bible: Polish citation -> English ref (with psalm versification)

2 files changed, 146 insertions, 0 deletions

diff --git a/internal/bible/convert.go b/internal/bible/convert.go
new file mode 100644
index 0000000..6181dbf
--- /dev/null
+++ b/internal/bible/convert.go
@@ -0,0 +1,127 @@
+package bible
+
+import (
+	"fmt"
+	"regexp"
+	"sort"
+	"strconv"
+	"strings"
+
+	"github.com/lukaszkasprzak/lectio/internal/psalter"
+)
+
+// polishToEn maps Polish Biblical abbreviations to the book name accepted by
+// the kjv-family tools. Gospels use the short forms (Mat/Mark/Luke/John);
+// other books use full names that the tools resolve by unique prefix.
+// Ported verbatim from ewangelia.py POLISH_TO_EN.
+var polishToEn = map[string]string{
+	// Old Testament
+	"Rdz": "Genesis", "Wj": "Exodus", "Kpł": "Leviticus", "Lb": "Numbers",
+	"Pwt": "Deuteronomy", "Joz": "Joshua", "Sdz": "Judges", "Rt": "Ruth",
+	"1 Sm": "1 Samuel", "2 Sm": "2 Samuel", "1 Krl": "1 Kings", "2 Krl": "2 Kings",
+	"1 Krn": "1 Chronicles", "2 Krn": "2 Chronicles", "Ezd": "Ezra",
+	"Ne": "Nehemiah", "Tb": "Tobit", "Jdt": "Judith", "Est": "Esther",
+	"1 Mch": "1 Maccabees", "2 Mch": "2 Maccabees", "Hi": "Job", "Ps": "Psalms",
+	"Prz": "Proverbs", "Koh": "Ecclesiastes", "Pnp": "Song of Solomon",
+	"Mdr": "Wisdom", "Syr": "Sirach", "Iz": "Isaiah", "Jr": "Jeremiah",
+	"Lm": "Lamentations", "Ba": "Baruch", "Ez": "Ezekiel", "Dn": "Daniel",
+	"Oz": "Hosea", "Jl": "Joel", "Am": "Amos", "Ab": "Obadiah", "Jon": "Jonah",
+	"Mi": "Micah", "Na": "Nahum", "Ha": "Habakkuk", "So": "Zephaniah",
+	"Ag": "Haggai", "Za": "Zechariah", "Ml": "Malachi",
+	// New Testament
+	"Mt": "Mat", "Mk": "Mark", "Łk": "Luke", "J": "John", "Dz": "Acts",
+	"Rz": "Romans", "1 Kor": "1 Corinthians", "2 Kor": "2 Corinthians",
+	"Ga": "Galatians", "Ef": "Ephesians", "Flp": "Philippians",
+	"Kol": "Colossians", "1 Tes": "1 Thessalonians", "2 Tes": "2 Thessalonians",
+	"1 Tm": "1 Timothy", "2 Tm": "2 Timothy", "Tt": "Titus", "Flm": "Philemon",
+	"Hbr": "Hebrews", "Jk": "James", "1 P": "1 Peter", "2 P": "2 Peter",
+	"1 J": "1 John", "2 J": "2 John", "3 J": "3 John", "Jud": "Jude",
+	"Ap": "Revelation",
+}
+
+var (
+	porRe         = regexp.MustCompile(`(?i)^\s*por\.\s*`)
+	refrainRe     = regexp.MustCompile(`\s*\(R\.:.*$`)
+	wsRe          = regexp.MustCompile(`\s+`)
+	iRe           = regexp.MustCompile(`\s+i\s+`)
+	dashRe        = regexp.MustCompile(`\s*[-–—]\s*`)
+	verseLetterRe = regexp.MustCompile(`(\d)[a-c]\b`)
+	psalmRestRe   = regexp.MustCompile(`(\d+)(?:\s*\((\d+)\))?(.*)$`)
+	digitsRe      = regexp.MustCompile(`\d+`)
+)
+
+// polishBookKeys holds the POLISH_TO_EN keys sorted longest-first, matching
+// the Python `sorted(POLISH_TO_EN, key=len, reverse=True)` traversal order.
+var polishBookKeys = sortedPolishBookKeys()
+
+func sortedPolishBookKeys() []string {
+	keys := make([]string, 0, len(polishToEn))
+	for k := range polishToEn {
+		keys = append(keys, k)
+	}
+	sort.SliceStable(keys, func(i, j int) bool { return len(keys[i]) > len(keys[j]) })
+	return keys
+}
+
+// psalmRef maps the psalm citation body ("63 (62), 2. 3-4. ...") to the target Psalter.
+func psalmRef(rest, system string) string {
+	m := psalmRestRe.FindStringSubmatch(rest)
+	if m == nil {
+		return rest
+	}
+	heb, _ := strconv.Atoi(m[1])
+	tail := m[3]
+	if system == "vulgate" {
+		ch := m[1]
+		if m[2] != "" {
+			ch = m[2]
+		}
+		return ch + tail
+	}
+	if system == "drb" {
+		tail = digitsRe.ReplaceAllStringFunc(tail, func(s string) string {
+			n, _ := strconv.Atoi(s)
+			return strconv.Itoa(psalter.DrbVerse(heb, n))
+		})
+	}
+	return strconv.Itoa(heb) + tail
+}
+
+// ToEnglishRef converts a Polish citation to kjv-tool style: "Mt 7, 1-5" -> "Mat 7:1-5".
+//
+// Handles the extra apparatus of non-gospel readings: a leading "por." (compare)
+// marker, a trailing "(R.: ...)" responsorial refrain, and psalm versification.
+// system selects the target Psalter: "vulgate" (vul/grb/wuj) uses the citation's
+// Vulgate chapter (62 of "63 (62)"); "drb" uses the Hebrew chapter (63) with the
+// DRB title-fold verse shift; other values use the Hebrew chapter unshifted.
+func ToEnglishRef(plRef, system string) (string, error) {
+	plRef = porRe.ReplaceAllString(plRef, "")     // 'compare' marker
+	plRef = refrainRe.ReplaceAllString(plRef, "") // responsorial refrain
+	plRef = wsRe.ReplaceAllString(strings.TrimSpace(plRef), " ")
+
+	var book string
+	found := false
+	for _, k := range polishBookKeys {
+		if plRef == k || strings.HasPrefix(plRef, k+" ") {
+			book = k
+			found = true
+			break
+		}
+	}
+	if !found {
+		return "", fmt.Errorf("unknown book in reference: %q", plRef)
+	}
+
+	rest := strings.TrimSpace(plRef[len(book):])
+	if polishToEn[book] == "Psalms" {
+		rest = psalmRef(rest, system)
+	}
+	rest = strings.ReplaceAll(rest, ", ", ":")        // chapter/verse separator
+	rest = strings.ReplaceAll(rest, ". ", ",")        // disjoint verse groups
+	rest = iRe.ReplaceAllString(rest, ",")            // Polish 'and'
+	rest = dashRe.ReplaceAllString(rest, "-")         // normalise ranges
+	rest = verseLetterRe.ReplaceAllString(rest, "$1") // drop verse-part letters (15a -> 15)
+	rest = strings.ReplaceAll(rest, " ", "")
+
+	return strings.TrimSpace(fmt.Sprintf("%s %s", polishToEn[book], rest)), nil
+}
diff --git a/internal/bible/convert_test.go b/internal/bible/convert_test.go
new file mode 100644
index 0000000..6f16c89
--- /dev/null
+++ b/internal/bible/convert_test.go
@@ -0,0 +1,19 @@
+package bible
+
+import "testing"
+
+func TestToEnglishRef(t *testing.T) {
+	cases := []struct{ in, system, want string }{
+		{"Mt 7, 1-5", "vulgate", "Mat 7:1-5"},
+		{"J 20, 1. 11-18", "vulgate", "John 20:1,11-18"},
+		{"por. J 20, 11", "vulgate", "John 20:11"},
+		{"Ps 63 (62), 2. 3-4. 5-6. 8-9 (R.: por. 2ab)", "vulgate", "Psalms 62:2,3-4,5-6,8-9"},
+		{"Ps 63 (62), 2. 3-4. 5-6. 8-9 (R.: por. 2ab)", "drb", "Psalms 63:1,2-3,4-5,7-8"},
+	}
+	for _, c := range cases {
+		got, err := ToEnglishRef(c.in, c.system)
+		if err != nil || got != c.want {
+			t.Errorf("ToEnglishRef(%q,%q)=%q,%v want %q", c.in, c.system, got, err, c.want)
+		}
+	}
+}