[FLINK-38822] Extend URL_DECODE with optional recursive decoding

[featzhang]

Background

Apache Flink SQL currently provides URL_DECODE(str) to decode strings in application/x-www-form-urlencoded format (introduced in FLINK-34108). In practice, log data, tracking data, or external system inputs may contain multi-level URL-encoded values, for example strings encoded multiple times by upstream systems or intermediate components such as redirects or proxies.

The existing function performs only a single decoding pass, which requires users to manually apply the function multiple times to fully decode such values.

Problem

Single-pass decoding is insufficient for handling multi-level URL-encoded values.

Repeated manual application of URL_DECODE reduces SQL readability and increases the risk of errors.

Solution

Extend URL decoding support by introducing an additional built-in function that accepts a boolean flag to control recursive decoding.

Signature

URL_DECODE(str)                     -- existing behavior (single-pass decoding)
URL_DECODE_RECURSIVE(str, recursive)

Behavior

• When recursive is FALSE or NULL, the function performs a single decoding pass.

• When recursive is TRUE, the function repeatedly applies URL decoding until:

  • the decoded result no longer changes, or
  • a maximum number of iterations is reached.

If the input value is NULL, the function returns NULL.

If a decoding error occurs:

• in non-recursive mode, the function returns NULL;
• in recursive mode, the function returns the last successfully decoded result, or NULL if no decoding step succeeds.

Examples

-- Single-pass decoding
SELECT URL_DECODE('%252Fpath%252Fto%252Fresource');
-- Result: '%2Fpath%2Fto%2Fresource'

-- Recursive decoding
SELECT URL_DECODE_RECURSIVE('%252Fpath%252Fto%252Fresource', TRUE);
-- Result: '/path/to/resource'

Compatibility

This change is fully backward-compatible. Existing queries using URL_DECODE(str) are unaffected.

Notes

Recursive decoding terminates when the decoded value stabilizes or when the maximum iteration limit is reached. This functionality is useful for data cleansing, normalization, and processing multi-level URL-encoded inputs.

[flinkbot]

CI report:

• c1e5dbe Azure: FAILURE

Bot commands

The @flinkbot bot supports the following commands:

• @flinkbot run azure re-run the last Azure build