move edge filtering earlier

biothings · Jun 11, 2024 · b2d1389 · b2d1389
1 parent 7aa6bd6
commit b2d1389
Show file tree

Hide file tree

Showing 4 changed files with 35 additions and 14 deletions.
diff --git a/src/edge_manager.ts b/src/edge_manager.ts
@@ -14,6 +14,8 @@ import { UnavailableAPITracker } from './types';
 import { RecordsByQEdgeID } from './results_assembly/query_results';
 import path from 'path';
 import { promises as fs } from 'fs';
+import KnowledgeGraph from './graph/knowledge_graph';
+import BTEGraph from './graph/graph';
 
 export default class QueryEdgeManager {
   private _qEdges: QEdge[];
@@ -211,7 +213,7 @@ export default class QueryEdgeManager {
         subjectIDs.some((curie) => execSubjectCuries.includes(curie)) || execSubjectCuries.length === 0;
       const objectMatch = objectIDs.some((curie) => execObjectCuries.includes(curie)) || execObjectCuries.length === 0;
 
-      //if both ends match then keep record
+      // if both ends match then keep record
 
       // Don't keep self-edges
       const selfEdge = [...subjectIDs].some((curie) => objectIDs.includes(curie));
@@ -230,6 +232,32 @@ export default class QueryEdgeManager {
     return keep;
   }
 
+  _constrainEdgeRecords(qEdge: QEdge, records: Record[]) {
+    const keep: Record[] = [];
+    const bte = new BTEGraph();
+    const kg = new KnowledgeGraph();
+    bte.update(records);
+    kg.update(bte);
+    records.forEach(record => {
+      const edge = kg.kg.edges[record.recordHash];
+      const sub = qEdge.reverse ? kg.kg.nodes[edge.object] : kg.kg.nodes[edge.subject];
+      const obj = qEdge.reverse ? kg.kg.nodes[edge.subject] : kg.kg.nodes[edge.object];
+      if (qEdge.meetsConstraints(edge, sub, obj)) {
+        keep.push(record);
+      }
+    });
+
+    debug(`'${qEdge.getID()}' dropped (${records.length - keep.length}) records based on edge/node constraints.`);
+    this.logs.push(
+      new LogEntry(
+        'DEBUG',
+        null,
+        `'${qEdge.getID()}' kept (${keep.length}) / dropped (${records.length - keep.length}) records (based on node/edge constraints).`,
+      ).getLog(),
+    );
+    return keep;
+  }
+
   collectRecords(): boolean {
     //go through edges and collect records organized by edge
     let recordsByQEdgeID: RecordsByQEdgeID = {};
@@ -296,8 +324,10 @@ export default class QueryEdgeManager {
 
   updateEdgeRecords(currentQEdge: QEdge): void {
     //1. filter edge records based on current status
-    const filteredRecords = this._filterEdgeRecords(currentQEdge);
-    //2.trigger node update / entity update based on new status
+    let filteredRecords = this._filterEdgeRecords(currentQEdge);
+    //2. make sure node/edge constraints are met
+    filteredRecords = this._constrainEdgeRecords(currentQEdge, filteredRecords); 
+    //3. trigger node update / entity update based on new status
     currentQEdge.storeRecords(filteredRecords);
   }
 

diff --git a/src/index.ts b/src/index.ts
@@ -724,8 +724,6 @@ export default class TRAPIQueryHandler {
     //update query results
     await this.trapiResultsAssembler.update(
       manager.getOrganizedRecords(),
-      this.queryGraphHandler,
-      this.knowledgeGraph.kg,
       !(this.options.smartAPIID || this.options.teamName)
     );
     this.logs = [...this.logs, ...this.trapiResultsAssembler.logs];

diff --git a/src/query_edge.ts b/src/query_edge.ts
@@ -356,7 +356,6 @@ export default class QEdge {
 
 
   storeRecords(records: Record[]): void {
-    debug((new Error()).stack)
     debug(`(6) Storing records...`);
     // store new records in current edge
     this.records = records;

diff --git a/src/results_assembly/query_results.ts b/src/results_assembly/query_results.ts
@@ -151,8 +151,6 @@ export default class TrapiResultsAssembler {
   _getQueryGraphSolutions(
     recordsByQEdgeID: RecordsByQEdgeID,
     qEdgeID: string,
-    queryGraph: QueryGraph,
-    kg: TrapiKnowledgeGraph,
     edgeCount: number,
     queryGraphSolutions: QueryGraphSolutionEdge[][],
     queryGraphSolution: QueryGraphSolutionEdge[],
@@ -192,7 +190,7 @@ export default class TrapiResultsAssembler {
 
     records
       .filter((record) => {
-        return [getMatchingPrimaryCurie(record), undefined].indexOf(primaryCurieToMatch) > -1 && queryGraph.edges[qEdgeID].meetsConstraints(kg.edges[record.recordHash], kg.nodes[kg.edges[record.recordHash].subject], kg.nodes[kg.edges[record.recordHash].object]);
+        return [getMatchingPrimaryCurie(record), undefined].indexOf(primaryCurieToMatch) > -1;
       })
       .forEach((record, i) => {
         // primaryCurie example: 'NCBIGene:1234'
@@ -225,8 +223,6 @@ export default class TrapiResultsAssembler {
           this._getQueryGraphSolutions(
             recordsByQEdgeID,
             connectedQEdgeID,
-            queryGraph,
-            kg,
             edgeCount,
             queryGraphSolutions,
             queryGraphSolution,
@@ -277,7 +273,7 @@ export default class TrapiResultsAssembler {
    * can safely assume every call to update contains all the records.
    *
    */
-  async update(recordsByQEdgeID: RecordsByQEdgeID, queryGraph: QueryGraph, kg: TrapiKnowledgeGraph, shouldScore = true): Promise<void> {
+  async update(recordsByQEdgeID: RecordsByQEdgeID, shouldScore = true): Promise<void> {
     debug(`Updating query results now!`);
 
     let scoreCombos: ScoreCombos;
@@ -322,8 +318,6 @@ export default class TrapiResultsAssembler {
     this._getQueryGraphSolutions(
       recordsByQEdgeID,
       initialQEdgeID,
-      queryGraph,
-      kg,
       qEdgeCount,
       queryGraphSolutions,
       [], // first queryGraphSolution