Skip to content
Visit Rules on GitHub
Set theme to dark (⇧+D)

How URL normalization works

Cloudflare URL normalization is similar to rfc3986, and modifies separators, encoded elements, and literal bytes in incoming URLs as follows:

  • The following unreserved characters are percent decoded:
    • Alphabetical characters: a-z, A-Z (decoded from %41-%5A and %61-%7A)
    • Digit characters: 0-9 (decoded from %30-%39)
    • hyphen - (%2D), period . (%2E), underscore _ (%5F), and tilde ~ (%7E)
  • These reserved characters are not encoded or decoded: : / ? # [ ] @ ! $ & ' ( ) * + , ; =
  • Other characters, for example literal byte values, are percent encoded.
  • Percent encoded representations are converted to upper case.
  • URL paths are normalized according to the Remove Dot Segments protocol. Deviations from this protocol include modifications to the following separators:
    • \ becomes /
    • // becomes /

Consider a Firewall Rule that blocks requests whose URLs match The rule would not block a request containing an encoded element Normalizing incoming URLs at the edge helps simplify Cloudflare Firewall Rules expressions that use URLs.

The following table shows some examples of URL normalization:

URLNormalized URL\

The exact URL normalization performed by Cloudflare varies according to the configured settings. For more information, refer to URL normalization settings.