Janady13/missionmatch / internal/integration/graph_wage_integration_test.go

1//go:build integration 2 3package integration 4 5import ( 6 "context" 7 "strings" 8 "testing" 9 10 "github.com/google/uuid" 11 "github.com/jackc/pgx/v5" 12 13 "github.com/missionworkforce/missionmatch/internal/modules/ai" 14 "github.com/missionworkforce/missionmatch/internal/platform/audit" 15) 16 17// seedWageBand inserts one bls_wages metric row as the migrator (the global public wage reference the app role 18// may only read), keyed by a unique series id per run so parallel runs do not collide. It carries the SOC 19// occupation code, the OEWS area code (HI or NM), the metric (annual_mean or annual_median), and the dollar 20// value, the columns the M5 wage band read pivots. 21func seedWageBand(t *testing.T, seriesID, soc, occTitle, role, area, areaName, metric, value string) { 22 t.Helper() 23 if _, err := migrator.Exec(context.Background(), 24 `INSERT INTO bls_wages 25 (series_id, occupation_code, occupation_title, role, area, area_name, metric, period_year, value, raw) 26 VALUES ($1, $2, $3, $4, $5, $6, $7, '2025', $8::numeric, '{}'::jsonb) 27 ON CONFLICT (occupation_code, area, metric) 28 DO UPDATE SET series_id = EXCLUDED.series_id, occupation_title = EXCLUDED.occupation_title, 29 role = EXCLUDED.role, area_name = EXCLUDED.area_name, period_year = EXCLUDED.period_year, 30 value = EXCLUDED.value`, 31 seriesID, soc, occTitle, role, area, areaName, metric, value); err != nil { 32 t.Fatalf("seed wage band %s: %v", seriesID, err) 33 } 34} 35 36// seedRequirementForWage creates a requirement for one org carrying a title (so resolveSOC can bind the role to 37// a SOC through the curated crosswalk) and a work_location (so resolveArea can bind the OEWS area), linked to no 38// opportunity (the market leg is not the subject here, the price to win leg is). It uses the migrator so the 39// seed crosses tenants for the RLS posture, and returns the requirement id. 40func seedRequirementForWage(t *testing.T, org uuid.UUID, title, workLocation, level string) uuid.UUID { 41 t.Helper() 42 var id uuid.UUID 43 if err := migrator.QueryRow(context.Background(), 44 `INSERT INTO requirements (organization_id, title, work_location, required_level, required_poly) 45 VALUES ($1, $2, $3, $4, 'none') RETURNING id`, 46 org, title, workLocation, level).Scan(&id); err != nil { 47 t.Fatalf("seed wage requirement: %v", err) 48 } 49 return id 50} 51 52// TestM5PriceToWinResolvesTheOEWSBand is the live keystone proof of M5: a requirement whose title resolves to a 53// Galapagos role (and thus a SOC through the curated crosswalk) and whose work_location resolves to a known 54// OEWS area, with a seeded wage band for that occupation and area, makes the G6 price to win leg carry the REAL 55// advisory OEWS annual mean and annual median dollars instead of the honest floor. It also proves the audit 56// records price_resolved true (a boolean only, never the dollar value). 57func TestM5PriceToWinResolvesTheOEWSBand(t *testing.T) { 58 org := seedOrg(t) 59 k := ai.ConfiguredSuppressionThreshold() 60 run := uuid.NewString()[:8] 61 62 // The Program Manager occupation (SOC 113021) in Hawaii, both metrics, unique series ids per run. 63 seedWageBand(t, "M5-MEAN-"+run, "113021", "Computer and Information Systems Managers", "Program Manager", "HI", "Hawaii", "annual_mean", "185000.00") 64 seedWageBand(t, "M5-MEDIAN-"+run, "113021", "Computer and Information Systems Managers", "Program Manager", "HI", "Hawaii", "annual_median", "172500.00") 65 66 // A requirement whose title carries the Program Manager role and whose work_location is Honolulu, Hawaii, so 67 // both resolvers bind and the wage band read finds the seeded HI band. 68 req := seedRequirementForWage(t, org, "Senior Program Manager, Space Systems", "Honolulu, Hawaii", "secret") 69 70 // A small bench so the bench legs run without changing the price to win subject of the test. 71 for i := 0; i < k; i++ { 72 seedCandidate(t, org, "secret", "active", true) 73 } 74 75 orch := ai.NewOrchestrator(appDB, nil, audit.New(appDB)) 76 resp, err := orch.Answer(ctxOrg(org), "", ai.IntentCapturePictureForRequirement, []any{req.String()}) 77 if err != nil { 78 t.Fatalf("capture picture answer: %v", err) 79 } 80 if resp.Refused || resp.Route != ai.RouteGraph { 81 t.Fatalf("a bound tenant capture question must answer on the graph route, got %+v", resp) 82 } 83 84 // The price to win leg carries the REAL advisory OEWS band, not the honest floor. 85 if strings.Contains(resp.Answer, "labor rate data not yet ingested") { 86 t.Errorf("a requirement with a resolvable role, state, and wage row must not show the honest floor, got %q", resp.Answer) 87 } 88 for _, want := range []string{"OEWS labor rate", "Program Manager", "Hawaii", "185000.00", "172500.00", "BLS OEWS", "advisory"} { 89 if !strings.Contains(resp.Answer, want) { 90 t.Errorf("the price to win leg must carry %q, got %q", want, resp.Answer) 91 } 92 } 93 94 // The audit records price_resolved true and carries no dollar value. 95 assertCaptureAudit(t, org, true, []string{"185000", "172500"}) 96} 97 98// TestM5PriceToWinKeepsTheFloorWithoutWageData proves the honest floor holds when NO occupation resolves, in two 99// cases: a non occupation title in a state with no state baseline (Virginia), and a non occupation title in a 100// state that does have a baseline (Hawaii). With the national OEWS baseline present, every RESOLVABLE occupation 101// now falls back to a US national band, so the floor is reached only when the title matches no occupation at all. 102// In both cases the price to win leg stays the honest floor, no rate is fabricated, and the audit records 103// price_resolved false. 104func TestM5PriceToWinKeepsTheFloorWithoutWageData(t *testing.T) { 105 k := ai.ConfiguredSuppressionThreshold() 106 107 // Case one: a title that resolves to no occupation, so no SOC binds and the floor holds. With the national 108 // OEWS baseline present every RESOLVABLE occupation now falls back to a US national band, so the honest floor 109 // is reached only when the title matches no occupation at all. A short non occupation title matches nothing. 110 t.Run("title_resolves_to_no_occupation", func(t *testing.T) { 111 org := seedOrg(t) 112 req := seedRequirementForWage(t, org, "Zephyr Wrangler", "Arlington, Virginia", "secret") 113 for i := 0; i < k; i++ { 114 seedCandidate(t, org, "secret", "active", true) 115 } 116 orch := ai.NewOrchestrator(appDB, nil, audit.New(appDB)) 117 resp, err := orch.Answer(ctxOrg(org), "", ai.IntentCapturePictureForRequirement, []any{req.String()}) 118 if err != nil { 119 t.Fatalf("capture picture answer: %v", err) 120 } 121 if !strings.Contains(resp.Answer, "labor rate data not yet ingested") { 122 t.Errorf("a requirement whose title resolves to no occupation must keep the honest floor, got %q", resp.Answer) 123 } 124 if strings.Contains(resp.Answer, "OEWS labor rate") { 125 t.Errorf("no OEWS rate must be fabricated when no occupation resolves, got %q", resp.Answer) 126 } 127 assertCaptureAudit(t, org, false, nil) 128 }) 129 130 // Case two: a known area (Hawaii) but a title that matches no Galapagos role, so the SOC does not resolve and 131 // the floor holds even though wage data exists for other occupations in Hawaii. 132 t.Run("title_matches_no_role", func(t *testing.T) { 133 org := seedOrg(t) 134 run := uuid.NewString()[:8] 135 // Seed a Hawaii band for an unrelated occupation so the state has data but not for this title's role. 136 seedWageBand(t, "M5-NOROLE-MEAN-"+run, "113021", "Computer and Information Systems Managers", "Program Manager", "HI", "Hawaii", "annual_mean", "185000.00") 137 req := seedRequirementForWage(t, org, "Underwater Welding Technician", "Honolulu, Hawaii", "secret") 138 for i := 0; i < k; i++ { 139 seedCandidate(t, org, "secret", "active", true) 140 } 141 orch := ai.NewOrchestrator(appDB, nil, audit.New(appDB)) 142 resp, err := orch.Answer(ctxOrg(org), "", ai.IntentCapturePictureForRequirement, []any{req.String()}) 143 if err != nil { 144 t.Fatalf("capture picture answer: %v", err) 145 } 146 if !strings.Contains(resp.Answer, "labor rate data not yet ingested") { 147 t.Errorf("a requirement whose title matches no role must keep the honest floor, got %q", resp.Answer) 148 } 149 assertCaptureAudit(t, org, false, nil) 150 }) 151} 152 153// TestM5LaborRateBandIntentReturnsTheSeededBand proves the labor_rate_band read directly: seeded with the two 154// metric rows for an occupation and area, LaborRateBand pivots them into one band carrying the exact mean and 155// median as decimals (never float), the occupation title, the role, and the area name, and returns ok false 156// for an occupation and area with no wage row so the honest floor holds upstream. 157func TestM5LaborRateBandIntentReturnsTheSeededBand(t *testing.T) { 158 org := seedOrg(t) 159 run := uuid.NewString()[:8] 160 161 // The Systems Engineer occupation (SOC 151252) in New Mexico, both metrics. 162 seedWageBand(t, "M5-SE-MEAN-"+run, "151252", "Software Developers", "Systems Engineer", "NM", "New Mexico", "annual_mean", "132000.00") 163 seedWageBand(t, "M5-SE-MEDIAN-"+run, "151252", "Software Developers", "Systems Engineer", "NM", "New Mexico", "annual_median", "128000.00") 164 165 band, ok, err := ai.LaborRateBand(ctxOrg(org), appDB, "151252", "NM") 166 if err != nil { 167 t.Fatalf("labor rate band read: %v", err) 168 } 169 if !ok { 170 t.Fatal("the labor rate band read must find the seeded New Mexico band") 171 } 172 if band.SOC != "151252" || band.Role != "Systems Engineer" || band.AreaName != "New Mexico" { 173 t.Errorf("the band must carry the seeded SOC, role, and area name, got %+v", band) 174 } 175 if band.AnnualMean.StringFixed(2) != "132000.00" { 176 t.Errorf("the annual mean must be the exact seeded decimal, got %s", band.AnnualMean.StringFixed(2)) 177 } 178 if band.AnnualMedian.StringFixed(2) != "128000.00" { 179 t.Errorf("the annual median must be the exact seeded decimal, got %s", band.AnnualMedian.StringFixed(2)) 180 } 181 182 // An occupation and area with no wage row returns ok false so the honest floor holds upstream. 183 _, ok2, err := ai.LaborRateBand(ctxOrg(org), appDB, "999999", "NM") 184 if err != nil { 185 t.Fatalf("labor rate band read for a missing occupation: %v", err) 186 } 187 if ok2 { 188 t.Error("the labor rate band read must return ok false for an occupation with no wage row") 189 } 190} 191 192// assertCaptureAudit reads the aedt.ai.capture_answered audit payloads for an org and asserts price_resolved is 193// the expected boolean, and that no forbidden dollar substring (a wage value) ever appears in the counts only 194// audit. It renders the boolean the way Postgres serializes jsonb to text (a space after the colon). 195func assertCaptureAudit(t *testing.T, org uuid.UUID, wantResolved bool, forbiddenValues []string) { 196 t.Helper() 197 var payloads []string 198 if err := appDB.RunInTenantTx(ctxOrg(org), func(ctx context.Context, tx pgx.Tx) error { 199 rows, qerr := tx.Query(ctx, 200 "SELECT new_values::text FROM audit_logs WHERE organization_id=$1 AND action='aedt.ai.capture_answered'", org) 201 if qerr != nil { 202 return qerr 203 } 204 defer rows.Close() 205 for rows.Next() { 206 var p string 207 if serr := rows.Scan(&p); serr != nil { 208 return serr 209 } 210 payloads = append(payloads, p) 211 } 212 return rows.Err() 213 }); err != nil { 214 t.Fatalf("read capture audit: %v", err) 215 } 216 if len(payloads) == 0 { 217 t.Fatal("a capture answer must emit an aedt.ai.capture_answered audit event") 218 } 219 wantBool := "\"price_resolved\": false" 220 if wantResolved { 221 wantBool = "\"price_resolved\": true" 222 } 223 var sawExpected bool 224 for _, p := range payloads { 225 if strings.Contains(p, wantBool) { 226 sawExpected = true 227 } 228 for _, forbidden := range forbiddenValues { 229 if strings.Contains(p, forbidden) { 230 t.Errorf("the capture audit must hold no dollar value, leaked %q in %q", forbidden, p) 231 } 232 } 233 } 234 if !sawExpected { 235 t.Errorf("the capture audit must record price_resolved %v", wantResolved) 236 } 237} 238