Documents ids with slashes cause issues in high-level client

[javanna]

This issue was originated from #36988 . When using the high-level REST client to get/index/update documents, slashes in ids can cause problems. This has to do with the logic that encodes each url part separately and then puts the result together in RequestConverters. The following test fails while I think it should succeed. What I think happens at the moment is that the document gets indexed with a different id than the one provided.

public void testIdWithSlashes() throws IOException {
       IndexRequest indexRequest = new IndexRequest("dev_product");
       indexRequest.id("/sku/test.com/48923AQ3");
       indexRequest.source("field", "value");
       IndexResponse indexResponse = highLevelClient().index(indexRequest, RequestOptions.DEFAULT);

       assertEquals("/sku/test.com/48923AQ3", indexResponse.getId());

       GetResponse getResponse = highLevelClient()
           .get(new GetRequest("dev_product", "/sku/test.com/48923AQ3"), RequestOptions.DEFAULT);
       assertTrue(getResponse.isExists());
   }

[elasticmachine]

Pinging @elastic/es-core-features

[martijnvg]

This is how the id is received when the index request is sent via the HLRC: test.com/48923AQ3
The /sku/ is missing.

I think the problem is in the EndpointBuilder#encodePart(...) method in RequestCoverters.java line 1106. I think in the case that the pathPart starts with a / then it should not add extra leading slash.

This is a more targeted test:

EndpointBuilder endpointBuilder = new EndpointBuilder().addPathPart("/sku/test.com/48923AQ3");
assertEquals("/%2Fsku%2Ftest.com%2F48923AQ3", endpointBuilder.build());

[iverase]

It seems this issue has been fixed in #34436 so it can be closed? the examples given here pass successfully.