func TestMultiDomainLogIter(t *testing.T) {
domains := []string{"test", "another_test"}
// Number of transactions
n := 100
// Number of facts per transaction
m := 100
engine := randMultidomainStorage(domains, n, m)
// Open and merge the commit logs.
log0, err := view.OpenLog(engine, domains[0], "commit")
if err != nil {
t.Fatal(err)
}
log1, err := view.OpenLog(engine, domains[1], "commit")
if err != nil {
t.Fatal(err)
}
now := time.Now()
mergedStreams := view.Merge(log0.Asof(now), log1.Asof(now))
if err = mergedStreams.Err(); err != nil {
t.Fatal(err)
}
i := 0
var prevTrnId uint64 = 1<<64 - 1
for {
f := mergedStreams.Next()
if err = mergedStreams.Err(); err != nil {
t.Fatal(err)
}
if f == nil {
break
} else {
if prevTrnId < f.Transaction {
t.Errorf("Transactions are ordered incorrectly")
}
i++
}
}
if i != n*m {
t.Errorf("expected %d facts, got %d", n*m, i)
}
}
func mergeDomains(engine storage.Engine, w origins.Writer, domains []string, since, asof time.Time) int {
var (
err error
count int
log *view.Log
)
iters := make([]origins.Iterator, len(domains))
// Merge and output facts across domains.
for i, d := range domains {
log, err = view.OpenLog(engine, d, "commit")
if err != nil {
logrus.Fatal(err)
}
iters[i] = log.View(since, asof)
}
if count, err = origins.Copy(view.Merge(iters...), w); err != nil {
logrus.Fatal(err)
}
return count
}
func concatDomains(engine storage.Engine, w origins.Writer, domains []string, since, asof time.Time) int {
var (
err error
n, count int
log *view.Log
)
// Output facts for each domain in the order they are supplied.
for _, d := range domains {
log, err = view.OpenLog(engine, d, "commit")
if err != nil {
logrus.Fatal(err)
}
v := log.View(since, asof)
n, err = origins.Copy(v, w)
if err != nil {
logrus.Fatal(err)
}
count += n
}
return count
}
// Encapsulates the logic for building a domain-based iterator.
func domainIteratorResource(domain string, r *http.Request, e storage.Engine) (origins.Iterator, int, error) {
var (
err error
since, asof time.Time
offset, limit int
)
if since, asof, err = parseTimeParams(r); err != nil {
return nil, StatusUnprocessableEntity, err
}
if offset, limit, err = parseSliceParams(r); err != nil {
return nil, StatusUnprocessableEntity, err
}
log, err := view.OpenLog(e, domain, "commit")
if err == view.ErrDoesNotExist {
return nil, http.StatusNotFound, err
}
if err != nil {
return nil, http.StatusInternalServerError, err
}
iter := log.View(since, asof)
if offset > 0 || limit > 0 {
iter = origins.Slice(iter, offset, limit)
}
return iter, http.StatusOK, nil
}
func TestLogIter(t *testing.T) {
domain := "test"
// Transactions
n := 100
// Size of write
m := 100
engine := randStorage(domain, n, m)
// Open the commit log.
log, err := view.OpenLog(engine, domain, "commit")
if err != nil {
t.Fatal(err)
}
num, err := testNext(log.Now())
if err != nil {
t.Fatal(err)
}
if num != n*m {
t.Errorf("expected %d facts, got %d", n*m, num)
}
}
func TestLogReader(t *testing.T) {
domain := "test"
// Transactions
n := 100
// Size of write
m := 100
engine := randStorage(domain, n, m)
// Open the commit log.
log, err := view.OpenLog(engine, domain, "commit")
if err != nil {
t.Fatal(err)
}
iter := log.Now()
facts, err := origins.ReadAll(iter)
if err != nil {
t.Fatal(err)
}
if len(facts) != n*m {
t.Errorf("expected %d facts, got %d", n*m, len(facts))
}
}
func benchmarkDeduplication(b *testing.B, numTrn, factsPerTrn, eLen, aLen, vLen int) {
domain := "test"
engine := randStorageWRepeats(domain, numTrn, factsPerTrn, eLen, aLen, vLen)
log, err := view.OpenLog(engine, domain, "commit")
if err != nil {
b.Fatal(err)
}
b.ResetTimer()
for i := 0; i < b.N; i++ {
now := log.Now()
iter := view.Deduplicate(now)
if err = iter.Err(); err != nil {
b.Fatal(err)
}
// The Deduplicate() operation is lazy, and most of the actual work happens during Next(),
// so to evaluate the true cost of deduplication we need to time how long it takes to step
// through the resulting iterator.
_, err = testNext(iter)
if err != nil {
b.Fatal(err)
}
}
}
// Load materializes the current state of a schema from the database.
func Load(engine storage.Engine, domain string) (*Schema, error) {
log, err := view.OpenLog(engine, domain, "commit")
if err != nil {
return nil, err
}
iter := log.Now()
return Init(domain, iter), nil
}
func TestLogExcludeDuplicates(t *testing.T) {
domain := "test"
// number of transactions
n := 100
// number of facts per transaction
m := 100
eLen, aLen, vLen := 2, 3, 4
engine := randStorageWRepeats(domain, n, m, eLen, aLen, vLen)
// Open the commit log.
log, err := view.OpenLog(engine, domain, "commit")
if err != nil {
t.Fatal(err)
}
// first check the total number of facts
num, err := testNext(log.Now())
if err != nil {
t.Fatal(err)
}
if num != n*m {
t.Errorf("expected %d total facts, got %d", n*m, num)
}
// Now check that Next() works on the deduplicated stream,
// and verify the number of unique facts.
iter := view.Deduplicate(log.Now())
if err := iter.Err(); err != nil {
t.Fatal(err)
}
num, err = testNext(iter)
if err != nil {
t.Fatal(err)
}
// With the dictionary size being very small (e.g. 24) compared to the number of generated facts (e.g. 10,000),
// the probability that any of the possible facts didn't get generated is negligible
if num != eLen*aLen*vLen {
t.Errorf("expected %d unique facts, got %d", eLen*aLen*vLen, num)
}
}
func benchmarkDomainMerge(b *testing.B, numTrn, factsPerTrn, numDomains int) {
var err error
domains := make([]string, numDomains)
iters := make([]origins.Iterator, len(domains))
logs := make([]*view.Log, len(domains))
for j := 0; j < numDomains; j++ {
domains[j] = "domain_" + strconv.Itoa(j+1)
}
engine := randMultidomainStorage(domains, numTrn, factsPerTrn)
for j := 0; j < len(domains); j++ {
logs[j], err = view.OpenLog(engine, domains[j], "commit")
if err != nil {
b.Fatal(err)
}
}
now := time.Now()
b.ResetTimer()
for i := 0; i < b.N; i++ {
for j := 0; j < len(domains); j++ {
iters[j] = logs[j].Asof(now)
}
iter := view.Merge(iters...)
if err := iter.Err(); err != nil {
b.Fatal(err)
}
// The Merge() operation is lazy, and most of the actual work happens during Next(),
// so to evaluate the true cost of merging we need to time how long it takes to step
// through the resulting iterator.
_, err = testNext(iter)
if err != nil {
b.Fatal(err)
}
}
}
func TestLogSince(t *testing.T) {
domain := "test"
// Transactions
n := 100
// Size of write
m := 100
engine := randStorage(domain, n, m)
// Open the commit log.
log, err := view.OpenLog(engine, domain, "commit")
// 1 minute before
min := time.Now().Add(-time.Minute)
iter := log.Since(min)
facts, err := origins.ReadAll(iter)
if err != nil {
t.Fatal(err)
}
if len(facts) != n*m {
t.Errorf("expected %d facts, got %d", n*m, len(facts))
}
// 1 minute later
min = time.Now().Add(time.Minute)
iter = log.Since(min)
facts, err = origins.ReadAll(iter)
if err != nil {
t.Fatal(err)
}
if len(facts) != 0 {
t.Errorf("expected 0 facts, got %d", len(facts))
}
}
func (p *Pipeline) initCache() error {
logrus.Debugf("transactor.Pipeline(%s): initializing cache", p.Domain)
p.initialized = true
log, err := view.OpenLog(p.engine, p.Domain, commitLogName)
// This denotes the domain is new.
if err == view.ErrDoesNotExist {
return nil
} else if err != nil {
return err
}
facts, err := origins.ReadAll(log.Asof(p.segment.Time))
if err != nil {
return err
}
// Sort facts by entity.
origins.Timsort(facts, origins.EAVTComparator)
// Group the facts by entity.
giter := origins.Groupby(origins.NewBuffer(facts), func(f1, f2 *origins.Fact) bool {
return f1.Entity.Is(f2.Entity)
})
// Initializing ctrie.
cache := ctrie.New(nil)
err = origins.MapFacts(giter, func(facts origins.Facts) error {
cache.Insert([]byte(facts[0].Entity.Name), facts)
return nil
})
p.cache = cache
logrus.Debugf("transactor.Pipeline(%s): cache initialized", p.Domain)
return err
}
"github.com/chop-dbhi/origins"
"github.com/chop-dbhi/origins/view"
"github.com/spf13/cobra"
)
var domainsCmd = &cobra.Command{
Use: "domains",
Short: "Outputs a list of domains.",
Run: func(cmd *cobra.Command, args []string) {
bindStorageFlags(cmd.Flags())
engine := initStorage()
log, err := view.OpenLog(engine, origins.DomainsDomain, "commit")
if err != nil {
logrus.Fatal(err)
}
idents, err := origins.Entities(log.Now())
if err != nil {
logrus.Fatal(err)
}
sort.Sort(idents)
for _, ident := range idents {
os.Stdout.Write([]byte(ident.Name + "\n"))
