import unittest, os, strutils, tables, sets
import ../src/nimpak/cas
import ../src/nip/types

suite "Cross-Format Deduplication Metrics Tests":
  
  var 
    cas: CasManager
    testRoot = getTempDir() / "nip_dedup_test_" & $getCurrentProcessId()
    
    # Test data
    chunk1 = @[1.byte, 2.byte, 3.byte] # 3 bytes
    chunk2 = @[4.byte, 5.byte, 6.byte] # 3 bytes
    chunk3 = @[7.byte, 8.byte, 9.byte] # 3 bytes
    
    hash1: string
    hash2: string
    hash3: string

  setup:
    createDir(testRoot)
    cas = initCasManager(testRoot)
    
    # Store chunks
    let res1 = cas.storeObject(chunk1)
    let res2 = cas.storeObject(chunk2)
    let res3 = cas.storeObject(chunk3)
    
    hash1 = res1.get().hash
    hash2 = res2.get().hash
    hash3 = res3.get().hash

  teardown:
    removeDir(testRoot)

  test "Basic Deduplication Stats":
    # Scenario:
    # NPK uses chunk1, chunk2
    # NIP uses chunk2, chunk3
    # NEXTER uses chunk1, chunk3
    
    # chunk1: NPK, NEXTER (Ref count 2)
    # chunk2: NPK, NIP (Ref count 2)
    # chunk3: NIP, NEXTER (Ref count 2)
    
    discard cas.addReference(hash1, NPK, "pkg1")
    discard cas.addReference(hash2, NPK, "pkg1")
    
    discard cas.addReference(hash2, NIP, "pkg2")
    discard cas.addReference(hash3, NIP, "pkg2")
    
    discard cas.addReference(hash1, NEXTER, "pkg3")
    discard cas.addReference(hash3, NEXTER, "pkg3")
    
    let statsResult = cas.getDeduplicationStats()
    check statsResult.isOk
    let stats = statsResult.get()
    
    # Physical size: 3 chunks * 3 bytes = 9 bytes
    check stats.totalPhysicalSize == 9
    
    # Logical size: 
    # pkg1: 3+3 = 6
    # pkg2: 3+3 = 6
    # pkg3: 3+3 = 6
    # Total: 18 bytes
    check stats.totalLogicalSize == 18
    
    # Deduplication ratio: 18 / 9 = 2.0
    check stats.deduplicationRatio == 2.0
    
    # Shared chunks: All 3 are shared
    check stats.sharedChunks == 3
    
    # Savings: 18 - 9 = 9 bytes
    check stats.savings == 9
    
    # Format Overlap
    # chunk1: NPK-NEXTER
    # chunk2: NIP-NPK
    # chunk3: NEXTER-NIP
    
    check stats.formatOverlap.hasKey("NEXTER-NPK")
    check stats.formatOverlap["NEXTER-NPK"] == 1
    
    check stats.formatOverlap.hasKey("NIP-NPK")
    check stats.formatOverlap["NIP-NPK"] == 1
    
    check stats.formatOverlap.hasKey("NEXTER-NIP")
    check stats.formatOverlap["NEXTER-NIP"] == 1

  test "No Deduplication":
    # Scenario: Unique chunks for each
    discard cas.addReference(hash1, NPK, "pkg1")
    discard cas.addReference(hash2, NIP, "pkg2")
    discard cas.addReference(hash3, NEXTER, "pkg3")
    
    let statsResult = cas.getDeduplicationStats()
    check statsResult.isOk
    let stats = statsResult.get()
    
    check stats.totalPhysicalSize == 9
    check stats.totalLogicalSize == 9
    check stats.deduplicationRatio == 1.0
    check stats.sharedChunks == 0
    check stats.savings == 0
    check stats.formatOverlap.len == 0

  test "High Redundancy":
    # Scenario: All formats use same chunk
    discard cas.addReference(hash1, NPK, "pkg1")
    discard cas.addReference(hash1, NIP, "pkg2")
    discard cas.addReference(hash1, NEXTER, "pkg3")
    
    let statsResult = cas.getDeduplicationStats()
    check statsResult.isOk
    let stats = statsResult.get()
    
    # Physical: 3 bytes (only chunk1 counted, others ignored if not referenced? No, we only iterate referenced hashes)
    # Wait, hash2 and hash3 exist in CAS but are NOT referenced.
    # getDeduplicationStats iterates over `cas.formatRefs`.
    # So unreferenced chunks are NOT included in stats.
    
    check stats.totalPhysicalSize == 3
    check stats.totalLogicalSize == 9 # 3 refs * 3 bytes
    check stats.deduplicationRatio == 3.0
    check stats.sharedChunks == 1
    
    check stats.formatOverlap.hasKey("NEXTER-NIP-NPK")
    check stats.formatOverlap["NEXTER-NIP-NPK"] == 1