ultimate-sitemap-parser 1.1.0__py3-none-any.whl → 1.1.1__py3-none-any.whl

{ultimate_sitemap_parser-1.1.0.dist-info → ultimate_sitemap_parser-1.1.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ultimate-sitemap-parser
-Version: 1.1.0
+Version: 1.1.1
 Summary: A performant library for parsing and crawling sitemaps
 License: GPL-3.0-or-later
 Keywords: sitemap,crawler,indexing,xml,rss,atom,google news

ultimate_sitemap_parser-1.1.1.dist-info/RECORD

@@ -0,0 +1,21 @@
+usp/__init__.py,sha256=_jshbOBBUHRZ5ko4SdI7GRFiF9xKGJVCEPgL9lZJ81o,124
+usp/cli/__init__.py,sha256=mGrjSftUYfM2SGp9yEN2dTJndl5thOdv77-EAe6ocWo,37
+usp/cli/_ls.py,sha256=BjF5bGuhe_E_Ak-yyY0cDM83LFstl5tA3XNIrGZJujs,2954
+usp/cli/_util.py,sha256=UL5WiRZlpiDOI_QvSU1PdjcS6iCmfcLQlO1Mm1wjSAw,505
+usp/cli/cli.py,sha256=ySNyYHoCQ440KfxmpTkzLXgqtbnt5ru-TgPs2Zw2-LI,592
+usp/exceptions.py,sha256=9KTgnocYYZCfyaCf9BrBN7Ok4cwn7_DlrNFbhUfFsGM,634
+usp/fetch_parse.py,sha256=VJrJSAG1X8oQyW2p9wSepuGWfHlMDNoJG8jn3an2XUY,41396
+usp/helpers.py,sha256=S9d8fEhHzZqVCx3SkcWVTgW1JYKujH-tM86urjORNWA,8482
+usp/objects/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+usp/objects/page.py,sha256=vz2QXC9Z3E65Cxf68tKfQkubIc_OB0m6pNYH146Qx_8,14253
+usp/objects/sitemap.py,sha256=yt5qe6fyKfmvJmV60mB8kc7yooGcpYhuIcNlmUqFGFA,11486
+usp/tree.py,sha256=AmK0TptwNAexwSBAjrziYvx9cueQDMt5w9_1m8d4edI,4055
+usp/web_client/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+usp/web_client/abstract_client.py,sha256=7MpIfqQpi1_yojEmuReT8iy9kFUWCD3i2LMpHmBOwV0,6291
+usp/web_client/requests_client.py,sha256=xxkBUHvakBN-Guw_DqGElZJVS42xgUwWHxM7jA_QEPI,5593
+ultimate_sitemap_parser-1.1.1.dist-info/LICENSE,sha256=ixuiBLtpoK3iv89l7ylKkg9rs2GzF9ukPH7ynZYzK5s,35148
+ultimate_sitemap_parser-1.1.1.dist-info/METADATA,sha256=9sK5LCSHHPuSvdDjakIqOm-Gv3_Lgm1tsZdDDFs8vSE,4447
+ultimate_sitemap_parser-1.1.1.dist-info/NOTICE,sha256=3ANZA5R9rYnCOnUoroGfFUOZ__ww_yG01NUAx0X6J7E,632
+ultimate_sitemap_parser-1.1.1.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
+ultimate_sitemap_parser-1.1.1.dist-info/entry_points.txt,sha256=v60w5WzqYlPOucntZUy0ydzlYwuAPSwoQY0KdT5ragQ,36
+ultimate_sitemap_parser-1.1.1.dist-info/RECORD,,

usp/cli/_ls.py

@@ -2,7 +2,7 @@ import argparse
 import sys
 from typing import Iterator
 
-from usp.cli._util import tabs, format_help
+from usp.cli._util import format_help, tabs
 from usp.objects.sitemap import AbstractSitemap
 from usp.tree import sitemap_tree_for_homepage
 

usp/cli/cli.py

@@ -1,7 +1,7 @@
 from argparse import ArgumentParser
 
-from usp.cli import _ls as ls_cmd
 from usp import __version__
+from usp.cli import _ls as ls_cmd
 
 
 def main():

usp/fetch_parse.py

@@ -8,49 +8,49 @@
 """
 
 import abc
+import logging
 import re
 import xml.parsers.expat
 from collections import OrderedDict
 from decimal import Decimal, InvalidOperation
-from typing import Optional, Dict, Union
-
+from typing import Dict, Optional, Union
 
 from .exceptions import SitemapException, SitemapXMLParsingException
 from .helpers import (
-    html_unescape_strip,
-    parse_iso8601_date,
     get_url_retry_on_client_errors,
-    ungzipped_response_content,
+    html_unescape_strip,
     is_http_url,
+    parse_iso8601_date,
     parse_rfc2822_date,
+    ungzipped_response_content,
 )
-from .log import create_logger
 from .objects.page import (
+    SITEMAP_PAGE_DEFAULT_PRIORITY,
     SitemapImage,
-    SitemapPage,
     SitemapNewsStory,
+    SitemapPage,
     SitemapPageChangeFrequency,
-    SITEMAP_PAGE_DEFAULT_PRIORITY,
 )
 from .objects.sitemap import (
     AbstractSitemap,
-    InvalidSitemap,
     IndexRobotsTxtSitemap,
     IndexXMLSitemap,
-    PagesXMLSitemap,
-    PagesTextSitemap,
-    PagesRSSSitemap,
+    InvalidSitemap,
     PagesAtomSitemap,
+    PagesRSSSitemap,
+    PagesTextSitemap,
+    PagesXMLSitemap,
 )
 from .web_client.abstract_client import (
     AbstractWebClient,
     AbstractWebClientSuccessResponse,
+    LocalWebClient,
+    NoWebClientException,
     WebClientErrorResponse,
 )
-from .web_client.abstract_client import LocalWebClient, NoWebClientException
 from .web_client.requests_client import RequestsWebClient
 
-log = create_logger(__name__)
+log = logging.getLogger(__name__)
 
 
 class SitemapFetcher:

usp/helpers.py

@@ -3,24 +3,25 @@
 import datetime
 import gzip as gzip_lib
 import html
+import logging
 import re
 import sys
 import time
 from typing import Optional
-from urllib.parse import urlparse, unquote_plus, urlunparse
-from dateutil.parser import parse as dateutil_parse
+from urllib.parse import unquote_plus, urlparse, urlunparse
+
 from dateutil.parser import isoparse as dateutil_isoparse
+from dateutil.parser import parse as dateutil_parse
 
-from .exceptions import SitemapException, GunzipException, StripURLToHomepageException
-from .log import create_logger
+from .exceptions import GunzipException, SitemapException, StripURLToHomepageException
 from .web_client.abstract_client import (
     AbstractWebClient,
+    AbstractWebClientResponse,
     AbstractWebClientSuccessResponse,
     WebClientErrorResponse,
-    AbstractWebClientResponse,
 )
 
-log = create_logger(__name__)
+log = logging.getLogger(__name__)
 
 __URL_REGEX = re.compile(r"^https?://[^\s/$.?#].[^\s]*$", re.IGNORECASE)
 """Regular expression to match HTTP(s) URLs."""
@@ -247,7 +248,7 @@ def ungzipped_response_content(
             data = gunzip(data)
         except GunzipException as ex:
             # In case of an error, just assume that it's one of the non-gzipped sitemaps with ".gz" extension
-            log.error(
+            log.warning(
                 f"Unable to gunzip response {response}, maybe it's a non-gzipped sitemap: {ex}"
             )
 

usp/objects/sitemap.py

@@ -9,11 +9,11 @@
 """
 
 import abc
-from functools import lru_cache
 import os
 import pickle
 import tempfile
-from typing import List, Iterator, Tuple
+from functools import lru_cache
+from typing import Iterator, List, Tuple
 
 from .page import SitemapPage
 
@@ -72,7 +72,7 @@ class AbstractSitemap(metaclass=abc.ABCMeta):
         return hash((self.url,))
 
     def __repr__(self):
-        return f"{self.__class__.__name__}(" f"url={self.url}" ")"
+        return f"{self.__class__.__name__}(url={self.url})"
 
     @property
     def url(self) -> str:
@@ -167,12 +167,7 @@ class InvalidSitemap(AbstractSitemap):
         return True
 
     def __repr__(self):
-        return (
-            f"{self.__class__.__name__}("
-            f"url={self.url}, "
-            f"reason={self.reason}"
-            ")"
-        )
+        return f"{self.__class__.__name__}(url={self.url}, reason={self.reason})"
 
     def to_dict(self, with_pages=True) -> dict:
         return {

usp/tree.py

@@ -1,19 +1,20 @@
 """Helpers to generate a sitemap tree."""
 
+import logging
 from typing import Optional
+
 from .exceptions import SitemapException
 from .fetch_parse import SitemapFetcher, SitemapStrParser
 from .helpers import is_http_url, strip_url_to_homepage
-from .log import create_logger
 from .objects.sitemap import (
     AbstractSitemap,
-    InvalidSitemap,
-    IndexWebsiteSitemap,
     IndexRobotsTxtSitemap,
+    IndexWebsiteSitemap,
+    InvalidSitemap,
 )
 from .web_client.abstract_client import AbstractWebClient
 
-log = create_logger(__name__)
+log = logging.getLogger(__name__)
 
 _UNPUBLISHED_SITEMAP_PATHS = {
     "sitemap.xml",

usp/web_client/abstract_client.py

@@ -2,8 +2,8 @@
 
 import abc
 import random
-from http import HTTPStatus
 import time
+from http import HTTPStatus
 from typing import Optional
 
 RETRYABLE_HTTP_STATUS_CODES = {

usp/web_client/requests_client.py

@@ -1,20 +1,23 @@
 """Implementation of :mod:`usp.web_client.abstract_client` with Requests."""
 
-from http import HTTPStatus
 import logging
-from typing import Optional, Dict, Tuple, Union
+from http import HTTPStatus
+from typing import Dict, Optional, Tuple, Union
 
 import requests
 
+from usp import __version__
+
 from .abstract_client import (
+    RETRYABLE_HTTP_STATUS_CODES,
     AbstractWebClient,
     AbstractWebClientResponse,
     AbstractWebClientSuccessResponse,
     RequestWaiter,
     WebClientErrorResponse,
-    RETRYABLE_HTTP_STATUS_CODES,
 )
-from usp import __version__
+
+log = logging.getLogger(__name__)
 
 
 class RequestsWebClientSuccessResponse(AbstractWebClientSuccessResponse):
@@ -153,7 +156,7 @@ class RequestsWebClient(AbstractWebClient):
                 )
             else:
                 message = f"{response.status_code} {response.reason}"
-                logging.info(f"Response content: {response.text}")
+                log.info(f"Response content: {response.text}")
 
                 if response.status_code in RETRYABLE_HTTP_STATUS_CODES:
                     return RequestsWebClientErrorResponse(

ultimate_sitemap_parser-1.1.0.dist-info/RECORD

@@ -1,22 +0,0 @@
-usp/__init__.py,sha256=_jshbOBBUHRZ5ko4SdI7GRFiF9xKGJVCEPgL9lZJ81o,124
-usp/cli/__init__.py,sha256=mGrjSftUYfM2SGp9yEN2dTJndl5thOdv77-EAe6ocWo,37
-usp/cli/_ls.py,sha256=YyDmtBjK02_26Qv8-3NLf87b1C4Wt0GzZ1XkdF2fllQ,2954
-usp/cli/_util.py,sha256=UL5WiRZlpiDOI_QvSU1PdjcS6iCmfcLQlO1Mm1wjSAw,505
-usp/cli/cli.py,sha256=D1tXZyhiG0sIwtepdPdglW5gUlPWyx4LNeBmaM700Yc,592
-usp/exceptions.py,sha256=9KTgnocYYZCfyaCf9BrBN7Ok4cwn7_DlrNFbhUfFsGM,634
-usp/fetch_parse.py,sha256=_AmrKQa_5dupqWWmXr3MXuPW3u_V4KBfLwCC0YaY3kk,41440
-usp/helpers.py,sha256=QJH3ETapqqbwRnjX_LM0EhWqeta9LTqVvW5OAkBKUOc,8491
-usp/log.py,sha256=BS0AtURK62TPGVqEuIu8kwGtIJDYoGsK5_N-b60VOpE,1631
-usp/objects/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-usp/objects/page.py,sha256=vz2QXC9Z3E65Cxf68tKfQkubIc_OB0m6pNYH146Qx_8,14253
-usp/objects/sitemap.py,sha256=XWgke1SJNA79qnOEvaY2nJbnlidWxqBvfuRcF4GhBHI,11564
-usp/tree.py,sha256=2cuHOpdYX5aKZ4XuUQPaKjILnoPnFKZwPNn0g8cxT18,4066
-usp/web_client/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-usp/web_client/abstract_client.py,sha256=m6glqfJyY_i63HhRb0SQLCx-i9yt-nwssiN_I-sCvQk,6291
-usp/web_client/requests_client.py,sha256=ywOovvBG4gnk0f_AJW8ZOv5r1DCrfk97fPjUT_f2US4,5561
-ultimate_sitemap_parser-1.1.0.dist-info/LICENSE,sha256=ixuiBLtpoK3iv89l7ylKkg9rs2GzF9ukPH7ynZYzK5s,35148
-ultimate_sitemap_parser-1.1.0.dist-info/METADATA,sha256=ya5PV0x-dJpeTpK4J5UUsgiMhzXlQ9DyH4QTv-ozQXk,4447
-ultimate_sitemap_parser-1.1.0.dist-info/NOTICE,sha256=3ANZA5R9rYnCOnUoroGfFUOZ__ww_yG01NUAx0X6J7E,632
-ultimate_sitemap_parser-1.1.0.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
-ultimate_sitemap_parser-1.1.0.dist-info/entry_points.txt,sha256=v60w5WzqYlPOucntZUy0ydzlYwuAPSwoQY0KdT5ragQ,36
-ultimate_sitemap_parser-1.1.0.dist-info/RECORD,,

usp/log.py

@@ -1,77 +0,0 @@
-"""Logging utilities."""
-
-import logging
-
-
-class Logger:
-    """
-    Logging helper class.
-    """
-
-    __LEVELS = {
-        "CRITICAL": logging.CRITICAL,
-        "ERROR": logging.ERROR,
-        "WARNING": logging.WARNING,
-        "INFO": logging.INFO,
-        "DEBUG": logging.DEBUG,
-    }
-    """Valid logging levels and their "logging" counterparts."""
-
-    __DEFAULT_LEVEL = "INFO"
-    """Default logging level."""
-
-    __slots__ = [
-        # "logging" object
-        "__l",
-    ]
-
-    def __init__(self, name: str):
-        """
-        Initialize logger object for a given name.
-
-        :param name: Module name that the logger should be initialized for.
-        """
-
-        self.__l = logging.getLogger(name)
-
-    def error(self, message: str) -> None:
-        """
-        Log error message.
-
-        :param message: Message to log.
-        """
-        self.__l.error(message)
-
-    def warning(self, message: str) -> None:
-        """
-        Log warning message.
-
-        :param message: Message to log.
-        """
-        self.__l.warning(message)
-
-    def info(self, message: str) -> None:
-        """
-        Log informational message.
-
-        :param message: Message to log.
-        """
-        self.__l.info(message)
-
-    def debug(self, message: str) -> None:
-        """
-        Log debugging message.
-
-        :param message: Message to log.
-        """
-        self.__l.debug(message)
-
-
-def create_logger(name: str) -> Logger:
-    """
-    Create and return Logger object.
-
-    :param name: Module name that the logger should be initialized for.
-    :return: Logger object.
-    """
-    return Logger(name=name)