Merge pull request #7 from RTIInternational/develop

Updated plater to handle biolink.

PLATER/services/util/bl_helper.py

@@ -1,5 +1,6 @@
 import asyncio
 from functools import reduce
+import re
 import httpx
 from PLATER.services.config import config
 
@@ -28,3 +29,62 @@ async def get_most_specific_concept(self, concept_list: list) -> list:
         response = await asyncio.gather(*tasks, return_exceptions=False)
         parents = list(reduce(lambda acc, value: acc + value, filter(lambda x: x, response), []))
         return list(filter(lambda x: x not in parents, concept_list))
+
+    @staticmethod
+    def upgrade_BiolinkEntity(entity):
+        if entity.startswith("biolink."):
+            return entity
+        return "biolink." + BLHelper._pascal_case(entity)
+
+    @staticmethod
+    def upgrade_BiolinkRelation(biolink_relation):
+        if biolink_relation is None:
+            return None
+        if biolink_relation.startswith("biolink."):
+            return biolink_relation
+        return "biolink." + BLHelper._snake_case(biolink_relation)
+
+
+    @staticmethod
+    def _pascal_case(arg: str):
+        """Convert string to PascalCase.
+
+        Non-alphanumeric characters are replaced with _.
+        "ThisCase" is replaced with "this_case".
+        """
+        # replace _x with X
+        tmp = re.sub(
+            r"(?<=[a-zA-Z])_([a-z])",
+            lambda c: c.group(1).upper(),
+            arg
+        )
+        # upper-case first character
+        tmp = re.sub(
+            r"^[a-z]",
+            lambda c: c.group(0).upper(),
+            tmp
+        )
+        return tmp
+
+
+    def _snake_case(arg: str):
+        """Convert string to snake_case.
+
+        Non-alphanumeric characters are replaced with _.
+        CamelCase is replaced with snake_case.
+        """
+        # replace non-alphanumeric characters with _
+        tmp = re.sub(r'\W', '_', arg)
+        # replace X with _x
+        tmp = re.sub(
+            r'(?<=[a-z])[A-Z](?=[a-z])',
+            lambda c: '_' + c.group(0).lower(),
+            tmp
+        )
+        # lower-case first character
+        tmp = re.sub(
+            r'^[A-Z](?=[a-z])',
+            lambda c: c.group(0).lower(),
+            tmp
+        )
+        return tmp

PLATER/services/util/drivers/redis_driver.py

@@ -6,7 +6,7 @@
 from redis.exceptions import ConnectionError
 from redis.backoff import NoBackoff
 from redis.retry import Retry
-from reasoner_converter.upgrading import upgrade_BiolinkRelation, upgrade_BiolinkEntity
+# from PLATER.services.util.bl_helper import BLHelper
 
 logger = LoggingUtil.init_logging(__name__,
                                   config.get('logging_level'),
@@ -144,7 +144,7 @@ def create_TRAPI_kg_response(self, query_graph , results_dict):
                     if node_id not in collected_nodes:
                         new_node = {}
                         collected_nodes.add(node_id)
-                        new_node['category'] = [upgrade_BiolinkEntity(x) for x in node['category']]
+                        # new_node['category'] = [BLHelper.upgrade_BiolinkEntity(x) for x in node['category']]
                         new_node['name'] = node.get('name', '')
                         new_node['attributes'] = []
                         for key, value in node.items():
@@ -160,6 +160,7 @@ def create_TRAPI_kg_response(self, query_graph , results_dict):
             for qg_id in edge_qg_ids:
                 edges = row[qg_id] if isinstance(row[qg_id], list) else [row[qg_id]]
                 edge_types = row[f'type__{qg_id}'] if isinstance(row[qg_id], list) else [row[f'type__{qg_id}']]
+                edge_types = [x.replace("biolink.", "biolink:") for x in edge_types]
                 id_pairs = row[f'id_pairs__{qg_id}']
                 index = 0
                 current_edge_binding = {qg_id: []}
@@ -169,7 +170,7 @@ def create_TRAPI_kg_response(self, query_graph , results_dict):
                     if edge_id not in collected_edges:
                         edge_in_query_graph = query_graph['edges'][qg_id]
                         source_real_id, target_real_id = id_pair
-                        edge_type = upgrade_BiolinkRelation(edge_type)
+                        # edge_type = BLHelper.upgrade_BiolinkRelation(edge_type)
                         new_edge = {
                             'subject': source_real_id,
                             'object': target_real_id,

PLATER/services/util/drivers/redis_trapi_cypher_compiler.py

@@ -1,7 +1,8 @@
 """Tools for compiling QGraph into Cypher query."""
 import re
-from reasoner_converter.downgrading import downgrade_BiolinkEntity, downgrade_BiolinkPredicate
-from reasoner_converter.upgrading import upgrade_BiolinkEntity, upgrade_BiolinkRelation
+#from reasoner_converter.downgrading import downgrade_BiolinkEntity, downgrade_BiolinkPredicate
+# from reasoner_converter.upgrading import upgrade_BiolinkEntity, upgrade_BiolinkRelation
+#from PLATER.services.util.bl_helper import BLHelper
 
 
 def cypher_prop_string(value):
@@ -21,7 +22,7 @@ def __init__(self, node, node_id ,anonymous=False):
         """Create a node reference."""
         node = dict(node)
         name = f'{node_id}' if not anonymous else ''
-        labels = node.pop('category', 'named_thing')
+        labels = node.pop('category', 'biolink.NamedThing')
         if not isinstance(labels, list):
             labels = [labels]
         props = {}
@@ -41,15 +42,9 @@ def __init__(self, node, node_id ,anonymous=False):
             self.has_curie = True
         label_filters = []
         if labels:
-            biolink_regex = "^biolink:[A-Z][a-zA-Z]*$"
             for label in labels:
-                label_filters.append(f"'{label}' in {name}.category")
-                is_biolinkified = re.match(biolink_regex, label)
-                if is_biolinkified:
-                    other_label =  downgrade_BiolinkEntity(label)
-                else:
-                    other_label = upgrade_BiolinkEntity(label)
-                label_filters.append(f"'{other_label}' in {name}.category")
+                other_label = label.replace('biolink:', 'biolink.')
+                label_filters.append(f"`{other_label}` in labels(`{name}`)")
         self._filters = ''
         if len(curie_filters):
             filters = '( ' + ' OR '.join(curie_filters) + ')'
@@ -72,7 +67,7 @@ def __str__(self):
         """Return the cypher node reference."""
         self._num += 1
         if self._num == 1:
-            label = f':`{self.labels[0]}`' if self.labels else ''
+            label = f':`{self.labels[0].replace("biolink:", "biolink.")}`' if self.labels else ''
             return f'{self.name}' + label + f'{self.prop_string}' # + ''.join(f':`{label}`' for label in self.labels)
         return self.name
 
@@ -109,14 +104,11 @@ def __init__(self, edge, edge_id, anonymous=False):
                 filters = ''
             elif isinstance(edge['predicate'], list):
                 filters = []
-                biolink_predicate_regex = "^biolink:[a-z][a-z_]*$"
+                # biolink_predicate_regex = "^biolink:[a-z][a-z_]*$"
                 for predicate in edge['predicate']:
-                    is_biolink_predicate = re.match(biolink_predicate_regex, predicate)
-                    filters.append(f'type({name}) = "{predicate}"')
-                    if is_biolink_predicate:
-                        other_predicate = downgrade_BiolinkPredicate(predicate)
-                    else:
-                        other_predicate = upgrade_BiolinkRelation(predicate)
+                    # is_biolink_predicate = re.match(biolink_predicate_regex, predicate)
+                    # filters.append(f'type({name}) = "{predicate}"')
+                    other_predicate = predicate.replace('biolink:', 'biolink.')
                     filters.append(f'type({name}) = "{other_predicate}" ')
                 filters = ' OR '.join(filters)
                 label = None
@@ -136,7 +128,7 @@ def __str__(self):
         """Return the cypher edge reference."""
         self._num += 1
         if self._num == 1:
-            label = f':`{self.label}`' if self.label else ''
+            label = f':`{self.label.replace("biolink:", "biolink.")}`' if self.label else ''
             innards = f'{self.name}{label}'
         else:
             innards = self.name