dropped dependency on requests

Author: xrotwang

CHANGELOG.md

@@ -7,6 +7,7 @@ The `pycldf` package adheres to [Semantic Versioning](http://semver.org/spec/v2.
 
 - Make sure all local media files are copied with `Dataset.copy` as well.
 - Allow passing a description to `Dataset.add_table` and `Dataset.add_component`.
+- Drop dependency on requests.
 
 
 ## [1.41.0] - 2025-02-15

setup.cfg

@@ -44,7 +44,6 @@ install_requires =
     # pybtex requires setuptools, but doesn't seem to declare this.
     setuptools
     pybtex
-    requests
     newick
     commonnexus>=1.2.0
     python-frontmatter

src/pycldf/cli_util.py

@@ -2,10 +2,10 @@
 Functionality to use in commandline tools which need to access CLDF datasets.
 """
 import argparse
+import urllib.request
 
 from clldutils.clilib import PathType, ParserError
 from csvw.utils import is_url
-import requests
 
 from pycldf import Dataset, Database
 from pycldf.ext import discovery
@@ -45,11 +45,23 @@ def __call__(self, string):
             return super().__call__(string)
 
 
+
+def head(url):  # pragma: no cover
+    class NoRedirection(urllib.request.HTTPErrorProcessor):
+        def http_response(self, request, response):
+            return response
+
+        https_response = http_response
+
+    opener = urllib.request.build_opener(NoRedirection)
+    return opener.open(urllib.request.Request(url, method="HEAD")).status
+
+
 class UrlOrPathType(PathType):
     def __call__(self, string):
         if is_url(string):
             if self._must_exist:
-                sc = requests.head(string).status_code
+                sc = head(string)
                 # We accept not only HTTP 200 as valid but also common redirection codes because
                 # these are used e.g. for DOIs.
                 if sc not in {200, 301, 302}:

src/pycldf/dataset.py

@@ -1026,6 +1026,8 @@ def validate(
         :raises ValueError: if a validation error is encountered (and `log` is `None`).
         :return: Flag signaling whether schema and data are valid.
         """
+        # We must import components with custom validation to make sure they can be detected as
+        # subclasses of ComponentWithValidation.
         from pycldf.media import MediaTable
         from pycldf.trees import TreeTable
 
@@ -1311,6 +1313,9 @@ def sentences(self) -> typing.List[orm.Example]:
 
 
 class ComponentWithValidation:
+    """
+    A virtual base class for custom, component-centered validation.
+    """
     def __init__(self, ds: Dataset):
         self.ds = ds
         self.component = self.__class__.__name__
@@ -1347,11 +1352,11 @@ def sniff(p: pathlib.Path) -> bool:
     return d.get('dc:conformsTo', '').startswith(TERMS_URL)
 
 
-def iter_datasets(d: pathlib.Path) -> typing.Iterator[Dataset]:
+def iter_datasets(d: pathlib.Path) -> typing.Generator[Dataset, None, None]:
     """
     Discover CLDF datasets - by identifying metadata files - in a directory.
 
-    :param d: directory
+    :param d: directory in which to look for CLDF datasets (recursively).
     :return: generator of `Dataset` instances.
     """
     for p in walk(d, mode='files'):

tests/test_cli_util.py

@@ -6,14 +6,10 @@
 
 
 def test_UrlOrPathType(mocker):
-    mocker.patch(
-        'pycldf.cli_util.requests',
-        mocker.Mock(head=mocker.Mock(return_value=mocker.Mock(status_code=200))))
+    mocker.patch('pycldf.cli_util.head', mocker.Mock(return_value=200))
     url = 'http://example.com'
     assert UrlOrPathType()(url) == url
 
-    mocker.patch(
-        'pycldf.cli_util.requests',
-        mocker.Mock(head=mocker.Mock(return_value=mocker.Mock(status_code=404))))
+    mocker.patch('pycldf.cli_util.head', mocker.Mock(return_value=404))
     with pytest.raises(argparse.ArgumentTypeError):
         _ = UrlOrPathType()(url)