How to assign default values when getting data elements from an object?

[Enet4]

Originally posted by @alanwilter in #446 (comment)

BTW, I'm a python guy and sorry if I'm using the wrong channel (if you know so, please tell me), but how do I do to change extract_dicom_data() function so if element is not found (for PatientID, ImageLaterality, etc.) I still want to assign a empty string value in the end. For ImageLaterality actually, I search for Laterality as well and only defaulting to "" if neither were found (this is how I do with pydicom).
Now that I put my code to run I'm seeing lots of No such data element ImageLaterality (with tag (0020,0062)) but I still want to fill my output CSV files with empty results if needed.

use dicom::dictionary_std::tags;
use dicom::object::OpenFileOptions;
...
fn extract_dicom_data(path: &Path) -> Result<DicomData, Box<dyn std::error::Error>> {
    let obj = OpenFileOptions::new()
        .read_until(tags::PIXEL_DATA)
        .open_file(path)?;

    let patient_id = obj.element_by_name("PatientID")?.to_str()?;
    let image_laterality = obj.element_by_name("ImageLaterality")?.to_str()?;
    let patient_sex = obj.element_by_name("PatientSex")?.to_str()?;
    let patient_dob = obj.element_by_name("PatientBirthDate")?.to_str()?;
    let content_date = obj.element_by_name("ContentDate")?.to_str()?;
    let modality = obj.element_by_name("Modality")?.to_str()?;
    let manufacturer = obj.element_by_name("Manufacturer")?.to_str()?;

    let formatted_patient_dob = format_dicom_date(&patient_dob)?;
    let formatted_content_date = format_dicom_date(&content_date)?;

    let file_path = path
        .canonicalize()?
        .to_str()
        .ok_or("Invalid file path")?
        .to_string();

    Ok(DicomData {
        patient_id: patient_id.to_string(),
        image_laterality: image_laterality.to_string(),
        patient_sex: patient_sex.to_string(),
        patient_dob: formatted_patient_dob,
        content_date: formatted_content_date,
        modality: modality.to_string(),
        manufacturer: manufacturer.to_string(),
        file_path,
    })
}

[Enet4]

Prefer using element_opt, so that you can handle missing elements more effectively.

use dicom_dictionary_std::tags;

let image_laterality = if let Some(elem) = obj.element_opt(tags::IMAGE_LATERALITY)? {
    elem.to_str()?
} else {
    "".into()
};

Note also the preference in the use of methods element, element_opt, or get, rather than element_by_name. The latter requires a data element dictionary lookup to resolve the keyword to a tag, which is not necessary because we know the attributes to retrieve at compile time.

[alanwilter]

Many thanks for that, this is the perfect place.
I did look into the docs but even that I'm learning how to fetch the right info.
For now, many of my dcms are failing with premature end of input.

I'm trying to process in parallel, though I don't see any speedup TBH, and I thought this could be the reason for the premature end of input but it is not AFAI can say.

Some details of my code:

fn main() -> Result<(), Box<dyn std::error::Error>> {
...
    // Process DICOM files in parallel
    let results: Vec<_> = dicom_files
        .par_chunks(num_jobs)
        .map(|chunk| process_dicom_files(chunk))
        .flatten()
        .collect();
...
    Ok(())
}
fn process_dicom_files(paths: &[PathBuf]) -> Vec<DicomData> {
    paths
        .iter()
        .filter_map(|path| match extract_dicom_data(path) {
            Ok(data) => Some(data),
            Err(e) => {
                eprintln!("Error processing {:?}: {}", path, e);
                None
            }
        })
        .collect()
}

I've updated extract_dicom_data() to use your suggested obj.element_opt().
Note that I do use:

fn extract_dicom_data(path: &Path) -> Result<DicomData, Box<dyn std::error::Error>> {
    let obj = OpenFileOptions::new()
        .read_until(tags::PIXEL_DATA)
        .open_file(path)?;

To avoid reading PIXEL_DATA as it makes a huge difference in performance. I thought it could be some dcms not having PIXEL_DATA, but that's not the cause either AFAI see. I used dicom-dump for some of the dcms that failed and I can clearly see the data looks all fine (BTW, my python code works fine with them).

I use vscode and I will look into how to use debug for rust projects.