Release 0.2.3

Author: christiam

CITATION.cff

@@ -1,8 +1,8 @@
 cff-version: "1.1.0"
 message: "If you use this software, please cite it using these metadata."
 title: ElasticBLAST
-version: "0.2.2"
-date-released: 2022-02-09
+version: "0.2.3"
+date-released: 2022-02-16
 license: "NCBI Public Domain"
 repository-code: "https://github.com/ncbi/elastic-blast/"
 authors: 

src/elastic_blast/elb_config.py

@@ -87,7 +87,7 @@
 from .util import validate_gke_cluster_name, ElbSupportedPrograms
 from .util import get_query_batch_size
 from .util import UserReportError, safe_exec
-from .util import gcp_get_regions
+from .util import gcp_get_regions, sanitize_for_k8s
 from .gcp_traits import get_machine_properties as gcp_get_machine_properties
 from .aws_traits import get_machine_properties as aws_get_machine_properties
 from .aws_traits import get_regions as aws_get_regions
@@ -989,7 +989,7 @@ def get_max_number_of_concurrent_blast_jobs(self) -> int:
 
 def generate_cluster_name(results: CloudURI) -> str:
     """ Returns the default cluster name """
-    username = sanitize_gcp_label(getpass.getuser().lower())
+    username = sanitize_for_k8s(sanitize_gcp_label(getpass.getuser().lower()))
     return f'elasticblast-{username}-{results.md5}'
 
 

src/elastic_blast/util.py

@@ -397,18 +397,6 @@ def sanitize_for_k8s(input_string: str) -> str:
     return re.sub(r'_', '-', input_string.lower(), flags=re.ASCII)
 
 
-def sanitize_gcp_labels(input_label: str) -> str:
-    """ Changes the input_label so that it is composed of valid GCP label characters"""
-    return re.sub(r'\W', '-', input_label.lower(), flags=re.ASCII)[:GCP_MAX_LABEL_LENGTH]
-
-
-def sanitize_aws_tag(input_label: str) -> str:
-    """ Changes the input_label so that it is composed of valid AWS tag characters"""
-    # NB: this AWS sanitizer is a bit more restrictive - it replaces '=' to
-    # simplify dataflow for GCP
-    return re.sub(r'[^\w_\.:/+@]', '-', input_label.strip(), flags=re.ASCII)[:AWS_MAX_TAG_LENGTH]
-
-
 def sanitize_aws_batch_job_name(input_name: str) -> str:
     """ Changes the input_name so that it is composed of valid AWS Batch job name characters"""
     return re.sub(r'[\W\-]', '-', input_name.strip(), flags=re.ASCII)[:AWS_MAX_JOBNAME_LENGTH]

tests/elb_config/test_elb_config.py

@@ -29,8 +29,8 @@
 import configparser
 from dataclasses import dataclass, fields
 from unittest.mock import MagicMock, patch
-import math
-from elastic_blast.constants import CSP
+import math, getpass
+from elastic_blast.constants import CSP, GCP_MAX_LABEL_LENGTH, AWS_MAX_TAG_LENGTH
 from elastic_blast.constants import CFG_CLOUD_PROVIDER
 from elastic_blast.constants import CFG_CP_GCP_PROJECT, CFG_CP_GCP_REGION, CFG_CP_GCP_ZONE
 from elastic_blast.constants import CFG_CP_GCP_NETWORK, CFG_CP_GCP_SUBNETWORK
@@ -67,6 +67,7 @@
 from elastic_blast.elb_config import CloudURI, GCPString, AWSRegion
 from elastic_blast.elb_config import GCPConfig, AWSConfig, BlastConfig, ClusterConfig
 from elastic_blast.elb_config import ElasticBlastConfig, get_instance_props
+from elastic_blast.elb_config import sanitize_gcp_label, sanitize_aws_tag
 from elastic_blast.constants import ElbCommand
 from elastic_blast.util import UserReportError, get_query_batch_size, ElbSupportedPrograms
 from elastic_blast.gcp_traits import get_machine_properties as gcp_get_machine_properties
@@ -638,6 +639,15 @@ def test_clusterconfig_from_configparser_errors():
         assert [message for message in errors if key in message and 'invalid value' in message and confpars[CFG_BLAST][key] in message]
 
 
+@patch(target='getpass.getuser', new=MagicMock(return_value='a-user-name_with_underscore'))
+def test_clusterconfig_username_with_underscore():
+    """Test that a username with an underscore is properly amended before
+       becoming part of a cluster name"""
+    assert '_' in getpass.getuser()
+    cfg = ClusterConfig(CloudURI('gs://some-bucket'))
+    assert '_' not in cfg.name
+
+
 TEST_MACHINE_TYPE = 'test-machine-type'
 
 @patch(target='elastic_blast.tuner.aws_get_machine_properties', new=MagicMock(return_value=InstanceProperties(32, 120)))
@@ -1707,3 +1717,20 @@ def test_mt_mode_and_batch_len_user_selected(gke_mock):
        assert cfg.blast.user_provided_batch_len
 
 
+def test_sanitize_gcp_label():
+    assert 'harry-potter' == sanitize_gcp_label('Harry.Potter')
+    assert 'macbook-pro-home' == sanitize_gcp_label('MacBook-Pro.Home')
+    label = sanitize_gcp_label('gs://tomcat-test/tc-elb-int-swissprot-psiblast-multi-node-sync-351')
+    assert len(label) <= GCP_MAX_LABEL_LENGTH
+    assert 'gs---tomcat-test-tc-elb-int-swissprot-psiblast-multi-node-sync-' == label
+
+
+def test_sanitize_gcp_user_name():
+    assert 'user-name' == sanitize_gcp_label('user.name')
+
+
+def test_sanitize_aws_tag():
+    assert 's3://abra-Cada-bra+-@.-' == sanitize_aws_tag('s3://abra;Cada#bra+-@.=')
+    label = sanitize_aws_tag('s3://tomcat-test/tc-elb-int-swissprot-psiblast-multi-node-sync-351')
+    assert len(label) <= AWS_MAX_TAG_LENGTH
+    assert 's3://tomcat-test/tc-elb-int-swissprot-psiblast-multi-node-sync-351' == label

tests/util/test_util.py

@@ -31,12 +31,11 @@
 
 from elastic_blast import util
 from elastic_blast.constants import ELB_DFLT_GCP_MACHINE_TYPE
-from elastic_blast.constants import GCP_MAX_LABEL_LENGTH, AWS_MAX_TAG_LENGTH
 from elastic_blast.constants import ElbCommand, MolType
 from elastic_blast.util import get_query_batch_size
 from elastic_blast.util import convert_memory_to_mb, get_blastdb_size, sanitize_aws_batch_job_name
 from elastic_blast.util import safe_exec, SafeExecError, convert_disk_size_to_gb
-from elastic_blast.util import sanitize_gcp_labels, sanitize_for_k8s, sanitize_aws_tag
+from elastic_blast.util import sanitize_for_k8s
 from elastic_blast.util import validate_gcp_string, convert_labels_to_aws_tags
 from elastic_blast.util import validate_gcp_disk_name, gcp_get_regions
 from elastic_blast.gcp_traits import get_machine_properties
@@ -139,33 +138,18 @@ def test_get_blastdb_size_invalid_database(self):
         with self.assertRaises(ValueError):
             get_blastdb_size(cfg.blast.db, cfg.cluster.db_source)
 
-    def test_sanitize_gcp_labels(self):
-        self.assertEqual('harry-potter', sanitize_gcp_labels('Harry.Potter'))
-        self.assertEqual('macbook-pro-home', sanitize_gcp_labels('MacBook-Pro.Home'))
-        label = sanitize_gcp_labels('gs://tomcat-test/tc-elb-int-swissprot-psiblast-multi-node-sync-351')
-        self.assertLessEqual(len(label), GCP_MAX_LABEL_LENGTH)
-        self.assertEqual('gs---tomcat-test-tc-elb-int-swissprot-psiblast-multi-node-sync-', label)
-
     def test_sanitize_for_k8s(self):
         self.assertEqual('ref-viruses-rep-genomes', sanitize_for_k8s('ref_viruses_rep_genomes'))
         self.assertEqual('betacoronavirus', sanitize_for_k8s('Betacoronavirus'))
         self.assertEqual('16s-ribosomal-rna', sanitize_for_k8s('16S_ribosomal_RNA'))
         self.assertEqual('gcf-000001405.38-top-level', sanitize_for_k8s('GCF_000001405.38_top_level'))
 
-    def test_sanitize_aws_tag(self):
-        self.assertEqual('s3://abra-Cada-bra+-@.-', sanitize_aws_tag('s3://abra;Cada#bra+-@.='))
-        label = sanitize_aws_tag('s3://tomcat-test/tc-elb-int-swissprot-psiblast-multi-node-sync-351')
-        self.assertLessEqual(len(label), AWS_MAX_TAG_LENGTH)
-        self.assertEqual('s3://tomcat-test/tc-elb-int-swissprot-psiblast-multi-node-sync-351', label)
-
     def test_sanitize_aws_batch_job_name(self):
         self.assertEqual('GCF_000001405-38_top_level', sanitize_aws_batch_job_name('GCF_000001405.38_top_level '))
 
     def test_sanitize_aws_user_name(self):
         self.assertEqual('user-name', sanitize_aws_batch_job_name('user.name'))
 
-    def test_sanitize_gcp_user_name(self):
-        self.assertEqual('user-name', sanitize_gcp_labels('user.name'))
 
 @patch(target='elastic_blast.elb_config.get_db_metadata', new=MagicMock(return_value=DB_METADATA))
 def create_config_for_db(dbname):