Adds lazy reader support for blobs #629
Conversation
| pub struct BytesRef<'data> { | ||
| data: Cow<'data, [u8]>, | ||
| } |
There was a problem hiding this comment.
🗺️ When there was only a binary reader, methods that read blobs could always return a &[u8]--a slice of the input buffer. Now that we also have a text reader, we need to accommodate base64-encoded blobs, which always require a new Vec to be allocated to hold the decoded data. BytesRef can hold either a borrowed &[u8] or an owned Vec<u8>, allowing it to be used in either situation.
This type is analogous to StrRef and SymbolRef but for blobs.
| @@ -18,7 +19,7 @@ pub enum RawValueRef<'data, D: LazyDecoder<'data>> { | |||
| Timestamp(Timestamp), | |||
| String(StrRef<'data>), | |||
| Symbol(RawSymbolTokenRef<'data>), | |||
| Blob(&'data [u8]), | |||
| Blob(BytesRef<'data>), | |||
There was a problem hiding this comment.
🗺️ RawValueRef now returns a BytesRef instead of a &[u8] so it has the option to allocate a Vec<u8> when the input encoding is base64 text. The binary reader can still return a slice of the input buffer.
| @@ -23,7 +24,7 @@ pub enum ValueRef<'top, 'data, D: LazyDecoder<'data>> { | |||
| Timestamp(Timestamp), | |||
| String(StrRef<'data>), | |||
| Symbol(SymbolRef<'top>), | |||
| Blob(&'data [u8]), | |||
| Blob(BytesRef<'data>), | |||
There was a problem hiding this comment.
🗺️ Like RawValueRef, ValueRef now returns a BytesRef instead of a &[u8] so it has the option to allocate a Vec<u8> when the input encoding is base64 text. The binary reader can still return a slice of the input buffer.
Codecov ReportPatch coverage is
📢 Thoughts on this report? Let us know!. |
Builds on outstanding PRs #612, #613, #614, #616, #617, #619, #620, #621, #622, #623, #627, and #628.
Adds lazy reader support for blobs.
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.