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

cobweb/equip/single/models.py

@@ -1,173 +0,0 @@
-import time
-from inspect import isgenerator
-# from pympler import asizeof
-from .. import log, ici
-from .. import DealModel, Queue, Seed, Setting
-
-
-class Scheduler:
-
-    def schedule_seed(self, ready_seed_length, get_scheduler_lock, add_seed):
-
-        inf_name = "SchedulerInterface"
-        if not ici(self.__class__, inf_name):
-            raise Exception("not have schedule function!")
-
-        if self.__class__.__name__ == "Default":
-            self.stop = True
-            return None
-
-        while not self.stop:
-            length = ready_seed_length()
-            if length > self.size:
-                time.sleep(15)
-
-            elif get_scheduler_lock():
-                seeds = self.schedule()
-                add_seed(seeds)
-
-        log.info(f"close thread: schedule_seed")
-
-    def schedule_task(self, stop, get_seed, ready_seed_length):
-        time.sleep(3)
-        while not stop.is_set():
-
-            if not ready_seed_length():
-                time.sleep(Setting.SCHEDULER_WAIT_TIME)
-                continue
-
-            if self.queue.length >= self.length:
-                time.sleep(Setting.SCHEDULER_BLOCK_TIME)
-                continue
-
-            seeds = get_seed(self.length)
-            self.queue.push(seeds)
-        log.info(f"close thread: schedule_task")
-
-
-class Spider:
-
-    def __init__(self, queue, storage, max_retries=5):
-        self.spider_in_progress = Queue()
-        self.max_retries = max_retries
-        self.storage = storage
-        self.queue = queue
-
-    def spider_task(self, stop, func, item, del_seed, add_seed):
-        while not stop.is_set():
-
-            seed = self.queue.pop()
-
-            if not seed:
-                time.sleep(Setting.SPIDER_WAIT_TIME)
-                continue
-
-            elif seed._retry >= self.max_retries:
-                del_seed(seed, spider_status=False)
-                continue
-
-            try:
-                self.spider_in_progress.push(1, direct_insertion=True)
-                log.info("spider seed: " + str(seed))
-
-                store_queue = None
-                store_data = list()
-                add_seed_list = list()
-                iterators = func(item, seed)
-
-                if not isgenerator(iterators):
-                    if not self.storage:
-                        del_seed(seed, spider_status=True)
-                        continue
-                    raise TypeError(f"{func.__name__} isn't a generator")
-
-                status = None
-                for it in iterators:
-                    status = True
-                    # if getattr(it, "table_name", None):
-                    #     store_queue = it.queue()
-                    #     store_queue.push(
-                    #         [seed, it.struct_data],
-                    #         direct_insertion=True
-                    #     )
-                    if getattr(it, "table_name", None):
-                        if not store_queue:
-                            store_queue = it.queue()
-                        store_data.append(it.struct_data)
-                    elif isinstance(it, Seed):
-                        add_seed_list.append(it)
-
-                    elif isinstance(it, str) and it == DealModel.polling:
-                        self.queue.push(seed)
-                        break
-                    elif isinstance(it, str) and it == DealModel.success:
-                        del_seed(seed, spider_status=True)
-                        break
-                    elif isinstance(it, str) and it == DealModel.failure:
-                        del_seed(seed, spider_status=False)
-                        break
-                    else:
-                        raise TypeError("yield value type error!")
-
-                if not status:
-                    raise Exception("yield value type error!")
-                if store_queue and store_data:
-                    store_data.append(seed)
-                    store_queue.push(store_data)
-                if add_seed_list:
-                    del_seed(seed, spider_status=True)
-                    add_seed(add_seed_list)
-
-            except Exception as e:
-                seed._retry += 1
-                self.queue.push(seed)
-                log.info(f"{str(seed)} -> {str(e)}")
-            finally:
-                self.spider_in_progress.pop()
-                time.sleep(Setting.SPIDER_SLEEP_TIME)
-        log.info(f"close thread: spider")
-
-
-class Storer:
-
-    def store_task(self, stop, last, reset_seed, del_seed):
-
-        inf_name = "StorerInterface"
-        if not ici(self.__class__, inf_name):
-            return None
-
-        if not getattr(self, "store", None):
-            raise Exception("not have store function!")
-
-        storer_name = self.__class__.__name__ + self.table
-
-        while not stop.is_set():
-
-            storer_length = self.queue.length
-            if not storer_length:
-                time.sleep(5)
-                continue
-            elif not last.is_set() and storer_length < self.length:
-                time.sleep(3)
-                continue
-
-            seeds, data_list = [], []
-
-            while True:
-                data = self.queue.pop()
-                if not data:
-                    break
-                if isinstance(data, Seed):
-                    seeds.append(data)
-                    if len(data_list) >= self.length:
-                        break
-                    continue
-                data_list.append(data)
-
-            if self.store(data_list):
-                del_seed(seeds)
-            else:
-                reset_seed(seeds)
-
-
-        log.info(f"close thread: {storer_name}")

cobweb/interface.py

@@ -1,34 +0,0 @@
-from abc import ABC, abstractmethod
-from .utils import parse_info
-
-
-class SchedulerInterface(ABC):
-
-    def __init__(self, table, sql, length, size, queue, config=None):
-        self.sql = sql
-        self.table = table
-        self.length = length
-        self.size = size
-        self.queue = queue
-        self.config = parse_info(config)
-        self.stop = False
-
-    @abstractmethod
-    def schedule(self, *args, **kwargs):
-        pass
-
-
-class StorerInterface(ABC):
-
-    def __init__(self, table, fields, length, queue, config=None):
-        self.table = table
-        self.fields = fields
-        self.length = length
-        self.queue = queue
-        self.config = parse_info(config)
-        # self.redis_db = redis_db
-
-    @abstractmethod
-    def store(self, *args, **kwargs):
-        pass
-

cobweb/log.py

@@ -1,96 +0,0 @@
-import logging
-
-
-class ColorCodes:
-    # Text Reset
-    RESET = "\033[0m"
-
-    # Regular Colors
-    RED = "\033[31m"
-    GREEN = "\033[32m"
-    YELLOW = "\033[33m"
-    BLUE = "\033[34m"
-    PURPLE = "\033[35m"
-    CYAN = "\033[36m"
-    WHITE = "\033[37m"
-
-    # Bright Colors
-    BRIGHT_RED = "\033[91m"
-    BRIGHT_GREEN = "\033[92m"
-    BRIGHT_YELLOW = "\033[93m"
-    BRIGHT_BLUE = "\033[94m"
-    BRIGHT_PURPLE = "\033[95m"
-    BRIGHT_CYAN = "\033[96m"
-    BRIGHT_WHITE = "\033[97m"
-
-    # Background Colors
-    BG_RED = "\033[41m"
-    BG_GREEN = "\033[42m"
-    BG_YELLOW = "\033[43m"
-    BG_BLUE = "\033[44m"
-    BG_PURPLE = "\033[45m"
-    BG_CYAN = "\033[46m"
-    BG_WHITE = "\033[47m"
-
-    # Bright Background Colors
-    BG_BRIGHT_RED = "\033[101m"
-    BG_BRIGHT_GREEN = "\033[102m"
-    BG_BRIGHT_YELLOW = "\033[103m"
-    BG_BRIGHT_BLUE = "\033[104m"
-    BG_BRIGHT_PURPLE = "\033[105m"
-    BG_BRIGHT_CYAN = "\033[106m"
-    BG_BRIGHT_WHITE = "\033[107m"
-
-    # Text Styles
-    BOLD = "\033[1m"
-    DIM = "\033[2m"
-    ITALIC = "\033[3m"
-    UNDERLINE = "\033[4m"
-    BLINK = "\033[5m"
-    REVERSE = "\033[7m"
-    HIDDEN = "\033[8m"
-
-
-class Log:
-    logging.basicConfig(
-        level=logging.INFO,
-        format=f'%(asctime)s %(name)s [%(filename)s:%(lineno)d %(funcName)s]'
-               f' %(levelname)s -> %(message)s'
-    )
-    log = logging.getLogger()
-
-    def set_log_name(self, name):
-        self.__class__.log = logging.getLogger(name)
-
-    @property
-    def debug(self):
-        return self.__class__.log.debug
-
-    @property
-    def info(self):
-        return self.__class__.log.info
-
-    @property
-    def warning(self):
-        return self.__class__.log.warning
-
-    @property
-    def exception(self):
-        return self.__class__.log.exception
-
-    @property
-    def error(self):
-        return self.__class__.log.error
-
-    @property
-    def critical(self):
-        return self.__class__.log.critical
-
-
-log = Log()
-# log.info("This text will be bold!")
-# print(ColorCodes.BOLD + "This text will be bold!" + ColorCodes.RESET)
-# print(ColorCodes.UNDERLINE + ColorCodes.BLUE + "This text will be underlined and blue!" + ColorCodes.RESET)
-# print(ColorCodes.BG_YELLOW + ColorCodes.RED + "This text will have a yellow background and red text!" + ColorCodes.RESET)
-# print(ColorCodes.BLINK + "This text will blink (if supported by the terminal)!" + ColorCodes.RESET)
-

cobweb/new.py

@@ -1,20 +0,0 @@
-
-
-
-
-class Launcher:
-
-    def __init__(self):
-        pass
-
-    def register(self, task_name, func):
-        pass
-
-    def launch(self, task_name):
-
-        def decorator(func):
-            # 注册爬虫程序
-            self.register(task_name, func)
-            return func
-
-        return decorator

cobweb/single/launcher.py

@@ -1,231 +0,0 @@
-import time
-import threading
-from threading import Thread
-
-from .models import Scheduler, Spider, Storer
-from cobweb import log, Queue, DBItem, RedisDB
-from cobweb.setting import MODEL, RESET_SCORE, CHECK_LOCK_TIME
-from cobweb.utils import (
-    struct_queue_name as sqn,
-    restore_table_name as rtn,
-    parse_import_model as pim,
-)
-
-
-def check(stop, last, spider, scheduler, storer, ready_seed_length, spider_queue_length):
-    log.info("run check thread after 30 seconds...")
-    time.sleep(30)
-    spider_info = """
-------------------- check: {0} ------------------
-            redis_spider_seed_length:  {1}
-            redis_ready_seed_length:   {2}
-            running_spider_thread_num: {3}
-            memory_seed_queue_length:  {4}
-            storer_queue_length_info:  {5}
------------------------  end  -----------------------"""
-    while True:
-        status = "running"
-        running_spider_thread_num = spider.spider_in_progress.length
-        redis_ready_seed_length = ready_seed_length()
-        redis_spider_seed_length = spider_queue_length()
-        memory_seed_queue_length = scheduler.queue.length
-        storer_upload_queue_length = storer.queue.length
-        if (
-                scheduler.stop and
-                # not redis_ready_seed_length and
-                not memory_seed_queue_length and
-                not running_spider_thread_num
-        ):
-            if not MODEL:
-                log.info("spider is done?")
-            last.set()
-            time.sleep(3)
-            storer_queue_empty = True
-            if storer.queue.length:
-                storer_queue_empty = False
-            storer_upload_queue_length = storer.queue.length
-            if (
-                    storer_queue_empty and
-                    not redis_ready_seed_length and
-                    not redis_spider_seed_length
-            ):
-                if MODEL:
-                    log.info("waiting for push seeds...")
-                    status = "waiting"
-                    time.sleep(30)
-                else:
-                    log.info("spider done!")
-                    break
-
-            last.clear()
-
-        log.info(spider_info.format(
-            status,
-            redis_spider_seed_length,
-            redis_ready_seed_length,
-            running_spider_thread_num,
-            memory_seed_queue_length,
-            storer_upload_queue_length
-        ))
-
-        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,
-            model=MODEL, cs_lct=CHECK_LOCK_TIME, rs_time=RESET_SCORE
-        )
-
-        # new item
-        item = type("Item", (object,), {"redis_client": redis_db.client})()
-
-        log.info("初始化cobweb!")
-
-        seed_queue = Queue()
-
-        scheduler_info = task.scheduler_info or dict()
-
-        # 调度器动态继承
-        sql = scheduler_info.get("sql")
-        table = scheduler_info.get("table")
-        size = scheduler_info.get("size")
-        scheduler_config = scheduler_info.get("config")
-        scheduler_db = scheduler_info.get("db", "default")
-        DB, class_name = pim(scheduler_db, "scheduler")
-        # SchedulerDB, table, sql, length, size, config = task.scheduler_info
-        SchedulerTmp = type(class_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)
-
-        storer = None
-
-        # 解析存储器信息
-        storer_info = task.storer_info or dict()
-
-        # for storer_info in storer_info_list:
-        if storer_info:
-            storer_db = storer_info["db"]
-            fields = storer_info["fields"]
-            storer_table = storer_info.get("table", "console")
-            storer_config = storer_info.get("config")
-
-            StorerDB, class_name = pim(storer_db, "storer")
-            StorerTmp = type(class_name, (Storer, StorerDB), {})
-
-            db_name = class_name.lower()
-            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 = sqn(db_name, storer_table)
-            queue = getattr(storer_item_instance, storer_queue)
-            # 初始话存储器
-            table_name = rtn(table_name=storer_table)
-            storer = StorerTmp(
-                table=table_name, fields=fields,
-                length=task.storer_queue_length,
-                queue=queue, config=storer_config
-            )
-
-        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
-        if storer:
-            Thread(
-                # name=f"xxxx_store_task:{storer.table}",
-                target=storer.store_task,
-                args=(
-                    stop, last,
-                    redis_db.reset_seed,
-                    redis_db.del_seed
-                )
-            ).start()
-
-        Thread(
-            # name="check_spider",
-            target=check,
-            args=(
-                stop, last, spider,
-                scheduler, storer,
-                redis_db.ready_seed_length,
-                redis_db.spider_queue_length,
-            )
-        ).start()
-
-    return decorator
-

cobweb/single/models.py

@@ -1,134 +0,0 @@
-import time
-from cobweb import log, Queue, Seed, Setting
-from cobweb.utils import issubclass_cobweb_inf
-# from pympler import asizeof
-
-
-class Scheduler:
-
-    def schedule_seed(self, ready_seed_length, get_scheduler_lock, add_seed):
-
-        inf_name = "SchedulerInterface"
-        if not issubclass_cobweb_inf(self.__class__, inf_name):
-            raise Exception("not have schedule function!")
-
-        if self.__class__.__name__ == "Default":
-            self.stop = True
-            return None
-
-        while not self.stop:
-            length = ready_seed_length()
-            if length > self.size:
-                time.sleep(15)
-
-            elif get_scheduler_lock():
-                seeds = self.schedule()
-                add_seed(seeds)
-
-        log.info(f"close thread: schedule_seed")
-
-    def schedule_task(self, stop, get_seed, ready_seed_length):
-        time.sleep(3)
-        while not stop.is_set():
-
-            if not ready_seed_length():
-                time.sleep(Setting.SCHEDULER_WAIT_TIME)
-                continue
-
-            if self.queue.length >= self.length:
-                time.sleep(Setting.SCHEDULER_BLOCK_TIME)
-                continue
-
-            seeds = get_seed(self.length)
-            self.queue.push(seeds)
-        log.info(f"close thread: schedule_task")
-
-
-class Spider:
-
-    def __init__(self, queue, max_retries=5):
-        self.spider_in_progress = Queue()
-        self.max_retries = max_retries
-        self.queue = queue
-
-    def spider_task(self, stop, func, item, del_seed):
-        while not stop.is_set():
-            seed = self.queue.pop()
-            if not seed:
-                time.sleep(Setting.SPIDER_WAIT_TIME)
-                continue
-            elif seed._retry >= self.max_retries:
-                del_seed(seed, spider_status=False)
-                continue
-            try:
-                self.spider_in_progress.push(1, direct_insertion=True)
-                # log.info("spider seed: " + str(seed))
-                ret_count = 0
-                status = None
-                for it in func(item, seed):
-                    ret_count += 1
-                    if getattr(it, "table_name", None):
-                        store_queue = it.queue()
-                        store_queue.push(
-                            [seed, it.struct_data],
-                            direct_insertion=True
-                        )
-                    elif isinstance(it, Seed):
-                        self.queue.push(it)
-                    elif any(isinstance(it, t) for t in (list, tuple)):
-                        self.queue.push([s if isinstance(s, Seed) else Seed(s) for s in it])
-                    elif isinstance(it, bool):
-                        status = it
-
-                if status:
-                    del_seed(seed, spider_status=True)
-                elif not ret_count or status is False:
-                    seed._retry += 1
-                    self.queue.push(seed)
-
-            except Exception as e:
-                seed._retry += 1
-                self.queue.push(seed)
-                log.info(f"{str(seed)} -> {str(e)}")
-            finally:
-                self.spider_in_progress.pop()
-        log.info(f"close thread: spider")
-
-
-class Storer:
-
-    def store_task(self, stop, last, reset_seed, del_seed):
-
-        inf_name = "StorerInterface"
-        if not issubclass_cobweb_inf(self.__class__, inf_name):
-            return None
-
-        if not getattr(self, "store", None):
-            raise Exception("not have store function!")
-
-        storer_name = self.__class__.__name__ + self.table
-
-        while not stop.is_set():
-
-            if last.is_set() or self.queue.length >= self.length:
-                seeds, data_list = [], []
-
-                for _ in range(self.length):
-                    items = self.queue.pop()
-                    if not items:
-                        break
-                    seed, data = items
-                    seeds.append(seed)
-                    data_list.append(data)
-
-                if data_list:
-                    if self.store(data_list):
-                        del_seed(seeds)
-                    else:
-                        reset_seed(seeds)
-                        log.info("reset seeds!")
-                    continue
-
-            time.sleep(3)
-
-        log.info(f"close thread: {storer_name}")