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
+}

internal/rasterize/rotate.go

@@ -0,0 +1,51 @@
+package rasterize
+
+import (
+	"bytes"
+	"fmt"
+
+	"github.com/pdfcpu/pdfcpu/pkg/api"
+	"github.com/pdfcpu/pdfcpu/pkg/pdfcpu/model"
+)
+
+// RotatePDFPages applies clockwise rotations (degrees, multiples of 90) to each page.
+// pageRotations is 1-indexed slice aligned with input pages: pageRotations[0] = page 1.
+func RotatePDFPages(in []byte, pageRotations []int) ([]byte, error) {
+	conf := model.NewDefaultConfiguration()
+	conf.ValidationMode = model.ValidationRelaxed
+
+	// pdfcpu's Rotate API takes a single rotation and a page selection. We group
+	// pages by rotation value to minimize passes.
+	groups := map[int][]int{}
+	for i, r := range pageRotations {
+		r = ((r % 360) + 360) % 360
+		if r == 0 {
+			continue
+		}
+		groups[r] = append(groups[r], i+1)
+	}
+	if len(groups) == 0 {
+		out := make([]byte, len(in))
+		copy(out, in)
+		return out, nil
+	}
+
+	current := bytes.NewReader(in)
+	var buf bytes.Buffer
+	for rot, pages := range groups {
+		buf.Reset()
+		selectedPages := make([]string, len(pages))
+		for i, p := range pages {
+			selectedPages[i] = fmt.Sprintf("%d", p)
+		}
+		if err := api.Rotate(current, &buf, rot, selectedPages, conf); err != nil {
+			return nil, fmt.Errorf("rotate %d°: %w", rot, err)
+		}
+		current = bytes.NewReader(append([]byte(nil), buf.Bytes()...))
+	}
+	out := make([]byte, current.Size())
+	if _, err := current.ReadAt(out, 0); err != nil {
+		return nil, err
+	}
+	return out, nil
+}