dcmnorm-encoding
dcmnorm's own fork of the DICOM encoding/decoding primitives
(TransferSyntax, Codec, PixelDataReader/PixelDataWriter,
EncodeOptions, plus the Implicit/Explicit VR Little/Big Endian dataset
encoders and decoders) that dcmnorm-transcode, dcmnorm-parser, and
dcmnorm-object build on.
Forked from dicom-rs's dicom-encoding
0.9.1 as a mechanical, byte-for-byte transcription. Still carries the
dataset-tokenizer/encoder machinery (decode/, encode/, text.rs) in
full, not trimmed to just the adapters.rs/transfer_syntax/ surface
dcmnorm's own code calls directly — dcmnorm-parser/dcmnorm-object
depend on that full surface. Trimming it to a thin, dcmnorm-scoped API is
a deliberate, not-yet-done follow-up.
Originally kept the package/lib name dicom-encoding/dicom_encoding
(patched in via [patch.crates-io]) because dicom-object and dicom-ul
both still depended on it. Fully renamed to dcmnorm-encoding/
dcmnorm_encoding once both were replaced by dcmnorm-object and
dcmnorm-dimse and nothing in the dependency graph needed the original name
anymore.
Upstream changes ported since the fork
Selectively ported from dicom-rs after 0.9.1 (each commit message names the upstream commit):
encode/explicit_{le,be}.rs: reject short-form VR values over 65535 bytes (upstream bff112e6) instead of truncating the length field.adapters.rs: a correct, cumulative Basic Offset Table from the defaultPixelDataWriter::encode(upstream 59ab4edb), which here also counts odd-length fragments' padding byte.decode/adaptive_le.rs:AdaptiveVRLittleEndianDecoder(upstream e5e2a99a), for files that declare Explicit VR Little Endian but write Implicit VR. dcmnorm decides a probed long-form VR that disagrees with the dictionary from its reserved bytes, so explicit files written as all-UNaren't misdetected.