initial commit

pkg/tesseract/tesseract.go

@@ -0,0 +1,60 @@
+package tesseract
+
+import (
+	"git.giftfish.de/ston1th/docstore/pkg/core"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"time"
+)
+
+type Tesseract struct {
+	cmd     string
+	args    []string
+	dir     string
+	timeout time.Duration
+}
+
+func NewTesseract(dir string, langs []string, timeout time.Duration) (t *Tesseract, err error) {
+	var (
+		cmd  string
+		lang string
+	)
+	cmd, err = exec.LookPath("tesseract")
+	if err != nil {
+		return
+	}
+	for i, l := range langs {
+		lang += l
+		if i < len(langs) {
+			lang += "+"
+		}
+	}
+	args := []string{"-l", lang, "--oem", "2"}
+	t = &Tesseract{cmd, args, dir, timeout}
+	return
+}
+
+func (t *Tesseract) Scan(file string) (filename, text string, err error) {
+	filename = core.FileName() + ".pdf"
+	txt, err := core.Timeout(exec.Command(t.cmd, append(t.args, []string{file, "stdout", "txt"}...)...), t.timeout)
+	if err != nil {
+		return
+	}
+	text = string(txt)
+	pdf, err := core.Timeout(exec.Command(t.cmd, append(t.args, []string{file, "stdout", "pdf"}...)...), t.timeout)
+	if err != nil {
+		return
+	}
+	f, err := os.OpenFile(filepath.Join(t.dir, filename), os.O_RDWR|os.O_CREATE, 0640)
+	if err != nil {
+		return
+	}
+	defer f.Close()
+	_, err = f.Write(pdf)
+	if err != nil {
+		return
+	}
+	err = os.Remove(file)
+	return
+}