Scanyonero/runner.go

package main

import (
	"log"
	"os"
	"os/exec"
	"path/filepath"
	"time"
)

const OCRMyPDFExecutable = "ocrmypdf"

type Runner struct {
	InputPatterns []string // The "Glob" patterns used for searching input files.
	OutputPath    string   // The output directory passed to OCRmyPDF.

	Interval time.Duration
}

func (r Runner) Run() {
	if r.Interval == 0 {
		r.Interval = 5 * time.Second
	}

	go func() {
		ticker := time.NewTicker(r.Interval)
		for range ticker.C {

			for _, inputPattern := range r.InputPatterns {
				filenames, err := filepath.Glob(inputPattern)
				if err != nil {
					log.Panicf("Failed to get input files: %v", err)
				}

				for _, filename := range filenames {

					outputPath := filepath.Join(r.OutputPath, filepath.Base(filename))

					args := []string{
						"-l deu+eng",
						"--rotate-pages",
						"--optimize 1",
					}
					args = append(args, filename, outputPath)
					cmd := exec.Command(OCRMyPDFExecutable, args...)

					if err := cmd.Run(); err != nil {
						log.Panicf("Failed to run OCRmyPDF: %v", err)
					}

					// Only delete the PDF if the previous steps did succeed.
					os.Remove(filename)
				}
			}

		}
	}()
}