nmdc-runtime 1.0.13__tar.gz → 1.0.14__tar.gz

{nmdc_runtime-1.0.13 → nmdc_runtime-1.0.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nmdc_runtime
-Version: 1.0.13
+Version: 1.0.14
 Summary: A runtime system for NMDC data management and orchestration
 Home-page: https://github.com/microbiomedata/nmdc-runtime
 Author: Donny Winston

{nmdc_runtime-1.0.13 → nmdc_runtime-1.0.14}/nmdc_runtime/api/endpoints/objects.py

@@ -123,7 +123,8 @@ def get_object_info(
         )  # TODO use HEAD when enabled upstream
         if rv.status_code != 404:
             return RedirectResponse(
-                url_to_try, status_code=status.HTTP_307_TEMPORARY_REDIRECT
+                f"https://data.microbiomedata.org/details/study/nmdc:{object_id}",
+                status_code=status.HTTP_307_TEMPORARY_REDIRECT,
             )
     elif object_id.startswith("bsm"):
         url_to_try = f"https://data.microbiomedata.org/api/biosample/nmdc:{object_id}"
@@ -132,7 +133,8 @@ def get_object_info(
         )  # TODO use HEAD when enabled upstream
         if rv.status_code != 404:
             return RedirectResponse(
-                url_to_try, status_code=status.HTTP_307_TEMPORARY_REDIRECT
+                f"https://data.microbiomedata.org/details/sample/nmdc:{object_id}",
+                status_code=status.HTTP_307_TEMPORARY_REDIRECT,
             )
 
     # If "sty" or "bsm" ID doesn't have preferred landing page (above), try for JSON payload

{nmdc_runtime-1.0.13 → nmdc_runtime-1.0.14}/nmdc_runtime/api/endpoints/queries.py

@@ -7,8 +7,8 @@ from pymongo.database import Database as MongoDatabase
 
 from nmdc_runtime.api.core.idgen import generate_one_id
 from nmdc_runtime.api.core.util import now, raise404_if_none
-from nmdc_runtime.api.db.mongo import get_mongo_db
-from nmdc_runtime.api.endpoints.util import permitted
+from nmdc_runtime.api.db.mongo import get_mongo_db, nmdc_schema_collection_names
+from nmdc_runtime.api.endpoints.util import permitted, users_allowed
 from nmdc_runtime.api.models.query import (
     Query,
     QueryResponseOptions,
@@ -17,6 +17,7 @@ from nmdc_runtime.api.models.query import (
     CommandResponse,
     command_response_for,
     QueryCmd,
+    UpdateCommand,
 )
 from nmdc_runtime.api.models.user import get_current_active_user, User
 
@@ -28,14 +29,12 @@ def unmongo(d: dict) -> dict:
     return json.loads(bson.json_util.dumps(d))
 
 
-def check_can_delete(user: User):
+def check_can_update_and_delete(user: User):
+    # update and delete queries require same level of permissions
     if not permitted(user.username, "/queries:run(query_cmd:DeleteCommand)"):
         raise HTTPException(
             status_code=status.HTTP_403_FORBIDDEN,
-            detail=(
-                f"Only specific users "
-                "are allowed to issue delete commands.",
-            ),
+            detail="Only specific users are allowed to issue update and delete commands."
         )
 
 
@@ -63,10 +62,15 @@ def run_query(
       "delete": "biosample_set",
       "deletes": [{"q": {"id": "NOT_A_REAL_ID"}, "limit": 1}]
     }
+
+    {
+        "update": "biosample_set",
+        "updates": [{"q": {"id": "YOUR_BIOSAMPLE_ID"}, "u": {"$set": {"name": "A_NEW_NAME"}}}]
+    }
     ```
     """
-    if isinstance(query_cmd, DeleteCommand):
-        check_can_delete(user)
+    if isinstance(query_cmd, (DeleteCommand, UpdateCommand)):
+        check_can_update_and_delete(user)
 
     qid = generate_one_id(mdb, "qy")
     saved_at = now()
@@ -104,7 +108,7 @@ def rerun_query(
     doc = raise404_if_none(mdb.queries.find_one({"id": query_id}))
     query = Query(**doc)
     if isinstance(query, DeleteCommand):
-        check_can_delete(user)
+        check_can_update_and_delete(user)
 
     cmd_response = _run_query(query, mdb)
     return unmongo(cmd_response.model_dump(exclude_unset=True))
@@ -115,10 +119,15 @@ def _run_query(query, mdb) -> CommandResponse:
     ran_at = now()
     if q_type is DeleteCommand:
         collection_name = query.cmd.delete
-        find_specs = [
-            {"filter": dcd.q, "limit": dcd.limit} for dcd in query.cmd.deletes
+        if collection_name not in nmdc_schema_collection_names(mdb):
+            raise HTTPException(
+                status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                detail="Can only delete documents in nmdc-schema collections.",
+            )
+        delete_specs = [
+            {"filter": del_statement.q, "limit": del_statement.limit} for del_statement in query.cmd.deletes
         ]
-        for spec in find_specs:
+        for spec in delete_specs:
             docs = list(mdb[collection_name].find(**spec))
             if not docs:
                 continue
@@ -131,6 +140,29 @@ def _run_query(query, mdb) -> CommandResponse:
                     detail="Failed to back up to-be-deleted documents. operation aborted.",
                 )
 
+    elif q_type is UpdateCommand:
+        collection_name = query.cmd.update
+        if collection_name not in nmdc_schema_collection_names(mdb):
+            raise HTTPException(
+                status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                detail="Can only update documents in nmdc-schema collections.",
+            )
+        update_specs = [
+            {"filter": up_statement.q, "limit": 0 if up_statement.multi else 1} for up_statement in query.cmd.updates
+        ]
+        for spec in update_specs:
+            docs = list(mdb[collection_name].find(**spec))
+            if not docs:
+                continue
+            insert_many_result = mdb.client["nmdc_updated"][
+                collection_name
+            ].insert_many({"doc": d, "updated_at": ran_at} for d in docs)
+            if len(insert_many_result.inserted_ids) != len(docs):
+                raise HTTPException(
+                    status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                    detail="Failed to back up to-be-updated documents. operation aborted.",
+                )
+
     q_response = mdb.command(query.cmd.model_dump(exclude_unset=True))
     cmd_response: CommandResponse = command_response_for(q_type)(**q_response)
     query_run = (
@@ -138,5 +170,14 @@ def _run_query(query, mdb) -> CommandResponse:
         if cmd_response.ok
         else QueryRun(qid=query.id, ran_at=ran_at, error=cmd_response)
     )
+    if q_type in (DeleteCommand, UpdateCommand) and cmd_response.n == 0:
+        raise HTTPException(
+            status_code=status.HTTP_418_IM_A_TEAPOT,
+            detail=(
+                f"{'update' if q_type is UpdateCommand else 'delete'} command modified zero documents."
+                " I'm guessing that's not what you expected. Check the syntax of your request."
+                " But what do I know? I'm just a teapot.",
+            ),
+        )
     mdb.query_runs.insert_one(query_run.model_dump(exclude_unset=True))
     return cmd_response

{nmdc_runtime-1.0.13 → nmdc_runtime-1.0.14}/nmdc_runtime/api/models/query.py

@@ -1,6 +1,7 @@
 import datetime
 from typing import Optional, Any, Dict, List, Union
 
+from bson import ObjectId
 from pydantic import (
     model_validator,
     Field,
@@ -72,7 +73,7 @@ class FindCommandResponse(CommandResponse):
     cursor: FindCommandResponseCursor
 
 
-class DeleteCommandDelete(BaseModel):
+class DeleteStatement(BaseModel):
     q: Document
     limit: OneOrZero
     hint: Optional[Dict[str, OneOrMinusOne]] = None
@@ -80,7 +81,7 @@ class DeleteCommandDelete(BaseModel):
 
 class DeleteCommand(CommandBase):
     delete: str
-    deletes: List[DeleteCommandDelete]
+    deletes: List[DeleteStatement]
 
 
 class DeleteCommandResponse(CommandResponse):
@@ -89,6 +90,34 @@ class DeleteCommandResponse(CommandResponse):
     writeErrors: Optional[List[Document]] = None
 
 
+# If `multi==True` all documents that meet the query criteria will be updated.
+# Else only a single document that meets the query criteria will be updated.
+class UpdateStatement(BaseModel):
+    q: Document
+    u: Document
+    upsert: bool = False
+    multi: bool = False
+    hint: Optional[Dict[str, OneOrMinusOne]] = None
+
+
+class UpdateCommand(CommandBase):
+    update: str
+    updates: List[UpdateStatement]
+
+
+class DocumentUpserted(BaseModel):
+    index: NonNegativeInt
+    _id: ObjectId
+
+
+class UpdateCommandResponse(CommandResponse):
+    ok: OneOrZero
+    n: NonNegativeInt
+    nModified: NonNegativeInt
+    upserted: Optional[List[DocumentUpserted]] = None
+    writeErrors: Optional[List[Document]] = None
+
+
 class GetMoreCommand(CommandBase):
     getMore: int
     collection: str
@@ -107,7 +136,12 @@ class GetMoreCommandResponse(CommandResponse):
 
 
 QueryCmd = Union[
-    CollStatsCommand, CountCommand, FindCommand, GetMoreCommand, DeleteCommand
+    CollStatsCommand,
+    CountCommand,
+    FindCommand,
+    GetMoreCommand,
+    DeleteCommand,
+    UpdateCommand,
 ]
 
 QueryResponseOptions = Union[
@@ -116,6 +150,7 @@ QueryResponseOptions = Union[
     FindCommandResponse,
     GetMoreCommandResponse,
     DeleteCommandResponse,
+    UpdateCommandResponse,
 ]
 
 
@@ -126,6 +161,7 @@ def command_response_for(type_):
         FindCommand: FindCommandResponse,
         GetMoreCommand: GetMoreCommandResponse,
         DeleteCommand: DeleteCommandResponse,
+        UpdateCommand: UpdateCommandResponse,
     }
     return d.get(type_)
 

{nmdc_runtime-1.0.13 → nmdc_runtime-1.0.14}/nmdc_runtime/site/translation/gold_translator.py

@@ -1,6 +1,6 @@
 import collections
 import re
-from typing import List, Union
+from typing import List, Tuple, Union
 from nmdc_schema import nmdc
 
 from nmdc_runtime.site.translation.translator import JSON_OBJECT, Translator
@@ -188,37 +188,48 @@ class GoldStudyTranslator(Translator):
     def _get_quantity_value(
         self,
         gold_entity: JSON_OBJECT,
-        gold_field: str,
+        gold_field: Union[str, Tuple[str, str]],
         unit: Union[str, None] = None,
     ) -> Union[nmdc.QuantityValue, None]:
         """Get any field of a GOLD entity object as a QuantityValue
 
         This method extracts any single field of a GOLD entity object (study, biosample, etc)
         and if it is not `None` returns it as an `nmdc:QuantityValue`. A has_numeric_value will
-        be inferred from the gold_field value in gold_entity. The inference is done only if the
-        unit is meters. Support for other units will be added incrementally. A unit can optionally
-        be provided, otherwise the unit will be `None`. If the value of the field is `None`,
-        `None` will be returned.
+        be inferred from the gold_field value in gold_entity if it is a simple string value. If
+        it is a tuple of two fields, a has_minimum_numeric_value and has_maximum_numeric_value
+        will be inferred from the gold_field values in gold_entity.
 
         :param gold_entity: GOLD entity object
-        :param gold_field: Name of the field to extract
+        :param gold_field: Name of the field to extract, or a tuple of two fields to extract a range
         :param unit: An optional unit as a string, defaults to None
         :return: QuantityValue object
         """
+        if isinstance(gold_field, tuple):
+            minimum_numeric_value = gold_entity.get(gold_field[0])
+            maximum_numeric_value = gold_entity.get(gold_field[1])
+
+            if minimum_numeric_value is None and maximum_numeric_value is None:
+                return None
+            elif minimum_numeric_value is not None and maximum_numeric_value is None:
+                return nmdc.QuantityValue(
+                    has_raw_value=field_value,
+                    has_numeric_value=nmdc.Double(minimum_numeric_value),
+                    has_unit=unit,
+                )
+            else:
+                return nmdc.QuantityValue(
+                    has_minimum_numeric_value=nmdc.Double(minimum_numeric_value),
+                    has_maximum_numeric_value=nmdc.Double(maximum_numeric_value),
+                    has_unit=unit,
+                )
+
         field_value = gold_entity.get(gold_field)
         if field_value is None:
             return None
 
-        numeric_value = None
-        # TODO: in the future we will need better handling
-        # to parse out the numerical portion of the quantity value
-        # ex. temp might be 3 C, and we will need to parse out 3.0 from it
-        if unit == "meters":
-            numeric_value = nmdc.Double(field_value)
-
         return nmdc.QuantityValue(
             has_raw_value=field_value,
-            has_numeric_value=numeric_value,
+            has_numeric_value=nmdc.Double(field_value),
             has_unit=unit,
         )
 
@@ -390,7 +401,7 @@ class GoldStudyTranslator(Translator):
         """
         return nmdc.Study(
             description=gold_study.get("description"),
-            gold_study_identifiers=self._get_curie("GOLD", gold_study["studyGoldId"]),
+            gold_study_identifiers=self._get_curie("gold", gold_study["studyGoldId"]),
             id=nmdc_study_id,
             name=gold_study.get("studyName"),
             principal_investigator=self._get_pi(gold_study),
@@ -427,7 +438,7 @@ class GoldStudyTranslator(Translator):
             collected_from=nmdc_field_site_id,
             collection_date=self._get_collection_date(gold_biosample),
             depth=self._get_quantity_value(
-                gold_biosample, "depthInMeters", unit="meters"
+                gold_biosample, ("depthInMeters", "depthInMeters2"), unit="meters"
             ),
             description=gold_biosample.get("description"),
             diss_oxygen=self._get_quantity_value(gold_biosample, "oxygenConcentration"),
@@ -440,7 +451,7 @@ class GoldStudyTranslator(Translator):
             env_local_scale=self._get_env_term_value(gold_biosample, "envoLocalScale"),
             env_medium=self._get_env_term_value(gold_biosample, "envoMedium"),
             geo_loc_name=self._get_text_value(gold_biosample, "geoLocation"),
-            gold_biosample_identifiers=self._get_curie("GOLD", gold_biosample_id),
+            gold_biosample_identifiers=self._get_curie("gold", gold_biosample_id),
             habitat=gold_biosample.get("habitat"),
             host_name=gold_biosample.get("hostName"),
             host_taxid=self._get_text_value(gold_biosample, "hostNcbiTaxid"),
@@ -498,7 +509,7 @@ class GoldStudyTranslator(Translator):
             id=nmdc_omics_processing_id,
             name=gold_project.get("projectName"),
             gold_sequencing_project_identifiers=self._get_curie(
-                "GOLD", gold_project_id
+                "gold", gold_project_id
             ),
             ncbi_project_name=gold_project.get("projectName"),
             type="nmdc:OmicsProcessing",

{nmdc_runtime-1.0.13 → nmdc_runtime-1.0.14}/nmdc_runtime.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nmdc-runtime
-Version: 1.0.13
+Version: 1.0.14
 Summary: A runtime system for NMDC data management and orchestration
 Home-page: https://github.com/microbiomedata/nmdc-runtime
 Author: Donny Winston

{nmdc_runtime-1.0.13 → nmdc_runtime-1.0.14}/tests/test_data/test_gold_translator.py

@@ -219,7 +219,7 @@ def test_get_quantity_value():
     value = translator._get_quantity_value(entity, "arbitraryField")
     assert value is not None
     assert value.has_raw_value == "7"
-    assert value.has_numeric_value is None
+    assert value.has_numeric_value == 7.0
     assert value.has_unit is None
 
     entity = {"arbitraryField": 0}
@@ -240,6 +240,17 @@ def test_get_quantity_value():
     value = translator._get_quantity_value(entity, "arbitraryField", unit="meters")
     assert value is None
 
+    entity = {"arbitraryField1": 1, "arbitraryField2": 10}
+    value = translator._get_quantity_value(
+        entity, ("arbitraryField1", "arbitraryField2"), unit="meters"
+    )
+    assert value is not None
+    assert value.has_minimum_numeric_value == 1.0
+    assert value.has_maximum_numeric_value == 10.0
+    assert value.has_raw_value is None
+    assert value.has_numeric_value is None
+    assert value.has_unit == "meters"
+
 
 def test_get_text_value():
     translator = GoldStudyTranslator()