bug: fail to read files with trailing whitespaces

[christer-eriksson]

Describe the bug

When using the read() method in opendal 0.54.0

opendal fail to read the content from the file

Steps to Reproduce

Create a file named "file with trailing white space " with some content in it

Use the read method to read the file content. It will fail with Error.

Expected Behavior

I expect to be able to read the content of a file even if the file name ends with a white space

Additional Context

The calls in this snippet shows that the normalize_path in deed trims the string from any leading or trailing white space.

pub async fn read(&self, path: &str) -> Result<Buffer> {
      self.read_options(path, options::ReadOptions::default())
          .await
  }

 pub async fn read_options(&self, path: &str, opts: options::ReadOptions) -> Result<Buffer> {
      let path = normalize_path(path);
      Self::read_inner(self.inner().clone(), path, opts).await
  }
pub fn normalize_path(path: &str) -> String {
  // - all whitespace has been trimmed.
  // - all leading `/` has been trimmed.
  let path = path.trim().trim_start_matches('/');

  // Fast line for empty path.
  if path.is_empty() {
      return "/".to_string();
  }

  let has_trailing = path.ends_with('/');

  let mut p = path
      .split('/')
      .filter(|v| !v.is_empty())
      .collect::<Vec<&str>>()
      .join("/");

  // Append trailing back if input path is endswith `/`.
  if has_trailing {
      p.push('/');
  }

  p
}

Are you willing to submit a PR to fix this bug?

• Yes, I would like to submit a PR.

[dosubot]

Hi @christer-eriksson! I'm Dosu and I’m helping the opendal team.

This behavior is caused by the normalize_path function, which trims all leading and trailing whitespace from file paths before performing operations. This is intentional and part of opendal's normalization rules, mainly for security reasons: it ensures all operations are scoped under the configured root_dir and helps prevent access outside this directory. As a result, files whose names end with whitespace can't be accessed, since the whitespace is stripped before the operation, altering the intended file name. You can see more about the rationale behind this design in this discussion.

Currently, there isn't a configuration option to disable this normalization. If you need to access files with trailing whitespace, you'll need to rename them to remove the whitespace or consider raising the need for more flexible path handling if this impacts your workflow.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}