Merge pull request #195 from bpow/branding-orion

Branding changes for the ORION name

.github/workflows/test.yml

@@ -20,8 +20,8 @@ jobs:
           echo "ROBOKOP_HOME=$PWD" >> $GITHUB_ENV
           mkdir -p $PWD/tests/logs
           mkdir -p $PWD/tests/storage
-          echo "DATA_SERVICES_LOGS=$PWD/tests/logs" >> $GITHUB_ENV
-          echo "DATA_SERVICES_STORAGE=$PWD/tests/storage" >> $GITHUB_ENV
+          echo "ORION_LOGS=$PWD/tests/logs" >> $GITHUB_ENV
+          echo "ORION_STORAGE=$PWD/tests/storage" >> $GITHUB_ENV
 
       - name: Set up Python ${{ matrix.python-version }}
         uses: actions/setup-python@v4

Common/build_manager.py

@@ -26,9 +26,9 @@ class GraphBuilder:
 
     def __init__(self):
 
-        self.logger = LoggingUtil.init_logging("Data_services.Common.GraphBuilder",
+        self.logger = LoggingUtil.init_logging("ORION.Common.GraphBuilder",
                                                line_format='medium',
-                                               log_file_path=os.environ['DATA_SERVICES_LOGS'])
+                                               log_file_path=os.environ['ORION_LOGS'])
 
         self.current_graph_versions = {}
         self.graphs_dir = self.init_graphs_dir()  # path to the graphs output directory
@@ -281,9 +281,9 @@ def run_qc(self,
         return qc_metadata
 
     def load_graph_specs(self):
-        if 'DATA_SERVICES_GRAPH_SPEC' in os.environ and os.environ['DATA_SERVICES_GRAPH_SPEC']:
+        if 'ORION_GRAPH_SPEC' in os.environ and os.environ['ORION_GRAPH_SPEC']:
             # this is a messy way to find the graph spec path, mainly for testing - URL is preferred
-            graph_spec_file = os.environ['DATA_SERVICES_GRAPH_SPEC']
+            graph_spec_file = os.environ['ORION_GRAPH_SPEC']
             graph_spec_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), '..', 'graph_specs', graph_spec_file)
             if os.path.exists(graph_spec_path):
                 self.logger.info(f'Loading graph spec: {graph_spec_file}')
@@ -292,15 +292,15 @@ def load_graph_specs(self):
                     return self.parse_graph_spec(graph_spec_yaml)
             else:
                 raise Exception(f'Configuration Error - Graph Spec could not be found: {graph_spec_file}')
-        elif 'DATA_SERVICES_GRAPH_SPEC_URL' in os.environ:
-            graph_spec_url = os.environ['DATA_SERVICES_GRAPH_SPEC_URL']
+        elif 'ORION_GRAPH_SPEC_URL' in os.environ:
+            graph_spec_url = os.environ['ORION_GRAPH_SPEC_URL']
             graph_spec_request = requests.get(graph_spec_url)
             graph_spec_request.raise_for_status()
             graph_spec_yaml = yaml.full_load(graph_spec_request.text)
             return self.parse_graph_spec(graph_spec_yaml)
         else:
             raise Exception(f'Configuration Error - No Graph Spec was configured. Set the environment variable '
-                            f'DATA_SERVICES_GRAPH_SPEC_URL to a URL with a valid Graph Spec yaml file. '
+                            f'ORION_GRAPH_SPEC_URL to a URL with a valid Graph Spec yaml file. '
                             f'See the README for more info.')
 
     def parse_graph_spec(self, graph_spec_yaml):
@@ -436,7 +436,7 @@ def get_graph_dir_path(self, graph_id: str, graph_version: str):
         return os.path.join(self.graphs_dir, graph_id, graph_version)
 
     def get_graph_output_URL(self, graph_id: str, graph_version: str):
-        graph_output_url = os.environ['DATA_SERVICES_OUTPUT_URL']
+        graph_output_url = os.environ['ORION_OUTPUT_URL']
         if graph_output_url[-1] != '/':
             graph_output_url += '/'
         return f'{graph_output_url}{graph_id}/{graph_version}/'
@@ -475,14 +475,14 @@ def generate_graph_version(graph_spec: GraphSpec):
 
     @staticmethod
     def init_graphs_dir():
-        # use the directory specified by the environment variable DATA_SERVICES_GRAPHS
-        if 'DATA_SERVICES_GRAPHS' in os.environ and os.path.isdir(os.environ['DATA_SERVICES_GRAPHS']):
-            return os.environ['DATA_SERVICES_GRAPHS']
+        # use the directory specified by the environment variable ORION_GRAPHS
+        if 'ORION_GRAPHS' in os.environ and os.path.isdir(os.environ['ORION_GRAPHS']):
+            return os.environ['ORION_GRAPHS']
         else:
             # if graph dir is invalid or not specified back out
             raise IOError(
                 'GraphBuilder graphs directory not found. '
-                'Specify a valid directory with environment variable DATA_SERVICES_GRAPHS.')
+                'Specify a valid directory with environment variable ORION_GRAPHS.')
 
 
 if __name__ == '__main__':

Common/containers.py

@@ -30,7 +30,7 @@ def __init__(self, container_name: str, logger):
         # placeholders for overriding
         self.default_image = None
         self.default_ports = None
-        self.default_volumes = [f"{os.environ['HOST_STORAGE_DIR']}:/Data_services_storage"]
+        self.default_volumes = [f"{os.environ['HOST_STORAGE_DIR']}:/ORION_storage"]
         self.environment_vars = {}
 
     def run(self,

Common/kgx_file_merger.py

@@ -18,9 +18,9 @@ class KGXFileMerger:
     def __init__(self,
                  output_directory: str):
         self.output_directory = output_directory
-        self.logger = LoggingUtil.init_logging("Data_services.Common.KGXFileMerger",
+        self.logger = LoggingUtil.init_logging("ORION.Common.KGXFileMerger",
                                                line_format='medium',
-                                               log_file_path=os.environ['DATA_SERVICES_LOGS'])
+                                               log_file_path=os.environ['ORION_LOGS'])
 
     def merge(self,
               graph_spec: GraphSpec,

Common/kgx_file_normalizer.py

@@ -35,10 +35,10 @@ def __init__(self, error_message: str, actual_error: Exception=None):
 #
 class KGXFileNormalizer:
 
-    logger = LoggingUtil.init_logging("Data_services.Common.KGXFileNormalizer",
+    logger = LoggingUtil.init_logging("ORION.Common.KGXFileNormalizer",
                                       line_format='medium',
                                       level=logging.INFO,
-                                      log_file_path=os.environ['DATA_SERVICES_LOGS'])
+                                      log_file_path=os.environ['ORION_LOGS'])
 
     def __init__(self,
                  source_nodes_file_path: str,

Common/kgx_file_writer.py

@@ -10,10 +10,10 @@
 
 class KGXFileWriter:
 
-    logger = LoggingUtil.init_logging("Data_services.Common.KGXFileWriter",
+    logger = LoggingUtil.init_logging("ORION.Common.KGXFileWriter",
                                       line_format='medium',
                                       level=logging.DEBUG,
-                                      log_file_path=os.environ.get('DATA_SERVICES_LOGS'))
+                                      log_file_path=os.environ.get('ORION_LOGS'))
     """
     constructor
     :param nodes_output_file_path: the file path for the nodes file

Common/load_manager.py

@@ -22,9 +22,9 @@ def __init__(self,
                  test_mode: bool = False,
                  fresh_start_mode: bool = False):
 
-        self.logger = LoggingUtil.init_logging("Data_services.Common.SourceDataManager",
+        self.logger = LoggingUtil.init_logging("ORION.Common.SourceDataManager",
                                                line_format='medium',
-                                               log_file_path=os.environ['DATA_SERVICES_LOGS'])
+                                               log_file_path=os.environ['ORION_LOGS'])
 
         self.test_mode = test_mode
         if test_mode:
@@ -682,15 +682,15 @@ def get_source_version_path(self, source_id: str, source_version: str):
         return os.path.join(self.storage_dir, source_id, source_version)
 
     def init_storage_dir(self):
-        # use the storage directory specified by the environment variable DATA_SERVICES_STORAGE
+        # use the storage directory specified by the environment variable ORION_STORAGE
         # check to make sure it's set and valid, otherwise fail
-        if "DATA_SERVICES_STORAGE" not in os.environ:
-            raise Exception(f'You must use the environment variable DATA_SERVICES_STORAGE '
+        if "ORION_STORAGE" not in os.environ:
+            raise Exception(f'You must use the environment variable ORION_STORAGE '
                             f'to specify a storage directory.')
-        if os.path.isdir(os.environ["DATA_SERVICES_STORAGE"]):
-            return os.environ["DATA_SERVICES_STORAGE"]
+        if os.path.isdir(os.environ["ORION_STORAGE"]):
+            return os.environ["ORION_STORAGE"]
         else:
-            raise IOError(f'Storage directory not valid: {os.environ["DATA_SERVICES_STORAGE"]}')
+            raise IOError(f'Storage directory not valid: {os.environ["ORION_STORAGE"]}')
 
     def init_source_output_dir(self, source_id: str):
         source_dir_path = os.path.join(self.storage_dir, source_id)
@@ -715,8 +715,8 @@ def init_source_output_dir(self, source_id: str):
                              'in the finalized kgx files.')
     args = parser.parse_args()
 
-    if 'DATA_SERVICES_TEST_MODE' in os.environ:
-        test_mode_from_env = os.environ['DATA_SERVICES_TEST_MODE']
+    if 'ORION_TEST_MODE' in os.environ:
+        test_mode_from_env = os.environ['ORION_TEST_MODE']
     else:
         test_mode_from_env = False
 

Common/loader_interface.py

@@ -34,7 +34,7 @@ def __init__(self, test_mode: bool = False, source_data_dir: str = None):
             if not os.path.exists(self.data_path):
                 os.mkdir(self.data_path)
         else:
-            self.data_path = os.environ.get("DATA_SERVICES_STORAGE")
+            self.data_path = os.environ.get("ORION_STORAGE")
 
         # the final output lists of nodes and edges
         self.final_node_list: list = []
@@ -44,10 +44,10 @@ def __init__(self, test_mode: bool = False, source_data_dir: str = None):
         self.output_file_writer: KGXFileWriter = None
 
         # create a logger
-        self.logger = LoggingUtil.init_logging(f"Data_services.parsers.{self.get_name()}",
+        self.logger = LoggingUtil.init_logging(f"ORION.parsers.{self.get_name()}",
                                                level=logging.INFO,
                                                line_format='medium',
-                                               log_file_path=os.environ.get('DATA_SERVICES_LOGS'))
+                                               log_file_path=os.environ.get('ORION_LOGS'))
 
     def get_latest_source_version(self):
         """Determine and return the latest source version ie. a unique identifier associated with the latest version."""

Common/neo4j_tools.py

@@ -19,13 +19,13 @@ def __init__(self,
         self.http_port = http_port
         self.https_port = https_port
         self.bolt_port = bolt_port
-        self.password = password if password else os.environ['DATA_SERVICES_NEO4J_PASSWORD']
+        self.password = password if password else os.environ['ORION_NEO4J_PASSWORD']
         self.graph_db_uri = f'bolt://{neo4j_host}:{bolt_port}'
         self.graph_db_auth = ("neo4j", self.password)
         self.neo4j_driver = neo4j.GraphDatabase.driver(self.graph_db_uri, auth=self.graph_db_auth)
-        self.logger = LoggingUtil.init_logging("Data_services.Common.neo4j_tools",
+        self.logger = LoggingUtil.init_logging("ORION.Common.neo4j_tools",
                                                line_format='medium',
-                                               log_file_path=os.environ['DATA_SERVICES_LOGS'])
+                                               log_file_path=os.environ['ORION_LOGS'])
 
     def import_csv_files(self,
                          graph_directory: str,

Common/normalization.py

@@ -37,10 +37,10 @@ def __init__(self,
         :param node_normalization_version - not implemented yet
         """
         # create a logger
-        self.logger = LoggingUtil.init_logging("Data_services.Common.NodeNormalizer",
+        self.logger = LoggingUtil.init_logging("ORION.Common.NodeNormalizer",
                                                level=log_level,
                                                line_format='medium',
-                                               log_file_path=os.environ.get('DATA_SERVICES_LOGS'))
+                                               log_file_path=os.environ.get('ORION_LOGS'))
         # storage for regular nodes that failed to normalize
         self.failed_to_normalize_ids = set()
         # storage for variant nodes that failed to normalize
@@ -361,7 +361,7 @@ def __init__(self,
         :param log_level - overrides default log level
         """
         # create a logger
-        self.logger = LoggingUtil.init_logging("Data_services.Common.EdgeNormalizer", level=log_level, line_format='medium', log_file_path=os.environ.get('DATA_SERVICES_LOGS'))
+        self.logger = LoggingUtil.init_logging("ORION.Common.EdgeNormalizer", level=log_level, line_format='medium', log_file_path=os.environ.get('ORION_LOGS'))
         # normalization map for future look up of all normalized predicates
         self.edge_normalization_lookup = {}
         self.cached_edge_norms = {}

Common/supplementation.py

@@ -56,10 +56,10 @@ class SequenceVariantSupplementation:
 
     def __init__(self):
 
-        self.logger = LoggingUtil.init_logging("Data_services.Common.SequenceVariantSupplementation",
+        self.logger = LoggingUtil.init_logging("ORION.Common.SequenceVariantSupplementation",
                                                line_format='medium',
-                                               log_file_path=environ['DATA_SERVICES_LOGS'])
-        workspace_dir = environ["DATA_SERVICES_STORAGE"]
+                                               log_file_path=environ['ORION_LOGS'])
+        workspace_dir = environ["ORION_STORAGE"]
 
         # if the snpEff dir exists, assume we already downloaded it
         self.snpeff_dir = path.join(workspace_dir, "snpEff")

Common/utils.py

@@ -60,7 +60,7 @@ def init_logging(name, level=logging.INFO, line_format='minimum', log_file_path=
         stream_handler.setFormatter(formatter)
 
         # set the logging level
-        if 'DATA_SERVICES_TEST_MODE' in os.environ and os.environ['DATA_SERVICES_TEST_MODE']:
+        if 'ORION_TEST_MODE' in os.environ and os.environ['ORION_TEST_MODE']:
             level = logging.DEBUG
         logger.setLevel(level)
 
@@ -116,7 +116,7 @@ def __init__(self, log_level=logging.INFO):
         :param log_level - overrides default log level
         """
         # create a logger
-        self.logger = LoggingUtil.init_logging("Data_services.Common.GetData", level=log_level, line_format='medium', log_file_path=os.environ.get('DATA_SERVICES_LOGS'))
+        self.logger = LoggingUtil.init_logging("ORION.Common.GetData", level=log_level, line_format='medium', log_file_path=os.environ.get('ORION_LOGS'))
 
     @staticmethod
     def pull_via_ftp_binary(ftp_site, ftp_dir, ftp_file):
@@ -436,7 +436,7 @@ def format_normalization_failures(data_set_name: str, node_norm_failures: list,
         :param edge_norm_failures: set of edge predicates
         :return:
         """
-        the_logger = LoggingUtil.init_logging(f"Data_services.Common.NormFailures.{data_set_name}", level=logging.INFO, line_format='medium', log_file_path=os.path.join(Path(__file__).parents[1], 'logs'))
+        the_logger = LoggingUtil.init_logging(f"ORION.Common.NormFailures.{data_set_name}", level=logging.INFO, line_format='medium', log_file_path=os.path.join(Path(__file__).parents[1], 'logs'))
 
         # get the list into a dataframe group
         df = pd.DataFrame(node_norm_failures, columns=['curie'])

Dockerfile

@@ -8,13 +8,13 @@ RUN apt-get update  \
     && apt-get -y install git \
     && apt-get -y install vim
 
-COPY ./requirements.txt /Data_services/requirements.txt
+COPY ./requirements.txt /ORION/requirements.txt
 
-RUN pip3 install -r /Data_services/requirements.txt
+RUN pip3 install -r /ORION/requirements.txt
 
-COPY . /Data_services/.
+COPY . /ORION/.
 
-RUN chmod -R 777 /Data_services
+RUN chmod -R 777 /ORION
 
 
-ENV PYTHONPATH "$PYTHONPATH:/Data_services"
+ENV PYTHONPATH "$PYTHONPATH:/ORION"