GEN2-009: add rebuildable retrieval projections

dev/ACTIVE_TASK.json

@@ -1 +1 @@
-{"task_file":"dev/tasks/GEN2-008-promote-episode-and-experience-v2-including-negative-experience.json","task_id":"GEN2-008","task_revision":1}
+{"task_file":"dev/tasks/GEN2-009-build-memory-retrieval-fabric-as-rebuildable-projections.json","task_id":"GEN2-009","task_revision":1}

runtime/retrieval/retrieval.go

@@ -0,0 +1,92 @@
+package retrieval
+
+import "sort"
+
+type Lane string
+
+const (
+	LaneSemantic           Lane = "SEMANTIC"
+	LaneEntity             Lane = "ENTITY"
+	LaneTemporal           Lane = "TEMPORAL"
+	LaneGoal               Lane = "GOAL"
+	LanePositiveExperience Lane = "POSITIVE_EXPERIENCE"
+	LaneNegativeExperience Lane = "NEGATIVE_EXPERIENCE"
+	LaneContradiction      Lane = "CONTRADICTION"
+	LaneSkill              Lane = "SKILL"
+)
+
+// RetrievalDocument is disposable read-side projection data. SourceRef and
+// Owner preserve provenance; this type never becomes canonical meaning.
+type RetrievalDocument struct {
+	ID              string
+	SourceRef       string
+	Owner           string
+	PrincipalID     string
+	ContextScopeRef string
+	VersionRef      string
+	Lane            Lane
+	Text            string
+	Relevance       int
+	Freshness       int
+	Confidence      int
+	Salience        int
+	Cost            int
+}
+
+type Query struct {
+	PrincipalID     string
+	ContextScopeRef string
+	VersionRef      string
+	Lanes           []Lane
+	Limit           int
+}
+
+func FilterAndRank(documents []RetrievalDocument, query Query) []RetrievalDocument {
+	filtered := make([]RetrievalDocument, 0, len(documents))
+	for _, doc := range documents {
+		if doc.PrincipalID != query.PrincipalID || (query.ContextScopeRef != "" && doc.ContextScopeRef != query.ContextScopeRef) || (query.VersionRef != "" && doc.VersionRef != query.VersionRef) || !containsLane(query.Lanes, doc.Lane) {
+			continue
+		}
+		filtered = append(filtered, doc)
+	}
+	sort.SliceStable(filtered, func(i, j int) bool {
+		left := filtered[i].Relevance + filtered[i].Freshness + filtered[i].Confidence + filtered[i].Salience - filtered[i].Cost
+		right := filtered[j].Relevance + filtered[j].Freshness + filtered[j].Confidence + filtered[j].Salience - filtered[j].Cost
+		if left != right {
+			return left > right
+		}
+		return filtered[i].ID < filtered[j].ID
+	})
+	if query.Limit > 0 && len(filtered) > query.Limit {
+		filtered = filtered[:query.Limit]
+	}
+	return filtered
+}
+
+func containsLane(lanes []Lane, lane Lane) bool {
+	if len(lanes) == 0 {
+		return true
+	}
+	for _, v := range lanes {
+		if v == lane {
+			return true
+		}
+	}
+	return false
+}
+
+type ProjectionBuilder interface {
+	Build() ([]RetrievalDocument, error)
+}
+
+func Rebuild(builders ...ProjectionBuilder) ([]RetrievalDocument, error) {
+	var result []RetrievalDocument
+	for _, builder := range builders {
+		docs, err := builder.Build()
+		if err != nil {
+			return nil, err
+		}
+		result = append(result, docs...)
+	}
+	return result, nil
+}

runtime/retrieval/retrieval_test.go

@@ -0,0 +1,14 @@
+package retrieval_test
+
+import (
+	"git.cin.su/m/Assistant/runtime/retrieval"
+	"testing"
+)
+
+func TestScopeAndVersionFilterBeforeRanking(t *testing.T) {
+	docs := []retrieval.RetrievalDocument{{ID: "wrong-company", PrincipalID: "a", ContextScopeRef: "b", VersionRef: "v", Lane: retrieval.LaneSemantic, Relevance: 100}, {ID: "right", PrincipalID: "a", ContextScopeRef: "a", VersionRef: "v", Lane: retrieval.LaneNegativeExperience, Relevance: 1}, {ID: "wrong-version", PrincipalID: "a", ContextScopeRef: "a", VersionRef: "old", Lane: retrieval.LaneSemantic, Relevance: 100}}
+	got := retrieval.FilterAndRank(docs, retrieval.Query{PrincipalID: "a", ContextScopeRef: "a", VersionRef: "v", Lanes: []retrieval.Lane{retrieval.LaneSemantic, retrieval.LaneNegativeExperience}, Limit: 10})
+	if len(got) != 1 || got[0].ID != "right" {
+		t.Fatalf("got %#v", got)
+	}
+}