Test fixes, getting dependencies for pinned requirements in advance.

Signed-off-by: Thomas Neidhart <thomas.neidhart@gmail.com>

requirements-dev.txt

@@ -38,4 +38,5 @@ tomli==1.2.3
 tqdm==4.64.0
 twine==3.8.0
 typed-ast==1.5.4
-webencodings==0.5.1
+webencodings==0.5.1
+pytest-asyncio==0.21.1

requirements.txt

@@ -25,3 +25,4 @@ toml==0.10.2
 urllib3==2.1.0
 zipp==3.17.0
 aiohttp==3.9.1
+aiofiles==23.2.1

setup.cfg

@@ -70,6 +70,7 @@ install_requires =
     mock >= 3.0.5
     packvers >= 21.5
     aiohttp >= 3.9
+    aiofiles >= 23.1
 
 [options.packages.find]
 where = src
@@ -88,6 +89,7 @@ testing =
     black
     isort
     pytest-rerunfailures
+    pytest-asyncio >= 0.21
 
 docs =
     Sphinx>=5.0.2

src/python_inspector/api.py

@@ -23,7 +23,7 @@
 
 from _packagedcode.models import DependentPackage
 from _packagedcode.models import PackageData
-from _packagedcode.pypi import PipRequirementsFileHandler
+from _packagedcode.pypi import PipRequirementsFileHandler, get_resolved_purl
 from _packagedcode.pypi import PythonSetupPyHandler
 from _packagedcode.pypi import can_process_dependent_package
 from python_inspector import dependencies
@@ -38,6 +38,7 @@
 from python_inspector.resolution import get_python_version_from_env_tag
 from python_inspector.resolution import get_reqs_insecurely
 from python_inspector.resolution import get_requirements_from_python_manifest
+from python_inspector.utils import Candidate
 from python_inspector.utils_pypi import PLATFORMS_BY_OS
 from python_inspector.utils_pypi import PYPI_SIMPLE_URL
 from python_inspector.utils_pypi import Environment
@@ -229,7 +230,7 @@ def resolve_dependencies(
     if not direct_dependencies:
         return Resolution(
             packages=[],
-            resolution={},
+            resolution=[],
             files=files,
         )
 
@@ -300,7 +301,7 @@ async def get_pypi_data(package):
             return data
 
         if verbose:
-            printer(f"retrieve data from pypi:")
+            printer(f"retrieve package data from pypi:")
 
         return await asyncio.gather(*[get_pypi_data(package) for package in purls])
 
@@ -390,6 +391,8 @@ def get_resolved_dependencies(
         ignore_errors=ignore_errors,
     )
 
+    # gather version data for all requirements concurrently in advance.
+
     async def gather_version_data():
         async def get_version_data(name: str):
             versions = await provider.fill_versions_for_package(name)
@@ -406,6 +409,28 @@ async def get_version_data(name: str):
 
     asyncio.run(gather_version_data())
 
+    # gather dependencies for all pinned requirements concurrently in advance.
+
+    async def gather_dependencies():
+        async def get_dependencies(requirement: Requirement):
+            purl = PackageURL(type="pypi", name=requirement.name)
+            resolved_purl = get_resolved_purl(purl=purl, specifiers=requirement.specifier)
+
+            if resolved_purl:
+                purl = resolved_purl.purl
+                candidate = Candidate(requirement.name, purl.version, requirement.extras)
+                await provider.fill_requirements_for_package(purl, candidate)
+
+                if verbose:
+                    printer(f"  retrieved dependencies for requirement '{str(purl)}'")
+
+        if verbose:
+            printer(f"dependencies:")
+
+        return await asyncio.gather(*[get_dependencies(requirement) for requirement in requirements])
+
+    asyncio.run(gather_dependencies())
+
     resolver = Resolver(
         provider=provider,
         reporter=BaseReporter(),

src/python_inspector/resolution.py

@@ -438,6 +438,7 @@ async def _get_versions_for_package_from_repo(
                 )
             if valid_wheel_present or pypi_valid_python_version:
                 versions.append(version)
+
         return versions
 
     async def _get_versions_for_package_from_pypi_json_api(self, name: str) -> List[Version]:
@@ -556,7 +557,7 @@ async def _get_requirements_for_package_from_pypi_json_api(
             return []
         info = resp.get("info") or {}
         requires_dist = info.get("requires_dist") or []
-        return requires_dist
+        return list(map(lambda r: Requirement(r), requires_dist))
 
     def get_candidates(
         self,

src/python_inspector/utils_pypi.py

@@ -18,13 +18,14 @@
 import tempfile
 import time
 from collections import defaultdict
-from typing import List, Dict
+from typing import List, Dict, Union, Tuple
 from typing import NamedTuple
 from urllib.parse import quote_plus
 from urllib.parse import unquote
 from urllib.parse import urlparse
 from urllib.parse import urlunparse
 
+import aiofiles
 import aiohttp
 import attr
 import packageurl
@@ -1593,7 +1594,7 @@ async def fetch_links(
         name using the `index_url` of this repository.
         """
         package_url = f"{self.index_url}/{normalized_name}"
-        text = await CACHE.get(
+        text, _ = await CACHE.get(
             path_or_url=package_url,
             credentials=self.credentials,
             as_text=True,
@@ -1671,7 +1672,7 @@ async def get(
         force=False,
         verbose=False,
         echo_func=None,
-    ):
+    ) -> Tuple[Union[str, bytes], str]:
         """
         Return the content fetched from a ``path_or_url`` through the cache.
         Raise an Exception on errors. Treats the content as text if as_text is
@@ -1692,13 +1693,13 @@ async def get(
                 echo_func=echo_func,
             )
             wmode = "w" if as_text else "wb"
-            with open(cached, wmode) as fo:
-                fo.write(content)
-            return content
+            async with aiofiles.open(cached, mode=wmode) as fo:
+                await fo.write(content)
+            return content, cached
         else:
             if TRACE_DEEP:
                 print(f"        FILE CACHE HIT: {path_or_url}")
-            return get_local_file_content(path=cached, as_text=as_text)
+            return await get_local_file_content(path=cached, as_text=as_text), cached
 
 
 CACHE = Cache()
@@ -1730,13 +1731,13 @@ async def get_file_content(
     elif path_or_url.startswith("file://") or (
         path_or_url.startswith("/") and os.path.exists(path_or_url)
     ):
-        return get_local_file_content(path=path_or_url, as_text=as_text)
+        return await get_local_file_content(path=path_or_url, as_text=as_text)
 
     else:
         raise Exception(f"Unsupported URL scheme: {path_or_url}")
 
 
-def get_local_file_content(path, as_text=True):
+async def get_local_file_content(path: str, as_text=True) -> str:
     """
     Return the content at `url` as text. Return the content as bytes is
     `as_text` is False.
@@ -1745,8 +1746,8 @@ def get_local_file_content(path, as_text=True):
         path = path[7:]
 
     mode = "r" if as_text else "rb"
-    with open(path, mode) as fo:
-        return fo.read()
+    async with aiofiles.open(path, mode=mode) as fo:
+        return await fo.read()
 
 
 class RemoteNotFetchedException(Exception):
@@ -1828,7 +1829,7 @@ async def fetch_and_save(
     errors. Treats the content as text if as_text is True otherwise as treat as
     binary.
     """
-    content = await CACHE.get(
+    content, path = await CACHE.get(
         path_or_url=path_or_url,
         credentials=credentials,
         as_text=as_text,
@@ -1837,7 +1838,8 @@ async def fetch_and_save(
     )
 
     output = os.path.join(dest_dir, filename)
-    wmode = "w" if as_text else "wb"
-    with open(output, wmode) as fo:
-        fo.write(content)
+    if os.path.exists(output):
+        os.remove(output)
+
+    os.symlink(os.path.abspath(path), output)
     return content

tests/test_utils.py

@@ -14,6 +14,7 @@
 from netrc import netrc
 from unittest import mock
 
+import pytest
 from commoncode.testcase import FileDrivenTesting
 from test_cli import check_json_file_results
 
@@ -47,22 +48,23 @@ def test_get_netrc_auth_with_no_matching_url():
     assert get_netrc_auth(url="https://pypi2.org/simple", netrc=parsed_netrc) == (None, None)
 
 
+@pytest.mark.asyncio
 @mock.patch("python_inspector.utils_pypi.CACHE.get")
-def test_fetch_links(mock_get):
+async def test_fetch_links(mock_get):
     file_name = test_env.get_test_loc("psycopg2.html")
     with open(file_name) as file:
         mock_get.return_value = file.read()
-    links = PypiSimpleRepository().fetch_links(normalized_name="psycopg2")
+    links = await PypiSimpleRepository().fetch_links(normalized_name="psycopg2")
     result_file = test_env.get_temp_file("json")
     expected_file = test_env.get_test_loc("psycopg2-links-expected.json", must_exist=False)
     with open(result_file, "w") as file:
         json.dump(links, file, indent=4)
     check_json_file_results(result_file, expected_file)
     # Testing relative links
-    realtive_links_file = test_env.get_test_loc("fetch_links_test.html")
-    with open(realtive_links_file) as realtive_file:
-        mock_get.return_value = realtive_file.read()
-    relative_links = PypiSimpleRepository().fetch_links(normalized_name="sources.whl")
+    relative_links_file = test_env.get_test_loc("fetch_links_test.html")
+    with open(relative_links_file) as relative_file:
+        mock_get.return_value = relative_file.read()
+    relative_links = await PypiSimpleRepository().fetch_links(normalized_name="sources.whl")
     relative_links_result_file = test_env.get_temp_file("json")
     relative_links_expected_file = test_env.get_test_loc(
         "relative-links-expected.json", must_exist=False
@@ -83,8 +85,9 @@ def test_parse_reqs():
     check_json_file_results(result_file, expected_file)
 
 
-def test_get_sdist_file():
-    sdist_file = fetch_and_extract_sdist(
+@pytest.mark.asyncio
+async def test_get_sdist_file():
+    sdist_file = await fetch_and_extract_sdist(
         repos=tuple([PypiSimpleRepository()]),
         candidate=Candidate(name="psycopg2", version="2.7.5", extras=None),
         python_version="3.8",