feat(scanner): fill sparse sidecars from the PDF Info dictionary

Same gap-fill rule as the EPUB path, against the PDF's embedded Info
dictionary via a readPDFInfoDict helper reusing extractPDFMetadata's
field conventions (creator falls back to author, subject maps to
description, producer to publisher, keywords to tags, plus page count).
Sidecar values always win; unopenable PDFs skip silently.

TestMergeMetadataPDFGapFill uses an Info-bearing hand-built PDF fixture
(shared with the sidecar-cover test) and asserts the sidecar title is
kept while author/description/publisher/tags/page count fill in.

internal/services/media_scanner.go

@@ -1221,6 +1221,31 @@ func (s *MediaScanner) mergeMetadata(path string, calibreMetadata *MediaMetadata
 		}
 	}
 
+	// For PDFs: fill blanks from the embedded Info dictionary. Same rule as
+	// EPUBs - sidecar values win, the file only fills gaps.
+	if ext == ".pdf" {
+		if embedded, err := s.readPDFInfoDict(path); err == nil && embedded != nil {
+			if metadata.Title == "" && embedded.Title != "" {
+				metadata.Title = embedded.Title
+			}
+			if metadata.Author == "" && embedded.Author != "" {
+				metadata.Author = embedded.Author
+			}
+			if metadata.Description == "" && embedded.Description != "" {
+				metadata.Description = embedded.Description
+			}
+			if metadata.Publisher == "" && embedded.Publisher != "" {
+				metadata.Publisher = embedded.Publisher
+			}
+			if len(metadata.Tags) == 0 && len(embedded.Tags) > 0 {
+				metadata.Tags = embedded.Tags
+			}
+			if metadata.PageCount == 0 && embedded.PageCount > 0 {
+				metadata.PageCount = embedded.PageCount
+			}
+		}
+	}
+
 	// For comic archives, try to extract ComicInfo.xml
 	if ext == ".cbz" || ext == ".cbr" || ext == ".cb7" || ext == ".cbt" {
 		comicInfo, cover, err := extractComicMetadata(path)
@@ -2268,6 +2293,51 @@ func findSidecarCover(mediaPath string) string {
 	return ""
 }
 
+// readPDFInfoDict reads a PDF's embedded Info dictionary into MediaMetadata,
+// using the same field conventions as extractPDFMetadata (creator falls back
+// to author, subject maps to description, producer to publisher, keywords to
+// tags). No cover or filename logic - purely the document's own metadata, for
+// filling sidecar gaps in mergeMetadata.
+func (s *MediaScanner) readPDFInfoDict(path string) (*MediaMetadata, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return nil, err
+	}
+	defer func() {
+		if err := f.Close(); err != nil {
+			fmt.Printf("Warning: failed to close PDF file %s: %v\n", path, err)
+		}
+	}()
+
+	pdfInfo, err := pdfcpuapi.PDFInfo(f, filepath.Base(path), nil, false, nil)
+	if err != nil {
+		return nil, err
+	}
+
+	metadata := &MediaMetadata{}
+	if pdfInfo.Title != "" {
+		metadata.Title = pdfInfo.Title
+	}
+	if pdfInfo.Author != "" {
+		metadata.Author = pdfInfo.Author
+	} else if pdfInfo.Creator != "" {
+		metadata.Author = pdfInfo.Creator
+	}
+	if pdfInfo.Subject != "" {
+		metadata.Description = pdfInfo.Subject
+	}
+	if pdfInfo.Producer != "" {
+		metadata.Publisher = pdfInfo.Producer
+	}
+	if len(pdfInfo.Keywords) > 0 {
+		metadata.Tags = utils.NormalizeTags(pdfInfo.Keywords)
+	}
+	if pdfInfo.PageCount > 0 {
+		metadata.PageCount = int32(pdfInfo.PageCount)
+	}
+	return metadata, nil
+}
+
 func (s *MediaScanner) extractPDFMetadata(path string) (*MediaMetadata, error) {
 	metadata := &MediaMetadata{}