scheduler_config
105
- )
106
-
107
- storer = None
108
- storer_info = task.storer_info or dict()
109
-
110
- if storer_info:
111
- storer_db = storer_info["db"]
112
- fields = storer_info["fields"]
113
- storer_table = storer_info.get("table", "console")
114
- storer_config = storer_info.get("config")
115
-
116
- StorerDB, class_name = pim(storer_db, "storer")
117
- StorerTmp = type(class_name, (Storer, StorerDB), {})
118
-
119
- db_name = class_name.lower()
120
- if not getattr(item, db_name, None):
121
- instance = type(db_name, (DBItem,), {})
122
- setattr(item, db_name, instance)
123
-
124
- storer_item_instance = getattr(item, db_name)
125
- storer_item_instance.init_item(storer_table, fields)
126
-
127
- storer_queue = sqn(db_name, storer_table)
128
- queue = getattr(storer_item_instance, storer_queue)
129
- # 初始话存储器
130
- table_name = rtn(table_name=storer_table)
131
- storer = StorerTmp(
132
- table=table_name, fields=fields,
133
- length=task.storer_queue_length,
134
- queue=queue, config=storer_config
135
- )
136
-
137
- # 初始化采集器
138
- spider = Spider(seed_queue, storer and True, task.max_retries)
139
-
140
- threading.Thread(target=redis_db.check_spider_queue, args=(stop, 0)).start()
141
- threading.Thread(target=redis_db.set_heartbeat, args=(stop,)).start()
142
-
143
- # 推送初始种子
144
- # seeds = start_seeds(task.start_seed)
145
- redis_db.add_seed(task.seeds)
146
- # 启动调度器, 调度至redis队列
147
- threading.Thread(
148
- # name="xxxx_schedule_seeds",
149
- target=scheduler.schedule_seed,
150
- args=(
151
- redis_db.ready_seed_length,
152
- redis_db.get_scheduler_lock,
153
- redis_db.add_seed
154
- )
155
- ).start()
156
-
157
- # 启动调度器, 调度任务队列
158
- threading.Thread(
159
- # name="xxxx_schedule_task",
160
- target=scheduler.schedule_task,
161
- args=(
162
- stop, redis_db.get_seed,
163
- redis_db.ready_seed_length
164
- )
165
- ).start()
166
-
167
- # 启动采集器
168
- for index in range(task.spider_num):
169
- threading.Thread(
170
- # name=f"xxxx_spider_task:{index}",
171
- target=spider.spider_task,
172
- args=(
173
- stop, func, item,
174
- redis_db.del_seed,
175
- redis_db.add_seed
176
- )
177
- ).start()
178
-
179
- # 启动存储器
180
- if storer:
181
- threading.Thread(
182
- # name=f"xxxx_store_task:{storer.table}",
183
- target=storer.store_task,
184
- args=(
185
- stop, last,
186
- redis_db.reset_seed,
187
- redis_db.del_seed
188
- )
189
- ).start()
190
-
191
- threading.Thread(
192
- # name="check_spider",
193
- target=check,
194
- args=(
195
- stop, last, spider,
196
- scheduler, storer,
197
- redis_db.ready_seed_length,
198
- redis_db.spider_queue_length,
199
- )
200
- ).start()
201
-
202
- return decorator
203
-