feat(gedcom): Add Note type

Also adjust unused or erroneously-added fields from SourceRecord.

internal/gedcom/event.go

@@ -22,6 +22,7 @@ type Event struct {
 	DateRange       *date.Range
 	Place           string
 	SourceCitations []*SourceCitation
+	Notes           []*Note
 }
 
 func parseEvent(ctx context.Context, line *gedcom7.Line, subnodes []*gedcom.Node) (out *Event, err error) {
@@ -67,6 +68,12 @@ func parseEvent(ctx context.Context, line *gedcom7.Line, subnodes []*gedcom.Node
 				return nil, fmt.Errorf("error parsing source citation: %w", err)
 			}
 			out.SourceCitations = append(out.SourceCitations, citation)
+		case "NOTE":
+			note, err := parseNote(ctx, subline, subnode.GetSubnodes())
+			if err != nil {
+				return nil, fmt.Errorf("error parsing note: %w", err)
+			}
+			out.Notes = append(out.Notes, note)
 		default:
 			log.Warn(ctx, fields, "unsupported Tag")
 		}

internal/gedcom/family_record.go

@@ -19,6 +19,7 @@ type FamilyRecord struct {
 	DivorcedAt      *Event
 	AnnulledAt      *Event
 	SourceCitations []*SourceCitation
+	Notes           []*Note
 }
 
 func parseFamilyRecord(ctx context.Context, i int, line *gedcom7.Line, subnodes []*gedcom.Node) (out *FamilyRecord, err error) {
@@ -74,6 +75,12 @@ func parseFamilyRecord(ctx context.Context, i int, line *gedcom7.Line, subnodes
 				return nil, fmt.Errorf("error parsing source citation: %w", err)
 			}
 			out.SourceCitations = append(out.SourceCitations, citation)
+		case "NOTE":
+			note, err := parseNote(ctx, subline, subnode.GetSubnodes())
+			if err != nil {
+				return nil, fmt.Errorf("error parsing note: %w", err)
+			}
+			out.Notes = append(out.Notes, note)
 		default:
 			log.Warn(ctx, fields, "unsupported Tag")
 		}

internal/gedcom/gedcom_test.go

@@ -146,6 +146,9 @@ func TestReadRecordsFields(t *testing.T) {
 2 DATE 12 JUN 1995
 1 DEAT
 2 DATE 1 JAN 2000
+1 NOTE The year 2000 problem, also commonly known as the Y2K problem, Y2K scare, millennium bug, Y2K bug, Y2K glitch, Y2K error, or simply Y2K,
+2 CONT refers to potential computer errors related to the formatting and storage of calendar data for dates in and after the year 2000.
+2 LANG en
 1 FAMC @F1@
 0 @F1@ FAM
 1 HUSB @I1@
@@ -162,6 +165,8 @@ func TestReadRecordsFields(t *testing.T) {
 1 ANUL
 2 TYPE annulment
 2 DATE 2001
+1 NOTE Test that the parser can also read th
+2 CONC e tag, CONC.
 0 @S1@ SOUR
 1 _UID 046A3AD191FF4DD3B0693F406E0A7FB87012
 1 DATA
@@ -196,6 +201,7 @@ func TestReadRecordsFields(t *testing.T) {
 				MarriedAt:   &gedcom.Event{Date: mustParseDate(t, "1985-01-18")},
 				DivorcedAt:  &gedcom.Event{Date: mustParseDate(t, "2000-01-01")},
 				AnnulledAt:  &gedcom.Event{Date: mustParseDate(t, "2001-01-01")},
+				Notes:       []*gedcom.Note{{Payload: "Test that the parser can also read the tag, CONC."}},
 			},
 		}
 		if len(records.Families) != len(expected) {
@@ -212,6 +218,7 @@ func TestReadRecordsFields(t *testing.T) {
 
 			cmpStringSlices(t, errMsgPrefix+".ParentXrefs", got.ParentXrefs, exp.ParentXrefs)
 			cmpStringSlices(t, errMsgPrefix+".ChildXrefs", got.ChildXrefs, exp.ChildXrefs)
+			testNotes(t, errMsgPrefix+".Notes", got.Notes, exp.Notes)
 		}
 	})
 
@@ -240,8 +247,13 @@ func TestReadRecordsFields(t *testing.T) {
 				Names: []gedcom.PersonalName{
 					{Payload: "Mike /Foxtrot/", Given: "Mike", Nickname: "Millennium Bug", Surname: "Foxtrot"},
 				},
-				Birth:           &gedcom.Event{Date: mustParseDate(t, "1995-06-12")},
-				Death:           &gedcom.Event{Date: mustParseDate(t, "2000-01-01")},
+				Birth: &gedcom.Event{Date: mustParseDate(t, "1995-06-12")},
+				Death: &gedcom.Event{Date: mustParseDate(t, "2000-01-01")},
+				Notes: []*gedcom.Note{
+					{Payload: `The year 2000 problem, also commonly known as the Y2K problem, Y2K scare, millennium bug, Y2K bug, Y2K glitch, Y2K error, or simply Y2K,
+refers to potential computer errors related to the formatting and storage of calendar data for dates in and after the year 2000.`,
+						Lang: "en"},
+				},
 				FamiliesAsChild: []string{"@F1@"},
 			},
 		}
@@ -268,6 +280,7 @@ func TestReadRecordsFields(t *testing.T) {
 
 			cmpStringSlices(t, errMsgPrefix+".FamiliesAsPartner", got.FamiliesAsPartner, exp.FamiliesAsPartner)
 			cmpStringSlices(t, errMsgPrefix+".FamiliesAsChild", got.FamiliesAsChild, exp.FamiliesAsChild)
+			testNotes(t, errMsgPrefix+".Notes", got.Notes, exp.Notes)
 		}
 	})
 
@@ -276,7 +289,7 @@ func TestReadRecordsFields(t *testing.T) {
 			{
 				Xref:          "@S1@",
 				Title:         "New York Times, March 4, 1946, pp. 1,3.",
-				Note:          "Geneanet Community Trees Index",
+				Notes:         []*gedcom.Note{{Payload: "Geneanet Community Trees Index"}},
 				Text:          "yes",
 				RepositoryIDs: []string{"@R0@"},
 			},
@@ -315,11 +328,9 @@ func TestReadRecordsFields(t *testing.T) {
 			if got.Text != exp.Text {
 				t.Errorf("%s; wrong Text, got %q, exp %q", errMsgPrefix, got.Text, exp.Text)
 			}
-			if got.Note != exp.Note {
-				t.Errorf("%s; wrong Note, got %q, exp %q", errMsgPrefix, got.Note, exp.Note)
-			}
 
 			cmpStringSlices(t, errMsgPrefix+".RepositoryIDs", got.RepositoryIDs, exp.RepositoryIDs)
+			testNotes(t, errMsgPrefix+".Notes", got.Notes, exp.Notes)
 		}
 	})
 }
@@ -406,3 +417,30 @@ func cmpStringMaps(t *testing.T, errMsgPrefix string, actual, expected map[strin
 		}
 	}
 }
+
+func testNotes(t *testing.T, errMsgPrefix string, actual, expected []*gedcom.Note) {
+	if len(actual) != len(expected) {
+		t.Errorf("%s; wrong length; got %d, exp %d", errMsgPrefix, len(actual), len(expected))
+	} else {
+		for i, got := range actual {
+			errMsgPrefix := fmt.Sprintf("%s[%d]", errMsgPrefix, i)
+			exp := expected[i]
+
+			if got.Payload != exp.Payload {
+				t.Errorf("%s; wrong Payload, got %q, exp %q", errMsgPrefix, got.Payload, exp.Payload)
+			}
+			if got.Lang != exp.Lang {
+				t.Errorf("%s; wrong Lang, got %q, exp %q", errMsgPrefix, got.Lang, exp.Lang)
+			}
+
+			if len(got.SourceCitations) != len(exp.SourceCitations) {
+				t.Errorf("%s; wrong number of SourceCitations; got %d, exp %d", errMsgPrefix, len(got.SourceCitations), len(exp.SourceCitations))
+			} else {
+				for j, got := range got.SourceCitations {
+					errMsgPrefix := fmt.Sprintf("%s[%d], ", errMsgPrefix, j)
+					cmpSourceCitation(t, errMsgPrefix, got, exp.SourceCitations[j])
+				}
+			}
+		}
+	}
+}

internal/gedcom/individual_record.go

@@ -22,6 +22,7 @@ type IndividualRecord struct {
 	FamiliesAsChild   []string // Xref IDs of families where the person is a child.
 	FamiliesAsPartner []string // Xref IDs of families where the person is a partner, such as a spouse.
 	SourceCitations   []*SourceCitation
+	Notes             []*Note
 }
 
 func parseIndividualRecord(ctx context.Context, i int, line *gedcom7.Line, subnodes []*gedcom.Node) (out *IndividualRecord, err error) {
@@ -78,6 +79,12 @@ func parseIndividualRecord(ctx context.Context, i int, line *gedcom7.Line, subno
 				return nil, fmt.Errorf("error parsing source citation: %w", err)
 			}
 			out.SourceCitations = append(out.SourceCitations, citation)
+		case "NOTE":
+			note, err := parseNote(ctx, subline, subnode.GetSubnodes())
+			if err != nil {
+				return nil, fmt.Errorf("error parsing note: %w", err)
+			}
+			out.Notes = append(out.Notes, note)
 		default:
 			log.Warn(ctx, fields, "unsupported Tag")
 		}

internal/gedcom/note.go

@@ -0,0 +1,62 @@
+package gedcom
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/funwithbots/go-gedcom/pkg/gedcom"
+	"github.com/funwithbots/go-gedcom/pkg/gedcom7"
+	"github.com/rafaelespinoza/ged/internal/log"
+)
+
+// A Note is a catch-all location for info that doesn't really fit within other
+// defined structures. The Payload field may contain extra research notes,
+// context, or alternative interpretations of other data. Its URI is g7:NOTE.
+type Note struct {
+	Payload string
+	// Lang is the primary language for which the Note is written.
+	Lang            string
+	SourceCitations []*SourceCitation
+}
+
+func parseNote(ctx context.Context, line *gedcom7.Line, subnodes []*gedcom.Node) (out *Note, err error) {
+	out = &Note{Payload: line.Payload}
+
+	var subline *gedcom7.Line
+
+	for _, subnode := range subnodes {
+		if subline, err = parseLine(subnode); err != nil {
+			return
+		}
+
+		fields := map[string]any{
+			"func":    "parseEvent",
+			"line":    line.Text,
+			"subtag":  subline.Tag,
+			"subline": subline.Text,
+		}
+
+		log.Debug(ctx, fields, "")
+
+		switch subline.Tag {
+		case "CONC":
+			// This tag was deprecated in GEDCOM v7. The last GEDCOM version for
+			// which it was valid was v5.5.1. This tag would only appear in the
+			// data if the library parser is configured to allow deprecated
+			// tags. See func gedcom7.WithMaxDeprecatedTags.
+			out.Payload += subline.Payload
+		case "LANG":
+			out.Lang = subline.Payload
+		case "SOUR":
+			citation, err := parseSourceCitation(ctx, subline, subnode.GetSubnodes())
+			if err != nil {
+				return nil, fmt.Errorf("error parsing source citation: %w", err)
+			}
+			out.SourceCitations = append(out.SourceCitations, citation)
+		default:
+			log.Warn(ctx, fields, "unsupported Tag")
+		}
+	}
+
+	return
+}

internal/gedcom/personal_name.go

@@ -18,6 +18,7 @@ type PersonalName struct {
 	Payload         string
 	Type            enumset.NameType
 	SourceCitations []*SourceCitation
+	Notes           []*Note
 
 	// The following fields are PERSONAL_NAME_PIECES.
 	NamePrefix    string // URI is g7:NPFX
@@ -70,6 +71,12 @@ func parsePersonalName(ctx context.Context, line *gedcom7.Line, subnodes []*gedc
 				return nil, fmt.Errorf("error parsing source citation: %w", err)
 			}
 			out.SourceCitations = append(out.SourceCitations, citation)
+		case "NOTE":
+			note, err := parseNote(ctx, subline, subnode.GetSubnodes())
+			if err != nil {
+				return nil, fmt.Errorf("error parsing note: %w", err)
+			}
+			out.Notes = append(out.Notes, note)
 		default:
 			// There may be some metadata-related tags such as NAME-TYPE , or
 			// NOTE. For now, not parsing those, but might try to do so later.

internal/gedcom/source.go

@@ -18,7 +18,8 @@ type SourceCitation struct {
 	Page string
 	// Data is meant to represent extra info about a source. Its URI is G7:SOUR-DATA.
 	// This field is rather free-form, there is no payload.
-	Data map[string]string
+	Data  map[string]string
+	Notes []*Note
 }
 
 func parseSourceCitation(ctx context.Context, line *gedcom7.Line, subnodes []*gedcom.Node) (out *SourceCitation, err error) {
@@ -53,6 +54,12 @@ func parseSourceCitation(ctx context.Context, line *gedcom7.Line, subnodes []*ge
 				}
 				out.Data[dataSubline.Tag] = dataSubline.Payload
 			}
+		case "NOTE":
+			note, err := parseNote(ctx, subline, subnode.GetSubnodes())
+			if err != nil {
+				return nil, fmt.Errorf("error parsing note: %w", err)
+			}
+			out.Notes = append(out.Notes, note)
 		default:
 			log.Warn(ctx, fields, "unsupported Tag")
 		}

internal/gedcom/source_record.go

@@ -2,6 +2,7 @@ package gedcom
 
 import (
 	"context"
+	"fmt"
 
 	"github.com/funwithbots/go-gedcom/pkg/gedcom"
 	"github.com/funwithbots/go-gedcom/pkg/gedcom7"
@@ -17,8 +18,7 @@ type SourceRecord struct {
 	Publication   string
 	Text          string
 	RepositoryIDs []string
-	Note          string
-	OriginalLines []string
+	Notes         []*Note
 
 	// TODO: add other fields such as Data, MultimediaLink, ChangeDate, CreationDate, as needed.
 }
@@ -58,7 +58,11 @@ func parseSourceRecord(ctx context.Context, i int, line *gedcom7.Line, subnodes
 		case "REPO":
 			out.RepositoryIDs = append(out.RepositoryIDs, subline.Payload)
 		case "NOTE":
-			out.Note = subline.Payload
+			note, err := parseNote(ctx, subline, subnode.GetSubnodes())
+			if err != nil {
+				return nil, fmt.Errorf("error parsing note: %w", err)
+			}
+			out.Notes = append(out.Notes, note)
 		default:
 			log.Warn(ctx, fields, "unsupported Tag")
 		}