cobweb-launcher 0.0.4__tar.gz → 0.0.6__tar.gz

{cobweb-launcher-0.0.4/cobweb_launcher.egg-info → cobweb-launcher-0.0.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cobweb-launcher
-Version: 0.0.4
+Version: 0.0.6
 Summary: spider_hole
 Home-page: https://github.com/Juannie-PP/cobweb
 Author: Juannie-PP

cobweb-launcher-0.0.6/cobweb/__init__.py

@@ -0,0 +1,8 @@
+from .bbb import Seed, Queue, DBItem
+from .task import Task
+from .log import log
+from .interface import SchedulerInterface, StorerInterface
+from .db.redis_db import RedisDB
+from .distributed.launcher import launcher
+
+

{cobweb-launcher-0.0.4/cobweb/base → cobweb-launcher-0.0.6/cobweb}/bbb.py

@@ -2,8 +2,8 @@
 import json
 import time
 import hashlib
-from log import log
-from utils import struct_queue_name
+from .log import log
+from .utils import struct_queue_name
 from collections import deque, namedtuple
 
 

cobweb-launcher-0.0.6/cobweb/db/__init__.py

@@ -0,0 +1,2 @@
+from . import oss_db, redis_db
+from . import scheduler, storer

{cobweb-launcher-0.0.4/cobweb/db/base → cobweb-launcher-0.0.6/cobweb/db}/oss_db.py

@@ -2,7 +2,7 @@ import oss2
 from typing import Union
 from oss2.models import PartInfo
 from requests import Response
-from base.log import log
+from cobweb import log
 
 
 class OssDB:

{cobweb-launcher-0.0.4/cobweb/db/base → cobweb-launcher-0.0.6/cobweb/db}/redis_db.py

@@ -1,6 +1,6 @@
 import time
 import redis
-from base.bbb import Seed
+from cobweb import Seed
 
 
 class RedisDB:
@@ -9,19 +9,9 @@ class RedisDB:
             self,
             project: str,
             task_name: str,
-            host=None,
-            port=None,
-            username=None,
-            password=None,
-            db=0
+            config: dict
     ):
-        pool = redis.ConnectionPool(
-            host=host,
-            port=port,
-            username=username,
-            password=password,
-            db=db
-        )
+        pool = redis.ConnectionPool(**config)
         self.heartbeat_key = f"{project}:{task_name}:heartbeat"  # redis type string
         self.spider_key = f"{project}:{task_name}:seed_info:spider"  # redis type zset, .format(priority)
         self.storer_key = f"{project}:{task_name}:seed_info:storer:%s"  # redis type set,

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb/db/scheduler/default.py

@@ -1,4 +1,4 @@
-from base.interface import SchedulerInterface
+from cobweb import SchedulerInterface
 
 
 class Default(SchedulerInterface):

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb/db/scheduler/textfile.py

@@ -1,6 +1,4 @@
-from base.log import log
-from base.bbb import Seed
-from base.interface import SchedulerInterface
+from cobweb import log, Seed, SchedulerInterface
 
 
 class Textfile(SchedulerInterface):

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb/db/storer/console.py

@@ -1,5 +1,4 @@
-from base.log import log
-from base.interface import StorerInterface
+from cobweb import log, StorerInterface
 
 
 class Console(StorerInterface):

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb/db/storer/loghub.py

@@ -1,13 +1,12 @@
 import json
-from base.log import log
-from base.interface import StorerInterface
 from aliyun.log import LogClient, LogItem, PutLogsRequest
+from cobweb import log, StorerInterface
 
 
 class Loghub(StorerInterface):
 
-    def __init__(self, table, fields, length, queue, config):
-        super().__init__(table, fields, length, queue, config)
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
         self.client = None
 
     def init_loghub_clint(self):

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb/db/storer/redis.py

@@ -1,5 +1,4 @@
-from base.log import log
-from base.interface import StorerInterface
+from cobweb import log, StorerInterface
 
 
 class Redis(StorerInterface):

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb/db/storer/textfile.py

@@ -1,5 +1,4 @@
-from base.log import log
-from base.interface import StorerInterface
+from cobweb import log, StorerInterface
 
 
 class Textfile(StorerInterface):

cobweb-launcher-0.0.6/cobweb/distributed/launcher.py

@@ -0,0 +1,273 @@
+import time
+import threading
+from threading import Thread
+from importlib import import_module
+
+from cobweb import log, Queue, DBItem, RedisDB, StorerInterface
+from cobweb.utils import struct_queue_name, restore_table_name
+from .models import Scheduler, Spider, Storer
+from collections import namedtuple
+
+# def start_seeds(seeds):
+#     if not seeds:
+#         return None
+#     if any(isinstance(seeds, t) for t in (list, tuple)):
+#         return [Seed(seed) for seed in seeds]
+#     elif any(isinstance(seeds, t) for t in (str, dict)):
+#         return Seed(seeds)
+
+
+# def parse_storer_info(storer_info):
+#     storer_data = {}
+#     storer_info_list = []
+#     if storer_info.__class__.__name__ == 'StorerInfo':
+#         storer_info_list.append(storer_info)
+#     elif any(isinstance(storer_info, t) for t in (list, tuple)):
+#         storer_info_list = storer_info
+#     for info in storer_info_list:
+#         db_name = info.DB.__name__
+#         storer_data.setdefault(db_name, {"StorerDB": info.DB, "db_args_list": []})
+#         storer_data[db_name]["db_args_list"].append(info[1:])
+#     return storer_data
+
+def get_scheduler_db(db):
+    if isinstance(db, str):
+        if "." in db:
+            model_path = db.split(".")
+            model = import_module(db)
+            obj = getattr(model, db)
+        else:
+            model = import_module(f"cobweb.db.scheduler.{db.lower()}")
+            obj = getattr(model, db.capitalize())
+        return obj
+        # if db.lower() in dir(StorerDB):
+        #     return getattr(StorerDB, db)
+        # else:
+        #     pass
+    elif issubclass(db, StorerInterface):
+        return db
+    raise TypeError()
+
+
+def get_storer_db(db):
+    if isinstance(db, str):
+        if "." in db:
+            model_path = db.split(".")
+            model = import_module(db)
+            obj = getattr(model, db)
+        else:
+            model = import_module(f"cobweb.db.storer.{db.lower()}")
+            obj = getattr(model, db.capitalize())
+        return obj, db.lower()
+        # if db.lower() in dir(StorerDB):
+        #     return getattr(StorerDB, db)
+        # else:
+        #     pass
+    elif issubclass(db, StorerInterface):
+        return db, db.__name__.lower()
+    raise TypeError()
+
+
+def check(stop, last, spider, scheduler, storer_list, ready_seed_length, spider_queue_length):
+    time.sleep(5)
+    while True:
+        if (
+                scheduler.stop and
+                not ready_seed_length() and
+                not scheduler.queue.length and
+                not spider.spider_in_progress.length
+        ):
+            log.info("spider is done?")
+            last.set()
+            time.sleep(5)
+            storer_queue_empty = True
+            for storer in storer_list:
+                if storer.queue.length:
+                    storer_queue_empty = False
+                    break
+            if storer_queue_empty and not spider_queue_length():
+                log.info("spider done!")
+                break
+        last.clear()
+        time.sleep(3)
+    stop.set()
+
+
+def launcher(task):
+    """
+    任务启动装饰器
+    :param task: 任务配置信息
+    """
+    def decorator(func):
+        """
+        Item:
+            Textfile()
+            Loghub()
+            Console()
+        e.g.
+        task.fields = "a,b"
+        func(item, seed)
+            a = "a"
+            b = "b"
+            data = {"a": "a", "b": "b"}
+            yield item.Loghub(**data)
+            yield item.Loghub(a=a, b=b)
+        """
+        storer_list = []
+
+        # 程序结束事件
+        last = threading.Event()
+        # 停止采集事件
+        stop = threading.Event()
+
+        # 初始化redis信息
+        redis_db = RedisDB(task.project, task.task_name, task.redis_info)
+
+        log.info("初始化cobweb!")
+
+        seed_queue = Queue()
+
+        if task.scheduler_info is None:
+            task.scheduler_info = dict()
+
+        # 调度器动态继承
+        sql = task.scheduler_info.get("sql")
+        table = task.scheduler_info.get("table")
+        size = task.scheduler_info.get("size")
+        scheduler_config = task.scheduler_info.get("config")
+        scheduler_db = task.scheduler_info.get("db", "default")
+        DB = get_scheduler_db(scheduler_db)
+        # SchedulerDB, table, sql, length, size, config = task.scheduler_info
+        SchedulerTmp = type(DB.__name__, (Scheduler, DB), {})
+
+        # 初始化调度器
+        scheduler = SchedulerTmp(
+            table=table, sql=sql, size=size, queue=seed_queue,
+            length=task.scheduler_queue_length, config=scheduler_config
+        )
+
+        # 初始化采集器
+        spider = Spider(seed_queue, task.max_retries)
+
+        # 解析存储器信息
+        storer_info_list = task.storer_info
+        if not isinstance(storer_info_list, list):
+            storer_info_list = [storer_info_list]
+
+        # new item
+        item = type("Item", (object,), {"redis_client": redis_db})()
+
+        for storer_info in storer_info_list:
+            storer_db = storer_info["db"]
+            fields = storer_info["fields"]
+            storer_table = storer_info.get("table", "console")
+            storer_config = storer_info.get("config")
+
+            StorerDB, db_name = get_storer_db(storer_db)
+            StorerTmp = type(StorerDB.__name__, (Storer, StorerDB), {})
+
+            if not getattr(item, db_name, None):
+                instance = type(db_name, (DBItem,), {})
+                setattr(item, db_name, instance)
+
+            storer_item_instance = getattr(item, db_name)
+            storer_item_instance.init_item(storer_table, fields)
+
+            storer_queue = struct_queue_name(db_name, storer_table)
+            queue = getattr(storer_item_instance, storer_queue)
+            # 初始话存储器
+            table_name = restore_table_name(table_name=storer_table)
+            storer = StorerTmp(
+                table=table_name, fields=fields,
+                length=task.storer_queue_length,
+                queue=queue, config=storer_config
+            )
+            storer_list.append(storer)
+
+        # for db_name in storer_data.keys():
+        #     # 存储器动态继承
+        #     StorerDB = storer_data[db_name]["StorerDB"]
+        #     StorerTmp = type(StorerDB.__name__, (Storer, StorerDB), {})
+        #     db_args_list = storer_data[db_name]["db_args_list"]
+        #     for storer_db_args in db_args_list:
+        #         table, fields, length, config = storer_db_args
+        #         if not getattr(item, db_name, None):
+        #             instance = type(db_name, (DBItem,), {})
+        #             setattr(item, db_name, instance)
+        #         # 创建存储xxx, 创建存储队列
+        #         storer_item_instance = getattr(item, db_name)
+        #         storer_item_instance.init_item(table, fields)
+        #         #
+        #         storer_queue = struct_queue_name(db_name, table)
+        #         queue = getattr(storer_item_instance, storer_queue)
+        #         # 初始话存储器
+        #         table_name = restore_table_name(table_name=table)
+        #         storer = StorerTmp(table_name, fields, length, queue, config)
+        #         storer_list.append(storer)
+
+        Thread(target=redis_db.check_spider_queue, args=(stop, len(storer_list))).start()
+        Thread(target=redis_db.set_heartbeat, args=(stop,)).start()
+
+        # 推送初始种子
+        # seeds = start_seeds(task.start_seed)
+        redis_db.add_seed(task.seeds)
+        # 启动调度器, 调度至redis队列
+        Thread(
+            # name="xxxx_schedule_seeds",
+            target=scheduler.schedule_seed,
+            args=(
+                redis_db.ready_seed_length,
+                redis_db.get_scheduler_lock,
+                redis_db.add_seed
+            )
+        ).start()
+
+        # 启动调度器, 调度任务队列
+        Thread(
+            # name="xxxx_schedule_task",
+            target=scheduler.schedule_task,
+            args=(
+                stop, redis_db.get_seed,
+                redis_db.ready_seed_length
+            )
+        ).start()
+
+        # 启动采集器
+        for index in range(task.spider_num):
+            Thread(
+                # name=f"xxxx_spider_task:{index}",
+                target=spider.spider_task,
+                args=(
+                    stop, func, item,
+                    redis_db.del_seed
+                )
+            ).start()
+
+        # 启动存储器
+        for storer in storer_list:
+            Thread(
+                # name=f"xxxx_store_task:{storer.table}",
+                target=storer.store_task,
+                args=(
+                    stop, last,
+                    redis_db.reset_seed,
+                    redis_db.set_storer
+                )
+            ).start()
+
+        Thread(
+            # name="check_spider",
+            target=check,
+            args=(
+                stop, last, spider,
+                scheduler, storer_list,
+                redis_db.ready_seed_length,
+                redis_db.spider_queue_length,
+            )
+        ).start()
+
+    return decorator
+
+
+# model = get_storer_db("console")
+# print()

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb/distributed/models.py

@@ -1,8 +1,6 @@
 import time
 from hashlib import md5
-from base.log import log
-from base.bbb import Queue, Seed
-from base.interface import SchedulerInterface, StorerInterface
+from cobweb import log, Queue, Seed, StorerInterface, SchedulerInterface
 # from pympler import asizeof
 
 

{cobweb-launcher-0.0.4/cobweb/base → cobweb-launcher-0.0.6/cobweb}/interface.py

@@ -1,15 +1,5 @@
-import json
 from abc import ABC, abstractmethod
-
-
-def parse_config(config):
-    if not config:
-        return None
-    if isinstance(config, str):
-        return json.loads(config)
-    if isinstance(config, dict):
-        return config
-    raise TypeError("config type is not in [string, dict]!")
+from .utils import parse_info
 
 
 class SchedulerInterface(ABC):
@@ -20,7 +10,7 @@ class SchedulerInterface(ABC):
         self.length = length
         self.size = size
         self.queue = queue
-        self.config = parse_config(config)
+        self.config = parse_info(config)
         self.stop = False
 
     @abstractmethod
@@ -35,7 +25,7 @@ class StorerInterface(ABC):
         self.fields = fields
         self.length = length
         self.queue = queue
-        self.config = parse_config(config)
+        self.config = parse_info(config)
         # self.redis_db = redis_db
 
     @abstractmethod

cobweb-launcher-0.0.6/cobweb/task.py

@@ -0,0 +1,44 @@
+from .utils import parse_info, struct_start_seeds
+
+
+class Task:
+
+    def __init__(
+            self,
+            seeds=None,
+            project=None,
+            task_name=None,
+            redis_info=None,
+            storer_info=None,
+            scheduler_info=None,
+            spider_num=None,
+            max_retries=None,
+            storer_queue_length=None,
+            scheduler_queue_length=None,
+    ):
+        """
+
+        :param seeds:
+        :param project:
+        :param task_name:
+        :param redis_info:
+        :param storer_info:
+        :param scheduler_info: dict(DB="", table="", size="", config="")
+        :param spider_num:
+        :param max_retries:
+        :param storer_queue_length:
+        :param scheduler_queue_length:
+        """
+        self.seeds = struct_start_seeds(seeds)
+        self.project = project or "test"
+        self.task_name = task_name or "spider"
+
+        self.redis_info = parse_info(redis_info)
+        self.storer_info = parse_info(storer_info)
+        self.scheduler_info = parse_info(scheduler_info)
+
+        self.spider_num = spider_num or 1
+        self.max_retries = max_retries or 5
+        self.storer_queue_length = storer_queue_length or 100
+        self.scheduler_queue_length = scheduler_queue_length or 100
+

cobweb-launcher-0.0.6/cobweb/utils.py

@@ -0,0 +1,85 @@
+import json
+import sys
+from typing import Iterable
+
+# from cobweb import Seed
+
+
+def struct_table_name(table_name):
+    return table_name.replace(".", "__p__").replace(":", "__c__")
+
+
+def restore_table_name(table_name):
+    return table_name.replace("__p__", ".").replace("__c__", ":")
+
+
+def struct_queue_name(db_name, table_name):
+    return sys.intern(f"__{db_name}_{table_name}_queue__")
+
+
+# class StorerDB:
+#
+#     @staticmethod
+#     def console(self):
+#         from db.storer.console import Console
+#         table = struct_table_name(table)
+#         return StorerInfo(DB=Console, table=table, length=length, config=None)
+#
+#     @staticmethod
+#     def textfile(table, length=200):
+#         from db.storer.textfile import Textfile
+#         table = struct_table_name(table)
+#         return StorerInfo(DB=Textfile, table=table, length=length, config=None)
+#
+#     @staticmethod
+#     def loghub(table, length=200, config=None):
+#         from db.storer.loghub import Loghub
+#         table = struct_table_name(table)
+#         return StorerInfo(DB=Loghub, table=table, length=length, config=config)
+
+
+def parse_info(info):
+    if not info:
+        return info
+
+    if isinstance(info, dict):
+        return info
+
+    if isinstance(info, str):
+        return json.loads(info)
+
+    if isinstance(info, Iterable):
+        result = list()
+        for ii in info:
+            if isinstance(ii, str):
+                result.append(json.loads(ii))
+            elif isinstance(ii, dict):
+                result.append(ii)
+            else:
+                raise TypeError("must be in [str, dict]")
+
+        return result
+
+
+def struct_start_seeds(seeds):
+    from .bbb import Seed
+    if not seeds:
+        return None
+    if any(isinstance(seeds, t) for t in (list, tuple)):
+        return [Seed(seed) for seed in seeds]
+    elif any(isinstance(seeds, t) for t in (str, dict)):
+        return Seed(seeds)
+
+
+# def get_storer_db(db):
+#
+#     if isinstance(db, str):
+#         model = import_module(f" db.storer.{db.lower()}")
+#
+#         # if db.lower() in dir(StorerDB):
+#         #     return getattr(StorerDB, db)
+#         # else:
+#         #     pass
+
+
+

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6/cobweb_launcher.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cobweb-launcher
-Version: 0.0.4
+Version: 0.0.6
 Summary: spider_hole
 Home-page: https://github.com/Juannie-PP/cobweb
 Author: Juannie-PP

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/cobweb_launcher.egg-info/SOURCES.txt

@@ -2,23 +2,14 @@ LICENSE
 README.md
 setup.py
 cobweb/__init__.py
-cobweb/base/__init__.py
-cobweb/base/bbb.py
-cobweb/base/config.py
-cobweb/base/decorators.py
-cobweb/base/hash_table.py
-cobweb/base/interface.py
-cobweb/base/log.py
-cobweb/base/queue_tmp.py
-cobweb/base/request.py
-cobweb/base/task.py
-cobweb/base/utils.py
+cobweb/bbb.py
+cobweb/interface.py
+cobweb/log.py
+cobweb/task.py
+cobweb/utils.py
 cobweb/db/__init__.py
-cobweb/db/base/__init__.py
-cobweb/db/base/client_db.py
-cobweb/db/base/oss_db.py
-cobweb/db/base/redis_db.py
-cobweb/db/base/redis_dbv3.py
+cobweb/db/oss_db.py
+cobweb/db/redis_db.py
 cobweb/db/scheduler/__init__.py
 cobweb/db/scheduler/default.py
 cobweb/db/scheduler/textfile.py

{cobweb-launcher-0.0.4 → cobweb-launcher-0.0.6}/setup.py

@@ -5,7 +5,7 @@ with open("README.md", "r", encoding="utf-8") as fh:
 
 setup(
     name="cobweb-launcher",
-    version="0.0.4",
+    version="0.0.6",
     packages=find_packages(),
     url="https://github.com/Juannie-PP/cobweb",
     license="MIT",

cobweb-launcher-0.0.4/cobweb/__init__.py

@@ -1,10 +0,0 @@
-from .base.config import StorerInfo, SchedulerInfo, RedisInfo, SchedulerDB, StorerDB
-from .base.interface import StorerInterface, SchedulerInterface
-from .distributed.launcher import launcher
-from .distributed import models
-from .base.task import Task
-from .base.bbb import Seed
-from .db.base import *
-from .db.storer import *
-from .db.scheduler import *
-