Initial orientator

internal/rasterize/pdf.go

@@ -0,0 +1,51 @@
+// Package rasterize converts PDF pages to in-memory images using MuPDF.
+package rasterize
+
+import (
+	"fmt"
+	"image"
+
+	"github.com/gen2brain/go-fitz"
+)
+
+// DPI used when rasterizing pages for the orientation classifier.
+// 150 mirrors the Ghostscript -r150 flag used in the original Python pipeline.
+const DPI = 150
+
+type Page struct {
+	Index int
+	Img   image.Image
+}
+
+// PDF wraps an open MuPDF document.
+type PDF struct {
+	doc *fitz.Document
+}
+
+func Open(data []byte) (*PDF, error) {
+	doc, err := fitz.NewFromMemory(data)
+	if err != nil {
+		return nil, fmt.Errorf("fitz open: %w", err)
+	}
+	return &PDF{doc: doc}, nil
+}
+
+func (p *PDF) Close() error {
+	if p == nil || p.doc == nil {
+		return nil
+	}
+	return p.doc.Close()
+}
+
+func (p *PDF) NumPages() int {
+	return p.doc.NumPage()
+}
+
+// RenderPage rasterizes a single page at the configured DPI.
+func (p *PDF) RenderPage(idx int) (image.Image, error) {
+	img, err := p.doc.ImageDPI(idx, DPI)
+	if err != nil {
+		return nil, fmt.Errorf("render page %d: %w", idx, err)
+	}
+	return img, nil
+}