EVA-3590 update initial check for validation required or not (#40)

* update validation required or not logic

bin/eva-sub-cli.py

@@ -1,6 +1,9 @@
 #!/usr/bin/env python
 import sys
 
+from eva_sub_cli.exceptions.submission_not_found_exception import SubmissionNotFoundException
+from eva_sub_cli.exceptions.submission_status_exception import SubmissionStatusException
+
 if not sys.warnoptions:
     import warnings
     warnings.simplefilter("ignore")
@@ -83,3 +86,7 @@ def get_version():
         main.orchestrate_process(**args.__dict__)
     except FileNotFoundError as fne:
         print(fne)
+    except SubmissionNotFoundException as snfe:
+        print(f'{snfe}. Please contact EVA Helpdesk')
+    except SubmissionStatusException as sse:
+        print(f'{sse}. Please try again later. If the problem persists, please contact EVA Helpdesk')

eva_sub_cli/exceptions/submission_not_found_exception.py

@@ -0,0 +1,4 @@
+class SubmissionNotFoundException(Exception):
+    def __init__(self, message):
+        self.message = message
+        super().__init__(self.message)

eva_sub_cli/exceptions/submission_status_exception.py

@@ -0,0 +1,5 @@
+
+class SubmissionStatusException(Exception):
+    def __init__(self, message):
+        self.message = message
+        super().__init__(self.message)

eva_sub_cli/main.py

@@ -4,20 +4,26 @@
 import os
 from collections import defaultdict
 
+import requests
 from ebi_eva_common_pyutils.config import WritableConfig
+from ebi_eva_common_pyutils.logger import logging_config
 from openpyxl.reader.excel import load_workbook
 
 from eva_sub_cli import SUB_CLI_CONFIG_FILE, __version__
+from eva_sub_cli.exceptions.submission_not_found_exception import SubmissionNotFoundException
+from eva_sub_cli.exceptions.submission_status_exception import SubmissionStatusException
+from eva_sub_cli.submission_ws import SubmissionWSClient
+from eva_sub_cli.submit import StudySubmitter, SUB_CLI_CONFIG_KEY_SUBMISSION_ID
 from eva_sub_cli.validators.docker_validator import DockerValidator
 from eva_sub_cli.validators.native_validator import NativeValidator
 from eva_sub_cli.validators.validator import READY_FOR_SUBMISSION_TO_EVA
-from eva_sub_cli.submit import StudySubmitter
 
 VALIDATE = 'validate'
 SUBMIT = 'submit'
 DOCKER = 'docker'
 NATIVE = 'native'
 
+logger = logging_config.get_logger(__name__)
 
 def get_vcf_files(mapping_file):
     vcf_files = []
@@ -110,6 +116,32 @@ def get_project_and_vcf_fasta_mapping_from_metadata_xlsx(metadata_xlsx, mapping_
     return project_title, vcf_fasta_report_mapping
 
 
+def check_validation_required(tasks, sub_config):
+    # Validation is mandatory so if submit is requested then VALIDATE must have run before or be requested as well
+    if SUBMIT in tasks:
+        if not sub_config.get(READY_FOR_SUBMISSION_TO_EVA, False):
+            return True
+        submission_id = sub_config.get(SUB_CLI_CONFIG_KEY_SUBMISSION_ID, None)
+        if submission_id:
+            try:
+                submission_status = SubmissionWSClient().get_submission_status(submission_id)
+                if submission_status == 'FAILED':
+                    return True
+                else:
+                    return False
+            except requests.HTTPError as ex:
+                if ex.response.status_code == 404:
+                    logger.error(
+                        f'Submission with id {submission_id} could not be found: statuc code: {ex.response.status_code} response: {ex.response.text}')
+                    raise SubmissionNotFoundException(f'Submission with id {submission_id} could not be found')
+                else:
+                    logger.error(f'Error occurred while getting status of the submission with Id {submission_id}: status code: {ex.response.status_code} response: {ex.response.text}')
+                    raise SubmissionStatusException(f'Error occurred while getting status of the submission with Id {submission_id}')
+
+        logger.info(f'submission id not found in config. This might be the first time user is submitting')
+        return False
+
+
 def orchestrate_process(submission_dir, vcf_files, reference_fasta, metadata_json, metadata_xlsx,
                         tasks, executor, username=None, password=None, **kwargs):
     # load config
@@ -124,10 +156,8 @@ def orchestrate_process(submission_dir, vcf_files, reference_fasta, metadata_jso
     project_title, vcf_files_mapping = get_project_title_and_create_vcf_files_mapping(submission_dir, vcf_files, reference_fasta, metadata_json, metadata_xlsx)
     vcf_files = get_vcf_files(vcf_files_mapping)
 
-    # Validation is mandatory so if submit is requested then VALIDATE must have run before or be requested as well
-    if SUBMIT in tasks and not sub_config.get(READY_FOR_SUBMISSION_TO_EVA, False):
-        if VALIDATE not in tasks:
-            tasks.append(VALIDATE)
+    if VALIDATE not in tasks and check_validation_required(tasks, sub_config):
+        tasks.append(VALIDATE)
 
     if VALIDATE in tasks:
         if executor == DOCKER:

eva_sub_cli/submission_ws.py

@@ -0,0 +1,61 @@
+import os
+
+import requests
+from ebi_eva_common_pyutils.logger import AppLogger
+from requests import HTTPError
+from retry import retry
+
+from eva_sub_cli import SUBMISSION_WS_VAR
+from eva_sub_cli.auth import get_auth
+
+
+class SubmissionWSClient(AppLogger):
+    """
+    Python client for interfacing with the Submission WS API.
+    """
+
+    def __init__(self, username=None, password=None):
+        self.auth = get_auth(username, password)
+        self.base_url = self._submission_ws_url
+
+    SUBMISSION_WS_URL = 'https://www.ebi.ac.uk/eva/webservices/submission-ws/v1/'
+    SUBMISSION_INITIATE_PATH = 'submission/initiate'
+    SUBMISSION_UPLOADED_PATH = 'submission/{submissionId}/uploaded'
+    SUBMISSION_STATUS_PATH = 'submission/{submissionId}/status'
+
+    @property
+    def _submission_ws_url(self):
+        """Retrieve the base URL for the submission web services.
+        In order of preference from the environment variable or the hardcoded value."""
+        if os.environ.get(SUBMISSION_WS_VAR):
+            return os.environ.get(SUBMISSION_WS_VAR)
+        else:
+            return self.SUBMISSION_WS_URL
+
+    def _submission_initiate_url(self):
+        return os.path.join(self.base_url, self.SUBMISSION_INITIATE_PATH)
+
+    def _submission_uploaded_url(self, submission_id):
+        return os.path.join(self.base_url, self.SUBMISSION_UPLOADED_PATH.format(submissionId=submission_id))
+
+    def _submission_status_url(self, submission_id):
+        return os.path.join(self.base_url, self.SUBMISSION_STATUS_PATH.format(submissionId=submission_id))
+
+    def mark_submission_uploaded(self, submission_id, metadata_json):
+        response = requests.put(self._submission_uploaded_url(submission_id),
+                                headers={'Accept': 'application/hal+json', 'Authorization': 'Bearer ' + self.auth.token},
+                                data=metadata_json)
+        response.raise_for_status()
+        return response.json()
+
+    def initiate_submission(self):
+        response = requests.post(self._submission_initiate_url(), headers={'Accept': 'application/hal+json',
+                                                                           'Authorization': 'Bearer ' + self.auth.token})
+        response.raise_for_status()
+        return response.json()
+
+    @retry(exceptions=(HTTPError,), tries=3, delay=2, backoff=1.2, jitter=(1, 3))
+    def get_submission_status(self, submission_id):
+        response = requests.get(self.get_submission_status_url(submission_id))
+        response.raise_for_status()
+        return response.text

eva_sub_cli/submit.py

@@ -7,22 +7,18 @@
 from ebi_eva_common_pyutils.logger import AppLogger
 from retry import retry
 
-from eva_sub_cli import SUB_CLI_CONFIG_FILE, __version__, SUBMISSION_WS_VAR
-from eva_sub_cli.auth import get_auth
+from eva_sub_cli import SUB_CLI_CONFIG_FILE, __version__
+from eva_sub_cli.submission_ws import SubmissionWSClient
 from eva_sub_cli.validators.validator import READY_FOR_SUBMISSION_TO_EVA
 
 SUB_CLI_CONFIG_KEY_SUBMISSION_ID = "submission_id"
 SUB_CLI_CONFIG_KEY_SUBMISSION_UPLOAD_URL = "submission_upload_url"
 SUB_CLI_CONFIG_KEY_COMPLETE = "submission_complete"
 
-SUBMISSION_WS_URL = "https://www.ebi.ac.uk/eva/webservices/submission-ws/v1/"
-
 
 class StudySubmitter(AppLogger):
-    def __init__(self, submission_dir, submission_ws_url=None,
-                 submission_config: WritableConfig = None, username=None, password=None):
-        self.auth = get_auth(username, password)
-        self.submission_ws_url = submission_ws_url
+    def __init__(self, submission_dir, submission_config: WritableConfig = None, username=None, password=None):
+        self.submission_ws_client = SubmissionWSClient(username, password)
         self.submission_dir = submission_dir
         if submission_config:
             self.sub_config = submission_config
@@ -50,24 +46,6 @@ def metadata_json(self):
     def vcf_files(self):
         return self.sub_config.get('vcf_files')
 
-    def _get_submission_ws_url(self):
-        """Retrieve the base URL for the submission web services. In order of preference from the user of this class,
-        the environment variable or the hardcoded value."""
-        if self.submission_ws_url:
-            return self.submission_ws_url
-        if os.environ.get(SUBMISSION_WS_VAR):
-            return os.environ.get(SUBMISSION_WS_VAR)
-        else:
-            return SUBMISSION_WS_URL
-
-    @property
-    def submission_initiate_url(self):
-        return os.path.join(self._get_submission_ws_url(), 'submission/initiate')
-
-    @property
-    def submission_uploaded_url(self):
-        return os.path.join(self._get_submission_ws_url(), 'submission/{submissionId}/uploaded')
-
     def update_config_with_submission_id_and_upload_url(self, submission_id, upload_url):
         self.sub_config.set(SUB_CLI_CONFIG_KEY_SUBMISSION_ID, value=submission_id)
         self.sub_config.set(SUB_CLI_CONFIG_KEY_SUBMISSION_UPLOAD_URL, value=upload_url)
@@ -91,21 +69,13 @@ def _upload_file(self, submission_upload_url, input_file):
         self.debug(f'Upload of {base_name} completed')
 
     def _initiate_submission(self):
-        response = requests.post(self.submission_initiate_url,
-                                 headers={'Accept': 'application/hal+json',
-                                          'Authorization': 'Bearer ' + self.auth.token})
-        response.raise_for_status()
-        response_json = response.json()
+        response_json = self.submission_ws_client.initiate_submission()
         self.debug(f'Submission ID {response_json["submissionId"]} received!!')
         # update config with submission id and upload url
         self.update_config_with_submission_id_and_upload_url(response_json["submissionId"], response_json["uploadUrl"])
 
     def _complete_submission(self):
-        response = requests.put(
-            self.submission_uploaded_url.format(submissionId=self.sub_config.get(SUB_CLI_CONFIG_KEY_SUBMISSION_ID)),
-            headers={'Accept': 'application/hal+json', 'Authorization': 'Bearer ' + self.auth.token}, data=self.metadata_json
-        )
-        response.raise_for_status()
+        self.submission_ws_client.mark_submission_uploaded(self.sub_config.get(SUB_CLI_CONFIG_KEY_SUBMISSION_ID), self.metadata_json)
         self.debug("Submission ID {} Complete".format(self.sub_config.get(SUB_CLI_CONFIG_KEY_SUBMISSION_ID)))
         # update config with completion of the submission
         self.sub_config.set(SUB_CLI_CONFIG_KEY_COMPLETE, value=True)

tests/test_main.py

@@ -2,12 +2,16 @@
 import os
 import shutil
 import unittest
-from unittest.mock import patch
+from unittest.mock import patch, Mock
 
 from ebi_eva_common_pyutils.config import WritableConfig
+from requests import HTTPError
 
 from eva_sub_cli import SUB_CLI_CONFIG_FILE
-from eva_sub_cli.main import orchestrate_process, VALIDATE, SUBMIT, DOCKER
+from eva_sub_cli.exceptions.submission_not_found_exception import SubmissionNotFoundException
+from eva_sub_cli.exceptions.submission_status_exception import SubmissionStatusException
+from eva_sub_cli.main import orchestrate_process, VALIDATE, SUBMIT, DOCKER, check_validation_required
+from eva_sub_cli.submit import SUB_CLI_CONFIG_KEY_SUBMISSION_ID
 from eva_sub_cli.validators.validator import READY_FOR_SUBMISSION_TO_EVA
 
 
@@ -33,6 +37,37 @@ def setUp(self) -> None:
     def tearDown(self) -> None:
         shutil.rmtree(self.test_sub_dir)
 
+    def test_check_validation_required(self):
+        tasks = ['submit']
+
+        sub_config = {READY_FOR_SUBMISSION_TO_EVA: False}
+        self.assertTrue(check_validation_required(tasks, sub_config))
+
+        sub_config = {READY_FOR_SUBMISSION_TO_EVA: True}
+        self.assertFalse(check_validation_required(tasks, sub_config))
+
+        with patch('eva_sub_cli.submission_ws.SubmissionWSClient.get_submission_status') as get_submission_status_mock:
+            sub_config = {READY_FOR_SUBMISSION_TO_EVA: True,
+                          SUB_CLI_CONFIG_KEY_SUBMISSION_ID: 'test123'}
+
+            get_submission_status_mock.return_value = 'OPEN'
+            self.assertFalse(check_validation_required(tasks, sub_config))
+
+            get_submission_status_mock.return_value = 'FAILED'
+            self.assertTrue(check_validation_required(tasks, sub_config))
+
+            mock_response = Mock()
+            mock_response.status_code = 404
+            get_submission_status_mock.side_effect = HTTPError(response=mock_response)
+            with self.assertRaises(SubmissionNotFoundException):
+                check_validation_required(tasks, sub_config)
+
+            mock_response = Mock()
+            mock_response.status_code = 500
+            get_submission_status_mock.side_effect = HTTPError(response=mock_response)
+            with self.assertRaises(SubmissionStatusException):
+                check_validation_required(tasks, sub_config)
+
     def test_orchestrate_validate(self):
         with patch('eva_sub_cli.main.get_vcf_files') as m_get_vcf,  \
                 patch('eva_sub_cli.main.WritableConfig') as m_config, \