Do more complete normalization of URIs & headers #11
Conversation
| @@ -288,6 +300,7 @@ function getCanonicalQueryString(url) { | |||
| function getCanonicalHeaders(headers) { | |||
| var aggregatedHeaders = new Array(); | |||
| for (var i=0; i<headers.length; i++) { | |||
| headers[i].value = headers[i].value.trim().replace(/ +/g, ' '); | |||
There was a problem hiding this comment.
This addresses AWS's rule that consecutive spaces in header values should be collapsed to one space.
|
Thanks for the pull request - sorry I've broken it by merging an earlier one - please hold off on fixing it for now as I'm going to write some tests which may break it further. |
|
Thanks @david-poirier-csn , keep me posted. |
|
I need to make another change anyway; I just realized my canonicalization code leaves off the trailing slash in the request path if there was one, which AWS also does not like. |
|
I hadn't actually looked at the other pull request; I just realized that these 2 PRs actually do literally conflict, as they deal with essentially the same issue. With all due respect, I think mine is more comprehensive and simpler. |
steven-collins-omega
changed the title
This is an attempt to implement more of AWS's canonicalization rules that are part of their signing algorithm. In particular (among other things) it addresses some specific problems I've run across:
/foo//bar(needs to be canonicalized to/foo/bar)/_cat/shards?v(needs to be/_cat/shards?v=)/_plugin/kibana/api/kibana/settings/discover:aggs:terms:size(needs to be/_plugin/kibana/api/kibana/settings/discover%3Aaggs%3Aterms%3Asize)