fix: pypi decorator handling packages with special characters

metaflow/plugins/pypi/conda_environment.py

@@ -20,7 +20,7 @@
 from metaflow.metaflow_profile import profile
 
 from . import MAGIC_FILE, _datastore_packageroot
-from .utils import conda_platform
+from .utils import conda_platform, generate_cache_path, parse_filename_from_url
 
 
 class CondaEnvironmentException(MetaflowException):
@@ -107,7 +107,7 @@ def cache(storage, results, type_):
  local_packages = {
  url: {
  # Path to package in datastore.
- "path": urlparse(url).netloc + urlparse(url).path,
+ "path": generate_cache_path(url, local_path),
  # Path to package on local disk.
  "local_path": local_path,
  }
@@ -122,9 +122,8 @@ def cache(storage, results, type_):
  # Cache only those packages that manifest is unaware of
  local_packages.pop(package["url"], None)
  else:
- package["path"] = (
- urlparse(package["url"]).netloc
- + urlparse(package["url"]).path
+ package["path"] = generate_cache_path(
+ package["url"], parse_filename_from_url(package["url"])
  )
  dirty.add(id_)
 

metaflow/plugins/pypi/pip.py

@@ -9,7 +9,7 @@
 from metaflow.util import which
 
 from .micromamba import Micromamba
-from .utils import pip_tags
+from .utils import parse_filename_from_url, pip_tags
 
 
 class PipException(MetaflowException):
@@ -118,7 +118,7 @@ def download(self, id_, packages, python, platform):
  for package in packages:
  cmd.append("{url}".format(**package))
  metadata["{url}".format(**package)] = "{prefix}/.pip/wheels/{wheel}".format(
- prefix=prefix, wheel=package["url"].split("/")[-1]
+ prefix=prefix, wheel=parse_filename_from_url(package["url"])
  )
  self._call(prefix, cmd)
  # write the url to wheel mappings in a magic location

metaflow/plugins/pypi/utils.py

@@ -1,3 +1,4 @@
+import os
 import platform
 import sys
 
@@ -13,6 +14,7 @@ class Tags:
  from metaflow._vendor.packaging import tags
 
 from metaflow.exception import MetaflowException
+from urllib.parse import unquote, urlparse
 
 
 def conda_platform():
@@ -73,3 +75,16 @@ def pip_tags(python_version, mamba_platform):
  supported.extend(tags.cpython_tags(py_version, abis, platforms))
  supported.extend(tags.compatible_tags(py_version, interpreter, platforms))
  return supported
+
+
+def parse_filename_from_url(url):
+ # Separate method as it might require additional checks for the parsing.
+ filename = url.split("/")[-1]
+ return unquote(filename)
+
+
+def generate_cache_path(url, local_path):
+ base, _ = os.path.split(urlparse(url).path)
+ _, localfile = os.path.split(local_path)
+ unquoted_base = unquote(base)
+ return urlparse(url).netloc + os.path.join(unquoted_base, localfile)