fix: remove cyclic imports (#458)

* fix: remove cyclic imports

* fix import

* lint code

setup.py

@@ -113,6 +113,7 @@ def get_long_description():
         "charset_normalizer >= 3.2.0; python_version >= '3.7'",
         "courlan >= 0.9.5",
         "htmldate >= 1.6.0",
+        "importlib_metadata; python_version < '3.8'",
         "justext >= 3.0.0",
         "lxml >= 4.9.3 ; platform_system != 'Darwin'",
         "lxml == 4.9.2 ; platform_system == 'Darwin'",

trafilatura/downloads.py

@@ -28,11 +28,20 @@
 from courlan import UrlStore
 from courlan.network import redirection_test
 
-from . import __version__
+try:  # Python 3.8+
+    from importlib.metadata import version
+except ImportError:
+    from importlib_metadata import version
+
+
 from .settings import DEFAULT_CONFIG
 from .utils import (URL_BLACKLIST_REGEX, decode_response, make_chunks,
                     uniquify_list)
 
+
+LOGGER = logging.getLogger(__name__)
+PKG_VERSION = version("trafilatura")
+
 NUM_CONNECTIONS = 50
 MAX_REDIRECTS = 2
 
@@ -42,11 +51,9 @@
 RETRY_STRATEGY = None
 
 DEFAULT_HEADERS = urllib3.util.make_headers(accept_encoding=True)
-USER_AGENT = 'trafilatura/' + __version__ + ' (+https://github.com/adbar/trafilatura)'
+USER_AGENT = 'trafilatura/' + PKG_VERSION + ' (+https://github.com/adbar/trafilatura)'
 DEFAULT_HEADERS['User-Agent'] = USER_AGENT
 
-LOGGER = logging.getLogger(__name__)
-
 RawResponse = namedtuple('RawResponse', ['data', 'status', 'url'])
 
 

trafilatura/xml.py

@@ -8,19 +8,28 @@
 
 import logging
 import lzma
+
 from html import unescape
 from json import dumps as json_dumps
 from pathlib import Path
 from pickle import load as load_pickle
 
+try:  # Python 3.8+
+    from importlib.metadata import version
+except ImportError:
+    from importlib_metadata import version
+
 from lxml.etree import (Element, RelaxNG, SubElement, XMLParser, fromstring,
                         tostring)
 
-from . import __version__
+
 from .filters import text_chars_test
 from .utils import sanitize, sanitize_tree
 
+
 LOGGER = logging.getLogger(__name__)
+PKG_VERSION = version("trafilatura")
+
 # validation
 TEI_SCHEMA = str(Path(__file__).parent / 'data/tei-schema-pickle.lzma')
 TEI_VALID_TAGS = {'ab', 'body', 'cell', 'code', 'del', 'div', 'graphic', 'head', 'hi', \
@@ -395,7 +404,7 @@ def write_fullheader(teidoc, docmeta):
             tags_list.text = ','.join(docmeta.tags)
     encodingdesc = SubElement(header, 'encodingDesc')
     appinfo = SubElement(encodingdesc, 'appInfo')
-    application = SubElement(appinfo, 'application', version=__version__, ident='Trafilatura')
+    application = SubElement(appinfo, 'application', version=PKG_VERSION, ident='Trafilatura')
     label = SubElement(application, 'label')
     label.text = 'Trafilatura'
     pointer = SubElement(application, 'ptr', target='https://github.com/adbar/trafilatura')