fix(download): parse quoted and extended response filenames

plugins/tools/_download_runner.py

@@ -5,6 +5,7 @@
 import argparse
 import contextlib
 import email.message
+import email.utils
 import fcntl
 import hashlib
 import http.client
@@ -56,10 +57,6 @@
     "image/",
     "text/",
 )
-_CONTENT_DISPOSITION_NAME_RE = re.compile(
-    r"""filename\*?=(?:UTF-8''|["']?)([^;"']+)""",
-    re.IGNORECASE,
-)
 
 
 class DownloadError(RuntimeError):
@@ -270,11 +267,21 @@ def _preflight(
 
 
 def _header_filename(headers: email.message.Message) -> str:
-    disposition = headers.get("Content-Disposition") or ""
-    match = _CONTENT_DISPOSITION_NAME_RE.search(disposition)
-    if not match:
-        return ""
-    return urllib.parse.unquote(match.group(1)).strip()
+    extended = ""
+    fallback = ""
+    for name, value in headers.get_params(header="content-disposition", failobj=[]):
+        if name != "filename":
+            continue
+        # The parser represents filename* as an RFC 2231 tuple. Prefer it
+        # over the ordinary filename, regardless of parameter order, but only
+        # when it decodes to something usable: an empty filename* must not
+        # discard a valid ordinary filename.
+        if isinstance(value, tuple):
+            if not extended:
+                extended = email.utils.collapse_rfc2231_value(value).strip()
+        elif not fallback:
+            fallback = value.strip()
+    return extended or fallback
 
 
 def _safe_filename(candidate: str) -> str:

tests/test_download_filenames.py

@@ -0,0 +1,66 @@
+from __future__ import annotations
+
+from email.message import Message
+
+import pytest
+
+from plugins.tools._download_runner import _header_filename, _resolve_basename
+
+
+@pytest.mark.parametrize(
+    ("disposition", "expected"),
+    [
+        ('attachment; filename="report.pdf"', "report.pdf"),
+        ('attachment; filename="report;final.pdf"', "report;final.pdf"),
+        ('attachment; filename="O\'Reilly.pdf"', "O'Reilly.pdf"),
+        ('attachment; filename = "report.pdf"', "report.pdf"),
+        ('attachment; filename="report%20draft.pdf"', "report%20draft.pdf"),
+        ("attachment; filename*=UTF-8''report%20final.pdf", "report final.pdf"),
+        ("attachment; filename*=UTF-8'en'report%20final.pdf", "report final.pdf"),
+        ("attachment; filename*=UTF-8''caf%C3%A9.pdf", "café.pdf"),
+        ("attachment; filename*=ISO-8859-1''caf%E9.pdf", "café.pdf"),
+        (
+            'attachment; filename="fallback.txt"; filename*=UTF-8\'\'actual.pdf',
+            "actual.pdf",
+        ),
+        (
+            'attachment; filename*=UTF-8\'\'actual.pdf; filename="fallback.txt"',
+            "actual.pdf",
+        ),
+        ('attachment; filename="paper.pdf"; filename*=UTF-8\'\'', "paper.pdf"),
+        ("attachment; filename*=UTF-8''; filename=\"paper.pdf\"", "paper.pdf"),
+        ("attachment; filename*=UTF-8''%20%20; filename=\"paper.pdf\"", "paper.pdf"),
+        ("attachment", ""),
+    ],
+)
+def test_response_filename(disposition: str, expected: str) -> None:
+    headers = Message()
+    headers["Content-Disposition"] = disposition
+
+    assert _header_filename(headers) == expected
+
+
+@pytest.mark.parametrize(
+    ("path_arg", "disposition", "expected"),
+    [
+        ("", 'attachment; filename="report;final.pdf"', "report_final.pdf"),
+        ("", 'attachment; filename="O\'Reilly.pdf"', "O_Reilly.pdf"),
+        (
+            "",
+            'attachment; filename="fallback.txt"; filename*=UTF-8\'\'actual.pdf',
+            "actual.pdf",
+        ),
+        ("", 'attachment; filename="../../outside.pdf"', "outside.pdf"),
+        ("", "attachment; filename*=UTF-8''..%2Foutside.pdf", "outside.pdf"),
+        ("", "attachment", "url name.pdf"),
+        ("", 'attachment; filename="paper.pdf"; filename*=UTF-8\'\'', "paper.pdf"),
+        ("", "", "url name.pdf"),
+        ("chosen.pdf", 'attachment; filename="report;final.pdf"', "chosen.pdf"),
+    ],
+)
+def test_download_basename(path_arg: str, disposition: str, expected: str) -> None:
+    headers = Message()
+    if disposition:
+        headers["Content-Disposition"] = disposition
+
+    assert _resolve_basename(path_arg, "https://example.com/url%20name.pdf", headers) == expected