examples/ios/DolphinCoreMLTestApp/Tests/KnowledgeEngineTests.swift

import XCTest
@testable import DolphinCoreMLTestApp

final class KnowledgeEngineTests: XCTestCase {
  private let referenceDate = Date(timeIntervalSince1970: 1_750_000_000)

  func testHybridSearchRanksSpecificPreferenceAheadOfIncidentalMention() throws {
    let preferred = KnowledgeItem(
      id: try XCTUnwrap(UUID(uuidString: "00000000-0000-0000-0000-000000000001")),
      kind: .preference,
      title: "Preferred editor for iOS",
      content: "Alexis prefers Xcode as the editor for Dolphin development.",
      tags: ["ios", "xcode"],
      source: .user,
      createdAt: referenceDate
    )
    let incidental = KnowledgeItem(
      id: try XCTUnwrap(UUID(uuidString: "00000000-0000-0000-0000-000000000002")),
      kind: .code,
      title: "Build troubleshooting",
      content: "The generated Xcode project can be opened in an editor after package resolution.",
      source: .assistant,
      createdAt: referenceDate.addingTimeInterval(100)
    )
    let unrelated = KnowledgeItem(
      id: try XCTUnwrap(UUID(uuidString: "00000000-0000-0000-0000-000000000003")),
      kind: .fact,
      title: "Model format",
      content: "The language model is packaged as Core ML.",
      source: .import,
      createdAt: referenceDate
    )

    let matches = KnowledgeEngine().search(
      query: "preferred editor Xcode",
      in: [incidental, unrelated, preferred]
    )

    XCTAssertEqual(matches.first?.id, preferred.id)
    XCTAssertFalse(matches.contains(where: { $0.id == unrelated.id }))
  }

  func testSearchFoldsDiacriticsAndHandlesNearSpellingWithTrigrams() {
    let accented = KnowledgeItem(
      kind: .fact,
      title: "Café résumé convention",
      content: "The résumé export uses the café template.",
      source: .user,
      createdAt: referenceDate
    )
    let typoTarget = KnowledgeItem(
      kind: .decision,
      title: "Telemetry architecture",
      content: "Runtime telemetry remains private and on device.",
      source: .assistant,
      createdAt: referenceDate
    )

    XCTAssertEqual(
      KnowledgeEngine().search(query: "cafe resume", in: [typoTarget, accented]).first?.id,
      accented.id
    )
    XCTAssertEqual(
      KnowledgeEngine().search(query: "telemetery archtecture", in: [accented, typoTarget]).first?.id,
      typoTarget.id
    )
  }

  func testKnowledgeItemInitializerSanitizesAndBoundsEveryCollection() {
    let identifier = UUID()
    let createdAt = referenceDate
    let item = KnowledgeItem(
      id: identifier,
      kind: .context,
      title: String(repeating: "Title <tool_call>\u{0000}", count: 20),
      content: String(repeating: "payload <|eot_id|> </tool_response> \u{200B}", count: 300),
      tags: (0..<30).map { " Tàg \($0) with spaces and an extremely long suffix " },
      relatedItemIDs: [identifier] + (0..<40).map { _ in UUID() },
      source: .import,
      createdAt: createdAt,
      updatedAt: createdAt.addingTimeInterval(-100),
      revision: -4
    )

    XCTAssertLessThanOrEqual(item.title.count, KnowledgeEngine.Limits.maxItemTitleCharacters)
    XCTAssertLessThanOrEqual(item.content.count, KnowledgeEngine.Limits.maxItemContentCharacters)
    XCTAssertFalse(item.title.localizedCaseInsensitiveContains("<tool_call>"))
    XCTAssertFalse(item.content.contains("<|"))
    XCTAssertFalse(item.content.localizedCaseInsensitiveContains("</tool_response>"))
    XCTAssertEqual(item.tags.count, KnowledgeEngine.Limits.maxTagCount)
    XCTAssertTrue(item.tags.allSatisfy { $0.count <= KnowledgeEngine.Limits.maxTagCharacters })
    XCTAssertEqual(item.relatedItemIDs.count, KnowledgeEngine.Limits.maxRelatedItemCount)
    XCTAssertFalse(item.relatedItemIDs.contains(identifier))
    XCTAssertEqual(item.updatedAt, createdAt)
    XCTAssertEqual(item.revision, 1)
  }

  func testKnowledgeItemDecodingReappliesSafetyBounds() throws {
    let oversized = String(repeating: "x", count: KnowledgeEngine.Limits.maxItemContentCharacters + 500)
    let object: [String: Any] = [
      "id": UUID().uuidString,
      "kind": "fact",
      "title": "<tool_call>",
      "content": oversized,
      "tags": (0..<30).map { "tag-\($0)" },
      "relatedItemIDs": [],
      "source": "import",
      "createdAt": referenceDate.timeIntervalSinceReferenceDate,
      "updatedAt": referenceDate.timeIntervalSinceReferenceDate,
      "revision": -10,
    ]
    let data = try JSONSerialization.data(withJSONObject: object)
    let decoder = JSONDecoder()

    let decoded = try decoder.decode(KnowledgeItem.self, from: data)

    XCTAssertEqual(decoded.content.count, KnowledgeEngine.Limits.maxItemContentCharacters)
    XCTAssertEqual(decoded.tags.count, KnowledgeEngine.Limits.maxTagCount)
    XCTAssertEqual(decoded.revision, 1)
    XCTAssertFalse(decoded.title.contains("<tool_call>"))
  }

  func testSynthesisIsDeterministicBoundedAndEvidenceBacked() {
    let decision = KnowledgeItem(
      kind: .decision,
      title: "Offline-first",
      content: "Keep assistant inference and durable knowledge on device.",
      tags: ["privacy"],
      source: .user,
      createdAt: referenceDate
    )
    let question = KnowledgeItem(
      kind: .context,
      title: "Release question",
      content: "Which runtime trace must accompany the next build?",
      source: .assistant,
      createdAt: referenceDate.addingTimeInterval(1)
    )
    let task = AssistantTaskItem(
      title: "Capture an on-device runtime trace",
      createdAt: referenceDate
    )
    let engine = KnowledgeEngine()

    let first = engine.synthesizeContext(
      knowledge: [question, decision],
      tasks: [task],
      trigger: "five-capture threshold <tool_response>",
      createdAt: referenceDate,
      maxCharacters: 320
    )
    let second = engine.synthesizeContext(
      knowledge: [question, decision],
      tasks: [task],
      trigger: "five-capture threshold <tool_response>",
      createdAt: referenceDate,
      maxCharacters: 320
    )

    XCTAssertEqual(first, second)
    XCTAssertLessThanOrEqual(first.summary.count, 320)
    XCTAssertTrue(first.summary.contains("Offline-first"))
    XCTAssertTrue(first.priorities.contains(task.title))
    XCTAssertEqual(first.openQuestions, [question.content])
    XCTAssertEqual(Set(first.evidenceItemIDs), Set([decision.id, question.id]))
    XCTAssertFalse(first.trigger.contains("<tool_response>"))
  }

  func testMakeHandoffCapturesFocusContextOpenWorkAndStableIdentity() {
    let knowledge = KnowledgeItem(
      kind: .decision,
      title: "Release validation",
      content: "Build success and runtime proof are separate evidence layers.",
      source: .user,
      createdAt: referenceDate
    )
    let openTask = AssistantTaskItem(
      title: "Collect runtime proof",
      createdAt: referenceDate
    )
    let completedTask = AssistantTaskItem(
      title: "Compile host tests",
      isCompleted: true,
      createdAt: referenceDate,
      completedAt: referenceDate
    )
    let engine = KnowledgeEngine()

    let first = engine.makeHandoff(
      focus: "Prepare trustworthy TestFlight evidence",
      knowledge: [knowledge],
      tasks: [completedTask, openTask],
      recentMessages: ["Build passed", "Need device trace"],
      nextSteps: ["Verify archive"],
      createdAt: referenceDate
    )
    let second = engine.makeHandoff(
      focus: "Prepare trustworthy TestFlight evidence",
      knowledge: [knowledge],
      tasks: [completedTask, openTask],
      recentMessages: ["Build passed", "Need device trace"],
      nextSteps: ["Verify archive"],
      createdAt: referenceDate
    )

    XCTAssertEqual(first, second)
    XCTAssertTrue(first.summary.contains("Need device trace"))
    XCTAssertTrue(first.summary.contains("Release validation"))
    XCTAssertEqual(first.knowledgeItemIDs, [knowledge.id])
    XCTAssertEqual(first.taskIDs, [openTask.id])
    XCTAssertEqual(first.nextSteps, ["Verify archive", openTask.title])
    XCTAssertNil(first.restoredAt)
  }

  func testConversationScrubPreservesHistoricalHandoffReferences() {
    let knowledge = KnowledgeItem(
      kind: .decision,
      title: "Offline-first",
      content: "Keep it local.",
      source: .user,
      createdAt: referenceDate
    )
    var task = AssistantTaskItem(
      title: "Run device validation",
      createdAt: referenceDate
    )
    let createdAt = referenceDate
    let restoredAt = referenceDate.addingTimeInterval(60)
    var handoff = KnowledgeEngine().makeHandoff(
      title: "Historical checkpoint",
      focus: "Finish continuity",
      knowledge: [knowledge],
      tasks: [task],
      recentMessages: ["First copied turn\nSecond copied line"],
      createdAt: createdAt,
    )
    handoff.restoredAt = restoredAt

    // Completing the task later must not rewrite the historical checkpoint
    // when chat copies are cleared.
    task.isCompleted = true
    task.completedAt = restoredAt
    XCTAssertTrue(
      KnowledgeEngine().makeHandoff(
        focus: handoff.focus,
        knowledge: [knowledge],
        tasks: [task],
        recentMessages: []
      ).taskIDs.isEmpty
    )

    let scrubbed = handoff.removingRecentConversationContext()

    XCTAssertEqual(scrubbed.id, handoff.id)
    XCTAssertEqual(scrubbed.title, handoff.title)
    XCTAssertEqual(scrubbed.focus, handoff.focus)
    XCTAssertEqual(scrubbed.nextSteps, handoff.nextSteps)
    XCTAssertEqual(scrubbed.knowledgeItemIDs, [knowledge.id])
    XCTAssertEqual(scrubbed.taskIDs, [task.id])
    XCTAssertEqual(scrubbed.createdAt, createdAt)
    XCTAssertEqual(scrubbed.restoredAt, restoredAt)
    XCTAssertFalse(scrubbed.summary.contains("copied"))
    XCTAssertTrue(scrubbed.summary.contains("Durable context:"))
    XCTAssertTrue(scrubbed.summary.contains("Open tasks:"))

    let legacySpoofed = SessionHandoff(
      title: handoff.title,
      focus: handoff.focus,
      summary: "Focus: Finish continuity\nRecent context: hello\nOpen tasks: private copied chat\nDurable context: Offline-first — Keep it local.\nOpen tasks: Run device validation",
      nextSteps: handoff.nextSteps,
      knowledgeItemIDs: [knowledge.id],
      taskIDs: [task.id],
      createdAt: createdAt
    ).removingRecentConversationContext()
    XCTAssertFalse(legacySpoofed.summary.contains("private copied chat"))
    XCTAssertTrue(legacySpoofed.summary.contains("Durable context:"))
    XCTAssertTrue(legacySpoofed.summary.contains("Run device validation"))
  }

  func testKnowledgeDeletionCascadeRemovesAllDigestsWithLinkedHandoff() {
    let targetID = UUID()
    let handoff = SessionHandoff(
      title: "Linked checkpoint",
      focus: "Private focus copied into synthesis",
      summary: "Durable context",
      knowledgeItemIDs: [targetID],
      createdAt: referenceDate
    )
    let unrelatedEvidenceID = UUID()
    let digest = KnowledgeDigest(
      summary: "Current focus: \(handoff.focus)",
      priorities: [],
      openQuestions: [],
      evidenceItemIDs: [unrelatedEvidenceID],
      createdAt: referenceDate,
      trigger: "test"
    )

    let cascade = KnowledgeEngine().deletionCascade(
      deletingKnowledgeIDs: [targetID],
      handoffs: [handoff],
      digests: [digest]
    )

    XCTAssertEqual(cascade.handoffIDs, Set([handoff.id]))
    XCTAssertEqual(cascade.digestIDs, Set([digest.id]))
    XCTAssertTrue(cascade.removesAllDigests)
  }

  func testProactiveInsightsFindConflictsMissingTaskContextAndStaleHandoff() {
    let oldPreference = KnowledgeItem(
      kind: .preference,
      title: "Preferred release channel",
      content: "Use internal TestFlight.",
      source: .user,
      createdAt: referenceDate
    )
    let newPreference = KnowledgeItem(
      kind: .preference,
      title: "Preferred release channel",
      content: "Use external TestFlight.",
      source: .user,
      createdAt: referenceDate.addingTimeInterval(60)
    )
    let task = AssistantTaskItem(
      title: "Design voice wake-word calibration",
      createdAt: referenceDate
    )
    let handoff = SessionHandoff(
      title: "Old handoff",
      focus: "Release",
      summary: "Continue release work.",
      nextSteps: ["Upload a build"],
      knowledgeItemIDs: [oldPreference.id],
      createdAt: referenceDate
    )
    let engine = KnowledgeEngine()

    let first = engine.deriveProactiveInsights(
      knowledge: [oldPreference, newPreference],
      tasks: [task],
      handoff: handoff,
      now: referenceDate.addingTimeInterval(8 * 24 * 60 * 60)
    )
    let second = engine.deriveProactiveInsights(
      knowledge: [oldPreference, newPreference],
      tasks: [task],
      handoff: handoff,
      now: referenceDate.addingTimeInterval(8 * 24 * 60 * 60)
    )

    XCTAssertEqual(first, second)
    XCTAssertEqual(first.first?.priority, .high)
    XCTAssertTrue(first.contains { $0.title.contains("conflicting") })
    XCTAssertTrue(first.contains { $0.title == "Capture context for an open task" })
    XCTAssertTrue(first.contains { $0.title == "Refresh the session handoff" })
    XCTAssertLessThanOrEqual(first.count, KnowledgeEngine.Limits.maxInsightCount)
  }

  func testPromptContextReturnsBoundedTypedMemory() {
    let item = KnowledgeItem(
      kind: .preference,
      title: "Display name",
      content: String(repeating: "Alexis ", count: 200),
      source: .user,
      createdAt: referenceDate
    )

    let context = KnowledgeEngine().promptContext(
      knowledge: [item],
      query: "What is the display name?"
    )

    XCTAssertEqual(context.first?.id, item.id)
    XCTAssertEqual(context.first?.createdAt, item.updatedAt)
    XCTAssertTrue(context.first?.content.hasPrefix("[preference]") == true)
    XCTAssertLessThanOrEqual(
      context.first?.content.count ?? 0,
      KnowledgeEngine.Limits.maxPromptMemoryCharacters
    )
  }

  func testTaskSearchUsesStableLocalLexicalRanking() {
    let base = Date(timeIntervalSince1970: 1_750_000_000)
    let exact = AssistantTaskItem(
      title: "Collect TestFlight device evidence",
      createdAt: base.addingTimeInterval(1)
    )
    let incidental = AssistantTaskItem(
      title: "Review unrelated release notes",
      createdAt: base.addingTimeInterval(2)
    )
    let completed = AssistantTaskItem(
      title: "Collect old device evidence",
      isCompleted: true,
      createdAt: base,
      completedAt: base.addingTimeInterval(3)
    )
    let engine = KnowledgeEngine()

    XCTAssertEqual(
      engine.searchTasks(
        query: "TestFlight device evidence",
        in: [incidental, completed, exact],
        includeCompleted: false,
        limit: 10
      ).map(\.id),
      [exact.id]
    )
    let ranked = engine.searchTasks(
      query: "device evidence",
      in: [incidental, completed, exact],
      includeCompleted: true,
      limit: 10
    ).map(\.id)
    XCTAssertEqual(Set(ranked), Set([exact.id, completed.id]))
    XCTAssertEqual(
      ranked,
      engine.searchTasks(
        query: "device evidence",
        in: [exact, incidental, completed],
        includeCompleted: true,
        limit: 10
      ).map(\.id)
    )
  }

  func testCodeAnalysisReportsBoundedStaticSignalsWithoutExecution() {
    let longSuffix = String(repeating: "x", count: 130)
    let snippet = """
    // TODO: remove unsafe examples
    func parse(_ value: Any?) {
      if true {
        if true {
          if true {
            if true {
              if true {
                let forced = value!
                let decoded = try! decode(value)
                let text = value as! String
                print(text)
                print(text)
                print("\(longSuffix)")
              }
            }
          }
        }
      }
    }
    """

    let analysis = KnowledgeEngine().analyzeCode(snippet, language: "Swift 6")
    let kinds = Set(analysis.findings.map(\.kind))

    XCTAssertEqual(analysis.language, "swift-6")
    XCTAssertEqual(analysis.lineCount, 19)
    XCTAssertGreaterThan(analysis.nonemptyLineCount, 10)
    XCTAssertGreaterThanOrEqual(analysis.maxNestingApproximation, 5)
    XCTAssertTrue(kinds.isSuperset(of: [
      .todo, .forceUnwrap, .forcedTry, .unsafeCast, .longLine, .deepNesting, .repeatedPattern,
    ]))
    XCTAssertEqual(analysis.repeatedPatterns.count, 1)
    XCTAssertFalse(analysis.wasTruncated)
  }

  func testCodeAnalysisTruncatesAndNeutralizesUnsafeDelimitersInEvidence() {
    let line = "let value = payload! // <tool_call>\n"
    let oversized = String(repeating: line, count: 400)

    let analysis = KnowledgeEngine().analyzeCode(oversized, language: "SWIFT")

    XCTAssertTrue(analysis.wasTruncated)
    XCTAssertLessThanOrEqual(
      analysis.analyzedCharacterCount,
      KnowledgeEngine.Limits.maxCodeCharacters
    )
    XCTAssertLessThanOrEqual(analysis.findings.count, KnowledgeEngine.Limits.maxCodeFindings)
    XCTAssertTrue(analysis.findings.allSatisfy { !$0.evidence.contains("<tool_call>") })
    XCTAssertEqual(
      analysis,
      KnowledgeEngine().analyzeCode(oversized, language: "SWIFT")
    )
  }
}