1 files changed, 90 insertions, 0 deletions
diff --git a/pdf.go b/pdf.go
new file mode 100644
index 0000000..e7480fe
--- /dev/null
+++ b/pdf.go
@@ -0,0 +1,90 @@
+package bookpipeline
+
+import (
+	"errors"
+	"fmt"
+	"html"
+	"image"
+	_ "image/jpeg"
+	_ "image/png"
+	"io/ioutil"
+	"os"
+
+	//"rescribe.xyz/gofpdf@addtextrenderingmode"
+	"rescribe.xyz/gofpdf"
+	"rescribe.xyz/utils/pkg/hocr"
+)
+
+const pageWidth = 5 // pageWidth in inches
+
+// pxToPt converts a pixel value into a pt value (72 pts per inch)
+// This uses pageWidth to determine the appropriate value
+func pxToPt(i int) float64 {
+	return float64(i) / pageWidth
+}
+
+type Fpdf struct {
+	fpdf *gofpdf.Fpdf
+}
+
+// Setup creates a new PDF with appropriate settings and fonts
+// TODO: find a font that's closer to the average dimensions of the
+//       text we're dealing with
+// TODO: once we have a good font, embed it in the binary as bytes
+func (p *Fpdf) Setup() error {
+	p.fpdf = gofpdf.New("P", "pt", "A4", "")
+	// Even though it's invisible, we need to add a font which can do
+	// UTF-8 so that text renders correctly.
+	p.fpdf.AddUTF8Font("dejavu", "", "DejaVuSansCondensed.ttf")
+	p.fpdf.SetFont("dejavu", "", 10)
+	p.fpdf.SetAutoPageBreak(false, float64(0))
+	return p.fpdf.Error()
+}
+
+// AddPage adds a page to the pdf with an image and (invisible)
+// text from an hocr file
+func (p *Fpdf) AddPage(imgpath, hocrpath string) error {
+	file, err := ioutil.ReadFile(hocrpath)
+	if err != nil {
+		return errors.New(fmt.Sprintf("Could not read file %s: %v", hocrpath, err))
+	}
+	// TODO: change hocr.Parse to take a Reader rather than []byte
+	h, err := hocr.Parse(file)
+	if err != nil {
+		return errors.New(fmt.Sprintf("Could not parse hocr in file %s: %v", hocrpath, err))
+	}
+
+	f, err := os.Open(imgpath)
+	defer f.Close()
+	if err != nil {
+		return errors.New(fmt.Sprintf("Could not open file %s: %v", imgpath, err))
+	}
+	img, _, err := image.Decode(f)
+	if err != nil {
+		return errors.New(fmt.Sprintf("Could not decode image: %v", err))
+	}
+	b := img.Bounds()
+	p.fpdf.AddPageFormat("P", gofpdf.SizeType{Wd: pxToPt(b.Dx()), Ht: pxToPt(b.Dy())})
+
+	// TODO: check for errors in pdf as going through
+
+	_ = p.fpdf.RegisterImageOptions(imgpath, gofpdf.ImageOptions{})
+	p.fpdf.ImageOptions(imgpath, 0, 0, pxToPt(b.Dx()), pxToPt(b.Dy()), false, gofpdf.ImageOptions{}, 0, "")
+
+	p.fpdf.SetTextRenderingMode(gofpdf.TextRenderingModeInvisible)
+
+	for _, l := range h.Lines {
+		coords, err := hocr.BoxCoords(l.Title)
+		if err != nil {
+			continue
+		}
+		p.fpdf.SetXY(pxToPt(coords[0]), pxToPt(coords[1]))
+		p.fpdf.CellFormat(pxToPt(coords[2]), pxToPt(coords[3]), html.UnescapeString(hocr.LineText(l)), "", 0, "T", false, 0, "")
+	}
+	return p.fpdf.Error()
+}
+
+// Save saves the PDF to the file at path
+func (p *Fpdf) Save(path string) error {
+	return p.fpdf.OutputFileAndClose(path)
+}