// Import imports triples from an Turtle stream, in batches of given size.
// It will ignore triples with blank nodes and errors.
// It returns the total number of triples imported.
func (db *DB) Import(r io.Reader, batchSize int) (int, error) {
dec := rdf.NewDecoder(r)
g := rdf.NewGraph()
c := 0 // totalt count
i := 0 // current batch count
for tr, err := dec.Decode(); err != io.EOF; tr, err = dec.Decode() {
if err != nil {
// log.Println(err.Error())
continue
}
g.Insert(tr)
i++
if i == batchSize {
err = db.ImportGraph(g)
if err != nil {
return c, err
}
c += i
i = 0
g = rdf.NewGraph()
}
}
if len(g.Nodes()) > 0 {
err := db.ImportGraph(g)
if err != nil {
return c, err
}
c += i
}
return c, nil
}
func TestImportDumpGraph_Quick(t *testing.T) {
f := func(items testdata) bool {
db := newTestDB()
defer db.Close()
RemoveDuplicates(&items)
input := items.Graph().Serialize(rdf.Turtle, "http://test.org/")
n, err := db.Import(bytes.NewBufferString(input), rand.Intn(1000))
if err != nil {
t.Logf("DB.Import() failed: %v", err)
t.FailNow()
}
if n != len(items) {
t.Logf("DB.Import() => %d; want %d", n, len(items))
t.FailNow()
}
want := items.Graph()
var b bytes.Buffer
err = db.Dump(&b)
if err != nil {
t.Logf("DB.Dump() failed: %v", err)
t.FailNow()
}
dec := rdf.NewDecoder(&b)
got, err := dec.DecodeGraph()
if err != nil {
t.Logf("Decoding dump of DB failed: %v", err)
t.FailNow()
}
if !got.Eq(want) {
t.Log("Dump of graph not equal inserted graph")
t.Log("got:")
t.Log(got.Triples())
t.Log("want:")
t.Log(want.Triples())
t.FailNow()
}
print(".")
return true
}
if err := quick.Check(f, qconfig()); err != nil {
t.Error(err)
}
}
