Merge branch 'interface-updates-epic' into cli-refactor

docs/scripts/generate_normalize_figure.py

@@ -12,8 +12,8 @@
 
 import gravis as gv
 
-from gene import APP_ROOT
 from gene.database import create_db
+from gene.etl.base import APP_ROOT
 from gene.query import QueryHandler
 from gene.schemas import UnmergedNormalizationService
 

src/gene/__init__.py

@@ -1,65 +1,4 @@
 """The VICC library for normalizing genes."""
-import logging
-from os import environ
-from pathlib import Path
+from .version import __version__
 
-from .version import __version__  # noqa: F401
-
-APP_ROOT = Path(__file__).resolve().parent
-
-logging.basicConfig(
-    filename="gene.log", format="[%(asctime)s] - %(name)s - %(levelname)s : %(message)s"
-)
-logger = logging.getLogger("gene")
-logger.setLevel(logging.DEBUG)
-logger.handlers = []
-
-logging.getLogger("boto3").setLevel(logging.INFO)
-logging.getLogger("botocore").setLevel(logging.INFO)
-logging.getLogger("urllib3").setLevel(logging.INFO)
-logging.getLogger("python_jsonschema_objects").setLevel(logging.INFO)
-logging.getLogger("biocommons.seqrepo.seqaliasdb.seqaliasdb").setLevel(logging.INFO)
-logging.getLogger("biocommons.seqrepo.fastadir.fastadir").setLevel(logging.INFO)
-
-
-SEQREPO_ROOT_DIR = Path(
-    environ.get("SEQREPO_ROOT_DIR", "/usr/local/share/seqrepo/latest")
-)
-
-
-class DownloadException(Exception):  # noqa: N818
-    """Exception for failures relating to source file downloads."""
-
-
-from gene.schemas import (  # noqa: E402
-    NamespacePrefix,
-    RefType,
-    SourceIDAfterNamespace,
-    SourceName,
-)
-
-ITEM_TYPES = {k.lower(): v.value for k, v in RefType.__members__.items()}
-
-# Sources we import directly (HGNC, Ensembl, NCBI)
-SOURCES = {
-    source.value.lower(): source.value for source in SourceName.__members__.values()
-}
-
-# Set of sources we import directly
-XREF_SOURCES = {src.lower() for src in SourceName.__members__}
-
-# use to fetch source name from schema based on concept id namespace
-# e.g. {"hgnc": "HGNC"}
-PREFIX_LOOKUP = {
-    v.value: SourceName[k].value
-    for k, v in NamespacePrefix.__members__.items()
-    if k in SourceName.__members__.keys()
-}
-
-# use to generate namespace prefix from source ID value
-# e.g. {"ensg": "ensembl"}
-NAMESPACE_LOOKUP = {
-    v.value.lower(): NamespacePrefix[k].value
-    for k, v in SourceIDAfterNamespace.__members__.items()
-    if v.value != ""
-}
+__all__ = ["__version__"]

src/gene/database/dynamodb.py

@@ -11,7 +11,6 @@
 from boto3.dynamodb.conditions import Key
 from botocore.exceptions import ClientError
 
-from gene import ITEM_TYPES, PREFIX_LOOKUP
 from gene.database.database import (
     AWS_ENV_VAR_NAME,
     SKIP_AWS_DB_ENV_NAME,
@@ -23,7 +22,14 @@
     DatabaseWriteException,
     confirm_aws_db_use,
 )
-from gene.schemas import RecordType, RefType, SourceMeta, SourceName
+from gene.schemas import (
+    ITEM_TYPES,
+    PREFIX_LOOKUP,
+    RecordType,
+    RefType,
+    SourceMeta,
+    SourceName,
+)
 
 logger = logging.getLogger(__name__)
 

src/gene/etl/base.py

@@ -6,7 +6,7 @@
 import shutil
 from abc import ABC, abstractmethod
 from ftplib import FTP
-from os import remove
+from os import environ, remove
 from pathlib import Path
 from typing import Callable, Dict, List, Optional
 
@@ -15,14 +15,19 @@
 from dateutil import parser
 from gffutils.feature import Feature
 
-from gene import ITEM_TYPES, SEQREPO_ROOT_DIR
 from gene.database import AbstractDatabase
-from gene.schemas import Gene, GeneSequenceLocation, MatchType, SourceName
+from gene.schemas import ITEM_TYPES, Gene, GeneSequenceLocation, MatchType, SourceName
 
 logger = logging.getLogger("gene")
 logger.setLevel(logging.DEBUG)
 
 
+APP_ROOT = Path(__file__).resolve().parent
+SEQREPO_ROOT_DIR = Path(
+    environ.get("SEQREPO_ROOT_DIR", "/usr/local/share/seqrepo/latest")
+)
+
+
 class Base(ABC):
     """The ETL base class."""
 

src/gene/etl/ensembl.py

@@ -10,9 +10,8 @@
 import requests
 from gffutils.feature import Feature
 
-from gene import APP_ROOT
 from gene.database import AbstractDatabase
-from gene.etl.base import Base
+from gene.etl.base import APP_ROOT, Base
 from gene.etl.exceptions import (
     GeneFileVersionError,
     GeneNormalizerEtlError,

src/gene/etl/hgnc.py

@@ -10,15 +10,15 @@
 
 from dateutil import parser
 
-from gene import APP_ROOT, PREFIX_LOOKUP
 from gene.database import AbstractDatabase
-from gene.etl.base import Base
+from gene.etl.base import APP_ROOT, Base
 from gene.etl.exceptions import (
     GeneFileVersionError,
     GeneNormalizerEtlError,
     GeneSourceFetchError,
 )
 from gene.schemas import (
+    PREFIX_LOOKUP,
     Annotation,
     Chromosome,
     NamespacePrefix,

src/gene/etl/ncbi.py

@@ -9,15 +9,15 @@
 
 import gffutils
 
-from gene import APP_ROOT, PREFIX_LOOKUP
 from gene.database import AbstractDatabase
-from gene.etl.base import Base
+from gene.etl.base import APP_ROOT, Base
 from gene.etl.exceptions import (
     GeneFileVersionError,
     GeneNormalizerEtlError,
     GeneSourceFetchError,
 )
 from gene.schemas import (
+    PREFIX_LOOKUP,
     Annotation,
     Chromosome,
     NamespacePrefix,

src/gene/main.py

@@ -6,8 +6,14 @@
 
 from gene import __version__
 from gene.database import create_db
-from gene.query import InvalidParameterException, QueryHandler
-from gene.schemas import NormalizeService, SearchService, UnmergedNormalizationService
+from gene.query import QueryHandler
+from gene.schemas import (
+    SOURCES,
+    NormalizeService,
+    SearchService,
+    SourceName,
+    UnmergedNormalizationService,
+)
 
 db = create_db()
 query_handler = QueryHandler(db)
@@ -42,14 +48,10 @@
 read_query_summary = "Given query, provide best-matching source records."
 response_description = "A response to a validly-formed query"
 q_descr = "Gene to normalize."
-incl_descr = """Optional. Comma-separated list of source names to include in
-             response. Will exclude all other sources. Returns HTTP status code
-             422: Unprocessable Entity if both 'incl' and 'excl' parameters
-             are given."""
-excl_descr = """Optional. Comma-separated list of source names to exclude in
-             response. Will include all other sources. Returns HTTP status
-             code 422: Unprocessable Entity if both 'incl' and 'excl'
-             parameters are given."""
+sources_descr = (
+    "Optional. Comma-separated list of source names to include in response, if given. "
+    "Will exclude all other sources."
+)
 search_description = (
     "For each source, return strongest-match concepts "
     "for query string provided by user"
@@ -66,24 +68,29 @@
 )
 def search(
     q: str = Query(..., description=q_descr),  # noqa: D103
-    incl: Optional[str] = Query(None, description=incl_descr),
-    excl: Optional[str] = Query(None, description=excl_descr),
+    sources: Optional[str] = Query(None, description=sources_descr),
 ) -> SearchService:
     """Return strongest match concepts to query string provided by user.
 
-    :param str q: gene search term
-    :param Optional[str] incl: comma-separated list of sources to include,
-        with all others excluded. Raises HTTPException if both `incl` and
-        `excl` are given.
-    :param Optional[str] excl: comma-separated list of sources exclude, with
-        all others included. Raises HTTPException if both `incl` and `excl`
-        are given.
+    :param q: gene search term
+    :param sources: If given, search only for records from these sources.
+        Provide as string of source names separated by commas.
     :return: JSON response with matched records and source metadata
     """
-    try:
-        resp = query_handler.search(html.unescape(q), incl=incl, excl=excl)
-    except InvalidParameterException as e:
-        raise HTTPException(status_code=422, detail=str(e))
+    parsed_sources = []
+    if sources:
+        for candidate_source in sources.split(","):
+            try:
+                parsed_source = SourceName[
+                    SOURCES[candidate_source.strip().lower()].upper()
+                ]
+            except KeyError:
+                raise HTTPException(
+                    status_code=422,
+                    detail=f"Unable to parse source name: {candidate_source}",
+                )
+            parsed_sources.append(parsed_source)
+    resp = query_handler.search(html.unescape(q), sources=parsed_sources)
     return resp