Fix zip-glob for files made by CLI (#1454)

Author: martindurant

fsspec/archive.py

@@ -66,8 +66,8 @@ def ls(self, path, detail=True, **kwargs):
                 if ppath not in paths:
                     out = {"name": ppath, "size": 0, "type": "directory"}
                     paths[ppath] = out
-        out = sorted(paths.values(), key=lambda _: _["name"])
         if detail:
+            out = sorted(paths.values(), key=lambda _: _["name"])
             return out
         else:
-            return [f["name"] for f in out]
+            return sorted(paths)

fsspec/implementations/tests/out.zip

@@ -1,4 +1,5 @@
 import collections.abc
+import os.path
 
 import pytest
 
@@ -95,6 +96,17 @@ def test_zip_glob_star(m):
     outfiles = fs.glob("*")
     assert len(outfiles) == 1
 
+    fs = fsspec.filesystem("zip", fo="memory://out.zip", mode="w")
+    fs.mkdir("adir")
+    fs.pipe("adir/afile", b"data")
+    outfiles = fs.glob("*")
+    assert len(outfiles) == 1
+
+    fn = f"{os.path.dirname(os.path.abspath((__file__)))}/out.zip"
+    fs = fsspec.filesystem("zip", fo=fn, mode="r")
+    outfiles = fs.glob("*")
+    assert len(outfiles) == 1
+
 
 def test_append(m, tmpdir):
     fs = fsspec.filesystem("zip", fo="memory://out.zip", mode="w")

fsspec/implementations/tests/test_zip.py

@@ -87,14 +87,18 @@ def _get_dirs(self):
             # not read from the file.
             files = self.zip.infolist()
             self.dir_cache = {
-                dirname: {"name": dirname, "size": 0, "type": "directory"}
+                dirname.rstrip("/"): {
+                    "name": dirname.rstrip("/"),
+                    "size": 0,
+                    "type": "directory",
+                }
                 for dirname in self._all_dirnames(self.zip.namelist())
             }
             for z in files:
                 f = {s: getattr(z, s, None) for s in zipfile.ZipInfo.__slots__}
                 f.update(
                     {
-                        "name": z.filename,
+                        "name": z.filename.rstrip("/"),
                         "size": z.file_size,
                         "type": ("directory" if z.is_dir() else "file"),
                     }