[Copilot]

DICOM queries previously required one explicit dicom_query block per search (date range, study description, etc.), with no way to target a known list of subjects while excluding everything else — and PACS can't be queried for multiple subjects in a single request.

Query expansion

• Added expand_search_specs() in workflow/lib/query_filter.py: detects a select key on a search spec and expands it into one explicit per-subject dicom_query, merging shared filters with the subject ID under patient_id (configurable via subject_query_field).
• query_dicoms() now runs expansion first, so each subject is queried individually and results are concatenated as before — downstream behavior is unchanged.

Backward compatibility

• Specs without select pass through untouched; select-list specs can be freely mixed with legacy explicit-query specs in the same search_specs list.

Validation

• Raises ValueError with clear messages for missing/non-list/empty select, and for blank or non-string select entries.

Docs & tests

• Updated README.md and config/config.yml with an example of the new mode.
• Added tests covering legacy mode, expansion, select_from_field override, mixed specs, and all validation error paths.

Example config:

search_specs:
  - dicom_query:
      study_description: YourStudy^*
      study_date: 20230101-
    select:
      - SUBJ001
      - SUBJ002
      - SUBJ003
    metadata_mappings:
      subject:
        source: PatientID
      session:
        source: StudyDate

This expands internally into three per-subject queries (each with patient_name: SUBJ00N merged into dicom_query), with results concatenated into a single study list.

[akhanf]

@copilot: works great - but upon reflection this config field should be named "select" instead of "subjects", and the "subject_query_field" can be called "select_from_field". Also make the default select_from_field to be "patient_name" instead of "patient_id". Make those changes throughout.