phc-ingestion 0.10.11__py3-none-any.whl → 0.10.13__py3-none-any.whl

ingestion/nebula/manifest_assembler.py

@@ -1,6 +1,10 @@
-from typing import Optional, TypedDict, Any
-from datetime import datetime
+import re
+import gzip
+from typing import Optional, TypedDict, Any, Callable
+from datetime import datetime, timezone
+from dateutil import parser
 from ingestion.shared_util.lambda_client import LambdaClient
+from ingestion.vcf_standardization.util.read_write import read_headers
 from ingestion.nebula.constants import (
     DATASET_SYSTEM,
     NEBULA_KIT_ID_SYSTEM,
@@ -9,7 +13,6 @@ from ingestion.nebula.constants import (
     NEBULA_BODY_SITE,
     NEBULA_INDICATION,
 )
-import re
 
 
 class HumanName(TypedDict):
@@ -49,11 +52,14 @@ class Practitioner(Resource):
 
 
 class ManifestAssembler:
-    def __init__(self, ingestion_id: str, account_id: str, project_id: str, kit_id: str):
+    def __init__(
+        self, ingestion_id: str, account_id: str, project_id: str, kit_id: str, vcf_file_path: str
+    ):
         self.ingestion_id = ingestion_id
         self.account_id = account_id
         self.project_id = project_id
         self.kit_id = kit_id
+        self.vcf_file_path = vcf_file_path
         self.client = LambdaClient(
             "patient-service",
             {
@@ -142,6 +148,37 @@ class ManifestAssembler:
             "fullName": f"{first_name} {last_name}",
         }
 
+    def _safe(self, lambda_func: Callable[[], Any]) -> Any:
+        try:
+            return lambda_func()
+        except Exception as e:
+            return None
+
+    def __extract_report_date(self) -> dict[str, str]:
+        in_file = self.vcf_file_path
+        with gzip.open(in_file, "rt") if in_file.endswith(".gz") else open(in_file, "r") as f:
+            headers = read_headers(f)
+            for header in headers:
+                epoch_parts = re.search(r"Epoch=(\d+)", header)
+                epoch_time_raw = epoch_parts.group(1) if epoch_parts else None
+                epoch_time = self._safe(
+                    lambda: datetime.fromtimestamp(int(epoch_time_raw) / 1000.0, tz=timezone.utc)
+                )
+                date_parts = re.search(r"Date=\"(.*)\"", header)
+                date_string_raw = date_parts.group(1) if date_parts else None
+                date_string = self._safe(
+                    lambda: parser.parse(date_string_raw, tzinfos={"GST": "UTC+4"})
+                )
+
+                report_date = epoch_time or date_string
+                if report_date:
+                    return {
+                        "reportDate": report_date.astimezone(tz=timezone.utc)
+                        .isoformat(timespec="milliseconds")
+                        .replace("+00:00", "Z")
+                    }
+        return {}
+
     def create_manifest(self) -> dict[str, Any]:
         patient = self.__fetch_patient_by_kit_id()
 
@@ -167,6 +204,7 @@ class ManifestAssembler:
         return {
             "name": "Nebula",
             "indexedDate": datetime.now().strftime("%Y-%m-%d"),
+            **self.__extract_report_date(),
             "reference": "GRCh38",
             "patientId": patient.get("id"),
             "mrn": self.__extract_elation_mrn(patient),

ingestion/nebula/process.py

@@ -15,8 +15,9 @@ def process(vcf_file, out_path, file_name, source_file_id, ingestion_id, account
         )
 
         case_id = file_name
-        manifest_assembler = ManifestAssembler(ingestion_id, account_id, project_id, case_id)
-        manifest = manifest_assembler.create_manifest()
+        manifest = ManifestAssembler(
+            ingestion_id, account_id, project_id, case_id, vcf_file
+        ).create_manifest()
         base_vcf_file = os.path.basename(vcf_file)
         vcf_out = base_vcf_file.replace(".vcf", ".modified.vcf")
         vcf_final = base_vcf_file.replace(".vcf", ".modified.nrm.filtered.vcf")
@@ -35,10 +36,10 @@ def process(vcf_file, out_path, file_name, source_file_id, ingestion_id, account
         manifest["testType"] = NEBULA_TEST_TYPE
         manifest["reportID"] = case_id
         manifest["sourceFileId"] = source_file_id
-        manifest["resources"] = [{"fileName": f".lifeomic/vcf-ingest/{case_id}/{base_vcf_file}"}]
+        manifest["resources"] = [{"fileName": f".lifeomic/nebula/{case_id}/{base_vcf_file}"}]
         manifest["files"] = [
             {
-                "fileName": f".lifeomic/vcf-ingest/{case_id}/{vcf_final}",
+                "fileName": f".lifeomic/nebula/{case_id}/{vcf_final}",
                 "sequenceType": "germline",
                 "type": "shortVariant",
             }

{phc_ingestion-0.10.11.dist-info → phc_ingestion-0.10.13.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 0.10.11
+Version: 0.10.13
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>

{phc_ingestion-0.10.11.dist-info → phc_ingestion-0.10.13.dist-info}/RECORD

@@ -28,8 +28,8 @@ ingestion/generic/process.py,sha256=ZaVnZ_gx9faDUsuresI1A0oCegTa-dPQT7DBFMeZGyY,
 ingestion/generic/utils.py,sha256=1MEIru7uq38IjUdL8lcHqDH0oTki9uWrz1f2e-pmRoU,2814
 ingestion/nebula/__init__.py,sha256=VauK-rup_N8ZXVohx3HYqHX_PE_WoPyMUhdv2R7al4o,45
 ingestion/nebula/constants.py,sha256=thKqSwemdaAwAmKvF4FEVI9l1Ph5ergsnMlx6nWte7E,357
-ingestion/nebula/manifest_assembler.py,sha256=Og1g_HD0cR9osH5eWdttgFEEayimaZRO7QckC9U9YKM,6383
-ingestion/nebula/process.py,sha256=vQZIyQH3iL2PG1xyyKLv5PsnnVsQ_fBBGzrP8oRJ5FA,2347
+ingestion/nebula/manifest_assembler.py,sha256=4lVZGiUc8DBkxb2W6zWOoX3azJf0VUMWYC20-T7wDNA,7989
+ingestion/nebula/process.py,sha256=N9OuipynGV_XgEL3nO5I_-di1tk9szOy8LqsyNTw0E0,2323
 ingestion/nextgen/__init__.py,sha256=7LQ-h_Bvc5P1QcHMdzsqi1Qm4fTJn04-ozar2ty9wSc,59
 ingestion/nextgen/process.py,sha256=5Z0RfclwTAYZruGDiLPutjPCYFh1DJpoWY9dnttghT4,3993
 ingestion/nextgen/util/alteration_table.py,sha256=JTWBL1Fqj_pGsH5vwuVEnCUJle2wOBk6VYImHYCF9vg,6129
@@ -59,6 +59,6 @@ ingestion/vcf_standardization/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQe
 ingestion/vcf_standardization/util/af_helpers.py,sha256=dpTzoeIQVeBRt0ETF3a9rp5ojZqznHg4x_hCZ8OPcOg,1061
 ingestion/vcf_standardization/util/dp_helpers.py,sha256=Nq8oLOLObu4_pv16qwwgpALRlUoJVCULrd9cFOD-eoI,823
 ingestion/vcf_standardization/util/read_write.py,sha256=x3Pf6Dq8tmolblbCS5CrNmrcHS3FGfqBSFpFgvFGC4g,2526
-phc_ingestion-0.10.11.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
-phc_ingestion-0.10.11.dist-info/METADATA,sha256=tRkIKGt0Yqk_NmBaE0qtD-tko63m3LozUr2V7ycoYzs,678
-phc_ingestion-0.10.11.dist-info/RECORD,,
+phc_ingestion-0.10.13.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
+phc_ingestion-0.10.13.dist-info/METADATA,sha256=7CQQcYy4OIzsmQ-0mv5ZiO1KV9npZF62M61s2633IS0,678
+phc_ingestion-0.10.13.dist-info/RECORD,,