add geo_to_h3 function

examples/EOI00002H.json

@@ -69,6 +69,21 @@
             ]
         }
     },
+    "producedBy_samplingSite_location_h3_0": "8029fffffffffff",
+    "producedBy_samplingSite_location_h3_1": "8128fffffffffff",
+    "producedBy_samplingSite_location_h3_10": "8a28ed791baffff",
+    "producedBy_samplingSite_location_h3_11": "8b28ed791ba9fff",
+    "producedBy_samplingSite_location_h3_12": "8c28ed791ba97ff",
+    "producedBy_samplingSite_location_h3_13": "8d28ed791ba967f",
+    "producedBy_samplingSite_location_h3_14": "8e28ed791ba9647",
+    "producedBy_samplingSite_location_h3_2": "8228effffffffff",
+    "producedBy_samplingSite_location_h3_3": "8328edfffffffff",
+    "producedBy_samplingSite_location_h3_4": "8428ed7ffffffff",
+    "producedBy_samplingSite_location_h3_5": "8528ed7bfffffff",
+    "producedBy_samplingSite_location_h3_6": "8628ed797ffffff",
+    "producedBy_samplingSite_location_h3_7": "8728ed791ffffff",
+    "producedBy_samplingSite_location_h3_8": "8828ed791bfffff",
+    "producedBy_samplingSite_location_h3_9": "8928ed791bbffff",
     "registrant": "Andra Bobbitt",
     "relatedResource": [],
     "sampleidentifier": "igsn:10.58052/EOI00002H",

examples/IEDUT103B.json

@@ -77,6 +77,21 @@
             ]
         }
     },
+    "producedBy_samplingSite_location_h3_0": "8067fffffffffff",
+    "producedBy_samplingSite_location_h3_1": "8145bffffffffff",
+    "producedBy_samplingSite_location_h3_10": "8a45b29884e7fff",
+    "producedBy_samplingSite_location_h3_11": "8b45b29884e1fff",
+    "producedBy_samplingSite_location_h3_12": "8c45b29884e15ff",
+    "producedBy_samplingSite_location_h3_13": "8d45b29884e147f",
+    "producedBy_samplingSite_location_h3_14": "8e45b29884e1457",
+    "producedBy_samplingSite_location_h3_2": "8245b7fffffffff",
+    "producedBy_samplingSite_location_h3_3": "8345b2fffffffff",
+    "producedBy_samplingSite_location_h3_4": "8445b29ffffffff",
+    "producedBy_samplingSite_location_h3_5": "8545b29bfffffff",
+    "producedBy_samplingSite_location_h3_6": "8645b298fffffff",
+    "producedBy_samplingSite_location_h3_7": "8745b2988ffffff",
+    "producedBy_samplingSite_location_h3_8": "8845b29885fffff",
+    "producedBy_samplingSite_location_h3_9": "8945b29884fffff",
     "registrant": "Andrea Dutton",
     "relatedResource": [],
     "sampleidentifier": "igsn:10.58052/IEDUT103B",

examples/IEEJR000M.json

@@ -73,6 +73,21 @@
             ]
         }
     },
+    "producedBy_samplingSite_location_h3_0": "8029fffffffffff",
+    "producedBy_samplingSite_location_h3_1": "8129bffffffffff",
+    "producedBy_samplingSite_location_h3_10": "8a29aa8f66dffff",
+    "producedBy_samplingSite_location_h3_11": "8b29aa8f66dbfff",
+    "producedBy_samplingSite_location_h3_12": "8c29aa8f66dbbff",
+    "producedBy_samplingSite_location_h3_13": "8d29aa8f66dbaff",
+    "producedBy_samplingSite_location_h3_14": "8e29aa8f66dbaef",
+    "producedBy_samplingSite_location_h3_2": "8229affffffffff",
+    "producedBy_samplingSite_location_h3_3": "8329aafffffffff",
+    "producedBy_samplingSite_location_h3_4": "8429aa9ffffffff",
+    "producedBy_samplingSite_location_h3_5": "8529aa8ffffffff",
+    "producedBy_samplingSite_location_h3_6": "8629aa8f7ffffff",
+    "producedBy_samplingSite_location_h3_7": "8729aa8f2ffffff",
+    "producedBy_samplingSite_location_h3_8": "8829aa8f2dfffff",
+    "producedBy_samplingSite_location_h3_9": "8929aa8f2d7ffff",
     "registrant": "Evan Ramos",
     "relatedResource": [],
     "sampleidentifier": "igsn:10.58052/IEEJR000M",

examples/IEJEN0040.json

@@ -65,6 +65,21 @@
             ]
         }
     },
+    "producedBy_samplingSite_location_h3_0": "800dfffffffffff",
+    "producedBy_samplingSite_location_h3_1": "810c7ffffffffff",
+    "producedBy_samplingSite_location_h3_10": "8a0c42d0e837fff",
+    "producedBy_samplingSite_location_h3_11": "8b0c42d0e832fff",
+    "producedBy_samplingSite_location_h3_12": "8c0c42d0e8327ff",
+    "producedBy_samplingSite_location_h3_13": "8d0c42d0e8326ff",
+    "producedBy_samplingSite_location_h3_14": "8e0c42d0e8326e7",
+    "producedBy_samplingSite_location_h3_2": "820c47fffffffff",
+    "producedBy_samplingSite_location_h3_3": "830c42fffffffff",
+    "producedBy_samplingSite_location_h3_4": "840c42dffffffff",
+    "producedBy_samplingSite_location_h3_5": "850c42d3fffffff",
+    "producedBy_samplingSite_location_h3_6": "860c42d0fffffff",
+    "producedBy_samplingSite_location_h3_7": "870c42d0effffff",
+    "producedBy_samplingSite_location_h3_8": "880c42d0e9fffff",
+    "producedBy_samplingSite_location_h3_9": "890c42d0e83ffff",
     "registrant": "Jonathan Nichols",
     "relatedResource": [],
     "sampleidentifier": "igsn:10.58052/IEJEN0040",

examples/IERVTL1I7.json

@@ -67,6 +67,21 @@
             "placeName": []
         }
     },
+    "producedBy_samplingSite_location_h3_0": "8027fffffffffff",
+    "producedBy_samplingSite_location_h3_1": "8126bffffffffff",
+    "producedBy_samplingSite_location_h3_10": "8a26b56364a7fff",
+    "producedBy_samplingSite_location_h3_11": "8b26b56364a3fff",
+    "producedBy_samplingSite_location_h3_12": "8c26b56364a31ff",
+    "producedBy_samplingSite_location_h3_13": "8d26b56364a30ff",
+    "producedBy_samplingSite_location_h3_14": "8e26b56364a30c7",
+    "producedBy_samplingSite_location_h3_2": "8226b7fffffffff",
+    "producedBy_samplingSite_location_h3_3": "8326b5fffffffff",
+    "producedBy_samplingSite_location_h3_4": "8426b57ffffffff",
+    "producedBy_samplingSite_location_h3_5": "8526b57bfffffff",
+    "producedBy_samplingSite_location_h3_6": "8626b5637ffffff",
+    "producedBy_samplingSite_location_h3_7": "8726b5636ffffff",
+    "producedBy_samplingSite_location_h3_8": "8826b56365fffff",
+    "producedBy_samplingSite_location_h3_9": "8926b56364bffff",
     "registrant": "SLAC SFA",
     "relatedResource": [],
     "sampleidentifier": "igsn:10.58052/IERVTL1I7",

examples/ODP02Q1IZ.json

@@ -67,6 +67,21 @@
             "placeName": []
         }
     },
+    "producedBy_samplingSite_location_h3_0": "80dffffffffffff",
+    "producedBy_samplingSite_location_h3_1": "81df3ffffffffff",
+    "producedBy_samplingSite_location_h3_10": "8adf35928cf7fff",
+    "producedBy_samplingSite_location_h3_11": "8bdf35928cf0fff",
+    "producedBy_samplingSite_location_h3_12": "8cdf35928cf05ff",
+    "producedBy_samplingSite_location_h3_13": "8ddf35928cf0cff",
+    "producedBy_samplingSite_location_h3_14": "8edf35928cf0cdf",
+    "producedBy_samplingSite_location_h3_2": "82df37fffffffff",
+    "producedBy_samplingSite_location_h3_3": "83e89bfffffffff",
+    "producedBy_samplingSite_location_h3_4": "84df359ffffffff",
+    "producedBy_samplingSite_location_h3_5": "85df3593fffffff",
+    "producedBy_samplingSite_location_h3_6": "86df3592fffffff",
+    "producedBy_samplingSite_location_h3_7": "87df35928ffffff",
+    "producedBy_samplingSite_location_h3_8": "88df35928dfffff",
+    "producedBy_samplingSite_location_h3_9": "89df35928cfffff",
     "registrant": "Integrated Ocean Drilling Program (TAMU)",
     "relatedResource": [],
     "sampleidentifier": "igsn:10.60471/ODP02Q1IZ",

isamples_sesar/sesar_transformer.py

@@ -1,6 +1,7 @@
 import typing
 from typing import Optional
 import logging
+import h3
 from .sample import Sample
 
 from .mapper import (
@@ -19,6 +20,8 @@ class Transformer():
 
     FEET_PER_METER = 3.28084
 
+    DEFAULT_H3_RESOLUTION = 15
+
     def __init__(self, sample: Sample):
         self.sample = sample
         self._material_prediction_results: Optional[list] = None
@@ -79,6 +82,10 @@ def transform(self) -> typing.Dict:
             "authorizedBy": self.authorized_by(),
             "compliesWith": self.complies_with(),
         }
+        for index in range(0, 15):
+            h3_at_resolution = self.h3_function()(self.sample.latitude, self.sample.longitude, index)
+            field_name = f"producedBy_samplingSite_location_h3_{index}"
+            transformed_record[field_name] = h3_at_resolution
         return transformed_record
 
     def has_context_categories(self) -> typing.List[str]:
@@ -406,6 +413,9 @@ def curation_responsibility(self) -> list[dict]:
         responsibility.append(metadata_publisher)
         return responsibility
 
+    def h3_function(self) -> typing.Callable:
+        return geo_to_h3
+
 
 class MaterialCategoryMetaMapper(AbstractCategoryMetaMapper):
     _endsWithRockMapper = StringEndsWithCategoryMapper("Rock", "Rock")
@@ -627,3 +637,14 @@ def categories_mappers(cls) -> typing.List[AbstractCategoryMapper]:
             cls._floodplainAquiferMapper,
             cls._creekBankMapper,
         ]
+
+
+def geo_to_h3(
+    latitude: typing.Optional[float],
+    longitude: typing.Optional[float],
+    resolution: int = Transformer.DEFAULT_H3_RESOLUTION
+) -> typing.Optional[str]:
+    if latitude is not None and longitude is not None:
+        return h3.latlng_to_cell(latitude, longitude, resolution)
+    else:
+        return None

scripts/sesar_things.py

@@ -5,7 +5,7 @@
 
 from isamples_sesar.sesar_adapter import SESARItem
 from isamples_sesar.sqlmodel_database import SQLModelDAO as SESAR_SQLModelDAO, get_sample_rows
-from isamples_sesar.sesar_transformer import Transformer
+from isamples_sesar.sesar_transformer import Transformer, geo_to_h3
 from isb_web.sqlmodel_database import SQLModelDAO as iSB_SQLModelDAO, all_thing_primary_keys, DatabaseBulkUpdater  # type: ignore
 
 BATCH_SIZE = 10000
@@ -34,9 +34,9 @@ def load_sesar_entries(sesar_db_session, isb_db_session, start_from=None):
                 num_newer += 1
                 thing_id = f"igsn:{sample.igsn}"
                 resolved_url = f"doi.org/{sample.igsn}"
-                # h3 = Transformer.geo_to_h3(current_record)
+                h3 = geo_to_h3(sample.latitude, sample.longitude)
                 t_created = sample.registration_date
-                bulk_updater.add_thing(current_record, thing_id, resolved_url, 200, "h3", t_created)
+                bulk_updater.add_thing(current_record, thing_id, resolved_url, 200, h3, t_created)
             offset += BATCH_SIZE
             bulk_updater.finish()
     print(f"Num newer={num_newer}\n\n")

tests/test_isamples_metadata.py

@@ -55,6 +55,7 @@ def test_example(sesar_session: Session, igsn):
     check_related_resource(transformed_test_data, expected_data)
     check_authorized_by(transformed_test_data, expected_data)
     check_complies_with(transformed_test_data, expected_data)
+    check_geo_to_h3(transformed_test_data, expected_data)
 
 
 def check_id(test_data, expected_data):
@@ -188,3 +189,21 @@ def check_authorized_by(test_data, expected_data):
 
 def check_complies_with(test_data, expected_data):
     assert test_data["compliesWith"] == expected_data["compliesWith"]
+
+
+def check_geo_to_h3(test_data, expected_data):
+    assert test_data["producedBy_samplingSite_location_h3_0"] == expected_data["producedBy_samplingSite_location_h3_0"]
+    assert test_data["producedBy_samplingSite_location_h3_1"] == expected_data["producedBy_samplingSite_location_h3_1"]
+    assert test_data["producedBy_samplingSite_location_h3_2"] == expected_data["producedBy_samplingSite_location_h3_2"]
+    assert test_data["producedBy_samplingSite_location_h3_3"] == expected_data["producedBy_samplingSite_location_h3_3"]
+    assert test_data["producedBy_samplingSite_location_h3_4"] == expected_data["producedBy_samplingSite_location_h3_4"]
+    assert test_data["producedBy_samplingSite_location_h3_5"] == expected_data["producedBy_samplingSite_location_h3_5"]
+    assert test_data["producedBy_samplingSite_location_h3_6"] == expected_data["producedBy_samplingSite_location_h3_6"]
+    assert test_data["producedBy_samplingSite_location_h3_7"] == expected_data["producedBy_samplingSite_location_h3_7"]
+    assert test_data["producedBy_samplingSite_location_h3_8"] == expected_data["producedBy_samplingSite_location_h3_8"]
+    assert test_data["producedBy_samplingSite_location_h3_9"] == expected_data["producedBy_samplingSite_location_h3_9"]
+    assert test_data["producedBy_samplingSite_location_h3_10"] == expected_data["producedBy_samplingSite_location_h3_10"]
+    assert test_data["producedBy_samplingSite_location_h3_11"] == expected_data["producedBy_samplingSite_location_h3_11"]
+    assert test_data["producedBy_samplingSite_location_h3_12"] == expected_data["producedBy_samplingSite_location_h3_12"]
+    assert test_data["producedBy_samplingSite_location_h3_13"] == expected_data["producedBy_samplingSite_location_h3_13"]
+    assert test_data["producedBy_samplingSite_location_h3_14"] == expected_data["producedBy_samplingSite_location_h3_14"]