Merge remote-tracking branch 'origin/pr/5236'

qutebrowser/browser/downloads.py

@@ -230,9 +230,16 @@ def suggested_fn_from_title(url_path, title=None):
 
     suggested_fn = None  # type: typing.Optional[str]
     if ext.lower() in ext_whitelist and title:
-        suggested_fn = utils.sanitize_filename(title)
+        suggested_fn = utils.sanitize_filename(
+            title,
+            max_bytes=(255 - len("(123).download"))
+        )
         if not suggested_fn.lower().endswith((".html", ".htm")):
             suggested_fn += ".html"
+            suggested_fn = utils.sanitize_filename(
+                suggested_fn,
+                max_bytes=(255 - len("(123).download"))
+            )
 
     return suggested_fn
 

qutebrowser/utils/utils.py

@@ -520,7 +520,8 @@ def force_encoding(text: str, encoding: str) -> str:
 
 
 def sanitize_filename(name: str,
-                      replacement: typing.Optional[str] = '_') -> str:
+                      replacement: typing.Optional[str] = '_',
+                      max_bytes: typing.Optional[int] = None) -> str:
     """Replace invalid filename characters.
 
     Note: This should be used for the basename, as it also removes the path
@@ -529,6 +530,7 @@ def sanitize_filename(name: str,
     Args:
         name: The filename.
         replacement: The replacement character (or None).
+        max_bytes: The biggest length in bytes allowed (or None, which is ∞).
     """
     if replacement is None:
         replacement = ''
@@ -550,6 +552,34 @@ def sanitize_filename(name: str,
 
     for bad_char in bad_chars:
         name = name.replace(bad_char, replacement)
+
+    if max_bytes:
+        # Truncate the filename if it's too long.
+        root, ext = os.path.splitext(name)
+        root = root[:max_bytes - len(ext)]
+        excess = len(os.fsencode(root + ext)) - max_bytes
+
+        while excess > 0 and root:
+            # Max 4 bytes per character is assumed.
+            # Integer division floors to -∞, not to 0.
+            root = root[:(-excess // 4)]
+            excess = len(os.fsencode(root + ext)) - max_bytes
+
+        if not root:
+            # Trimming the root is not enough. We must trim the extension.
+            # We leave one character in the root, so that the filename
+            # doesn't start with a dot, which makes the file hidden.
+            root = name[0]
+            excess = len(os.fsencode(root + ext)) - max_bytes
+            while excess > 0 and ext:
+                ext = ext[:(-excess // 4)]
+                excess = len(os.fsencode(root + ext)) - max_bytes
+
+            if not ext:
+                raise ValueError("max_bytes is too small")
+
+        name = root + ext
+
     return name
 
 

tests/unit/utils/test_utils.py

@@ -644,6 +644,13 @@ def test_sanitize_filename_empty_replacement():
     assert utils.sanitize_filename(name, replacement=None) == 'Bad File'
 
 
+@hypothesis.given(filename=strategies.text(min_size=100),
+                  max_bytes=strategies.integers(min_value=50))
+def test_sanitize_filename_invariants(filename, max_bytes):
+    sanitized = utils.sanitize_filename(filename, max_bytes=max_bytes)
+    assert len(os.fsencode(sanitized)) <= max_bytes
+
+
 class TestGetSetClipboard:
 
     @pytest.fixture(autouse=True)