feat: add filter parameter to file_browser for flexible filtering by by22Jy · Pull Request #8751 · marimo-team/marimo

by22Jy · 2026-03-18T04:06:04Z

Summary

Adds a filter parameter to mo.ui.file_browser() for more flexible file filtering beyond simple file extensions.

Changes

Added filter parameter to file_browser.__init__() that accepts:
- Regex pattern string (matched against file.name using re.match)
- Callable that takes a Path and returns bool
Added _should_include_file() method to handle filtering logic
Updated _list_directory() to use the new unified filtering method
Added examples to docstring demonstrating regex and callback filters
Added import re at the top of the file

Implementation Details

Filter Priority: filter > filetypes

When filter is provided, it takes precedence over filetypes. If filter is None, the existing filetypes behavior is used (backward compatible).

Error Handling:

Invalid regex patterns: logged as warning, file excluded
Callback exceptions: logged as warning, file excluded

Backward Compatibility: ✅

filetypes parameter retained
Default behavior unchanged (filter=None)

Examples

Regex Filter

# Only show Python files
file_browser = mo.ui.file_browser(
    initial_path=Path("."),
    filter=r".*\.py$"
)

# Only show files starting with "test_"
file_browser = mo.ui.file_browser(
    initial_path=Path("."),
    filter=r"^test_.*"
)

Callback Filter

# Only show files larger than 1MB
def large_files(path: Path) -> bool:
    return path.is_file() and path.stat().st_size > 1_000_000

file_browser = mo.ui.file_browser(
    initial_path=Path("."),
    filter=large_files
)

Testing

This PR includes the core functionality. Tests should cover:

Regex filter with valid patterns
Regex filter with invalid patterns (error handling)
Callback filter with normal functions
Callback filter with exceptions (error handling)
Filter priority over filetypes
Backward compatibility (filetypes still works)

Notes

Does not modify frontend code (all filtering is backend-side)
Recursive directory checking (_has_files_recursive) still uses filetypes only, not the filter parameter (for simplicity and consistency)
Uses re.match() for regex matching (not re.search()), matching against the full filename

Pull request overview

Adds a new backend-side filter parameter to mo.ui.file_browser() to support flexible file inclusion rules (regex or callback) beyond the existing filetypes extension filtering.

Changes:

Introduces a filter parameter to file_browser.__init__ (regex string via re.match or Callable[[Path], bool]).
Adds _should_include_file() and updates _list_directory() to use unified filtering logic.
Expands the file_browser docstring with usage examples for regex and callback filtering.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-20T17:49:18Z

+    def _should_include_file(self, file: Path, is_directory: bool) -> bool:
+        """Determine if a file should be included based on filter/filetypes.
+
+        Args:
+            file: The file path to check
+            is_directory: Whether the path is a directory
+
+        Returns:
+            bool: True if the file should be included, False otherwise
+        """
+        # Priority: filter > filetypes
+        if self._filter is not None:
+            # Apply custom filter
+            if isinstance(self._filter, str):
+                # Regex filter (match against file.name)
+                try:
+                    return bool(re.match(self._filter, file.name))
+                except re.error:
+                    # Invalid regex pattern, treat as no match
+                    LOGGER.warning(
+                        f"Invalid regex pattern in filter: {self._filter}"
+                    )
+                    return False
+            else:
+                # Callback filter
+                try:
+                    return self._filter(file)
+                except Exception as e:
+                    # Callback raised exception, log and exclude file
+                    LOGGER.warning(
+                        f"Filter callback raised exception for {file}: {e}"
+                    )
+                    return False
+
+        # Fall back to filetypes
+        if self._filetypes and not is_directory:
+            return file.suffix.lower() in self._filetypes
+
+        return True


New filter behavior is not covered by the existing tests/_plugins/ui/_impl/test_file_browser.py suite (which currently focuses on filetypes and ignore_empty_dirs). Adding tests for regex filters (valid/invalid), callback filters (normal/exception), and filter precedence over filetypes would help prevent regressions—especially around directory visibility/navigation when filter is set.

Copilot · 2026-03-20T17:49:19Z

+        # Priority: filter > filetypes
+        if self._filter is not None:
+            # Apply custom filter
+            if isinstance(self._filter, str):
+                # Regex filter (match against file.name)
+                try:
+                    return bool(re.match(self._filter, file.name))
+                except re.error:


When filter is provided, _should_include_file applies it to directories too (no is_directory guard). This can hide folders and prevent users from navigating into them, unlike the existing filetypes behavior which only filters non-directories. Consider always including directories (or only applying filter when not is_directory, unless selection_mode intentionally wants directory-name filtering).

Copilot · 2026-03-20T17:49:19Z

+                try:
+                    return bool(re.match(self._filter, file.name))
+                except re.error:
+                    # Invalid regex pattern, treat as no match
+                    LOGGER.warning(
+                        f"Invalid regex pattern in filter: {self._filter}"
+                    )
+                    return False


Regex filtering currently calls re.match(self._filter, file.name) for every entry, and an invalid pattern will raise/log on every file, potentially spamming logs and adding avoidable overhead. Consider compiling/validating the regex once in __init__ (store a compiled re.Pattern or disable filtering after one warning) and then using pattern.match(...) during listing.

Suggested change

try:

return bool(re.match(self._filter, file.name))

except re.error:

# Invalid regex pattern, treat as no match

LOGGER.warning(

f"Invalid regex pattern in filter: {self._filter}"

)

return False

# Compile the pattern once and cache it; if invalid, log once.

invalid = getattr(self, "_filter_invalid", False)

if invalid:

# Previously determined to be invalid; treat as no match.

return False

pattern = getattr(self, "_compiled_filter", None)

if pattern is None:

try:

pattern = re.compile(self._filter)

except re.error:

# Invalid regex pattern, treat as no match and log once.

LOGGER.warning(

f"Invalid regex pattern in filter: {self._filter}"

)

setattr(self, "_filter_invalid", True)

return False

else:

setattr(self, "_compiled_filter", pattern)

return bool(pattern.match(file.name))

Copilot · 2026-03-20T17:49:19Z

+            else:
+                # Callback filter
+                try:
+                    return self._filter(file)
+                except Exception as e:
+                    # Callback raised exception, log and exclude file
+                    LOGGER.warning(
+                        f"Filter callback raised exception for {file}: {e}"
+                    )
+                    return False


If filter is neither a str nor a callable, the code falls into the callback branch and will raise TypeError at call time, which gets caught and logged as a warning and silently excludes the file. Since filter is part of the public API, it would be better to validate the type in __init__ and raise a ValueError for unsupported values so misconfiguration is detected early.

dmadisetti · 2026-04-07T22:25:05Z

@by22Jy Were you still interested in contributing this? Happy to give a deeper look once you sign the CLA

feat: add filter parameter to file_browser for flexible filtering

f6d8f6e

vercel bot deployed to Preview March 18, 2026 04:07 View deployment

[pre-commit.ci] auto fixes from pre-commit.com hooks

a82079c

for more information, see https://pre-commit.ci

vercel bot deployed to Preview March 18, 2026 04:09 View deployment

mscolnick requested review from Copilot and dmadisetti March 20, 2026 17:46

Copilot started reviewing on behalf of mscolnick March 20, 2026 17:46 View session

Copilot AI reviewed Mar 20, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: add filter parameter to file_browser for flexible filtering#8751

feat: add filter parameter to file_browser for flexible filtering#8751
by22Jy wants to merge 2 commits intomarimo-team:mainfrom
by22Jy:feat-file-browser-filter

by22Jy commented Mar 18, 2026

Uh oh!

vercel bot commented Mar 18, 2026 •

edited

Loading

Uh oh!

github-actions bot commented Mar 18, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

dmadisetti commented Apr 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

by22Jy commented Mar 18, 2026

Summary

Changes

Implementation Details

Examples

Regex Filter

Callback Filter

Testing

Notes

Related

Uh oh!

vercel bot commented Mar 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Mar 18, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

dmadisetti commented Apr 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

vercel bot commented Mar 18, 2026 •

edited

Loading