Merge branch 'of-offline-migration': offline daily view, remove scrapers

4 files changed, 126 insertions, 137 deletions

diff --git a/internal/bible/convert.go b/internal/bible/convert.go
index e7d5944..6b9a663 100644
--- a/internal/bible/convert.go
+++ b/internal/bible/convert.go
@@ -1,128 +1,15 @@
 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",
-}
+import "regexp"
 
+// Shared citation-normalisation regexes used by the book table (normalizeSigla)
+// and the Ordinary Form reference renumbering (OFRef). The niedziela.pl Polish
+// citation converter that once lived here (ToEnglishRef) was removed with the
+// scraper: readings are now authored in English-canonical form.
 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`)
 	bookDotRe     = regexp.MustCompile(`(\p{L})\.`) // abbreviation dot after a letter ("Cor." -> "Cor")
-	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, "$1") // drop verse-part letters (15a -> 15)
-	rest = strings.ReplaceAll(rest, " ", "")
-
-	return strings.TrimSpace(fmt.Sprintf("%s %s", polishToEn[book], rest)), nil
-}
diff --git a/internal/bible/convert_test.go b/internal/bible/convert_test.go
deleted file mode 100644
index 6f16c89..0000000
--- a/internal/bible/convert_test.go
+++ /dev/null
@@ -1,19 +0,0 @@
-package bible
-
-import "testing"
-
-func TestToEnglishRef(t *testing.T) {
-	cases := []struct{ in, system, want string }{
-		{"Mt 7, 1-5", "vulgate", "Mat 7:1-5"},
-		{"J 20, 1. 11-18", "vulgate", "John 20:1,11-18"},
-		{"por. J 20, 11", "vulgate", "John 20:11"},
-		{"Ps 63 (62), 2. 3-4. 5-6. 8-9 (R.: por. 2ab)", "vulgate", "Psalms 62:2,3-4,5-6,8-9"},
-		{"Ps 63 (62), 2. 3-4. 5-6. 8-9 (R.: por. 2ab)", "drb", "Psalms 63:1,2-3,4-5,7-8"},
-	}
-	for _, c := range cases {
-		got, err := ToEnglishRef(c.in, c.system)
-		if err != nil || got != c.want {
-			t.Errorf("ToEnglishRef(%q,%q)=%q,%v want %q", c.in, c.system, got, err, c.want)
-		}
-	}
-}
diff --git a/internal/bible/ofref.go b/internal/bible/ofref.go
new file mode 100644
index 0000000..8a72342
--- /dev/null
+++ b/internal/bible/ofref.go
@@ -0,0 +1,71 @@
+package bible
+
+import (
+	"regexp"
+	"strconv"
+	"strings"
+
+	"github.com/lukaszkasprzak/lectio/internal/psalter"
+)
+
+var (
+	// ofCiteRe splits an English-canonical citation into book, chapter and the
+	// verse tail: "2 Samuel 6:12b-15,17-19" -> ("2 Samuel", "6", "12b-15,17-19").
+	// The book is non-greedy so a leading ordinal ("2 Samuel") stays with it.
+	ofCiteRe = regexp.MustCompile(`^(.*?)\s+(\d+):(.+)$`)
+	// 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..4b9cb1e
--- /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)
+			}
+		})
+	}
+}