More MITAardvark methods

tests/fixtures/aardvark/aardvark_record_all_fields.jsonl

@@ -1 +1 @@
-{"id": "123", "dcat_bbox": "ENVELOPE(-111.1, -104.0, 45.0, 40.9)", "dcat_keyword_sm": ["Country"], "dcat_theme_sm": ["Political boundaries"], "dct_accessRights_s": "Access note", "dct_alternative_sm": ["Alternate title"], "dct_creator_sm": ["Smith, Jane", "Smith, John"], "dct_description_sm": ["A description"], "dct_format_s": "Shapefile", "dct_language_sm": ["eng"], "dct_license_sm": ["http://license.license", "http://another_license.another_license"], "dct_publisher_sm": ["ML InfoMap (Firm)"], "dct_rights_sm": ["Some person has the rights"], "dct_rightsHolder_sm": ["The person with the rights", "Another person with the rights"], "dct_spatial_sm": ["Some city, Some country"], "dct_subject_sm": ["Geography", "Earth"], "dct_title_s": "Test title 1", "gbl_displayNote_sm": ["Danger: This text will be displayed in a red box","Info: This text will be displayed in a blue box","Tip: This text will be displayed in a green box","Warning: This text will be displayed in a yellow box","This is text without a tag and it will be assigned default 'note' style"], "gbl_resourceClass_sm": ["Dataset"], "gbl_resourceType_sm": ["Vector data"], "locn_geometry": "POLYGON((-80 25, -65 18, -64 33, -80 25))", "schema_provider_s": "MIT"}
+{"id": "mit:123", "dcat_bbox": "ENVELOPE(-111.1, -104.0, 45.0, 40.9)", "dcat_keyword_sm": ["Country"], "dcat_theme_sm": ["Political boundaries"], "dct_accessRights_s": "Access note", "dct_alternative_sm": ["Alternate title"], "dct_creator_sm": ["Smith, Jane", "Smith, John"], "dct_description_sm": ["A description"], "dct_format_s": "Shapefile", "dct_identifier_sm": ["abc123"], "dct_issued_s": "2003-10-23", "dct_language_sm": ["eng"], "dct_license_sm": ["http://license.license", "http://another_license.another_license"], "dct_publisher_sm": ["ML InfoMap (Firm)"], "dct_references_s": "{\"https://schema.org/downloadUrl\": [{\"label\": \"Source Metadata\", \"protocol\": \"Download\", \"url\": \"https://example.com/GISPORTAL_GISOWNER01_BOSTONWATER95.source.fgdc.xml\"}, {\"label\": \"Normalized Metadata\", \"protocol\": \"Download\", \"url\": \"https://example.com/GISPORTAL_GISOWNER01_BOSTONWATER95.normalized.aardvark.json\"}, {\"label\": \"Data Zipfile\", \"protocol\": \"Download\", \"url\": \"https://example.com/GISPORTAL_GISOWNER01_BOSTONWATER95.zip\"}]}", "dct_rights_sm": ["Some person has the rights"], "dct_rightsHolder_sm": ["The person with the rights", "Another person with the rights"], "dct_spatial_sm": ["Some city, Some country"], "dct_subject_sm": ["Geography", "Earth"], "dct_temporal_sm": ["1943", "1979"], "dct_title_s": "Test title 1", "gbl_dateRange_drsim": ["[1943 TO 1946]"], "gbl_displayNote_sm": ["Danger: This text will be displayed in a red box","Info: This text will be displayed in a blue box","Tip: This text will be displayed in a green box","Warning: This text will be displayed in a yellow box","This is text without a tag and it will be assigned default 'note' style"], "gbl_indexYear_im": [1943,1944,1945,1946], "gbl_resourceClass_sm": ["Dataset"], "gbl_resourceType_sm": ["Vector data"], "locn_geometry": "ENVELOPE(-111.1, -104.0, 45.0, 40.9)", "schema_provider_s": "MIT"}

tests/fixtures/aardvark_records.jsonl

@@ -1,2 +1,2 @@
-{"id": "123", "dct_title_s": "Test title 1"}
-{"id": "456", "dct_title_s": "Test title 2"}
+{"dct_accessRights_s": "Access rights", "dct_references_s": "", "dct_title_s": "Test title 1", "gbl_mdModified_dt": "", "gbl_mdVersion_s": "", "gbl_resourceClass_sm": "", "id": "mit:123", "locn_geometry": ""}
+{"dct_accessRights_s": "Access rights", "dct_references_s": "", "dct_title_s": "Test title 2", "gbl_mdModified_dt": "", "gbl_mdVersion_s": "", "gbl_resourceClass_sm": "", "id": "ogm:456", "locn_geometry": ""}

tests/sources/json/test_aardvark.py

@@ -1,3 +1,5 @@
+import pytest
+
 import transmogrifier.models as timdex
 from transmogrifier.sources.json.aardvark import MITAardvark
 
@@ -21,6 +23,7 @@ def test_aardvark_transform_returns_timdex_record(aardvark_records):
         title="Test title 1",
         citation="Test title 1. Geospatial data. https://example.com/123",
         content_type=["Geospatial data"],
+        rights=[timdex.Rights(description="Access rights", kind="Access")],
     )
 
 
@@ -63,6 +66,76 @@ def test_aardvark_get_contributors_success(aardvark_record_all_fields):
     ]
 
 
+def test_aardvark_get_dates_success(aardvark_record_all_fields):
+    assert MITAardvark.get_dates(next(aardvark_record_all_fields), "123") == [
+        timdex.Date(kind="Issued", value="2003-10-23"),
+        timdex.Date(kind="Coverage", value="1943"),
+        timdex.Date(kind="Coverage", value="1979"),
+        timdex.Date(kind="Coverage", value="1944"),
+        timdex.Date(kind="Coverage", value="1945"),
+        timdex.Date(kind="Coverage", value="1946"),
+        timdex.Date(
+            range=timdex.Date_Range(gte="1943", lte="1946"),
+        ),
+    ]
+
+
+def test_aardvark_parse_solr_date_range_string_success():
+    assert MITAardvark.parse_solr_date_range_string("[1932 TO 1937]", "123") == (
+        "1932",
+        "1937",
+    )
+
+
+def test_parse_solr_date_range_invalid_date_range_string_raises_error():
+    with pytest.raises(
+        ValueError,
+        match="Record ID '123': Unable to parse date range string 'Invalid'",
+    ):
+        MITAardvark.parse_solr_date_range_string("Invalid", "123")
+
+
+def test_aardvark_get_identifiers_success(aardvark_record_all_fields):
+    assert MITAardvark.get_identifiers(next(aardvark_record_all_fields)) == [
+        timdex.Identifier(value="abc123")
+    ]
+
+
+def test_aardvark_get_links_success(aardvark_record_all_fields):
+    assert MITAardvark.get_links(next(aardvark_record_all_fields), "123") == [
+        timdex.Link(
+            url="https://example.com/GISPORTAL_GISOWNER01_BOSTONWATER95.source.fgdc.xml",
+            kind="Download",
+            text="Source Metadata",
+        ),
+        timdex.Link(
+            url="https://example.com/GISPORTAL_GISOWNER01_BOSTONWATER95."
+            "normalized.aardvark.json",
+            kind="Download",
+            text="Normalized Metadata",
+        ),
+        timdex.Link(
+            url="https://example.com/GISPORTAL_GISOWNER01_BOSTONWATER95.zip",
+            kind="Download",
+            text="Data Zipfile",
+        ),
+    ]
+
+
+def test_aardvark_get_links_logs_warning_for_invalid_json(caplog):
+    assert MITAardvark.get_links({"dct_references_s": "Invalid"}, "123") == []
+    assert (
+        "Record ID '123': Unable to parse links string 'Invalid' as JSON" in caplog.text
+    )
+
+
+def test_aardvark_get_locations_success(aardvark_record_all_fields):
+    assert MITAardvark.get_locations(next(aardvark_record_all_fields), "123") == [
+        timdex.Location(kind="Bounding Box", geodata=[-111.1, -104.0, 45.0, 40.9]),
+        timdex.Location(kind="Geometry", geodata=[-111.1, -104.0, 45.0, 40.9]),
+    ]
+
+
 def test_aardvark_get_notes_success(aardvark_record_all_fields):
     assert MITAardvark.get_notes(next(aardvark_record_all_fields)) == [
         timdex.Note(

tests/test_helpers.py

@@ -1,9 +1,12 @@
 from datetime import datetime
 
+import pytest
+
 import transmogrifier.models as timdex
 from transmogrifier.helpers import (
     generate_citation,
     parse_date_from_string,
+    parse_geodata_string,
     validate_date,
     validate_date_range,
 )
@@ -256,6 +259,23 @@ def test_parse_date_from_string_invalid_date_returns_none():
     assert parse_date_from_string("circa 1930s") is None
 
 
+def test_parse_geodata_string_success():
+    assert parse_geodata_string("ENVELOPE(-111.1, -104.0, 45.0, 40.9)", "123") == [
+        -111.1,
+        -104.0,
+        45.0,
+        40.9,
+    ]
+
+
+def test_parse_geodata_string_invalid_geodata_string_raises_error():
+    with pytest.raises(
+        ValueError,
+        match="Record ID '123': Unable to parse geodata string 'Invalid'",
+    ):
+        parse_geodata_string("Invalid", "123")
+
+
 def test_validate_date_success():
     assert validate_date("1930", "1234") is True
 

transmogrifier/config.py

@@ -103,6 +103,16 @@
         "base-url": "https://libguides.mit.edu/",
         "transform-class": "transmogrifier.sources.xml.springshare.SpringshareOaiDc",
     },
+    "gismit": {
+        "name": "MIT GIS Resources",
+        "base-url": "https://search.libraries.mit.edu/record/",
+        "transform-class": "transmogrifier.sources.json.aardvark.MITAardvark",
+    },
+    "gisogm": {
+        "name": "OpenGeoMetadata GIS Resources",
+        "base-url": "https://search.libraries.mit.edu/record/",
+        "transform-class": "transmogrifier.sources.json.aardvark.OGMAardvark",
+    },
     "researchdatabases": {
         "name": "Research Databases",
         "base-url": "https://libguides.mit.edu/",

transmogrifier/helpers.py

@@ -75,6 +75,31 @@ def parse_date_from_string(
     return None
 
 
+def parse_geodata_string(geodata_string: str, source_record_id: str) -> list[float]:
+    """Get list of values from a formatted geodata string.
+
+    Example:
+     - "ENVELOPE(-111.1, -104.0, 45.0, 40.9)"
+
+     Args:
+        geodata_string: Formatted geodata string to parse.
+        source_record_id: The ID of the record containing the string to parse.
+    """
+    geodata_points = []
+    try:
+        raw_geodata_points = geodata_string.split("(")[-1].split(")")[0].split(",")
+        stripped_geodata_points = map(str.strip, raw_geodata_points)
+        geodata_floats = list(map(float, stripped_geodata_points))
+        geodata_points.extend(geodata_floats)
+    except ValueError:
+        message = (
+            f"Record ID '{source_record_id}': "
+            f"Unable to parse geodata string '{geodata_string}'"
+        )
+        raise ValueError(message)
+    return geodata_points
+
+
 def validate_date(
     date_string: str,
     source_record_id: str,

transmogrifier/models.py

@@ -123,7 +123,7 @@ class Link:
 
 @define
 class Location:
-    value: Optional[str] = field(validator=optional(instance_of(str)))
+    value: Optional[str] = field(default=None, validator=optional(instance_of(str)))
     kind: Optional[str] = field(default=None, validator=optional(instance_of(str)))
     geodata: Optional[list[float]] = field(
         default=None, validator=optional(list_of(float))