Files
docconv/image_ocr.go
T
Hamish c4a9e3d578 ocr: update gosseract to v2
This also allows multiple gosseract languages to be provided.

Co-authored-by: Jonathan Ingram <jonathan.b.ingram@gmail.com>
2020-07-02 07:46:51 +10:00

52 lines
826 B
Go

// +build ocr
package docconv
import (
"fmt"
"io"
"sync"
"github.com/otiai10/gosseract/v2"
)
var config = struct {
langs []string
sync.Mutex
}{
langs: []string{"eng"},
}
func SetImageLanguages(l ...string) {
config.Lock()
config.langs = l
config.Unlock()
}
// ConvertImage converts images to text.
// Requires gosseract.
func ConvertImage(r io.Reader) (string, map[string]string, error) {
f, err := NewLocalFile(r)
if err != nil {
return "", nil, fmt.Errorf("error creating local file: %v", err)
}
defer f.Done()
meta := make(map[string]string)
client := gosseract.NewClient()
defer client.Close()
config.Lock()
defer config.Unlock()
client.SetLanguage(config.langs...)
client.SetImage(f.Name())
text, err := client.Text()
if err != nil {
return "", nil, err
}
return text, meta, nil
}