From 25568f30bf2544a8c7d4f757528fc0f4bb459c03 Mon Sep 17 00:00:00 2001
From: Tony Tung <ttung@chanzuckerberg.com>
Date: Tue, 3 Sep 2019 14:58:13 -0700
Subject: [PATCH] Refactor file-url-to-path

This is the same issue as #112, so I added a general method to do this conversion.
---
 slicedimage/io/_base.py    |  7 +++----
 slicedimage/url/path.py    | 15 +++++++++++++++
 slicedimage/url/resolve.py |  9 ++-------
 3 files changed, 20 insertions(+), 11 deletions(-)

diff --git a/slicedimage/io/_base.py b/slicedimage/io/_base.py
index 5a84611..b4309a6 100644
--- a/slicedimage/io/_base.py
+++ b/slicedimage/io/_base.py
@@ -20,6 +20,7 @@
 
 from packaging import version
 
+from slicedimage.url.path import get_path_from_parsed_file_url
 from slicedimage.url.resolve import resolve_url
 from slicedimage._collection import Collection
 from slicedimage._formats import ImageFormat
@@ -283,8 +284,7 @@ def partition_url_generator(self, parent_partition_url: str, partition_name: str
         if self.partition_path_generator is None:
             return super().partition_url_generator(parent_partition_url, partition_name)
         parent_parsed_url = urllib.parse.urlparse(parent_partition_url)
-        assert parent_parsed_url.scheme == "file"
-        parent_path = PurePosixPath(parent_parsed_url.path)
+        parent_path = get_path_from_parsed_file_url(parent_parsed_url)
         partition_path = self.partition_path_generator(parent_path, partition_name)
         partition_parsed_url = parent_parsed_url._replace(path=str(partition_path))
         return urllib.parse.urlunparse(partition_parsed_url)
@@ -310,8 +310,7 @@ def tile_url_generator(self, tileset_url: str, tile: Tile, ext: str) -> str:
         if self.tile_opener is None:
             return super().tile_url_generator(tileset_url, tile, ext)
         tileset_parsed_url = urllib.parse.urlparse(tileset_url)
-        assert tileset_parsed_url.scheme == "file"
-        tileset_path = PurePosixPath(tileset_parsed_url.path)
+        tileset_path = get_path_from_parsed_file_url(tileset_parsed_url)
         with self.tile_opener(tileset_path, tile, ext) as open_fh:
             tile_path = open_fh.name
 
diff --git a/slicedimage/url/path.py b/slicedimage/url/path.py
index 71cb9b1..c0b5fbf 100644
--- a/slicedimage/url/path.py
+++ b/slicedimage/url/path.py
@@ -1,3 +1,4 @@
+import os
 import pathlib
 import posixpath
 import urllib.parse
@@ -104,3 +105,17 @@ def calculate_relative_url(baseurl: str, name_or_url: str) -> str:
         result = result + "#" + absolute_url_parsed.fragment
 
     return result
+
+
+def get_path_from_parsed_file_url(parsed_file_url: urllib.parse.ParseResult) -> pathlib.Path:
+    """If parsed_file_url is the result of parsing a URL using urllib.parse.unparse, and the URL is
+    a "file:" URL, then extract the local filesystem path.  Handles idiosyncrasies such as pathlib
+    and Windows paths.
+    """
+    assert parsed_file_url.scheme == "file"
+    if os.name == "nt":
+        # pathlib can parse c:/windows/xxx, but not /c:/windows/xxx.  however, url paths always
+        # start with a /
+        return pathlib.Path(urllib.parse.unquote(parsed_file_url.path[1:]))
+    else:
+        return pathlib.Path(urllib.parse.unquote(parsed_file_url.path))
diff --git a/slicedimage/url/resolve.py b/slicedimage/url/resolve.py
index df3677f..553fdbc 100644
--- a/slicedimage/url/resolve.py
+++ b/slicedimage/url/resolve.py
@@ -5,7 +5,7 @@
 
 from slicedimage._compat import fspath
 from slicedimage.backends import CachingBackend, DiskBackend, HttpBackend, S3Backend, SIZE_LIMIT
-from .path import get_absolute_url
+from .path import get_absolute_url, get_path_from_parsed_file_url
 
 
 def infer_backend(baseurl, backend_config=None):
@@ -26,12 +26,7 @@ def infer_backend(baseurl, backend_config=None):
     parsed = urllib.parse.urlparse(baseurl)
 
     if parsed.scheme == "file":
-        if os.name == "nt":
-            # pathlib can parse c:/windows/xxx, but not /c:/windows/xxx.  however, url paths always
-            # start with a /
-            local_path = pathlib.Path(urllib.parse.unquote(parsed.path[1:]))
-        else:
-            local_path = pathlib.Path(urllib.parse.unquote(parsed.path))
+        local_path = get_path_from_parsed_file_url(parsed)
         return DiskBackend(fspath(local_path))
 
     if parsed.scheme in ("http", "https"):