aio-scrapy 2.0.8__py3-none-any.whl → 2.0.10__py3-none-any.whl

{aio_scrapy-2.0.8.dist-info → aio_scrapy-2.0.10.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: aio-scrapy
-Version: 2.0.8
+Version: 2.0.10
 Summary: A high-level Web Crawling and Web Scraping framework based on Asyncio
 Home-page: https://github.com/conlin-huang/aio-scrapy.git
 Author: conlin

{aio_scrapy-2.0.8.dist-info → aio_scrapy-2.0.10.dist-info}/RECORD

@@ -1,4 +1,4 @@
-aioscrapy/VERSION,sha256=FHhKxnUnBTC0gfINdE8plREGVAupdclqCULpdz5FL3U,5
+aioscrapy/VERSION,sha256=bkksF7-FeZMTR8EfltCUKJZNQaHaQkySSXYbwvc2qdw,6
 aioscrapy/__init__.py,sha256=esJeH66Mz9WV7XbotvZEjNn49jc589YZ_L2DKoD0JvA,858
 aioscrapy/__main__.py,sha256=rvTdJ0cQwbi29aucPj3jJRpccx5SBzvRcV7qvxvX2NQ,80
 aioscrapy/cmdline.py,sha256=1qhNg2Edl-Obmf2re2K4V8pJG7ubGfZZCzcHdKtdE_s,5159
@@ -21,10 +21,10 @@ aioscrapy/commands/settings.py,sha256=sc0rwwfBQNySKX8uV3iJqv3i7SelFwNcrlHYxDupKO
 aioscrapy/commands/startproject.py,sha256=Rcc7JkN75Jp2t2aZIxBzPsWbLXChNAUSByDhcW_6Ig8,4001
 aioscrapy/commands/version.py,sha256=yqqTMlZkkiQhtbU9w_IqUWLMOAjqYlv24friEkPRQYM,485
 aioscrapy/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-aioscrapy/core/engine.py,sha256=YJC597TZwhX81qjPHBH2mkRi4oVGNImzSz58_sFiXPA,10925
+aioscrapy/core/engine.py,sha256=zW3GPigPqyrWJ_Jk7SUxD0ueV1HTuxUvweFUU4WFG-0,10926
 aioscrapy/core/scheduler.py,sha256=YCRw9j79ZOL8bijDa3IdRaw0YlMTrwXuJGzaApkN7lc,5737
 aioscrapy/core/scraper.py,sha256=M_bcizLUzWuECe7sIIZ_HJLNrPzL7dX2o-tN5nvFnCs,10304
-aioscrapy/core/downloader/__init__.py,sha256=XRpiDJg9Yc47SoFy-48HxFyuV3sZkjE30sK3DA80YJQ,9465
+aioscrapy/core/downloader/__init__.py,sha256=22TC0z49BX3YvDUPl6DKMrOonECpY5tjaWJGGEV7RbU,9574
 aioscrapy/core/downloader/handlers/__init__.py,sha256=CriaX2Cp4jUqzDDGZDB7HiIEgUWt2pnYVho6HMV6sJ0,3198
 aioscrapy/core/downloader/handlers/aiohttp.py,sha256=dFVVeGgJ1WZcE1zI4fQOZIzmrkC6l1WZcYstHmB3qYg,3942
 aioscrapy/core/downloader/handlers/httpx.py,sha256=-DfjYgfrjxMhaMpTgEOFlQRONasCXV0g6UgH3WmWcfs,3041
@@ -40,9 +40,9 @@ aioscrapy/db/aiomysql.py,sha256=-xCLfeH7RzvghY1jqREAb_Qnz9q_dVjxoHGfz7sCqbU,3799
 aioscrapy/db/aiopg.py,sha256=WG4s_2X0b8LQHbZpoIrwZeuGHNolKj-SvmvAZQlCk00,3213
 aioscrapy/db/aiorabbitmq.py,sha256=tNKl4Kx7KM7H_lOj8xfeA0uD8PuBTVzySApTEn5TyAE,5583
 aioscrapy/db/aioredis.py,sha256=UOoTRTQUvghnq29bVL8v1HvksMXYOzHaS8Btgbpn0bY,2966
-aioscrapy/dupefilters/__init__.py,sha256=bxOMWqqXuOZS8r5OXqK9JSB8Sf-LNDqfKtllTCi-Gls,725
-aioscrapy/dupefilters/disk.py,sha256=z4PcayzR1zMxXkCR1pJAJvzcvzsfpHM64ibYQltnOMo,2178
-aioscrapy/dupefilters/redis.py,sha256=rYmKnjoSZlOmAABkdrFa9HUSy3F6VBre_7z6u8YpMCs,5160
+aioscrapy/dupefilters/__init__.py,sha256=17s6Hyr_lWDFPto6wLEvRfT2TbGU2RIssTDuChzrDNA,1498
+aioscrapy/dupefilters/disk.py,sha256=EMgxeC2a6aYCGKgp4QOs5xwHp33LUsOZ8pliKBTFx1c,1551
+aioscrapy/dupefilters/redis.py,sha256=cUuM68dEM1_ki2eOzZ6pAvmLZlAP_tC4lx73Ufmg_Bs,4812
 aioscrapy/http/__init__.py,sha256=yeQTT5W1iwr6dKznTS5d9vnx2hsB47i9roPM57wQp_0,597
 aioscrapy/http/headers.py,sha256=H-RJ6KqOsFFFAXORfvoyz3V-ud0I8TAj5Jt5fAACcLc,1573
 aioscrapy/http/request/__init__.py,sha256=PFoFU3ncTN-gj6Rx01rjVa_744Qfv3EH29mooW6JX9U,7121
@@ -58,7 +58,7 @@ aioscrapy/libs/downloader/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJW
 aioscrapy/libs/downloader/defaultheaders.py,sha256=tg_ULA0Y-41bZKG607mowFJQGVfnZ45LdR044DsjA_A,563
 aioscrapy/libs/downloader/downloadtimeout.py,sha256=hNh3OEj7rC0ceQrv_yrhR5lb5AvfxJ6cspj3qsQWj4o,704
 aioscrapy/libs/downloader/ja3fingerprint.py,sha256=DgTw74GXC_Bp94eD_bwoG6A_DphUHTt7bH4glBNXyV8,1058
-aioscrapy/libs/downloader/retry.py,sha256=onItmt4s_diH3v1iMJiTitMn12DflwerSK2fIgJyBsY,5082
+aioscrapy/libs/downloader/retry.py,sha256=eaMig7JpSyr6QQBD6FNYpcttuGK811Dm4tJGTUIi3q8,5191
 aioscrapy/libs/downloader/stats.py,sha256=FlkS8Zm4j3SBjHb6caXwq08HvvZ37VKORGCAjlA2U38,1376
 aioscrapy/libs/downloader/useragent.py,sha256=E5x5dk9AxsSCGDDICJlTXwWXRkqAibWgesqG0VhAG8M,743
 aioscrapy/libs/extensions/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -110,7 +110,7 @@ aioscrapy/utils/curl.py,sha256=I8eZWFNgvyUiJ2YS9-s3HltGNVG8XMMU0HPhlMxuxdA,3295
 aioscrapy/utils/decorators.py,sha256=gMQArNxF9QQc1bENA0IqDchAjqmfWvHGKOyUdjXdg6A,794
 aioscrapy/utils/deprecate.py,sha256=STy55Q8kZI8q8CQUfxK4QQVu1Rs4En3rlhzWc7p7T00,5467
 aioscrapy/utils/httpobj.py,sha256=ytec7IZzsQY_GwR___051hdbOWs1ZM6S57HwcNiu2es,708
-aioscrapy/utils/log.py,sha256=U4BqJQ3p-s_lYbqgln_9FHeUegKF7T1MTrN1KTtn92o,1592
+aioscrapy/utils/log.py,sha256=NRDivw8w21J77qEUeqqLdC4sgdIKaj2UAP6lDvWGotM,1697
 aioscrapy/utils/misc.py,sha256=9NOssEl7CP_c6R9skxyXwmz4bd-nZ_gkw6F0EybeLTQ,3509
 aioscrapy/utils/ossignal.py,sha256=jAsCIKu17KV45-9dZwEkFJHF31Y13KP_zxY0x49j1jo,896
 aioscrapy/utils/project.py,sha256=cT98HaR5JaNmm-Y1UzSuzXj6B5S7GlmMshUfMhjpjJY,2905
@@ -124,9 +124,9 @@ aioscrapy/utils/template.py,sha256=HR97X4lpv2WuqhuPfzTgaBN66fYnzHVpP6zQ5IoTwcI,8
 aioscrapy/utils/tools.py,sha256=WJowViZB8XEs2CFqjVvbqXK3H5Uvf4BgWgBD_RcHMaM,2319
 aioscrapy/utils/trackref.py,sha256=0nIpelT1d5WYxALl8SGA8vHNYsh-jS0Z2lwVEAhwx8E,2019
 aioscrapy/utils/url.py,sha256=8W8tAhU7lgfPOfzKp3ejJGEcLj1i_PnA_53Jv5LpxiY,5464
-aio_scrapy-2.0.8.dist-info/LICENSE,sha256=L-UoAEM3fQSjKA7FVWxQM7gwSCbeue6gZRAnpRS_UCo,1088
-aio_scrapy-2.0.8.dist-info/METADATA,sha256=AzuL_eshj6vpMA1gAPpc18nTm96ptt5Las7oJ1s3U38,6384
-aio_scrapy-2.0.8.dist-info/WHEEL,sha256=oiQVh_5PnQM0E3gPdiz09WCNmwiHDMaGer_elqB3coM,92
-aio_scrapy-2.0.8.dist-info/entry_points.txt,sha256=WWhoVHZvqhW8a5uFg97K0EP_GjG3uuCIFLkyqDICgaw,56
-aio_scrapy-2.0.8.dist-info/top_level.txt,sha256=8l08KyMt22wfX_5BmhrGH0PgwZdzZIPq-hBUa1GNir4,10
-aio_scrapy-2.0.8.dist-info/RECORD,,
+aio_scrapy-2.0.10.dist-info/LICENSE,sha256=L-UoAEM3fQSjKA7FVWxQM7gwSCbeue6gZRAnpRS_UCo,1088
+aio_scrapy-2.0.10.dist-info/METADATA,sha256=qMfSjJmZpj8xAaoGdjEC-oNULa4wYcWFwgJJm8wBQ3U,6385
+aio_scrapy-2.0.10.dist-info/WHEEL,sha256=oiQVh_5PnQM0E3gPdiz09WCNmwiHDMaGer_elqB3coM,92
+aio_scrapy-2.0.10.dist-info/entry_points.txt,sha256=WWhoVHZvqhW8a5uFg97K0EP_GjG3uuCIFLkyqDICgaw,56
+aio_scrapy-2.0.10.dist-info/top_level.txt,sha256=8l08KyMt22wfX_5BmhrGH0PgwZdzZIPq-hBUa1GNir4,10
+aio_scrapy-2.0.10.dist-info/RECORD,,

aioscrapy/VERSION

@@ -1 +1 @@
-2.0.8
+2.0.10

aioscrapy/core/downloader/__init__.py

@@ -126,6 +126,7 @@ class Downloader(BaseDownloader):
         self.dupefilter = dupefilter
 
         self.total_concurrency: int = self.settings.getint('CONCURRENT_REQUESTS')
+        self.get_requests_count: int = self.settings.getint('GET_REQUESTS_COUNT') or self.total_concurrency
         self.domain_concurrency: int = self.settings.getint('CONCURRENT_REQUESTS_PER_DOMAIN')
         self.ip_concurrency: int = self.settings.getint('CONCURRENT_REQUESTS_PER_IP')
         self.randomize_delay: bool = self.settings.getbool('RANDOMIZE_DOWNLOAD_DELAY')

aioscrapy/core/engine.py

@@ -138,7 +138,7 @@ class ExecutionEngine(object):
         while self.unlock and not self._needs_backout() and self.unlock:
             self.unlock = False
             try:
-                async for request in self.scheduler.next_request(self.downloader.total_concurrency):
+                async for request in self.scheduler.next_request(self.downloader.get_requests_count):
                     if request:
                         self.slot.add_request(request)
                         await self.downloader.fetch(request)

aioscrapy/dupefilters/__init__.py

@@ -1,6 +1,7 @@
 from abc import ABCMeta, abstractmethod
 
 from aioscrapy import Request, Spider
+from aioscrapy.utils.log import logger
 
 
 class DupeFilterBase(metaclass=ABCMeta):
@@ -19,6 +20,20 @@ class DupeFilterBase(metaclass=ABCMeta):
     async def close(self, reason: str = '') -> None:
         """ Delete data on close """
 
-    @abstractmethod
-    def log(self, request: Request, spider: Spider) -> None:
-        """ Logs given request """
+    def log(self, request: Request, spider: Spider):
+        if self.info:
+            logger.info("Filtered duplicate request: %(request)s" % {
+                'request': request.meta.get('dupefilter_msg') or request
+            })
+        elif self.debug:
+            logger.debug("Filtered duplicate request: %(request)s" % {
+                'request': request.meta.get('dupefilter_msg') or request
+            })
+        elif self.logdupes:
+            msg = ("Filtered duplicate request: %(request)s"
+                   " - no more duplicates will be shown"
+                   " (see DUPEFILTER_DEBUG to show all duplicates)")
+            logger.debug(msg % {'request': request.meta.get('dupefilter_msg') or request})
+            self.logdupes = False
+
+        spider.crawler.stats.inc_value('dupefilter/filtered', spider=spider)

aioscrapy/dupefilters/disk.py

@@ -1,20 +1,19 @@
 import os
 from typing import Optional, Set
 
-from aioscrapy import Request, Spider
+from aioscrapy import Request
 from aioscrapy.dupefilters import DupeFilterBase
-from aioscrapy.utils.log import logger
-from aioscrapy.utils.request import referer_str
 
 
 class DiskRFPDupeFilter(DupeFilterBase):
     """Request Fingerprint duplicates filter built with Disk storage"""
 
-    def __init__(self, path: Optional[str] = None, debug: bool = False):
+    def __init__(self, path: Optional[str] = None, debug: bool = False, info: bool = False):
         self.file: Optional["File object"] = None
         self.debug = debug
         self.fingerprints: Set = set()
         self.logdupes: bool = True
+        self.info: bool = info
         if path:
             self.file = open(os.path.join(path, 'requests.seen'), 'a+')
             self.file.seek(0)
@@ -23,10 +22,11 @@ class DiskRFPDupeFilter(DupeFilterBase):
     @classmethod
     def from_crawler(cls, crawler: "aioscrapy.crawler.Crawler"):
         debug = crawler.settings.getbool('DUPEFILTER_DEBUG')
+        info = crawler.settings.getbool('DUPEFILTER_INFO')
         path = crawler.settings.get('JOBDIR', './job_dir')
         if path and not os.path.exists(path):
             os.makedirs(path)
-        return cls(path, debug)
+        return cls(path, debug, info)
 
     async def request_seen(self, request: Request) -> bool:
         if request.fingerprint in self.fingerprints:
@@ -40,19 +40,5 @@ class DiskRFPDupeFilter(DupeFilterBase):
         if self.file:
             self.file.close()
 
-    def log(self, request: Request, spider: Spider):
-        if self.debug:
-            logger.debug("Filtered duplicate request: %(request)s (referer: %(referer)s)" % {
-                'request': request, 'referer': referer_str(request)
-            })
-        elif self.logdupes:
-            msg = ("Filtered duplicate request: %(request)s"
-                   " - no more duplicates will be shown"
-                   " (see DUPEFILTER_DEBUG to show all duplicates)")
-            logger.debug(msg % {'request': request})
-            self.logdupes = False
-
-        spider.crawler.stats.inc_value('dupefilter/filtered', spider=spider)
-
 
 RFPDupeFilter = DiskRFPDupeFilter

aioscrapy/dupefilters/redis.py

@@ -2,8 +2,6 @@ from aioscrapy import Request
 from aioscrapy.db import db_manager
 from aioscrapy.dupefilters import DupeFilterBase
 
-from aioscrapy.utils.log import logger
-
 
 class RedisRFPDupeFilter(DupeFilterBase):
     """Request Fingerprint duplicates filter built with Set of Redis"""
@@ -13,13 +11,15 @@ class RedisRFPDupeFilter(DupeFilterBase):
             server: "redis.asyncio.Redis",
             key: str,
             debug: bool = False,
-            keep_on_close: bool = True
+            keep_on_close: bool = True,
+            info: bool = False,
     ):
         self.server = server
         self.key = key
         self.debug = debug
         self.keep_on_close = keep_on_close
         self.logdupes: bool = True
+        self.info: bool = info
 
     @classmethod
     def from_crawler(cls, crawler: "aioscrapy.crawler.Crawler"):
@@ -28,7 +28,8 @@ class RedisRFPDupeFilter(DupeFilterBase):
         keep_on_close = crawler.settings.getbool("KEEP_DUPEFILTER_DATA_ON_CLOSE", True)
         key = dupefilter_key % {'spider': crawler.spider.name}
         debug = crawler.settings.getbool('DUPEFILTER_DEBUG', False)
-        instance = cls(server, key=key, debug=debug, keep_on_close=keep_on_close)
+        info = crawler.settings.getbool('DUPEFILTER_INFO', False)
+        instance = cls(server, key=key, debug=debug, keep_on_close=keep_on_close, info=info)
         return instance
 
     async def request_seen(self, request: Request):
@@ -41,17 +42,6 @@ class RedisRFPDupeFilter(DupeFilterBase):
     async def clear(self):
         await self.server.delete(self.key)
 
-    def log(self, request, spider):
-        if self.debug:
-            logger.debug("Filtered duplicate request: %(request)s" % {'request': request})
-        elif self.logdupes:
-            msg = ("Filtered duplicate request %(request)s"
-                   " - no more duplicates will be shown"
-                   " (see DUPEFILTER_DEBUG to show all duplicates)")
-            logger.debug(msg % {'request': request})
-            self.logdupes = False
-        spider.crawler.stats.inc_value('dupefilter/filtered', spider=spider)
-
 
 class HashMap(object):
     def __init__(self, m, seed):
@@ -112,8 +102,8 @@ class BloomFilter(object):
 class RedisBloomDupeFilter(RedisRFPDupeFilter):
     """Bloom filter built with the bitis bitmap of redis"""
 
-    def __init__(self, server, key, debug, bit, hash_number, keep_on_close):
-        super().__init__(server, key, debug, keep_on_close)
+    def __init__(self, server, key, debug, bit, hash_number, keep_on_close, info):
+        super().__init__(server, key, debug, keep_on_close, info)
         self.bit = bit
         self.hash_number = hash_number
         self.bf = BloomFilter(server, self.key, bit, hash_number)
@@ -125,9 +115,10 @@ class RedisBloomDupeFilter(RedisRFPDupeFilter):
         keep_on_close = crawler.settings.getbool("KEEP_DUPEFILTER_DATA_ON_CLOSE", True)
         key = dupefilter_key % {'spider': crawler.spider.name}
         debug = crawler.settings.getbool('DUPEFILTER_DEBUG', False)
+        info = crawler.settings.getbool('DUPEFILTER_INFO', False)
         bit = crawler.settings.getint('BLOOMFILTER_BIT', 30)
         hash_number = crawler.settings.getint('BLOOMFILTER_HASH_NUMBER', 6)
-        return cls(server, key=key, debug=debug, bit=bit, hash_number=hash_number, keep_on_close=keep_on_close)
+        return cls(server, key=key, debug=debug, bit=bit, hash_number=hash_number, keep_on_close=keep_on_close, info=info)
 
     async def request_seen(self, request: Request) -> bool:
         fp = await self.bf.exists(request.fingerprint)

aioscrapy/libs/downloader/retry.py

@@ -26,6 +26,13 @@ try:
 except ImportError:
     pass
 
+try:
+    from anyio import EndOfStream
+
+    NEED_RETRY_ERROR += (EndOfStream,)
+except ImportError:
+    pass
+
 try:
     from httpx import HTTPError as HttpxError
 

aioscrapy/utils/log.py

@@ -7,7 +7,9 @@ from loguru import logger as _logger
 
 from aioscrapy.settings import Settings
 
-_logger.remove(0)
+for _handler in _logger._core.handlers.values():
+    if _handler._name == '<stderr>':
+        _logger.remove(_handler._id)
 
 
 def configure_logging(spider: Type["Spider"], settings: Settings):