uri_parts ingest processor should not URLdecode path and query

Elasticsearch Version

8.15.0

Installed Plugins

No response

Java Version

bundled

OS Version

RockyLinux 9

Problem Description

URLencoded characters in query and path properties should not be decoded because it modifies the real URL :

decoding percent-encoded %xx in query it makes it impossible to properly decode query parameters, eg: %3d becoming = (equal)
decoding for instance %2f to / (slash) in path breaks the real path of the URL

Ones could argue decoding of the path is more suject to discussion... However, as a rule it should be possible to recreate the original URL by reassembling the parts resulting of the processor. Also for consistency, it is bad idea to have different reasoning for defferent parts or the URL. Currently the processor is mixing parsing and URLdecod'ing whereas its goal is to parse (extract parts). Decoding percent-encoded sequences in query may return illegal characters, hence making real (as in proper) decoding of query parameters impossible.

I believe getRaw functions (https://docs.oracle.com/javase/8/docs/api/java/net/URI.html) should be used in main/java/org/elasticsearch/ingest/common/UriPartsProcessor.java.

Sample 1

Input URL : http://www.acme.com/some/thing?a=123&b=x%26c%3dy Should give query : a=123&b=x%26c%3dy (which decode in a 123, b x&c=y) But currently gives query : a=123&b=x&c=y (which would decode in a 123, b x, c y)

Sample 2

Input URL : http://www.acme.com/some%2fthing Should give path : /some%2fthing Currently gives path : /some/thing

Steps to Reproduce

- uri_parts:
    if: 'ctx.url?.original != null'
    field: url.original
    target_field: url
    keep_original: true

With inputs :

{"url":{"original":"http://www.acme.com/some/thing?a=123&b=x%26c%3dy"}}
{"url":{"original":"http://www.acme.com/some%2fthing"}}

Logs (if relevant)

No response

elastic / elasticsearch