Merge pull request #17 from cedadev/v0.1.3

V0.1.3

facet_scanner/__init__.py

@@ -5,3 +5,12 @@
 __license__ = 'BSD - see LICENSE file in top-level package directory'
 __contact__ = '[email protected]'
 
+# Logger setup
+import logging
+
+logging.basicConfig(level=logging.DEBUG)
+logstream = logging.StreamHandler()
+
+formatter = logging.Formatter('%(levelname)s [%(name)s]: %(message)s')
+logstream.setFormatter(formatter)
+

facet_scanner/collection_handlers/base.py

@@ -17,6 +17,13 @@
 from facet_scanner.utils import generator_grouper, Singleton
 import time
 
+from facet_scanner import logstream
+import logging
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
+
 
 class CollectionHandler(metaclass=Singleton):
     """

facet_scanner/collection_handlers/cci.py

@@ -11,12 +11,19 @@
 from tqdm import tqdm
 import hashlib
 import json
+import logging
 
 from facet_scanner.collection_handlers.base import CollectionHandler
 from facet_scanner.collection_handlers.utils import CatalogueDatasets
 from facet_scanner.utils import parse_key
 from tag_scanner.tagger import ProcessDatasets
 
+from facet_scanner import logstream
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
+
 def nested_get(key_list, input_dict, default=None):
     """
     Takes an iterable of keys and returns none if not found or the value
@@ -113,8 +120,11 @@ def get_facets(self, path):
         :return: Dict  Facet:value pairs
         """
 
+        logger.debug('Getting facets for CCI-type path')
+
         tagged_dataset = self.pds.get_file_tags(path)
 
+        logger.debug('Translating tagging code to facet map')
         # Translate between output from tagging code to map to named facets
         mapped_facets = {}
         for tag_name, tag_value in tagged_dataset.labels.items():
@@ -128,6 +138,7 @@ def get_facets(self, path):
                 if tag_name == facet and tag_name_mapping is None:
                     mapped_facets[facet] = tag_value
 
+        logger.debug('Obtaining moles record metadata')
         # Get MOLES catalogue
         moles_info = self.catalogue.get_moles_record_metadata(path)
 
@@ -137,6 +148,7 @@ def get_facets(self, path):
         if moles_info:
             mapped_facets['datasetId'] = moles_info['url'].split('uuid/')[-1]
 
+        logger.debug('Completed facet mapping')
         return mapped_facets
 
     @staticmethod
@@ -256,7 +268,6 @@ def _get_collection_variables(self, results, file_index):
 
         if values:
             ids = [x['key'] for x in values]
-            print(ids)
 
             # Sample 1 netCDF file from each DRS
             for id in ids:

facet_scanner/collection_handlers/cmip5.py

@@ -10,6 +10,13 @@
 
 from facet_scanner.collection_handlers.base import CollectionHandler
 import os
+import logging
+
+from facet_scanner import logstream
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
 
 class CMIP5(CollectionHandler):
     project_name = 'opensearch'

facet_scanner/core/elasticsearch_connection.py

@@ -11,6 +11,12 @@
 from elasticsearch.helpers import scan, bulk
 from ceda_elasticsearch_tools.elasticsearch import CEDAElasticsearchClient
 
+import logging
+from facet_scanner import logstream
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
 
 class ElasticsearchConnection:
     """

facet_scanner/core/facet_scanner.py

@@ -11,7 +11,11 @@
 from facet_scanner.collection_handlers.utils import FacetFactory
 import logging
 
+from facet_scanner import logstream
+
 logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
 
 
 class FacetScanner:
@@ -33,8 +37,10 @@ def get_handler(self, path, **kwargs):
         :return: Mapped collection handler
         :rtype: CollectionHandler
         """
+        logger.debug("Obtaining handler")
         handler, collection_root = self.handler_factory.get_handler(path)
 
+        logger.debug('Handler Obtainment complete')
         # Handle situation where handler not found
         if handler is None:
             logger.error(f'Unable to find a handler for: {path} in facet_scanner.collection_handlers.utils.collection_map.'

facet_scanner/scripts/facet_scanner_cmd.py

@@ -24,6 +24,13 @@
 from configparser import RawConfigParser
 from facet_scanner.utils import query_yes_no
 from facet_scanner.core.facet_scanner import FacetScanner
+import logging
+
+from facet_scanner import logstream
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
 
 
 class FacetExtractor(FacetScanner):

facet_scanner/scripts/lotus_facet_scanner.py

@@ -23,6 +23,13 @@
 import os
 import json
 
+import logging
+from facet_scanner import logstream
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
+
 
 class LotusFacetScanner(FacetExtractor):
 

facet_scanner/utils/snippets.py

@@ -11,6 +11,13 @@
 import sys
 import itertools
 from collections import OrderedDict
+import logging
+
+from facet_scanner import logstream
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logstream)
+logger.propagate = False
 
 
 def query_yes_no(question, default="yes"):

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "facet-scanner"
-version = "0.1.1"
+version = "0.1.3"
 description = "Extracts facets from datasets to add to elasticsearch."
 authors = ["Daniel Westwood <[email protected]>"]
 license = "{file='LICENSE'}"
@@ -24,10 +24,10 @@ python = "^3.12"
 elasticsearch = "^7"
 requests = "^2.32.3"
 tqdm = "^4.66.6"
-sphinx = "^8.1.3"
-directory-tree = { git = "https://github.com/cedadev/ceda-directory-tree.git", tag = "v1.1.2" }
+sphinx = "^7"
+directory-tree = { git = "https://github.com/cedadev/ceda-directory-tree.git", tag = "v1.1.3" }
 ceda-elasticsearch-tools = { git = "https://github.com/cedadev/ceda-elasticsearch-tools.git", tag = "v2.4.0" }
-tag-scanner = { git = "https://github.com/cedadev/cci-tag-scanner.git", tag = "v2.1.3" }
+tag-scanner = { git = "https://github.com/cedadev/cci-tag-scanner.git", tag = "v2.1.6" }
 pytest = "^8.3.3"