eicr metadata (#338)

## Description

Add a property to eICRProcessor to get the eICR metadata: The eICR ID
and vender.

## Related Issues

Closes #328

---------

Co-authored-by: Joshua Nygaard <jnygaard@mac.myfiosgateway.com>

Author: JNygaard-Skylight

packages/shared-models/src/shared_models/__init__.py

@@ -4,6 +4,19 @@
 from pydantic import ConfigDict
 
 
+class CdaInstanceIdentifier(BaseModel):
+    """CDA Instance Identifier (II) data type.
+
+    https://build.fhir.org/ig/HL7/CDA-core-2.0/StructureDefinition-II.html
+    """
+
+    null_flavor: str | None = None
+    assigning_authority_name: str | None = None
+    displayable: bool | None = None
+    root: str | None = None
+    extension: str | None = None
+
+
 class DataField(StrEnum):
     """Enum for eICR data fields relevant to the TTC module."""
 

packages/text-to-code/src/text_to_code/models/eicr.py

@@ -1,6 +1,7 @@
 from enum import StrEnum
 
 from pydantic import BaseModel
+from shared_models import CdaInstanceIdentifier
 
 
 class LabXPaths(StrEnum):
@@ -19,3 +20,10 @@ class Candidate(BaseModel):
     value: str
     xpath: LabXPaths
     system: str | None = None
+
+
+class Metadata(BaseModel):
+    """Model representing metadata about the eICR."""
+
+    eicr_id: CdaInstanceIdentifier
+    eicr_vendor: str | None = None

packages/text-to-code/src/text_to_code/services/eicr_processor.py

@@ -1,8 +1,10 @@
 from lxml import etree
 from lxml.etree import Element
+from shared_models import CdaInstanceIdentifier
 from shared_models import DataField
 
 from text_to_code.models import Candidate
+from text_to_code.models.eicr import Metadata
 from text_to_code.services.utils import get_config_for_data_field
 
 
@@ -102,6 +104,36 @@ def _extract_text_from_element(self, element: Element) -> str:
 
         return " ".join(filter(None, text_parts))
 
+    @property
+    def eicr_metadata(self) -> Metadata:
+        """Get the eICR ID from the XML."""
+        id_element = self._xml_root.find(".//id")
+        if id_element is None or id_element.get("nullFlavor") is not None:
+            raise ValueError("No ID element found in eICR XML.")
+        instance_identifer = CdaInstanceIdentifier(
+            root=id_element.get("root"),
+            extension=id_element.get("extension"),
+            assigning_authority_name=id_element.get("assigningAuthorityName"),
+            displayable=_to_bool(id_element.get("displayable")),
+            null_flavor=id_element.get("nullFlavor"),
+        )
+        vendor = self._xml_root.find(
+            ".//author/assignedAuthor/assignedAuthoringDevice/softwareName"
+        )
+        return Metadata(
+            eicr_id=instance_identifer, eicr_vendor=vendor.text if vendor is not None else None
+        )
+
+
+def _to_bool(value: str | None) -> bool | None:
+    if value is None:
+        return None
+    if value.lower() == "true":
+        return True
+    if value.lower() == "false":
+        return False
+    return None
+
 
 def _create_xml_tree(xml: str) -> Element:
     """Remove all namespaces from an XML tree."""

packages/text-to-code/tests/assets/basic_test_eicr.xml

@@ -3,6 +3,7 @@
   xmlns:voc="http://www.lantanagroup.com/voc"
   xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
   xsi:schemaLocation="urn:hl7-org:v3 ../../schema/infrastructure/cda/CDA_SDTC.xsd">
+  <id root="c8516bdc-8bb2-40aa-8dae-20a77546488f" />
   <component>
     <structuredBody>
       <component>
@@ -22,4 +23,11 @@
       </component>
     </structuredBody>
   </component>
+  <author>
+    <assignedAuthor>
+      <assignedAuthoringDevice>
+        <softwareName>Test eCR Vendor Name</softwareName>
+      </assignedAuthoringDevice>
+    </assignedAuthor>
+  </author>
 </ClinicalDocument>

packages/text-to-code/tests/unit/test_eicr_processor.py

@@ -2,9 +2,11 @@
 
 import pytest
 from lxml.etree import XMLSyntaxError
+from shared_models import CdaInstanceIdentifier
 from shared_models import DataField
 from text_to_code.models import Candidate
 from text_to_code.models import LabXPaths
+from text_to_code.models.eicr import Metadata
 from text_to_code.services.eicr_processor import EicrProcessor
 
 EXAMPLE_EICRS_DIRECTORY = Path(__file__).parent.parent / "assets"
@@ -40,28 +42,40 @@ def test_bad_eicr(self):
 
 class TestBasicEicrProcessor:
     @pytest.fixture(scope="class")
-    def result(self) -> list[Candidate]:
+    def eicr_processor(self) -> EicrProcessor:
         eicr_path = EXAMPLE_EICRS_DIRECTORY / "basic_test_eicr.xml"
         with eicr_path.open() as f:
             eicr_output = f.read()
 
-        return EicrProcessor(eicr_output).get_text_candidates(
-            BASE_XPATH, DataField.LAB_TEST_NAME_RESULTED
-        )
+        return EicrProcessor(eicr_output)
+
+    @pytest.fixture(scope="class")
+    def eicr_metadata(self, eicr_processor: EicrProcessor) -> Metadata:
+        return eicr_processor.eicr_metadata
+
+    @pytest.fixture(scope="class")
+    def candidates(self, eicr_processor: EicrProcessor) -> list[Candidate]:
+        return eicr_processor.get_text_candidates(BASE_XPATH, DataField.LAB_TEST_NAME_RESULTED)
 
-    def test_attribute_candidate(self, result: list[Candidate]):
-        assert result[0] == Candidate(
+    def test_attribute_candidate(self, candidates: list[Candidate]):
+        assert candidates[0] == Candidate(
             value="A custom code in display name.", xpath=LabXPaths.CODE_DISPLAY_NAME
         )
 
-    def test_text_candidate(self, result: list[Candidate]):
-        assert result[1] == Candidate(
+    def test_text_candidate(self, candidates: list[Candidate]):
+        assert candidates[1] == Candidate(
             value="A custom code in original text.", xpath=LabXPaths.CODE_ORIGINAL_TEXT
         )
 
-    def test_candidate_count(self, result: list[Candidate]):
+    def test_candidate_count(self, candidates: list[Candidate]):
         expected = 2
-        assert len(result) == expected
+        assert len(candidates) == expected
+
+    def test_metadata(self, eicr_metadata: Metadata):
+        assert eicr_metadata == Metadata(
+            eicr_id=CdaInstanceIdentifier(root="c8516bdc-8bb2-40aa-8dae-20a77546488f"),
+            eicr_vendor="Test eCR Vendor Name",
+        )
 
 
 class TestReferences: