What is format preservation, and how can I disable it?

[jennydaman]

I see here that some element values are represented as PrimitiveValue::Strs even if their VR is something like "Decimal String".

dicom-rs/core/src/value/primitive.rs

Lines 226 to 230 in 41e87f3

	/// A sequence of strings.
	/// Used for AE, AS, PN, SH, CS, LO, UI and UC.
	/// Can also be used for IS, SS, DS, DA, DT and TM when decoding
	/// with format preservation.
	Strs(C<String>),

How can I read a DICOM file without format preservation?

[jennydaman]

I see that it's related to ValueReadStrategy

dicom-rs/parser/src/dataset/read.rs

Lines 89 to 124 in 41e87f3

	/// The value reading strategy for the data set reader.
	///
	/// It defines how the `PrimitiveValue`s in value tokens are constructed.
	#[derive(Debug, Default, Copy, Clone, Eq, Hash, PartialEq)]
	pub enum ValueReadStrategy {
	/// Textual values will be decoded according to their value representation.
	///
	/// Word-sized binary values are read according to
	/// the expected byte order.
	/// Dates, times, and date-times (DA, DT, TM) are parsed
	/// into their more specific variants,
	/// leading to parser failure if they are not valid DICOM.
	/// String numbers (IS, FD) are also converted into binary representations.
	/// For the case of floats, this may introduce precision errors.
	Interpreted,
	/// Values will be stored without decoding dates or textual numbers.
	///
	/// Word-sized binary values are read according to
	/// the expected byte order.
	/// Date-time values and numbers are kept in their original string
	/// representation as string objects.
	/// All text is still decoded into Rust string values,
	/// in accordance to the standard,
	/// unless its value representation is unknown to the decoder.
	///
	/// This is the default strategy.
	#[default]
	Preserved,
	/// All primitive values are fetched as raw byte buffers,
	/// without any form of decoding or interpretation.
	/// Not even byte order conversions are made.
	///
	/// This strategy is not recommended,
	/// as it makes the retrieval of important textual data more difficult.
	Raw,
	}

Is there any way I can pass a strategy to dicom::object::open_file?

[Enet4]

Yes, "format preservation" refers to the parser's value reading strategy that preserves textual values (numbers and dates) as strings. It is currently only available as an option in the dicom_parser data set reader API, as the use of any other strategy in the object API could bring other issues in the loading process. Namely, the strategy that reads everything as bytes has never really been tried here, and the strategy to convert textual numbers and dates would often choke on non-compliant or unsupported DICOM data.

It would be great if you could share the use case for changing the value reading strategy. It is generally sufficient to use one of the various to_* methods in the DICOM value types, which convert the underlying representation on demand into another, more usable type.