WebDAV: File paths with "#" characters are truncated

[nextgovitAS]

Bug Report

Q	A
Flysystem Version	3.29.0
Adapter Name	webdav
Adapter version	3.29.0

Description

When using league/flysystem-webdav, file paths that include the # character are truncated, causing unexpected behavior. This issue arises because the library applies parse_url() to paths, which interprets # as a fragment identifier and discards everything following it. For example, a file named file#example.txt is processed as file, resulting in data loss and errors when attempting to interact with files that contain # in their names.

Steps to Reproduce

1. Create a file in WebDAV storage with a # in its name, e.g., file#example.txt.
2. Attempt to list the contents of the directory using the listContents function.
3. Observe that the file name is truncated at the #, leading to failed operations.

Expected Behavior

The library should treat # as a valid character within the file path and not interpret it as a URL fragment identifier. The entire file name, including the #, should be preserved.

Actual Behavior

The library truncates the path at the # character due to parse_url() processing, which results in incorrect file paths.

Suggested Fix

I modified the code in WebDAVAdapter as follows to address this issue:

Old:

public function listContents(string $path, bool $deep): iterable
{
  [...]
  $path = (string) parse_url(rawurldecode($path), PHP_URL_PATH);
  [...]

New:

public function listContents(string $path, bool $deep): iterable
{
  [...]
  $rawUrl = rawurldecode($path);
  $path = (string) parse_url($rawUrl, PHP_URL_PATH);
  $fragment = parse_url($rawUrl, PHP_URL_FRAGMENT);
  if (is_string($fragment)) {
    $path .= '#' . $fragment;
  }
  [...]

With this change, paths are correctly reconstructed to include the # fragment, preserving the original file names.