encode_uri/1 should html encode with uppercase to handle special chars such as +

[onno-vos-dev]

Description:

encode_uri/1 should html encode with uppercase to handle special chars such as +

• Preserve old behavior for aws_util:base16/1

This is essentially the same code as can be seen in aws_signature: aws_signature_utils.erl#L44

Certain special characters such as + would still lead to SignatureDoesNotMatch errors even after #73 since the signature would uppercase encoded characters whereas the encoded uri would not. Leading to SignatureDoesNotMatch errors since the signature would be signed with an uppercase html encoded char whereas the url that would be called would have the lowercased version.

1> s3_wrapper:write(Bucket, <<"file(+1)">>, <<>>).
ok
2> s3_wrapper:read(Bucket, <<"file(+1)">>).
{ok,<<>>}
3> s3_wrapper:list_objects(Bucket, <<"file">>).
[<<"file(+1)">>]
4> s3_wrapper:delete(Bucket, <<"file(+1)">>).
ok
5> s3_wrapper:write(Bucket, <<"file%#&(+1)">>, <<>>).
ok
6> s3_wrapper:read(Bucket, <<"file%#&(+1)">>).
{ok,<<>>}
7> s3_wrapper:delete(Bucket, <<"file%#&(+1)">>).
ok

Added a unit test with the most screwed up filename possible:

1> s3_wrapper:write(Bucket, <<"file_!-_.(*)&=;:+ ,?{^}%]>[~<#`|.content">>, <<"content">>).
ok
2> s3_wrapper:read(Bucket, <<"file_!-_.(*)&=;:+ ,?{^}%]>[~<#`|.content">>).
{ok,<<"content">>}
3> s3_wrapper:list_objects(Bucket, <<"file">>).
[<<"file_!-_.(*)&=;:+ ,?{^}%]>[~<#`|.content">>]
4> s3_wrapper:delete(Bucket, <<"file_!-_.(*)&=;:+ ,?{^}%]>[~<#`|.content">>).
ok

[onno-vos-dev]

@philss Sorry for the mess but it's really annoying to have all tests towards S3Mock from Adobe pass whereas things end up exploding when moving to my staging environment 😢 Hence, I tested this in staging towards AWS itself and now every bit of weirdness seems to be resolved. 🤞

[danfilip]

here's the link to the object key names documentation: https://docs.aws.amazon.com/AmazonS3/latest/userguide/object-keys.html

[philss]

@onno-vos-dev no problem, and thanks for the PR! 💜