cobweb-launcher 0.1.8__py3-none-any.whl → 1.2.41__py3-none-any.whl

cobweb/__init__.py

@@ -1,11 +1,2 @@
-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 .db.oss_db import OssDB
-from .distributed.launcher import launcher
-from .single.launcher import launcher as single_launcher
-from . import setting
-
-
+from .launchers import LauncherAir, LauncherPro, LauncherApi
+from .constant import CrawlerModel

cobweb/base/__init__.py

@@ -0,0 +1,9 @@
+from .common_queue import Queue
+from .response import Response
+from .request import Request
+from .item import BaseItem, ConsoleItem
+from .seed import Seed
+
+from .log import logger
+from .decorators import decorator_oss_db
+

cobweb/base/basic.py

@@ -0,0 +1,297 @@
+import json
+import random
+import time
+import hashlib
+import requests
+
+
+class Params:
+
+    def __init__(self, retry=None, priority=None, version=None, status=None):
+        self.retry = retry or 0
+        self.priority = priority or 300
+        self.version = version or int(time.time())
+        self.status = status
+
+
+class Seed:
+    __SEED_PARAMS__ = [
+        "retry",
+        "priority",
+        "version",
+        "status"
+    ]
+
+    def __init__(
+            self,
+            seed,
+            sid=None,
+            retry=None,
+            priority=None,
+            version=None,
+            status=None,
+            **kwargs
+    ):
+        if any(isinstance(seed, t) for t in (str, bytes)):
+            try:
+                item = json.loads(seed)
+                self._init_seed(item)
+            except json.JSONDecodeError:
+                self.__setattr__("url", seed)
+        elif isinstance(seed, dict):
+            self._init_seed(seed)
+        else:
+            raise TypeError(Exception(
+                f"seed type error, "
+                f"must be str or dict! "
+                f"seed: {seed}"
+            ))
+
+        seed_params = {
+            "retry": retry,
+            "priority": priority,
+            "version": version,
+            "status": status,
+        }
+
+        if kwargs:
+            # for k, v in kwargs.items():
+            #     if k in seed_params.keys():
+            #         seed_params[k] = v
+            #     else:
+            #         self.__setattr__(k, v)
+            self._init_seed(kwargs)
+            seed_params.update({
+                k: v for k, v in kwargs.items()
+                if k in self.__SEED_PARAMS__
+            })
+        if sid or not getattr(self, "sid", None):
+            self._init_id(sid)
+        self.params = Params(**seed_params)
+
+    def __getattr__(self, name):
+        return None
+
+    def __setitem__(self, key, value):
+        setattr(self, key, value)
+
+    def __getitem__(self, item):
+        return getattr(self, item)
+
+    def __str__(self):
+        return json.dumps(self.__dict__, ensure_ascii=False)
+
+    def __repr__(self):
+        chars = [f"{k}={v}" for k, v in self.__dict__.items()]
+        return f'{self.__class__.__name__}({", ".join(chars)})'
+
+    def _init_seed(self, seed_info: dict):
+        for k, v in seed_info.items():
+            if k not in self.__SEED_PARAMS__:
+                self.__setattr__(k, v)
+
+    def _init_id(self, sid):
+        if not sid:
+            sid = hashlib.md5(self.to_string.encode()).hexdigest()
+        self.__setattr__("sid", sid)
+
+    @property
+    def to_dict(self) -> dict:
+        seed = self.__dict__.copy()
+        if seed.get("params"):
+            del seed["params"]
+        return seed
+
+    @property
+    def to_string(self) -> str:
+        return json.dumps(
+            self.to_dict,
+            ensure_ascii=False,
+            separators=(",", ":")
+        )
+
+    @property
+    def seed(self):
+        return self.to_string
+
+
+class Request:
+    __SEED_PARAMS__ = [
+        "retry",
+        "priority",
+        "version",
+        "status"
+    ]
+
+    __REQUEST_ATTRS__ = {
+        "params",
+        "headers",
+        "cookies",
+        "data",
+        "json",
+        "files",
+        "auth",
+        "timeout",
+        "proxies",
+        "hooks",
+        "stream",
+        "verify",
+        "cert",
+        "allow_redirects",
+    }
+
+    def __init__(
+            self,
+            # url,
+            seed,
+            random_ua=True,
+            check_status_code=True,
+            retry=None,
+            priority=None,
+            version=None,
+            status=None,
+            **kwargs
+    ):
+        # self.url = url
+        self.check_status_code = check_status_code
+        self.request_setting = {}
+
+        seed_params = {
+            "retry": retry,
+            "priority": priority,
+            "version": version,
+            "status": status,
+        }
+
+        if isinstance(seed, Seed):
+            kwargs.update(**seed.to_dict)
+        elif isinstance(seed, str):
+            kwargs.update(**json.loads(seed))
+        elif isinstance(seed, dict):
+            kwargs.update(**seed)
+
+        for k, v in kwargs.items():
+            if k in self.__class__.__REQUEST_ATTRS__:
+                self.request_setting[k] = v
+                continue
+            elif k in self.__SEED_PARAMS__:
+                seed_params[k] = v
+            self.__setattr__(k, v)
+
+        if not getattr(self, "method", None):
+            self.method = "POST" if self.request_setting.get("data") or self.request_setting.get("json") else "GET"
+
+        if random_ua:
+            self._build_header()
+
+        self.params = Params(**seed_params)
+        # self.seed = self.to_string
+
+    @property
+    def _random_ua(self) -> str:
+        v1 = random.randint(4, 15)
+        v2 = random.randint(3, 11)
+        v3 = random.randint(1, 16)
+        v4 = random.randint(533, 605)
+        v5 = random.randint(1000, 6000)
+        v6 = random.randint(10, 80)
+        user_agent = (f"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_{v1}_{v2}) AppleWebKit/{v4}.{v3} "
+                      f"(KHTML, like Gecko) Chrome/105.0.0.0 Safari/{v4}.{v3} Edg/105.0.{v5}.{v6}")
+        return user_agent
+
+    def _build_header(self) -> dict:
+        if not self.request_setting.get("headers"):
+            self.request_setting["headers"] = {"accept": "*/*", "user-agent": self._random_ua}
+        elif "user-agent" not in [key.lower() for key in self.request_setting["headers"].keys()]:
+            self.request_setting["headers"]["user-agent"] = self._random_ua
+
+    def download(self) -> requests.Response:
+        response = requests.request(self.method, self.url, **self.request_setting)
+        if self.check_status_code:
+            response.raise_for_status()
+        return response
+
+    def __getattr__(self, name):
+        return None
+
+    def __setitem__(self, key, value):
+        setattr(self, key, value)
+
+    def __getitem__(self, item):
+        return getattr(self, item)
+
+    @property
+    def to_dict(self):
+        _dict = self.__dict__.copy()
+        # _dict.pop('seed')
+        _dict.pop('params')
+        _dict.pop('check_status_code')
+        # _dict.pop('request_setting')
+        return _dict
+
+    @property
+    def to_string(self) -> str:
+        return json.dumps(
+            self.to_dict,
+            ensure_ascii=False,
+            separators=(",", ":")
+        )
+
+    @property
+    def seed(self):
+        return self.to_string
+
+
+class Response:
+
+    def __init__(
+            self,
+            seed,
+            response,
+            retry=None,
+            priority=None,
+            version=None,
+            status=None,
+            **kwargs
+    ):
+        self.seed = seed
+        self.response = response
+        seed_params = {
+            "retry": retry,
+            "priority": priority,
+            "version": version,
+            "status": status,
+        }
+        for k, v in kwargs.items():
+            if k in seed_params.keys():
+                seed_params[k] = v
+            else:
+                self.__setattr__(k, v)
+        self.params = Params(**seed_params)
+
+    @property
+    def to_dict(self):
+        _dict = self.__dict__.copy()
+        _dict.pop('seed')
+        _dict.pop('response')
+        _dict.pop('method')
+        _dict.pop('params')
+        _dict.pop('request_setting')
+        return _dict
+
+    @property
+    def to_string(self) -> str:
+        return json.dumps(
+            self.to_dict,
+            ensure_ascii=False,
+            separators=(",", ":")
+        )
+
+    def __getattr__(self, name):
+        return None
+
+    def __setitem__(self, key, value):
+        setattr(self, key, value)
+
+    def __getitem__(self, item):
+        return getattr(self, item)

cobweb/base/common_queue.py

@@ -0,0 +1,30 @@
+from collections import deque
+
+
+class Queue:
+
+    def __init__(self):
+        self._queue = deque()
+
+    @property
+    def length(self) -> int:
+        return len(self._queue)
+
+    def push(self, data, left: bool = False, direct_insertion: bool = False):
+        try:
+            if not data:
+                return None
+            if not direct_insertion and any(isinstance(data, t) for t in (list, tuple)):
+                self._queue.extendleft(data) if left else self._queue.extend(data)
+            else:
+                self._queue.appendleft(data) if left else self._queue.append(data)
+        except AttributeError:
+            pass
+
+    def pop(self, left: bool = True):
+        try:
+            return self._queue.popleft() if left else self._queue.pop()
+        except IndexError:
+            return None
+        except AttributeError:
+            return None

cobweb/base/decorators.py

@@ -0,0 +1,40 @@
+from functools import wraps
+
+
+# def check_redis_status(func):
+#     @wraps(func)
+#     def wrapper(*args, **kwargs):
+#         try:
+#             result = func(*args, **kwargs)
+#         except Exception:
+#             result = False
+#         return result
+#
+#     return wrapper
+
+
+def decorator_oss_db(exception, retries=3):
+    def decorator(func):
+        @wraps(func)
+        def wrapper(callback_func, *args, **kwargs):
+            result = None
+            for i in range(retries):
+                msg = None
+                try:
+                    return func(callback_func, *args, **kwargs)
+                except Exception as e:
+                    result = None
+                    msg = e
+                finally:
+                    if result:
+                        return result
+
+                    if i >= 2 and msg:
+                        raise exception(msg)
+
+        return wrapper
+
+    return decorator
+
+
+

cobweb/base/dotting.py

@@ -0,0 +1,35 @@
+import os
+import json
+from aliyun.log import LogClient, LogItem, PutLogsRequest
+
+
+class LoghubDot:
+
+    def __init__(self):
+        endpoint = os.getenv("DOTTING_ENDPOINT", "")
+        accessKeyId = os.getenv("DOTTING_ACCESS_KEY", "")
+        accessKey = os.getenv("DOTTING_SECRET_KEY", "")
+        self.client = LogClient(endpoint=endpoint, accessKeyId=accessKeyId, accessKey=accessKey) \
+            if endpoint and accessKeyId and accessKey else None
+
+    def build(self, topic, **kwargs):
+        if self.client:
+            temp = {}
+            log_items = []
+            log_item = LogItem()
+            for key, value in kwargs.items():
+                if not isinstance(value, str):
+                    temp[key] = json.dumps(value, ensure_ascii=False)
+                else:
+                    temp[key] = value
+            contents = sorted(temp.items())
+            log_item.set_contents(contents)
+            log_items.append(log_item)
+            request = PutLogsRequest(
+                project="databee-download-log",
+                logstore="download-logging",
+                topic=topic,
+                logitems=log_items,
+                compress=True
+            )
+            self.client.put_logs(request=request)

cobweb/base/item.py

@@ -0,0 +1,46 @@
+from .seed import Seed
+from collections import namedtuple
+
+
+class Item(type):
+
+    def __new__(cls, name, bases, dct):
+        new_class_instance = type.__new__(cls, name, bases, dct)
+        if name != "BaseItem":
+            table = getattr(new_class_instance, "__TABLE__")
+            fields = getattr(new_class_instance, "__FIELDS__")
+            new_class_instance.Data = namedtuple(table, fields)
+        return new_class_instance
+
+
+class BaseItem(metaclass=Item):
+
+    __TABLE__ = ""
+    __FIELDS__ = ""
+
+    def __init__(self, seed: Seed, **kwargs):
+        self.seed = seed
+
+        data = {}
+        for key, value in kwargs.items():
+            if key not in self.__FIELDS__:
+                self.__setattr__(key, value)
+            else:
+                data[key] = value
+
+        self.data = self.Data(**data)
+
+    @property
+    def to_dict(self):
+        return self.data._asdict()
+
+    @property
+    def table(self):
+        return self.Data.__name__
+
+
+class ConsoleItem(BaseItem):
+
+    __TABLE__ = "console"
+    __FIELDS__ = "data"
+

cobweb/{log.py → base/log.py}

@@ -52,6 +52,7 @@ class ColorCodes:
 
 
 class Log:
+    logging.getLogger('oss2.api').setLevel(logging.WARNING)
     logging.basicConfig(
         level=logging.INFO,
         format=f'%(asctime)s %(name)s [%(filename)s:%(lineno)d %(funcName)s]'
@@ -87,10 +88,7 @@ class Log:
         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)
+logger = Log()
+
+
 

cobweb/base/request.py

@@ -0,0 +1,82 @@
+import random
+import requests
+
+
+class Request:
+
+    __REQUEST_ATTRS__ = {
+        "params",
+        "headers",
+        "cookies",
+        "data",
+        "json",
+        "files",
+        "auth",
+        "timeout",
+        "proxies",
+        "hooks",
+        "stream",
+        "verify",
+        "cert",
+        "allow_redirects",
+    }
+
+    def __init__(
+            self,
+            url,
+            seed,
+            random_ua=True,
+            check_status_code=True,
+            **kwargs
+    ):
+        self.url = url
+        self.seed = seed
+        self.check_status_code = check_status_code
+        self.request_setting = {}
+
+        for k, v in kwargs.items():
+            if k in self.__class__.__REQUEST_ATTRS__:
+                self.request_setting[k] = v
+                continue
+            self.__setattr__(k, v)
+
+        if not getattr(self, "method", None):
+            self.method = "POST" if self.request_setting.get("data") or self.request_setting.get("json") else "GET"
+
+        if random_ua:
+            self._build_header()
+
+    @property
+    def _random_ua(self) -> str:
+        v1 = random.randint(4, 15)
+        v2 = random.randint(3, 11)
+        v3 = random.randint(1, 16)
+        v4 = random.randint(533, 605)
+        v5 = random.randint(1000, 6000)
+        v6 = random.randint(10, 80)
+        user_agent = (f"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_{v1}_{v2}) AppleWebKit/{v4}.{v3} "
+                      f"(KHTML, like Gecko) Chrome/105.0.0.0 Safari/{v4}.{v3} Edg/105.0.{v5}.{v6}")
+        return user_agent
+
+    def _build_header(self) -> dict:
+        if not self.request_setting.get("headers"):
+            self.request_setting["headers"] = {"accept": "*/*", "user-agent": self._random_ua}
+        elif "user-agent" not in [key.lower() for key in self.request_setting["headers"].keys()]:
+            self.request_setting["headers"]["user-agent"] = self._random_ua
+
+    def download(self) -> requests.Response:
+        response = requests.request(self.method, self.url, **self.request_setting)
+        if self.check_status_code:
+            response.raise_for_status()
+        return response
+
+    @property
+    def to_dict(self):
+        _dict = self.__dict__.copy()
+        _dict.pop('url')
+        _dict.pop('seed')
+        _dict.pop('check_status_code')
+        _dict.pop('request_setting')
+        return _dict
+
+

cobweb/base/response.py

@@ -0,0 +1,23 @@
+
+
+class Response:
+
+    def __init__(
+            self,
+            seed,
+            response,
+            **kwargs
+    ):
+        self.seed = seed
+        self.response = response
+
+        for k, v in kwargs.items():
+            self.__setattr__(k, v)
+
+    @property
+    def to_dict(self):
+        _dict = self.__dict__.copy()
+        _dict.pop('seed')
+        _dict.pop('response')
+        return _dict
+