fix: Allow using local file:// locations with find_links and index_ur…

…ls (#132)

src/unearth/collector.py

@@ -176,9 +176,14 @@ def collect_links_from_location(
                 index_html = Link(path_to_url(path.joinpath("index.html").as_posix()))
                 yield from _collect_links_from_index(session, index_html, headers)
         else:
-            yield from _collect_links_from_index(session, location, headers)
+            if _is_html_file(str(path)):
+                yield from _collect_links_from_index(session, location, headers)
+            else:
+                yield location
 
-    else:
+    else:  # remote url, can be either a remote file or an index URL containing files
+        if is_secure_origin(session, location) and not location.is_vcs:
+            yield location
         yield from _collect_links_from_index(session, location)
 
 

src/unearth/finder.py

@@ -7,11 +7,11 @@
 import itertools
 import os
 import pathlib
+import posixpath
 import warnings
 from datetime import datetime
 from tempfile import TemporaryDirectory
 from typing import TYPE_CHECKING, Any, Iterable, NamedTuple, Sequence
-from urllib.parse import urljoin
 
 import packaging.requirements
 from packaging.utils import BuildTag, canonicalize_name, parse_wheel_filename
@@ -190,9 +190,8 @@ def build_evaluator(
         )
 
     def _build_index_page_link(self, index_url: str, package_name: str) -> Link:
-        return Link(
-            urljoin(index_url.rstrip("/") + "/", canonicalize_name(package_name) + "/")
-        )
+        url = posixpath.join(index_url, canonicalize_name(package_name)) + "/"
+        return self._build_find_link(url)
 
     def _build_find_link(self, find_link: str) -> Link:
         if os.path.exists(find_link):

tests/test_collector.py

@@ -25,22 +25,15 @@ def test_collector_skip_vcs_link(pypi_session, caplog):
 
 
 def test_collect_links_from_404_page(pypi_session):
-    collected = list(
-        collect_links_from_location(
-            pypi_session, Link("https://test.pypi.org/simple/not-found")
-        )
-    )
-    assert not collected
+    link = Link("https://test.pypi.org/simple/not-found")
+    collected = list(collect_links_from_location(pypi_session, link))
+    assert collected == [link]
 
 
 def test_skip_non_html_archive(pypi_session, caplog):
-    collected = list(
-        collect_links_from_location(
-            pypi_session,
-            Link("https://test.pypi.org/files/click-8.1.3-py3-none-any.whl"),
-        )
-    )
-    assert not collected
+    link = Link("https://test.pypi.org/files/click-8.1.3-py3-none-any.whl")
+    collected = list(collect_links_from_location(pypi_session, link))
+    assert collected == [link]
     assert "Content-Type unsupported" in caplog.records[0].message
 
 
@@ -52,7 +45,7 @@ def test_collect_links_from_index_page(pypi_session):
         ),
         key=lambda link: link.filename,
     )
-    assert len(collected) == 4
+    assert len(collected) == 5
     assert all(link.url.startswith("https://test.pypi.org") for link in collected)