la-baujmi: particle pi #2

3 files changed, 84 insertions, 0 deletions

diff --git a/la-baujmi/corpus.txt b/la-baujmi/corpus.txt
new file mode 100644
index 0000000..6d3ddda
--- /dev/null
+++ b/la-baujmi/corpus.txt
@@ -0,0 +1,5 @@
+sina li lukin e ni la sina pilin pona. sina li lukin e ni la pilin pona ni li suli suwi. sina li lukin e ni la sina li pona ale. sina li lukin e ni la sina li pali e ijo mute mute.
+
+sina li pona ale la sina pali ijo mute mute.
+
+sina li jan pi pali ijo mute mute. sina ken pali ijo mute mute.
diff --git a/la-baujmi/fact.go b/la-baujmi/fact.go
index 549c524..3caf53b 100644
--- a/la-baujmi/fact.go
+++ b/la-baujmi/fact.go
@@ -64,12 +64,69 @@ func SentenceToSelbris(s tokiponatokens.Sentence) ([]Selbri, error) {
 				continue
 			}
 
+			if len(pt.Parts) != 0 {
+				var sb strings.Builder
+				sb.WriteString("subject(")
+
+				for i, sp := range pt.Parts {
+					if i != 0 {
+						sb.WriteString(", ")
+					}
+					if sp.Sep != nil && *sp.Sep == "pi" {
+						sb.WriteString("pi(")
+						for j, tk := range sp.Tokens {
+							if j != 0 {
+								sb.WriteString(", ")
+							}
+
+							sb.WriteString(tk)
+						}
+						sb.WriteString(")")
+					} else {
+						sb.WriteString(strings.Join(sp.Tokens, "_"))
+					}
+				}
+
+				sb.WriteString(")")
+
+				subjects = append(objects, sb.String())
+				continue
+			}
+
 			subjects = append(subjects, strings.Join(pt.Tokens, "_"))
 
 		case tokiponatokens.PartVerbMarker:
 			verbs = append(verbs, strings.Join(pt.Tokens, "_"))
 
 		case tokiponatokens.PartObjectMarker:
+			if len(pt.Parts) != 0 {
+				var sb strings.Builder
+				sb.WriteString("object(")
+
+				for i, sp := range pt.Parts {
+					if i != 0 {
+						sb.WriteString(", ")
+					}
+					if sp.Sep != nil && *sp.Sep == "pi" {
+						sb.WriteString("pi(")
+						for j, tk := range sp.Tokens {
+							if j != 0 {
+								sb.WriteString(", ")
+							}
+
+							sb.WriteString(tk)
+						}
+						sb.WriteString(")")
+					} else {
+						sb.WriteString(strings.Join(sp.Tokens, "_"))
+					}
+				}
+
+				sb.WriteString(")")
+
+				objects = append(objects, sb.String())
+				continue
+			}
 			objects = append(objects, strings.Join(pt.Tokens, "_"))
 
 		case tokiponatokens.PartPunctuation:
diff --git a/la-baujmi/fact_test.go b/la-baujmi/fact_test.go
index 5199300..3e7c481 100644
--- a/la-baujmi/fact_test.go
+++ b/la-baujmi/fact_test.go
@@ -92,6 +92,28 @@ func TestSentenceToSelbris(t *testing.T) {
 			wantFacts: []string{"ala(tenpo_ni, A)."},
 		},
 		{
+			name: "pi_subject",
+			json: []byte(`[{"part":"subject","parts":[{"part":"subject","tokens":["ilo","mi"]},{"part":"subject","sep":"pi","tokens":["kasi","nasa"]}]},{"part":"verbMarker","sep":"li","tokens":["pona","ale"]}]`),
+			want: []Selbri{
+				{
+					Predicate: "pona_ale",
+					Arguments: []string{"subject(ilo_mi, pi(kasi, nasa))"},
+				},
+			},
+			wantFacts: []string{"pona_ale(subject(ilo_mi, pi(kasi, nasa)))."},
+		},
+		{
+			name: "pi_object",
+			json: []byte(`[{"part":"subject","tokens":["mi"]},{"part":"verbMarker","sep":"li","tokens":["esun"]},{"part":"objectMarker","sep":"e","parts":[{"part":"objectMarker","tokens":["ilo"]},{"part":"objectMarker","sep":"pi","tokens":["kalama","musi"]}]},{"part":"punctuation","tokens":["period"]}]`),
+			want: []Selbri{
+				{
+					Predicate: "esun",
+					Arguments: []string{"mi", "object(ilo, pi(kalama, musi))"},
+				},
+			},
+			wantFacts: []string{"esun(mi, object(ilo, pi(kalama, musi)))."},
+		},
+		{
 			name: "multiple verbs",
 			json: []byte(`[{"part":"subject","tokens":["ona"]},{"part":"verbMarker","sep":"li","tokens":["sona"]},{"part":"verbMarker","sep":"li","tokens":["pona"]},{"part":"objectMarker","sep":"e","tokens":["mute"]},{"part":"punctuation","tokens":["period"]}]`),
 			want: []Selbri{