path: root/internal/bible/booktable.go

package bible

import (
	"embed"
	"fmt"
	"sort"
	"strings"

	"github.com/lukaszkasprzak/lectio/internal/ini"
)

// booksFS embeds the built-in default book table (names + abbreviations per
// dialect: en/pl/la). Users override it per book via ~/.config/lectio/books.ini.
//
//go:embed books.ini
var booksFS embed.FS

// BookInfo is one book's display data in one dialect, for `lectio --list` and
// (later) the reader book pickers.
type BookInfo struct {
	Canonical string // English canonical name (the Bible-corpus key), e.g. "John"
	Name      string // full display name in the dialect, e.g. "John" / "Jana"
	Shortcut  string // primary abbreviation in the dialect, e.g. "Jn" / "J"
}

// BookTable resolves book references and lists books per "dialect" -- a books.ini
// section, keyed by dialect ("en"/"pl"/"la"). It is built
// from the embedded default merged with an optional user override.
type BookTable struct {
	forms   map[string]map[string][]string // dialect -> canonical -> forms (first=shortcut, last=name)
	resolve map[string]map[string]string   // dialect -> lowercased form -> canonical
	sorted  map[string][]string            // dialect -> lowercased forms, longest first (book matching)
}

// LoadBookTable builds the table from the embedded default, merging userTOML
// over it per book (a book the user lists replaces that book's default forms;
// unlisted books keep the default). userTOML may be nil/empty (defaults only).
// It always returns a usable table; a non-nil error means userTOML was present
// but unparseable (the returned table is defaults-only, so callers can warn and
// proceed).
func LoadBookTable(userINI []byte) (*BookTable, error) {
	def, err := parseBooks(mustReadEmbedded())
	if err != nil {
		return nil, fmt.Errorf("embedded books.ini: %w", err) // our bug, not the user's
	}
	var uerr error
	if len(strings.TrimSpace(string(userINI))) > 0 {
		user, perr := parseBooks(userINI)
		if perr != nil {
			uerr = fmt.Errorf("ignoring malformed books.ini: %w", perr)
		} else {
			for dialect, books := range user {
				if def[dialect] == nil {
					def[dialect] = map[string][]string{}
				}
				for canon, forms := range books {
					def[dialect][canon] = forms
				}
			}
		}
	}
	t := &BookTable{
		forms:   def,
		resolve: map[string]map[string]string{},
		sorted:  map[string][]string{},
	}
	for dialect, books := range def {
		t.resolve[dialect] = map[string]string{}
		var forms []string
		for canon, fs := range books {
			for _, f := range fs {
				k := strings.ToLower(strings.TrimSpace(f))
				if k == "" {
					continue
				}
				t.resolve[dialect][k] = canon
				forms = append(forms, k)
			}
		}
		sort.SliceStable(forms, func(i, j int) bool { return len(forms[i]) > len(forms[j]) })
		t.sorted[dialect] = forms
	}
	return t, uerr
}

func mustReadEmbedded() []byte {
	b, _ := booksFS.ReadFile("books.ini") // embedded: always present
	return b
}

// DefaultBooksTOML returns the embedded default books.ini (the seed the
// /settings editor shows when the user has no override yet).
func DefaultBooksINI() []byte { return mustReadEmbedded() }

func parseBooks(data []byte) (map[string]map[string][]string, error) {
	secs, err := ini.Parse(data)
	if err != nil {
		return nil, err
	}
	out := map[string]map[string][]string{}
	for _, s := range secs {
		if s.Name == "" { // no top-level pairs in the book table
			continue
		}
		if out[s.Name] == nil {
			out[s.Name] = map[string][]string{}
		}
		for _, p := range s.Pairs {
			out[s.Name][p.Key] = ini.List(p.Val)
		}
	}
	return out, nil
}

// Books returns every book of the dialect in scriptural (canonOrder) order.
// A dialect with no entry for a canonical book simply omits it.
func (t *BookTable) Books(dialect string) []BookInfo {
	books := t.forms[dialect]
	out := make([]BookInfo, 0, len(canonOrder))
	for _, canon := range canonOrder {
		forms := books[canon]
		if len(forms) == 0 {
			continue
		}
		out = append(out, BookInfo{
			Canonical: canon,
			Shortcut:  forms[0],
			Name:      forms[len(forms)-1],
		})
	}
	return out
}

// ParseRef parses a user-typed reference in the dialect into the English,
// colon-style reference bible.Lookup expects ("John 5:15-17,20-22"), or
// ok=false if the book is not recognised in this dialect or there is no
// chapter:verse tail. Book resolution is dialect-scoped (idiomatic): only the
// dialect's own forms match. The number syntax is dialect-specific -- see
// normalizeSigla.
func (t *BookTable) ParseRef(dialect, input string) (string, bool) {
	low := strings.ToLower(strings.Join(strings.Fields(input), " "))
	if low == "" {
		return "", false
	}
	for _, form := range t.sorted[dialect] {
		if low == form {
			return "", false // book only, no chapter:verse
		}
		if strings.HasPrefix(low, form+" ") {
			tail := normalizeSigla(dialect, low[len(form)+1:])
			if tail == "" || !strings.Contains(tail, ":") {
				return "", false
			}
			return t.resolve[dialect][form] + " " + tail, true
		}
	}
	return "", false
}

// shortcut returns the dialect's primary abbreviation (the first form) for a
// canonical book, or "" when the dialect has no entry for it.
func (t *BookTable) shortcut(dialect, canonical string) string {
	if forms := t.forms[dialect][canonical]; len(forms) > 0 {
		return forms[0]
	}
	return ""
}

// Abbrev returns the dialect's primary abbreviation for a canonical book,
// falling back to the canonical name when the dialect has no entry.
func (t *BookTable) Abbrev(dialect, canonical string) string {
	if a := t.shortcut(dialect, canonical); a != "" {
		return a
	}
	return canonical
}

// FormatRef renders a canonical English reference ("John 20:1,11-18") in the
// dialect's sigla: the dialect's book shortcut plus its number style -- English
// "Jn 20:1,11-18" (colon chapter:verse, comma groups); Polish "J 20, 1. 11-18"
// (comma chapter, verse; period groups). It is the inverse of ParseRef's
// normalizeSigla. If canonicalRef can't be parsed or the book is not in the
// dialect, it is returned unchanged (a safe, still-readable fallback).
func (t *BookTable) FormatRef(dialect, canonicalRef string) string {
	m := refRe.FindStringSubmatch(strings.TrimSpace(canonicalRef))
	if m == nil {
		return canonicalRef
	}
	canon, chap, rest := m[1], m[2], m[3]
	abbrev := t.shortcut(dialect, canon)
	if abbrev == "" {
		return canonicalRef
	}
	// rest is a comma-separated group list; a group with ":" opens a new chapter.
	var b strings.Builder
	b.WriteString(abbrev + " ")
	for i, g := range strings.Split(rest, ",") {
		g = strings.TrimSpace(g)
		if g == "" {
			continue
		}
		newChap := ""
		if idx := strings.IndexByte(g, ':'); idx >= 0 {
			newChap, g = g[:idx], g[idx+1:]
		}
		switch {
		case i == 0: // first group uses the chapter from refRe
			b.WriteString(chapVerse(dialect, chap, g))
		case newChap != "" && newChap != chap: // cross-chapter group
			chap = newChap
			b.WriteString("; " + chapVerse(dialect, chap, g))
		case dialect == "pl": // another verse group in the same chapter
			b.WriteString(". " + g)
		default:
			b.WriteString("," + g)
		}
	}
	return b.String()
}

// chapVerse renders one "chapter + verses" in the dialect's style: English
// "5:15-17", Polish "5, 15-17".
func chapVerse(dialect, chap, verses string) string {
	if dialect == "pl" {
		return chap + ", " + verses
	}
	return chap + ":" + verses
}

// normalizeSigla rewrites a dialect's verse-reference tail into the colon/comma
// form bible.Lookup expects: "<chap>:<groups>", groups comma-separated, ranges
// with "-". Polish uses a comma (with or without a space) as the chapter/verse
// separator and a period between disjoint groups (and " i " for "and");
// English uses a colon for chapter/verse and a period or comma between groups.
// Reuses the shared regexes iRe/dashRe/verseLetterRe from convert.go.
func normalizeSigla(dialect, tail string) string {
	tail = strings.TrimSpace(tail)
	if dialect == "pl" {
		tail = strings.ReplaceAll(tail, ",", ":") // chapter,verse -> chapter:verse
		tail = iRe.ReplaceAllString(tail, ",")    // Polish 'and'
		tail = strings.ReplaceAll(tail, ".", ",") // disjoint groups
	} else {
		tail = strings.ReplaceAll(tail, ".", ",") // disjoint groups (comma already fine)
	}
	// A semicolon separates cross-chapter groups ("13:22-25; 15:10"); treat it
	// as a group separator so bible.SplitRef resolves each chapter.
	tail = strings.ReplaceAll(tail, ";", ",")
	tail = dashRe.ReplaceAllString(tail, "-")
	tail = verseLetterRe.ReplaceAllString(tail, "$1")
	tail = strings.ReplaceAll(tail, " ", "")
	return tail
}