import "github.com/hundredwatt/difference_digest"

package difference_digest_test

import (
  "fmt"
  "math"
  "math/rand"
  "time"

  _ "github.com/lib/pq"

  "github.com/hundredwatt/difference_digest"
)

var (
  db *db.SQL
)

func Example_dbs() {
  db.Exec("CREATE TEMP TABLE sourcethings (id bigint)")
  db.Exec("CREATE TEMP TABLE sinkthings (id bigint)")
  db.Exec("INSERT INTO sourcethings (id) SELECT * from generate_series(1,9999900)") // 100 items missing out of 10,000,000
  db.Exec("INSERT INTO sinkthings (id) SELECT * from generate_series(1,10000000)")

  // 1. Use the Strata Estimator to get the approximate number of differences between the two tables
  sourceEstimator, _ := difference_digest.EncodeEstimatorDB(db, "sourcethings", "id")
  sinkEstimator, _ := difference_digest.EncodeEstimatorDB(db, "sinkthings", "id")
  estimatedDeletes := sinkEstimator.EstimateDifference(sourceEstimator)

  // 2. Compute an Invertible Bloom Filter (of the appropriate size) from each source
  alpha := 3.0
  cells := int(math.Ceil(float64(estimatedDeletes) * float64(alpha)))
  sourceIBF, _ := difference_digest.EncodeIBFDB(cells, db, "sourcethings", "id")
  sinkIBF, _ := difference_digest.EncodeIBFDB(cells, db, "sinkthings", "id")

  // 3. Compute the difference of the IBFs
  diff := sinkIBF.Subtract(sourceIBF)
  sinkWithoutSource, _, ok := diff.Decode()
  if !ok {
    fmt.Println("Invertible Bloom Filter failed to decode, please try again")
  } else {
    fmt.Printf("%d elements found in sink that are not in source", len(sinkWithoutSource))
    // Output: 100 elements found in sink that are not in source
  }
}