feat(scanner): Calibre-aligned OPF metadata extraction

Adopt Calibre's reading conventions for the Dublin Core metadata that
parseOPFContent now pulls from the structured OPF parse:

- Titles: EPUB3 title-type selection (prefer 'main', join a distinct
  subtitle with ': ' exactly as Calibre stores it). There is no separate
  subtitle column by design - Calibre-sidecar books arrive pre-joined,
  so a column would stay empty for most libraries and force every client
  to reimplement concatenation.
- Genre: first dc:subject, mirroring the existing processGenresAndTags
  behavior of the Calibre-sidecar path; the embedded path never
  populated Genre before. Subjects stay one-element-one-tag - Library
  of Congress headings legitimately contain commas ("Holmes, Sherlock
  (Fictitious character) -- Fiction") and must not be split.
- Identifiers: urn:isbn:/urn:asin: prefixed values parse in addition to
  opf:scheme attributes, and the scheme-less fallback now requires an
  ISBN-shaped value (10/13 digits, optional separators/trailing X) so
  URIs like the Gutenberg identifiers cannot masquerade as ISBNs -
  observed live on 'A Study in Scarlet'.
- Series: EPUB3 belongs-to-collection with collection-type=series and
  group-position refines, ahead of the classic calibre:series metas.
- Audiobookshelf metadata.json sidecars join their subtitle field into
  the title the same way.

Tests cover title-type main+subtitle joining, belongs-to-collection
series with fractional group-position, urn:isbn extraction, genre/tag
parity, and comma preservation inside subject headings.

internal/services/media_scanner_opf.go

@@ -3,6 +3,7 @@ package services
 import (
 	"bytes"
 	"encoding/xml"
+	"strconv"
 	"strings"
 )
 
@@ -77,6 +78,112 @@ func parseOPFXML(content []byte) (*opfDocument, error) {
 	return &doc, nil
 }
 
+// refinesFor maps an element id to its EPUB3 refining metas
+// (those whose refines attribute starts with '#').
+func (d *opfDocument) refinesFor(id string) []opfMeta {
+	var out []opfMeta
+	if id == "" {
+		return out
+	}
+	for _, m := range d.Metadata.Metas {
+		if strings.HasPrefix(m.Refines, "#") && m.Refines[1:] == id {
+			out = append(out, m)
+		}
+	}
+	return out
+}
+
+// refinesProperty returns the value of the first refining meta carrying the
+// given property (e.g. "title-type", "collection-type", "group-position").
+func refinesProperty(metas []opfMeta, property string) (string, bool) {
+	for _, m := range metas {
+		if strings.EqualFold(m.Property, property) {
+			if v := strings.TrimSpace(m.Value); v != "" {
+				return v, true
+			}
+		}
+	}
+	return "", false
+}
+
+// selectTitle ports Calibre's read_title: prefer the dc:title refined as
+// title-type "main"; fall back to the first non-empty title. A distinct
+// subtitle (title-type containing "subtitle"/"sub-title") is joined onto the
+// main title with ": ", exactly as Calibre stores it.
+func (d *opfDocument) selectTitle() string {
+	var first, main, subtitle string
+	for _, t := range d.Metadata.Titles {
+		v := strings.TrimSpace(t.Value)
+		if v == "" {
+			continue
+		}
+		if first == "" {
+			first = v
+		}
+		tt, ok := refinesProperty(d.refinesFor(t.ID), "title-type")
+		if !ok {
+			continue
+		}
+		switch strings.ToLower(tt) {
+		case "main":
+			if main == "" {
+				main = v
+			}
+		default:
+			l := strings.ToLower(tt)
+			if strings.Contains(l, "subtitle") || strings.Contains(l, "sub-title") {
+				if subtitle == "" {
+					subtitle = v
+				}
+			}
+		}
+	}
+	title := main
+	if title == "" {
+		title = first
+	}
+	if subtitle != "" && subtitle != title {
+		title = title + ": " + subtitle
+	}
+	return title
+}
+
+// readSeries ports Calibre's read_series: EPUB3 belongs-to-collection (with a
+// collection-type=series refine and group-position index) first, then the
+// classic calibre:series / calibre:series_index metas.
+func (d *opfDocument) readSeries() (series string, index float64) {
+	for _, m := range d.Metadata.Metas {
+		if !strings.EqualFold(m.Property, "belongs-to-collection") {
+			continue
+		}
+		name := strings.TrimSpace(m.Value)
+		if name == "" {
+			continue
+		}
+		refines := d.refinesFor(m.ID)
+		if ct, ok := refinesProperty(refines, "collection-type"); !ok || !strings.EqualFold(ct, "series") {
+			continue
+		}
+		if gp, ok := refinesProperty(refines, "group-position"); ok {
+			if v, err := strconv.ParseFloat(strings.TrimSpace(gp), 64); err == nil {
+				index = v
+			}
+		}
+		return name, index
+	}
+	for _, m := range d.Metadata.Metas {
+		switch m.Name {
+		case "calibre:series":
+			series = m.Content
+		case "calibre:series_index":
+			if v, err := strconv.ParseFloat(strings.TrimSpace(m.Content), 64); err == nil {
+				index = v
+			}
+		}
+	}
+	return series, index
+}
+
 // itemByID returns manifest items with id, href and media-type, keyed by id.
 func (d *opfDocument) itemByID() map[string]opfItem {
 	m := make(map[string]opfItem, len(d.Manifest.Items))
@@ -166,6 +273,27 @@ func (d *opfDocument) coverPageHref() string {
 	return ""
 }
 
+// isISBNLike reports whether a bare identifier value is shaped like an ISBN
+// (digits, optional hyphens/spaces, optional trailing X; 10 or 13
+// significant characters). Guards the scheme-less dc:identifier fallback
+// against URLs and UUIDs sharing the same slot.
+func isISBNLike(v string) bool {
+	digits := 0
+	for i, r := range v {
+		switch {
+		case r >= '0' && r <= '9':
+			digits++
+		case r == '-' || r == ' ':
+			// separator
+		case (r == 'X' || r == 'x') && i == len(v)-1:
+			digits++ // ISBN-10 check character
+		default:
+			return false
+		}
+	}
+	return digits == 10 || digits == 13
+}
+
 // findImageReferenceInPage extracts the first raster image reference from a
 // cover (X)HTML page: <img src="..."> or SVG <image xlink:href="...">.
 // Token-based parsing keeps it tolerant of mixed namespaces and fragments.