mtbls-mhd-integration 0.0.3__py3-none-any.whl → 0.0.5__py3-none-any.whl

mtbls2mhd/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "v0.0.3"
+__version__ = "v0.0.5"
 
 import pathlib
 import sys

mtbls2mhd/config.py

@@ -24,6 +24,7 @@ class Mtbls2MhdConfiguration(BaseSettings):
     default_dataset_licence_url: str = (
         "https://creativecommons.org/publicdomain/zero/1.0"
     )
+    default_mhd_model_version: str = "0.1"
     mtbls_studies_root_path: str
     model_config = SettingsConfigDict(env_file=".env", env_file_encoding="utf-8")
 

mtbls2mhd/v0_1/legacy/builder.py

@@ -1,3 +1,4 @@
+import datetime
 import enum
 import json
 import logging
@@ -52,7 +53,16 @@ logger = logging.getLogger(__name__)
 MTBLS_ASSAY_TYPES = {
     "LC-MS": COMMON_ASSAY_TYPES["OBI:0003097S"],
     "GC-MS": COMMON_ASSAY_TYPES["OBI:0003110"],
-    # TODO: Add CE-MS, FIA-MS, DI-MS for MetaboLights
+    "CE-MS": CvTerm(
+        source="OBI",
+        accession="OBI:0003741",
+        name="capillary electrophoresis mass spectrometry assay",
+    ),
+    "GCxGC-MS": COMMON_ASSAY_TYPES["OBI:0003110"],
+    "FIA-MS": COMMON_ASSAY_TYPES["OBI:0000470"],
+    "MALDI-MS": COMMON_ASSAY_TYPES["OBI:0000470"],
+    "DI-MS": COMMON_ASSAY_TYPES["OBI:0000470"],
+    "MS": COMMON_ASSAY_TYPES["OBI:0000470"],
 }
 MTBLS_MEASUREMENT_TYPES = {
     "targeted": COMMON_MEASUREMENT_TYPES["MSIO:0000100"],
@@ -73,13 +83,20 @@ COMMON_PROTOCOLS_MAP = {
     "Treatment": COMMON_PROTOCOLS["EFO:0003969"],
     "Flow Injection Analysis": COMMON_PROTOCOLS["MS:1000058"],
     "Capillary Electrophoresis": COMMON_PROTOCOLS["CHMO:0001024"],
-    "Direct infusion": COMMON_PROTOCOLS.get(
-        "CHMO:0001024"
-    ),  # TODO: Update after adding to managed CV terms
+    # TODO: Update after adding to managed CV terms
 }
 
 MTBLS_PROTOCOLS_MAP = COMMON_PROTOCOLS_MAP.copy()
 
+MTBLS_PROTOCOLS_MAP.update(
+    {
+        "Direct infusion": CvTerm(
+            source="MS",
+            accession="MS:1000060",
+            name="infusion",
+        ),
+    }
+)
 MANAGED_CHARACTERISTICS_MAP = {
     "organism": COMMON_CHARACTERISTIC_DEFINITIONS["NCIT:C14250"],
     "organism part": COMMON_CHARACTERISTIC_DEFINITIONS["NCIT:C103199"],
@@ -1877,7 +1894,23 @@ class MhdLegacyDatasetBuilder:
             error = f"{data.investigation_file_path} file does not have any study. Skipping..."
             logger.warning(error)
             return False, error
-
+        if not revision:
+            db_metadata = data.study_db_metadata
+            revision_date = (
+                datetime.datetime.strptime(db_metadata.revision_date, "%Y-%m-%d")
+                if db_metadata and db_metadata.revision_date
+                else None
+            )
+            if revision_date:
+                revision = Revision(
+                    revision_datetime=revision_date,
+                    revision=db_metadata.revision_number
+                    if db_metadata.revision_number
+                    else 0,
+                    comment=db_metadata.revision_comment
+                    if db_metadata.revision_comment
+                    else "",
+                )
         selected_assays: list[Assay] = []
         study = data.investigation.studies[0]
 
@@ -1913,9 +1946,14 @@ class MhdLegacyDatasetBuilder:
             repository_identifier=study.identifier,
             schema_name=target_mhd_model_schema_uri,
             profile_uri=target_mhd_model_profile_uri,
-            repository_revision=revision.revision if revision else 1,
+            repository_revision=revision.revision
+            if revision and revision and revision.revision
+            else 0,
             repository_revision_datetime=revision.revision_datetime
-            if revision
+            if revision and revision.revision_datetime
+            else None,
+            repository_revision_comment=revision.comment
+            if revision and revision.comment
             else None,
             change_log=[revision] if revision else None,
         )
@@ -1970,11 +2008,9 @@ class MhdLegacyDatasetBuilder:
         self.add_study_factor_definitions(mhd_builder, mhd_study, data, build_type)
         samples = self.add_samples(mhd_builder, mhd_study, sample_file, build_type)
         if build_type in (BuildType.FULL, BuildType.FULL_AND_CUSTOM_NODES):
-            mhd_study.license = (
-                HttpUrl(config.default_dataset_licence_url)
-                if config.default_dataset_licence_url
-                else None
-            )
+            mhd_study.license = data.study_db_metadata.dataset_license_url
+            if not mhd_study.license:
+                mhd_study.license = HttpUrl(config.default_dataset_licence_url) or ""
 
             self.add_publications(data, mhd_builder, mhd_study)
             self.add_protocols(mhd_builder, mhd_study, study)

mtbls2mhd/v0_1/legacy/db_metadata_collector.py

@@ -4,10 +4,11 @@ from functools import lru_cache
 from logging import getLogger
 from typing import Any, Dict, List
 
-import psycopg2
+import psycopg
 from metabolights_utils.models.common import ErrorMessage
 from metabolights_utils.models.metabolights.model import (
     CurationRequest,
+    StudyCategory,
     StudyDBMetadata,
     StudyStatus,
     Submitter,
@@ -15,7 +16,7 @@ from metabolights_utils.models.metabolights.model import (
     UserStatus,
 )
 from metabolights_utils.provider.study_provider import AbstractDbMetadataCollector
-from psycopg2.extras import DictCursor
+from psycopg.rows import dict_row
 from sqlalchemy import or_, select
 from sqlalchemy.ext.asyncio import AsyncSession, create_async_engine
 from sqlalchemy.orm import sessionmaker
@@ -59,22 +60,6 @@ def get_session_factory(mtbls2mhd_config: Mtbls2MhdConfiguration):
     return AsyncSessionFactory
 
 
-STUDY_FIELDS = [
-    "id",
-    "acc",
-    "obfuscationcode",
-    "submissiondate",
-    "releasedate",
-    "updatedate",
-    "studysize",
-    "status_date",
-    "studytype",
-    "status",
-    "override",
-    "comment",
-    "curation_request",
-]
-
 SUBMITTER_FIELDS = [
     "id",
     "orcid",
@@ -98,15 +83,16 @@ def create_postgresql_connection(mtbls2mhd_config: Mtbls2MhdConfiguration):
     Creates and returns a PostgreSQL connection.
     """
     try:
-        connection = psycopg2.connect(
+        connection = psycopg.connect(
             dbname=mtbls2mhd_config.database_name,
             user=mtbls2mhd_config.database_user,
             password=mtbls2mhd_config.database_user_password,
             host=mtbls2mhd_config.database_host,
             port=mtbls2mhd_config.database_host_port,
+            row_factory=dict_row,
         )
         return connection
-    except psycopg2.Error as e:
+    except psycopg.Error as e:
         logger.exception(e)
         raise e
 
@@ -118,10 +104,18 @@ class DbMetadataCollector(AbstractDbMetadataCollector):
     def get_study_metadata_from_db(self, study_id: str, connection):
         try:
             study = self._get_study_from_db(study_id, connection)
+            revision = None
+            if study["revision_number"] and study["revision_number"] > 0:
+                revision = self._get_study_revision_from_db(
+                    study_id, study["revision_number"], connection
+                )
             submitters = self._get_study_submitters_from_db(study_id, connection)
-            study_db_metadata = self._create_study_db_metadata(study, submitters)
+            study_db_metadata = self._create_study_db_metadata(
+                study, revision, submitters
+            )
             return study_db_metadata, []
         except Exception as ex:
+            logger.exception(ex)
             return StudyDBMetadata(), [
                 ErrorMessage(short="Error while loading db metadata", detail=str(ex))
             ]
@@ -178,7 +172,7 @@ class DbMetadataCollector(AbstractDbMetadataCollector):
         where_clause = " and ".join(_filter)
         _input = f"select acc from studies where {where_clause};"
         try:
-            cursor = connection.cursor(cursor_factory=DictCursor)
+            cursor = connection.cursor()
             cursor.execute(_input)
             data = cursor.fetchall()
             return data
@@ -201,11 +195,9 @@ class DbMetadataCollector(AbstractDbMetadataCollector):
             return study_ids
 
     def _get_study_from_db(self, study_id: str, connection):
-        _input = (
-            f"select {', '.join(STUDY_FIELDS)} from studies where acc = %(study_id)s;"
-        )
+        _input = "select * from studies where acc = %(study_id)s;"
         try:
-            cursor = connection.cursor(cursor_factory=DictCursor)
+            cursor = connection.cursor()
             cursor.execute(_input, {"study_id": study_id})
             data = cursor.fetchone()
             return data
@@ -213,13 +205,24 @@ class DbMetadataCollector(AbstractDbMetadataCollector):
         except Exception as ex:
             raise ex
 
+    def _get_study_revision_from_db(self, study_id: str, revision: int, connection):
+        _input = "select * from study_revisions where accession_number = %(study_id)s and revision_number = %(revision)s;"
+        try:
+            cursor = connection.cursor()
+            cursor.execute(_input, {"study_id": study_id, "revision": revision})
+            data = cursor.fetchone()
+            return data
+
+        except Exception as ex:
+            raise ex
+
     def _get_study_submitters_from_db(self, study_id: str, connection):
         submitter_fields = [f"u.{field}" for field in SUBMITTER_FIELDS]
 
         _input = f"select {', '.join(submitter_fields)} from studies as s, study_user as su, \
             users as u where su.userid = u.id and su.studyid = s.id and s.acc = %(study_id)s;"
         try:
-            cursor = connection.cursor(cursor_factory=DictCursor)
+            cursor = connection.cursor()
             cursor.execute(_input, {"study_id": study_id})
             data = cursor.fetchall()
             if data:
@@ -229,8 +232,19 @@ class DbMetadataCollector(AbstractDbMetadataCollector):
         except Exception as ex:
             raise ex
 
+    LICENSE_URLS = {
+        (
+            "CC0 1.0 UNIVERSAL",
+            "1.0",
+        ): "https://creativecommons.org/publicdomain/zero/1.0/",
+        (
+            "EMBL-EBI TERMS OF USE",
+            "5TH FEBRUARY 2024",
+        ): "https://www.ebi.ac.uk/about/terms-of-use/",
+    }
+
     def _create_study_db_metadata(
-        self, study, submitters: List[Dict[str, Any]]
+        self, study, revision, submitters: List[Dict[str, Any]]
     ) -> StudyDBMetadata:
         study_db_metadata: StudyDBMetadata = StudyDBMetadata()
         study_db_metadata.db_id = study["id"] or -1
@@ -248,26 +262,49 @@ class DbMetadataCollector(AbstractDbMetadataCollector):
         if study["studytype"] and len(study["studytype"].strip()) > 0:
             study_db_metadata.study_types = study["studytype"].strip().split(";")
 
-        if study["override"] and len(study["override"].strip()) > 0:
-            override_list = study["override"].strip().split("|")
-            overrides = {}
-            for item in override_list:
-                if item:
-                    key_value = item.split(":")
-                    if len(key_value) > 1:
-                        overrides[key_value[0]] = key_value[1] or ""
-            study_db_metadata.overrides = overrides
         study_db_metadata.study_size = int(study["studysize"])
-        study_db_metadata.submission_date = self._get_date_string(
-            study["submissiondate"]
+
+        study_db_metadata.first_private_date = self._get_date_string(
+            study["first_private_date"]
         )
+        study_db_metadata.submission_date = study_db_metadata.first_private_date
+
         study_db_metadata.curation_request = CurationRequest.get_from_int(
             study["curation_request"]
         )
-        study_db_metadata.release_date = self._get_date_string(study["releasedate"])
+        study_db_metadata.first_public_date = self._get_date_string(
+            study["first_public_date"]
+        )
+        study_db_metadata.release_date = study_db_metadata.first_public_date
         study_db_metadata.update_date = self._get_date_time_string(study["updatedate"])
         study_db_metadata.status_date = self._get_date_time_string(study["status_date"])
         study_db_metadata.submitters = self._create_submitters(submitters)
+        if revision:
+            study_db_metadata.revision_number = revision.get("revision_number", 0)
+            study_db_metadata.revision_comment = revision.get("revision_comment", "")
+            study_db_metadata.revision_date = self._get_date_string(
+                revision.get("revision_datetime", "")
+            )
+        study_db_metadata.dataset_license = study["dataset_license"] or ""
+        study_db_metadata.dataset_license_version = (
+            study["dataset_license_version"] or ""
+        )
+        study_db_metadata.dataset_license_url = self.LICENSE_URLS.get(
+            (
+                study_db_metadata.dataset_license.upper(),
+                study_db_metadata.dataset_license_version.upper(),
+            ),
+        )
+        study_db_metadata.study_category = StudyCategory(study["study_category"])
+        study_db_metadata.mhd_model_version = study["mhd_model_version"]
+        study_db_metadata.reserved_mhd_accession = study["mhd_accession"] or ""
+        study_db_metadata.created_at = (
+            self._get_date_time_string(study["created_at"]) or ""
+        )
+        study_db_metadata.study_template = study["study_template"] or ""
+        study_db_metadata.sample_template = study["sample_type"] or ""
+        study_db_metadata.template_version = study["template_version"] or ""
+
         return study_db_metadata
 
     def _create_submitters(self, submitters: List[Dict[str, Any]]) -> List[Submitter]:

{mtbls_mhd_integration-0.0.3.dist-info → mtbls_mhd_integration-0.0.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mtbls-mhd-integration
-Version: 0.0.3
+Version: 0.0.5
 Summary: MetaboLights - MetabolomicsHub Integration
 Author-email: MetaboLights Team <metabolights-help@ebi.ac.uk>
 License-Expression: Apache-2.0
@@ -9,8 +9,8 @@ Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: asyncpg>=0.30.0
 Requires-Dist: metabolights-utils>=1.4.16
-Requires-Dist: mhd-model>=0.1.37
-Requires-Dist: psycopg2-binary>=2.9.11
+Requires-Dist: mhd-model>=0.1.39
+Requires-Dist: psycopg[binary,pool]>=3.3.2
 Requires-Dist: pydantic>=2.12.4
 Requires-Dist: pydantic-settings>=2.10.1
 Requires-Dist: pyyaml>=6.0.3

{mtbls_mhd_integration-0.0.3.dist-info → mtbls_mhd_integration-0.0.5.dist-info}/RECORD

@@ -1,5 +1,5 @@
-mtbls2mhd/__init__.py,sha256=4oQxeAlcyP87VQOUq168Fi3sYGp0s98KvT6gDZltZWw,157
-mtbls2mhd/config.py,sha256=z_iN3rUolcLXz_Q55DraGmyOUhJhP0xK4zlmaj-VWxw,2267
+mtbls2mhd/__init__.py,sha256=3byMWcFYU6S6BKe-liUPRtmlSwQrjbuEljIatokiM3s,157
+mtbls2mhd/config.py,sha256=BjOqAyfDhp9byoFjJz70xh4HRR8pu1yrm_5jweqygSI,2310
 mtbls2mhd/convertor_factory.py,sha256=4loatqIRIvIhcaeIS0cSonJNYJu47o56ZllX6593ypk,1133
 mtbls2mhd/commands/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mtbls2mhd/commands/cli.py,sha256=XhLcqBxmBNX_nSEc4aU9q2D9VoYvqZKeRuYLbt-JwZ8,662
@@ -8,16 +8,16 @@ mtbls2mhd/commands/create_mhd_file.py,sha256=0sDr-Cm0JhhEB5V1g66uoag3rlcaAnGP8Md
 mtbls2mhd/commands/validate.py,sha256=iwIKegviRxdH0r8scRXbDISlwQUzAq5uVoCHinU7x6Q,473
 mtbls2mhd/v0_1/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mtbls2mhd/v0_1/legacy/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-mtbls2mhd/v0_1/legacy/builder.py,sha256=SdNvxZMMyDz0UteGn9b-syqrfVge7_sU4xQH38rXFm0,84029
+mtbls2mhd/v0_1/legacy/builder.py,sha256=Kz4bHoGIUFTaj3nJaHs5OFG095sgbHmOxd5r-rlCeL0,85414
 mtbls2mhd/v0_1/legacy/convertor.py,sha256=gewPleS8FH3TS63dALqtIUuX26dAkIFo701UMlKiOE8,2607
-mtbls2mhd/v0_1/legacy/db_metadata_collector.py,sha256=OFpF0L0ueFC90r3jq27RdHo86L7myJ0ah7CcAPwGLZQ,11346
+mtbls2mhd/v0_1/legacy/db_metadata_collector.py,sha256=4OyA_KD2X2zr7AHn8pQZJ0Y2_bW5r2_2wgTtQ93LM6A,13193
 mtbls2mhd/v0_1/legacy/folder_metadata_collector.py,sha256=1lELGwTsr12nBGwTog_Z8qi9dLt4awma56vBYoI678k,7439
 mtbls2mhd/v0_1/legacy/mtbls_study_schema.py,sha256=gUTbRmI8GfHI5leLiw8dxsmWnV3NnWw5RPX_LQWRFRQ,3162
 mtbls2mhd/v0_1/ms/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mtbls2mhd/v0_1/ms/convertor.py,sha256=kLIUpxOrH6hcs2Y9Bq1D0Mdvypg40pLyEJpHtGj6H_g,89
-mtbls_mhd_integration-0.0.3.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-mtbls_mhd_integration-0.0.3.dist-info/METADATA,sha256=J88s4MwzG1rogEzcJrgeBxwIv461t7UHD3If6ow8yiU,678
-mtbls_mhd_integration-0.0.3.dist-info/WHEEL,sha256=qELbo2s1Yzl39ZmrAibXA2jjPLUYfnVhUNTlyF1rq0Y,92
-mtbls_mhd_integration-0.0.3.dist-info/entry_points.txt,sha256=WQjM4flaYMyvHyv9zGKjCVk1i1_FGdNlhTmFVGgLgxs,61
-mtbls_mhd_integration-0.0.3.dist-info/top_level.txt,sha256=b7pI95n6HIQMFXDD0yL1NwldiDc-XdeWql4Iw-uYygQ,10
-mtbls_mhd_integration-0.0.3.dist-info/RECORD,,
+mtbls_mhd_integration-0.0.5.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+mtbls_mhd_integration-0.0.5.dist-info/METADATA,sha256=2BQEjtVChMbUMm737D2dUGEWB2kJoh0oUvmM0Dl57Hs,682
+mtbls_mhd_integration-0.0.5.dist-info/WHEEL,sha256=qELbo2s1Yzl39ZmrAibXA2jjPLUYfnVhUNTlyF1rq0Y,92
+mtbls_mhd_integration-0.0.5.dist-info/entry_points.txt,sha256=WQjM4flaYMyvHyv9zGKjCVk1i1_FGdNlhTmFVGgLgxs,61
+mtbls_mhd_integration-0.0.5.dist-info/top_level.txt,sha256=b7pI95n6HIQMFXDD0yL1NwldiDc-XdeWql4Iw-uYygQ,10
+mtbls_mhd_integration-0.0.5.dist-info/RECORD,,