Issue #438: Extended parsing compatibility of the FTP LIST command for Windows servers.

atollk · atollk · commit 19e4387cf9b2 · 2020-11-05T08:45:09.000+01:00
The parsing process now properly supports 24-hour time format, both with and without leading zeros.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -12,6 +12,8 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 - Missing `mode` attribute to `_MemoryFile` objects returned by `MemoryFS.openbin`.
 - Missing `readinto` method for `MemoryFS` and `FTPFS` file objects. Closes
   [#380](https://github.com/PyFilesystem/pyfilesystem2/issues/380).
+- Added compatibility if a Windows FTP server returns file information to the
+  `LIST` command with 24-hour times. Closes [#438](https://github.com/PyFilesystem/pyfilesystem2/issues/438).
 
 ### Changed
 
diff --git a/CONTRIBUTORS.md b/CONTRIBUTORS.md
@@ -13,3 +13,4 @@ Many thanks to the following developers for contributing to this project:
 - [Will McGugan](https://github.com/willmcgugan)
 - [Zmej Serow](https://github.com/zmej-serow)
 - [Morten Engelhardt Olsen](https://github.com/xoriath)
+- [Andreas Tollkötter](https://github.com/atollk)
diff --git a/fs/_ftp_parse.py b/fs/_ftp_parse.py
@@ -41,14 +41,17 @@
 RE_WINDOWSNT = re.compile(
     r"""
     ^
-    (?P<modified>.*?(AM|PM))
-    \s*
-    (?P<size>(<DIR>|\d*))
-    \s*
+    (?P<modified_date>\S+)
+    \s+
+    (?P<modified_time>\S+(AM|PM)?)
+    \s+
+    (?P<size>(<DIR>|\d+))
+    \s+
     (?P<name>.*)
     $
     """,
-    re.VERBOSE)
+    re.VERBOSE,
+)
 
 
 def get_decoders():
@@ -104,6 +107,10 @@ def _parse_time(t, formats):
     return epoch_time
 
 
+def _decode_linux_time(mtime):
+    return _parse_time(mtime, formats=["%b %d %Y", "%b %d %H:%M"])
+
+
 def decode_linux(line, match):
     perms, links, uid, gid, size, mtime, name = match.groups()
     is_link = perms.startswith("l")
@@ -114,7 +121,7 @@ def decode_linux(line, match):
         _link_name = _link_name.strip()
     permissions = Permissions.parse(perms[1:])
 
-    mtime_epoch = _parse_time(mtime, formats=["%b %d %Y", "%b %d %H:%M"])
+    mtime_epoch = _decode_linux_time(mtime)
 
     name = unicodedata.normalize("NFC", name)
 
@@ -138,12 +145,22 @@ def decode_linux(line, match):
     return raw_info
 
 
+def _decode_windowsnt_time(date, time):
+    while len(time.split(":")[0]) < 2:
+        time = "0" + time
+    return _parse_time(
+        date + " " + time, formats=["%d-%m-%y %I:%M%p", "%d-%m-%y %H:%M"]
+    )
+
+
 def decode_windowsnt(line, match):
     """
-    Decodes a Windows NT FTP LIST line like these two:
+    Decodes a Windows NT FTP LIST line like one of these:
 
     `11-02-18  02:12PM       <DIR>          images`
     `11-02-18  03:33PM                 9276 logo.gif`
+
+    Alternatively, the time (02:12PM) might also be present in 24-hour format (14:12).
     """
     is_dir = match.group("size") == "<DIR>"
 
@@ -161,7 +178,9 @@ def decode_windowsnt(line, match):
     if not is_dir:
         raw_info["details"]["size"] = int(match.group("size"))
 
-    modified = _parse_time(match.group("modified"), formats=["%d-%m-%y %I:%M%p"])
+    modified = _decode_windowsnt_time(
+        match.group("modified_date"), match.group("modified_time")
+    )
     if modified is not None:
         raw_info["details"]["modified"] = modified
 
diff --git a/tests/test_ftp_parse.py b/tests/test_ftp_parse.py
@@ -17,17 +17,18 @@ class TestFTPParse(unittest.TestCase):
     @mock.patch("time.localtime")
     def test_parse_time(self, mock_localtime):
         self.assertEqual(
-            ftp_parse._parse_time("JUL 05 1974", formats=["%b %d %Y"]),
-            142214400.0)
+            ftp_parse._parse_time("JUL 05 1974", formats=["%b %d %Y"]), 142214400.0
+        )
 
         mock_localtime.return_value = time2017
         self.assertEqual(
-            ftp_parse._parse_time("JUL 05 02:00", formats=["%b %d %H:%M"]),
-            1499220000.0)
+            ftp_parse._parse_time("JUL 05 02:00", formats=["%b %d %H:%M"]), 1499220000.0
+        )
 
         self.assertEqual(
             ftp_parse._parse_time("05-07-17  02:00AM", formats=["%d-%m-%y %I:%M%p"]),
-            1499220000.0)
+            1499220000.0,
+        )
 
         self.assertEqual(ftp_parse._parse_time("notadate", formats=["%b %d %Y"]), None)
 
@@ -164,39 +165,50 @@ def test_decode_linux(self, mock_localtime):
     def test_decode_windowsnt(self, mock_localtime):
         mock_localtime.return_value = time2017
         directory = """\
+unparsable line
 11-02-17  02:00AM       <DIR>          docs
 11-02-17  02:12PM       <DIR>          images
-11-02-17  02:12PM       <DIR>          AM to PM
+11-02-17 02:12PM <DIR> AM to PM
 11-02-17  03:33PM                 9276 logo.gif
+05-11-20   22:11  <DIR>       src
+11-02-17   01:23       1      12
+11-02-17    4:54                 0 icon.bmp
 """
         expected = [
             {
                 "basic": {"is_dir": True, "name": "docs"},
                 "details": {"modified": 1486778400.0, "type": 1},
-                "ftp": {
-                    "ls": "11-02-17  02:00AM       <DIR>          docs"
-                },
+                "ftp": {"ls": "11-02-17  02:00AM       <DIR>          docs"},
             },
             {
                 "basic": {"is_dir": True, "name": "images"},
                 "details": {"modified": 1486822320.0, "type": 1},
-                "ftp": {
-                    "ls": "11-02-17  02:12PM       <DIR>          images"
-                },
+                "ftp": {"ls": "11-02-17  02:12PM       <DIR>          images"},
             },
             {
                 "basic": {"is_dir": True, "name": "AM to PM"},
                 "details": {"modified": 1486822320.0, "type": 1},
-                "ftp": {
-                    "ls": "11-02-17  02:12PM       <DIR>          AM to PM"
-                },
+                "ftp": {"ls": "11-02-17 02:12PM <DIR> AM to PM"},
             },
             {
                 "basic": {"is_dir": False, "name": "logo.gif"},
                 "details": {"modified": 1486827180.0, "size": 9276, "type": 2},
-                "ftp": {
-                    "ls": "11-02-17  03:33PM                 9276 logo.gif"
-                },
+                "ftp": {"ls": "11-02-17  03:33PM                 9276 logo.gif"},
+            },
+            {
+                "basic": {"is_dir": True, "name": "src"},
+                "details": {"modified": 1604614260.0, "type": 1},
+                "ftp": {"ls": "05-11-20   22:11  <DIR>       src"},
+            },
+            {
+                "basic": {"is_dir": False, "name": "12"},
+                "details": {"modified": 1486776180.0, "size": 1, "type": 2},
+                "ftp": {"ls": "11-02-17   01:23       1      12"},
+            },
+            {
+                "basic": {"is_dir": False, "name": "icon.bmp"},
+                "details": {"modified": 1486788840.0, "size": 0, "type": 2},
+                "ftp": {"ls": "11-02-17    4:54                 0 icon.bmp"},
             },
         ]
 
