Avoid redundant lineage shadow scans

Author: iam-brain

Sources/CodexBarCore/Generated/CodexParserHash.generated.swift

@@ -1,5 +1,5 @@
 // Generated by Scripts/regenerate-codex-parser-hash.sh. Do not edit by hand.
 
 enum CodexParserHash {
-    static let value = "adea6d0a163c51e2"
+    static let value = "a22a4756d8f471d4"
 }

Sources/CodexBarCore/Providers/Codex/CodexLineageLedger.swift

@@ -96,18 +96,24 @@ enum CodexLineageLedger {
         }
 
         var acceptedByComponent: [String: [ObservationIdentity: AcceptedObservation]] = [:]
+        var acceptedFingerprintByComponentOwner: [String: [String: [Fingerprint: ObservationIdentity]]] = [:]
         var physicalObservationCount = 0
         for document in documents {
             try checkCancellation?()
             let componentID = graph.find(document.ownerID)
             var accepted = acceptedByComponent[componentID] ?? [:]
+            var acceptedFingerprintByOwner = acceptedFingerprintByComponentOwner[componentID] ?? [:]
             for observation in document.observations {
                 try checkCancellation?()
                 physicalObservationCount += 1
                 let date = try Self.date(from: observation.timestamp)
-                let identity = ObservationIdentity(
+                let fingerprint = Fingerprint(last: observation.last, total: observation.total)
+                let proposedIdentity = ObservationIdentity(
                     eventID: Self.nonEmpty(observation.eventID),
-                    fingerprint: Fingerprint(last: observation.last, total: observation.total))
+                    fingerprint: fingerprint)
+                let identity = accepted[proposedIdentity] == nil
+                    ? acceptedFingerprintByOwner[document.ownerID]?[fingerprint] ?? proposedIdentity
+                    : proposedIdentity
                 if let existing = accepted[identity] {
                     if existing.date < date {
                         continue
@@ -122,8 +128,12 @@ enum CodexLineageLedger {
                     date: date,
                     model: observation.model,
                     last: observation.last)
+                if identity == proposedIdentity {
+                    acceptedFingerprintByOwner[document.ownerID, default: [:]][fingerprint] = identity
+                }
             }
             acceptedByComponent[componentID] = accepted
+            acceptedFingerprintByComponentOwner[componentID] = acceptedFingerprintByOwner
         }
 
         var utcDays: [String: Totals] = [:]
@@ -159,11 +169,11 @@ enum CodexLineageLedger {
     }
 
     private enum ObservationIdentity: Equatable, Hashable {
-        case event(String)
+        case event(String, Fingerprint)
         case fingerprint(Fingerprint)
 
         init(eventID: String?, fingerprint: Fingerprint) {
-            self = eventID.map(Self.event) ?? .fingerprint(fingerprint)
+            self = eventID.map { .event($0, fingerprint) } ?? .fingerprint(fingerprint)
         }
     }
 

Sources/CodexBarCore/Vendored/CostUsage/CostUsageScanner.swift

@@ -80,6 +80,7 @@ enum CostUsageScanner {
         var contributingSessionIds: Set<String> = []
         var seenFileIds: Set<String> = []
         var seenCodexUsageRowKeys: Set<String> = []
+        var changedLineageInputs = false
     }
 
     struct CodexScannedSession {
@@ -2473,6 +2474,7 @@ enum CostUsageScanner {
         if Self.keepCachedCodexFileIfFresh(input: input, context: context, cache: &cache, state: &state) {
             return
         }
+        state.changedLineageInputs = true
         if try Self.appendCodexFileIncrementIfPossible(input: input, context: context, cache: &cache, state: &state) {
             return
         }
@@ -2682,6 +2684,7 @@ enum CostUsageScanner {
                 let shouldDrop = shouldDropAllUnscannedFiles ||
                     old.touchesCodexScanWindow(sinceKey: range.scanSinceKey, untilKey: range.scanUntilKey)
                 guard shouldDrop else { continue }
+                scanState.changedLineageInputs = true
                 Self.applyFileDays(cache: &cache, fileDays: old.days, sign: -1)
                 cache.files.removeValue(forKey: key)
             }
@@ -2692,6 +2695,7 @@ enum CostUsageScanner {
                     guard old.touchesCodexScanWindow(sinceKey: range.scanSinceKey, untilKey: range.scanUntilKey)
                     else { continue }
                     guard FileManager.default.fileExists(atPath: key) else {
+                        scanState.changedLineageInputs = true
                         Self.applyFileDays(cache: &cache, fileDays: old.days, sign: -1)
                         cache.files.removeValue(forKey: key)
                         continue
@@ -2708,12 +2712,14 @@ enum CostUsageScanner {
                 ? [cachedUntilKey, range.scanUntilKey].compactMap(\.self).max() ?? range.scanUntilKey
                 : range.scanUntilKey
             Self.pruneDays(cache: &cache, sinceKey: retainedSinceKey, untilKey: retainedUntilKey)
-            try Self.recordCodexLineageShadow(
-                files: files,
-                roots: plan.roots,
-                cache: cache,
-                range: range,
-                checkCancellation: checkCancellation)
+            if scanState.changedLineageInputs {
+                try Self.recordCodexLineageShadow(
+                    files: files,
+                    roots: plan.roots,
+                    cache: cache,
+                    range: range,
+                    checkCancellation: checkCancellation)
+            }
             cache.roots = plan.rootsFingerprint
             cache.scanSinceKey = retainedSinceKey
             cache.scanUntilKey = retainedUntilKey

Tests/CodexBarTests/CodexLineageLedgerTests.swift

@@ -244,6 +244,36 @@ struct CodexLineageLedgerTests {
         #expect(report.duplicateObservationCount == 2)
     }
 
+    @Test
+    func `matching event hints with different token states remain distinct`() throws {
+        let first = Self.observation(
+            eventID: "turn-a:0", timestamp: "2026-07-09T12:00:00Z", input: 100, totalInput: 100)
+        let forked = Self.observation(
+            eventID: "turn-a:0", timestamp: "2026-07-09T12:01:00Z", input: 25, totalInput: 125)
+        let report = try CodexLineageLedger.reconcile(
+            documents: [Self.document(owner: "root", observations: [first, forked])],
+            localTimeZone: .gmt)
+
+        #expect(report.utcDays["2026-07-09"]?.input == 125)
+        #expect(report.acceptedObservationCount == 2)
+        #expect(report.duplicateObservationCount == 0)
+    }
+
+    @Test
+    func `different event ordinals do not revive an unchanged owner state`() throws {
+        let first = Self.observation(
+            eventID: "turn-a:0", timestamp: "2026-07-09T12:00:00Z", input: 100, totalInput: 100)
+        let repeated = Self.observation(
+            eventID: "turn-a:1", timestamp: "2026-07-09T12:01:00Z", input: 100, totalInput: 100)
+        let report = try CodexLineageLedger.reconcile(
+            documents: [Self.document(owner: "root", observations: [first, repeated])],
+            localTimeZone: .gmt)
+
+        #expect(report.utcDays["2026-07-09"]?.input == 100)
+        #expect(report.acceptedObservationCount == 1)
+        #expect(report.duplicateObservationCount == 1)
+    }
+
     @Test
     func `complete token state distinguishes observations within a lineage`() throws {
         let first = Self.observation(
@@ -353,6 +383,7 @@ struct CodexLineageLedgerTests {
     }
 
     private static func observation(
+        eventID: String? = nil,
         timestamp: String,
         model: String = CostUsagePricing.codexUnattributedModel,
         input: Int,
@@ -361,6 +392,7 @@ struct CodexLineageLedgerTests {
         totalInput: Int) -> CodexLineageLedger.Observation
     {
         .init(
+            eventID: eventID,
             timestamp: timestamp,
             model: model,
             last: .init(input: input, cached: cached, output: output),