Vite app

.github/ISSUE_TEMPLATE/transfer.md

@@ -13,7 +13,7 @@ Fill this out if you would like to transfer ownership of an API entry to someone
 
 * **I'm am ...**
 
-- [ ] the person transferring ownership of my API
+- [ ] the current owner transferring ownership of my API
 
 - [ ] the person that this API is being transferred to
 
@@ -26,7 +26,7 @@ Fill this out if you would like to transfer ownership of an API entry to someone
 
 
 
-* What is the ***name*** and ***username*** of the person that this API currently belongs to? *eg. Jane Doe @janedoe1
+* What is the ***name*** and ***username*** of the Responsible Developer that currently maintains this API? *eg. Jane Doe @janedoe1
 
 
 

src/admin.py

@@ -195,6 +195,26 @@ def consolidate_metakg(reset=True):
     SmartAPI.index_metakg_consolidation()
 
 
+def refresh_has_metakg():
+    """
+    Refreshes the 'has_metakg' attribute for SmartAPI objects.
+    This function iterates through all SmartAPI objects, checks if there's a corresponding entry in the ConsolidatedMetaKGDoc 
+    collection based on the SmartAPI ID, and updates the 'has_metakg' attribute accordingly.
+    Note:
+    - This function assumes the existence of the SmartAPI and ConsolidatedMetaKGDoc classes.
+    - 'has_metakg' attribute is a boolean value indicating whether a SmartAPI has corresponding metadata in the Meta-Knowledge Graph.
+    Returns:
+    None
+    """
+    for smartapi in SmartAPI.get_all(1000):
+        value = ConsolidatedMetaKGDoc.exists(smartapi._id, field="api.smartapi.id")
+        if value:
+            smartapi.has_metakg = True
+        else:
+            smartapi.has_metakg = False
+        smartapi.save()
+
+
 restore = restore_from_file
 backup = backup_to_file
 
@@ -229,6 +249,8 @@ def routine():
     refresh_metakg()
     logger.info("consolidate_metakg()")
     consolidate_metakg()
+    logger.info("refresh_has_metakg()")
+    refresh_has_metakg()
 
 
 if __name__ == "__main__":

src/controller/base.py

@@ -377,4 +377,4 @@ def update(self, content):
                 self._status = self.STATUS.NOT_MODIFIED.value
             else:  # raw field changed
                 self._status = self.STATUS.UPDATED.value
-            self._entity.raw = content.raw
+                self._entity.raw = content.raw

src/controller/smartapi.py

@@ -100,6 +100,7 @@ def __init__(self, url):
         self.slug = None
         self.date_created = None
         self.last_updated = None
+        self.has_metakgq = None
 
         self.uptime = APIMonitorStatus(self)
         self.webdoc = APIRefreshStatus(self)
@@ -117,6 +118,8 @@ def get(cls, _id):
         obj.date_created = obj._doc._meta.date_created
         obj.last_updated = obj._doc._meta.last_updated
 
+        obj.has_metakg = obj._doc._meta.has_metakg
+
         obj.uptime = APIMonitorStatus(
             obj,
             (
@@ -289,6 +292,14 @@ def save(self, force_save=True):
         if self.date_created > self.last_updated:
             raise ControllerError("Invalid timestamps.")
 
+
+        if not self.has_metakg:
+            value = ConsolidatedMetaKGDoc.exists(self._id, field="api.smartapi.id")
+            if value:
+                self.has_metakg = True
+            else:
+                self.has_metakg = False
+
         # NOTE
         # if the slug of another document changed at this point
         # it's possible to have two documents with the same slug
@@ -326,7 +337,7 @@ def save(self, force_save=True):
         doc._meta.date_created = self.date_created
         doc._meta.last_updated = self.last_updated
         doc._meta.slug = self.slug
-
+        doc._meta.has_metakg = self.has_metakg
         doc.save(skip_empty=False)
 
         return self._id

src/handlers/api.py

@@ -21,6 +21,7 @@
 from utils.metakg.export import edges2graphml
 from utils.metakg.path_finder import MetaKGPathFinder
 from utils.metakg.cytoscape_formatter import CytoscapeDataFormatter
+from utils.metakg.biolink_helpers import get_expanded_values
 from utils.notification import SlackNewAPIMessage, SlackNewTranslatorAPIMessage
 
 logger = logging.getLogger("smartAPI")
@@ -433,20 +434,6 @@ def initialize(self, *args, **kwargs):
         self.pipeline = MetaKGQueryPipeline(ns=self.biothings)
         self.biolink_model_toolkit = bmt.Toolkit()
 
-    def get_expanded_values(self, value: Union[str, List[str]]) -> List[str]:
-        """return exapnded value list for a given biolink class name"""
-        if isinstance(value, str):
-            value = [value]
-        _out = []
-        for v in value:
-            try:
-                v = self.biolink_model_toolkit.get_descendants(v, reflexive=True, formatted=True)
-                v = [x.split(":")[-1] for x in v]  # remove biolink: prefix
-            except ValueError:
-                v = [v]
-            _out.extend(v)
-        return _out
-
     @capture_exceptions
     async def get(self, *args, **kwargs):
         expanded_fields = {"subject": False, "object": False, "predicate": False, "node": False}
@@ -463,7 +450,7 @@ async def get(self, *args, **kwargs):
             value_list = getattr(self.args, field)
             if not value_list:
                 continue
-            value_list = self.get_expanded_values(value_list) if expanded_fields[field] else value_list
+            value_list = get_expanded_values(value_list, self.biolink_model_toolkit) if expanded_fields[field] else value_list
             setattr(self.args, field, value_list)
 
         await super().get(*args, **kwargs)
@@ -539,22 +526,111 @@ class MetaKGPathFinderHandler(QueryHandler):
             **QUERY_KWARGS.get("GET", {}),
             "subject": {"type": str, "required": True, "max": 1000},
             "object": {"type": str, "required": True, "max": 1000},
+            "predicate": {"type": list, "max": 10, "default": []},
             "cutoff": {"type": int, "default": 3, "max": 5},
             "api_details": {"type": bool, "default": False},
+            "expand": {
+                "type": list,
+                "max": 6,
+                "default": [],
+                "enum": ["subject", "object", "predicate", "node", "edge", "all"]
+                },
+            "rawquery": {"type": bool, "default": False},
         },
     }
 
+    def initialize(self, *args, **kwargs):
+        super().initialize(*args, **kwargs)
+        # change the default query pipeline from self.biothings.pipeline
+        self.pipeline = MetaKGQueryPipeline(ns=self.biothings)
+        self.biolink_model_toolkit = bmt.Toolkit()
+
+    def setup_pathfinder_rawquery(self, expanded_fields):
+        # JSON-structured summary of operations and criteria applied
+        operations_summary = {
+            "input_parameters": {},
+            "expansion_logic": {},
+            "search_criteria": []
+        }
+
+        # Include original query parameters
+        operations_summary["input_parameters"] = {
+            "subject": self.args.subject,
+            "object": self.args.object,
+            "predicate": getattr(self.args, 'predicate', None)  # Including predicate if provided
+        }
+
+        # Detail the expansion logic in a way that explains what expansions are applied
+        operations_summary["expansion_logic"] = {
+            "expand_subject": "subject" in self.args.expand or "all" in self.args.expand or "node" in self.args.expand,
+            "expand_object": "object" in self.args.expand or "all" in self.args.expand or "node" in self.args.expand,
+            "expand_predicate": "predicate" in self.args.expand,
+        }
+
+        # Summarize the search criteria based on expanded fields
+        for field, values in expanded_fields.items():
+            if values:  # Ensure values exist for the field before adding
+                operations_summary["search_criteria"].append({
+                    "field": field,
+                    "description": f"Expanding '{field}' to include {len(values)} variant(s)",
+                    "values": values
+                })
+
+        # The operations_summary is already in a format that can be directly returned as JSON
+        return operations_summary
+
     @capture_exceptions
     async def get(self, *args, **kwargs):
         query_data = {"q": self.args.q}
-        pathfinder = MetaKGPathFinder(query_data=query_data)
+
+        # Initialize with the original subject and object, and setup for expansion
+        expanded_fields = {
+            "subject": [self.args.subject],
+            "object": [self.args.object],
+        }
+
+        # Check if expansion is requested
+        if self.args.expand:
+            # Define a set for fields affected by 'node' and 'all' for simpler updates
+            common_fields = {"subject", "object"}
+
+            # Initialize expandable_fields based on 'node' or 'all' presence
+            expandable_fields = set()
+            if "node" in self.args.expand or "all" in self.args.expand:
+                expandable_fields.update(common_fields)
+            if "edge" in self.args.expand or "all" in self.args.expand:
+                expandable_fields.add("predicate")
+
+            # Add specific fields if mentioned explicitly
+            expandable_fields.update({field for field in ["subject", "object", "predicate"] if field in self.args.expand})
+
+            # Expand the fields as required
+            for field in expandable_fields:
+                # Use the built-in utility function, get_expanded_values, to expand the fields
+                expanded_fields[field] = get_expanded_values(getattr(self.args, field), self.biolink_model_toolkit)
+
+        # Initalize pathfinder
+        pathfinder = MetaKGPathFinder(query_data=query_data, expanded_fields=expanded_fields)
+
+        # Initialize the pathfinder results list
+        paths_with_edges = []
+
+        # Run get_paths method to retrieve paths and edges
         paths_with_edges = pathfinder.get_paths(
-            subject=self.args.subject,
-            object=self.args.object,
+            expanded_fields=expanded_fields,
             cutoff=self.args.cutoff,
             api_details=self.args.api_details,
+            predicate_filter=self.args.predicate
         )
-        # Return the result in JSON format
-        res = {"paths_with_edges": paths_with_edges}
+
+        # Check if rawquery parameter is true -- respond with correct output
+        if self.args.rawquery:
+            raw_query_output = self.setup_pathfinder_rawquery(expanded_fields)
+            self.write(raw_query_output)
+            return
+        res = { 
+                "total": len(paths_with_edges), 
+                "paths": paths_with_edges,
+            }
         await asyncio.sleep(0.01)
         self.finish(res)

src/model/smartapi.py

@@ -1,7 +1,7 @@
 """
     Elasticsearch Document Object Model for SmartAPI
 """
-from elasticsearch_dsl import Binary, Date, InnerDoc, Integer, Keyword, Object, Text
+from elasticsearch_dsl import Binary, Date, InnerDoc, Integer, Keyword, Object, Text, Boolean
 
 from config import SMARTAPI_ES_INDEX
 
@@ -27,7 +27,7 @@ class UserMeta(InnerDoc):
     username = Keyword(required=True)
     date_created = Date(default_timezone="UTC")
     last_updated = Date(default_timezone="UTC")
-
+    has_metakg = Boolean()
 
 class SmartAPIDoc(BaseDoc):
     _status = Object(StatMeta)

src/tests/test_controller.py

@@ -4,7 +4,7 @@
 import json
 import os
 import time
-from datetime import datetime, timezone
+from datetime import datetime, timedelta, timezone
 
 import elasticsearch
 import pytest
@@ -380,6 +380,10 @@ def test_refresh_status():
     assert mygene.webdoc.timestamp > datetime(2020, 1, 1)
     _ts0 = mygene.webdoc.timestamp
 
+    original_last_updated = mygene.last_updated.replace(microsecond=0, tzinfo=None)
+    one_hour_before = (datetime.now() - timedelta(hours=1)).replace(microsecond=0)
+    assert original_last_updated > one_hour_before
+
     mygene.save()
     refresh()
 
@@ -393,9 +397,15 @@ def test_refresh_status():
     assert "components" in mygene
     assert mygene.webdoc.timestamp > _ts0
 
+    current_last_updated = mygene.last_updated.replace(microsecond=0, tzinfo=None)
+    assert current_last_updated == original_last_updated
+
     mygene.save()
     refresh()
 
+    # confirm last_updated is not changed after refresh
+    assert mygene.last_updated.replace(microsecond=0, tzinfo=None) == current_last_updated
+
     mygene_doc = SmartAPIDoc.get(MYGENE_ID)
     assert mygene_doc._status.refresh_status == 200
     assert "components" in mygene_doc
@@ -408,6 +418,9 @@ def test_refresh_status():
     mygene.save()
     refresh()
 
+    # confirm last_updated is not changed after refresh
+    assert mygene.last_updated.replace(microsecond=0, tzinfo=None) == current_last_updated
+
     mygene_doc = SmartAPIDoc.get(MYGENE_ID)
     assert mygene_doc._status.refresh_status == 404
     assert "components" in mygene_doc
@@ -432,6 +445,9 @@ def test_refresh_status():
     mygene.save()
     refresh()
 
+    # confirm last_updated is not changed after refresh
+    assert mygene.last_updated.replace(microsecond=0, tzinfo=None) == current_last_updated
+
     mygene_doc = SmartAPIDoc.get(MYGENE_ID)
     assert mygene_doc._status.refresh_status == 499
     assert "components" in mygene_doc

src/utils/metakg/biolink_helpers.py

@@ -0,0 +1,20 @@
+from typing import Union, List
+import bmt
+
+# Initialize the Biolink Model Toolkit instance globally if it's used frequently
+# or pass it as a parameter to functions that require it.
+toolkit = bmt.Toolkit()
+
+def get_expanded_values(value: Union[str, List[str]], toolkit_instance=toolkit) -> List[str]:
+    """Return expanded value list for a given Biolink class name."""
+    if isinstance(value, str):
+        value = [value]
+    _out = []
+    for v in value:
+        try:
+            v = toolkit_instance.get_descendants(v, reflexive=True, formatted=True)
+            v = [x.split(":")[-1] for x in v]  # Remove 'biolink:' prefix
+        except ValueError:
+            v = [v]
+        _out.extend(v)
+    return _out