2 files changed, 121 insertions, 0 deletions
diff --git a/internal/bible/ofref.go b/internal/bible/ofref.go
new file mode 100644
index 0000000..8a72342
--- /dev/null
+++ b/internal/bible/ofref.go
@@ -0,0 +1,71 @@
+package bible
+
+import (
+	"regexp"
+	"strconv"
+	"strings"
+
+	"github.com/lukaszkasprzak/lectio/internal/psalter"
+)
+
+var (
+	// ofCiteRe splits an English-canonical citation into book, chapter and the
+	// verse tail: "2 Samuel 6:12b-15,17-19" -> ("2 Samuel", "6", "12b-15,17-19").
+	// The book is non-greedy so a leading ordinal ("2 Samuel") stays with it.
+	ofCiteRe = regexp.MustCompile(`^(.*?)\s+(\d+):(.+)$`)
+	// ofVerseLetterRe drops sub-verse part letters so lookups resolve at whole
+	// verses: "12b" -> "12", "3ab" -> "3", "3cd" -> "3".
+	ofVerseLetterRe = regexp.MustCompile(`(\d)[a-z]+`)
+)
+
+// normalizeOFVerses rewrites a citation's verse tail into the shape
+// bible.Lookup parses: cross-chapter semicolons become commas, ranges use a
+// plain hyphen, sub-verse letters are dropped, and spaces are removed.
+func normalizeOFVerses(tail string) string {
+	tail = strings.ReplaceAll(tail, ";", ",")
+	tail = dashRe.ReplaceAllString(tail, "-")
+	tail = ofVerseLetterRe.ReplaceAllString(tail, "$1")
+	tail = strings.ReplaceAll(tail, " ", "")
+	return tail
+}
+
+// OFRef turns an English-canonical Ordinary Form citation (lectio's authored
+// form) into a bible.Lookup-ready reference for the target corpus's Psalter.
+//
+// Every citation's verse tail is normalized (see normalizeOFVerses). Only the
+// Psalms differ between Psalters, so their chapter/verse is additionally
+// renumbered:
+//
+//   - "vulgate" (vul/grb/wuj): shift the Hebrew chapter to its Vulgate
+//     equivalent (Ps 27 -> 26, Ps 147 -> 146). Verse numbers are kept, so the
+//     rare intra-psalm verse-boundary shifts are not corrected here.
+//   - "drb": the Douay-Rheims corpus keeps modern (Hebrew) chapter numbers, so
+//     the chapter is kept; only the DRB title-fold verse shift applies (see
+//     psalter.DrbVerse).
+//   - "hebrew" / anything else: already in the corpus's numbering.
+//
+// A citation OFRef cannot parse (no "chapter:verse") is returned unchanged for
+// bible.Lookup to accept or reject.
+func OFRef(citation, system string) string {
+	m := ofCiteRe.FindStringSubmatch(citation)
+	if m == nil {
+		return citation
+	}
+	book, chap, verses := m[1], m[2], normalizeOFVerses(m[3])
+
+	if book == "Psalms" {
+		heb, err := strconv.Atoi(chap)
+		if err == nil {
+			switch system {
+			case "vulgate":
+				chap = strconv.Itoa(psalter.HebrewToVulgateChapter(heb))
+			case "drb":
+				verses = digitsRe.ReplaceAllStringFunc(verses, func(s string) string {
+					n, _ := strconv.Atoi(s)
+					return strconv.Itoa(psalter.DrbVerse(heb, n))
+				})
+			}
+		}
+	}
+	return book + " " + chap + ":" + verses
+}
diff --git a/internal/bible/ofref_test.go b/internal/bible/ofref_test.go
new file mode 100644
index 0000000..fae593a
--- /dev/null
+++ b/internal/bible/ofref_test.go
@@ -0,0 +1,50 @@
+package bible
+
+import "testing"
+
+func TestOFRef(t *testing.T) {
+	cases := []struct {
+		name     string
+		citation string
+		system   string
+		want     string
+	}{
+		// Non-psalms pass through unchanged in every system.
+		{"gospel-vulgate", "Matthew 4:12-23", "vulgate", "Matthew 4:12-23"},
+		{"gospel-drb", "Matthew 4:12-23", "drb", "Matthew 4:12-23"},
+		{"epistle-hebrew", "1 Corinthians 1:10-13,17", "hebrew", "1 Corinthians 1:10-13,17"},
+
+		// Psalm chapter shift for the Vulgate family (Hebrew -> Vulgate).
+		{"ps27-vulgate", "Psalms 27:1,4,13-14", "vulgate", "Psalms 26:1,4,13-14"},
+		{"ps27-drb", "Psalms 27:1,4,13-14", "drb", "Psalms 27:1,4,13-14"}, // Ps 27 has no DRB title fold
+		{"ps27-hebrew", "Psalms 27:1,4,13-14", "hebrew", "Psalms 27:1,4,13-14"},
+
+		// Boundary cases of HebrewToVulgateChapter.
+		{"ps8-vulgate", "Psalms 8:2", "vulgate", "Psalms 8:2"},     // <=8 unchanged
+		{"ps9-vulgate", "Psalms 9:2", "vulgate", "Psalms 9:2"},     // 9,10 -> 9
+		{"ps10-vulgate", "Psalms 10:1", "vulgate", "Psalms 9:1"},   // 10 -> 9
+		{"ps11-vulgate", "Psalms 11:1", "vulgate", "Psalms 10:1"},  // 11..113 -> h-1
+		{"ps116-vulgate", "Psalms 116:12", "vulgate", "Psalms 114:12"},
+		{"ps147-vulgate", "Psalms 147:12", "vulgate", "Psalms 146:12"},
+		{"ps148-vulgate", "Psalms 148:1", "vulgate", "Psalms 148:1"}, // >=148 unchanged
+
+		// DRB title-fold verse shift: Ps 51 folds 2 title verses into verse 1.
+		{"ps51-drb", "Psalms 51:3-4,12-13", "drb", "Psalms 51:1-2,10-11"},
+		{"ps51-vulgate", "Psalms 51:3-4", "vulgate", "Psalms 50:3-4"},
+		// Ps 63 folds 1 title verse.
+		{"ps63-drb", "Psalms 63:2,3-4", "drb", "Psalms 63:1,2-3"},
+
+		// Tail normalization: sub-verse letters dropped, ";" -> ",".
+		{"verseletter-2sam", "2 Samuel 6:12b-15,17-19", "vulgate", "2 Samuel 6:12-15,17-19"},
+		{"semicolon-acts", "The Acts 11:21b-26;13:1-3", "drb", "The Acts 11:21-26,13:1-3"},
+		{"multiletter-psalm-drb", "Psalms 98:1,2-3ab,3cd-4,5-6", "drb", "Psalms 98:1,2-3,3-4,5-6"},
+		{"multiletter-psalm-vulgate", "Psalms 98:1,2-3ab,3cd-4,5-6", "vulgate", "Psalms 97:1,2-3,3-4,5-6"},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			if got := OFRef(c.citation, c.system); got != c.want {
+				t.Errorf("OFRef(%q, %q) = %q, want %q", c.citation, c.system, got, c.want)
+			}
+		})
+	}
+}