added tesseract threads

pkg/cmd/cmd.go

@@ -30,6 +30,7 @@ const (
 	defRunGroup = "docstore"
 
 	defTimeout       = 600
+	tesseractThreads = 1
 )
 
 var (
@@ -71,8 +72,11 @@ func initServer(cfg core.Config) (err error) {
 	log.InitLogger(cfg)
 
 	cfg.DataDir = filepath.Join(cfg.RunDir, core.DataDir)
+	debugCfg := cfg
+	debugCfg.Secret = "***hidden***"
+	log.Debugf("docstore config: %+v", debugCfg)
 
-	scanner, err := scan.New(cfg.DataDir, cfg.Langs, time.Second*time.Duration(cfg.CMDTimeout))
+	scanner, err := scan.New(cfg.DataDir, cfg.Langs, cfg.TesseractThreads, time.Second*time.Duration(cfg.CMDTimeout))
 	if err != nil {
 		return errors.New("scanner: " + err.Error())
 	}
@@ -255,6 +259,12 @@ func serverFlags() []cli.Flag {
 			Usage:       "timeout in seconds for executed teseract and pdftotext commands",
 			Destination: &config.CMDTimeout,
 		},
+		cli.IntFlag{
+			Name:        "threads",
+			Value:       tesseractThreads,
+			Usage:       "tesseract thread limit (1-4)",
+			Destination: &config.TesseractThreads,
+		},
 		cli.BoolFlag{
 			Name:        "foreground, f",
 			Usage:       "do not fork into the background",

pkg/core/types.go

@@ -44,6 +44,7 @@ type Config struct {
 	Secret           string   `json:"secret,omitempty"`
 	Langs            []string `json:"langs,omitempty"`
 	CMDTimeout       int      `json:"cmd_timeout,omitempty"`
+	TesseractThreads int      `json:"tesseract_threads,omitempty"`
 	Foreground       bool     `json:"foreground,omitempty"`
 	SecureCookie     bool     `json:"secure_cookie,omitempty"`
 	Debug            bool     `json:"debug,omitempty"`

pkg/scan/scanner.go

@@ -13,11 +13,15 @@ import (
 	"os/exec"
 	"path/filepath"
 	"regexp"
-	"runtime"
 	"time"
 )
 
-const pdfExt = ".pdf"
+const (
+	pdfExt = ".pdf"
+
+	minThreads = 1
+	maxThreads = 4
+)
 
 type PDF struct {
 	cmd  string
@@ -46,7 +50,7 @@ type Scanner struct {
 	p       PDF
 }
 
-func New(base string, langs []string, timeout time.Duration) (s *Scanner, err error) {
+func New(base string, langs []string, threads int, timeout time.Duration) (s *Scanner, err error) {
 	var (
 		tcmd string
 		pcmd string
@@ -77,11 +81,16 @@ func New(base string, langs []string, timeout time.Duration) (s *Scanner, err er
 			lang += "+"
 		}
 	}
+	if threads > maxThreads {
+		threads = maxThreads
+	} else if threads < minThreads {
+		threads = minThreads
+	}
 	args := []string{"-l", lang, "--oem", tesseract_oem}
 	s = &Scanner{
 		base,
 		timeout,
-		Tesseract{tcmd, args, []string{fmt.Sprintf("OMP_THREAD_LIMIT=%d", runtime.NumCPU())}},
+		Tesseract{tcmd, args, []string{fmt.Sprintf("OMP_THREAD_LIMIT=%d", threads)}},
 		PDF{pcmd, []string{"-layout", "-nopgbrk", "-eol", "unix"}},
 	}
 	return

pkg/server/templates.go

@@ -81,7 +81,7 @@ const (
             <td>File</td>
             <td><div class="btn-group">
                 {{if $item.Scan}}
-                <a href="#" class="btn btn-sm btn-warning">Scanning..</a>
+                <a href="#" class="btn btn-sm btn-warning">Scanning</a>
                 {{else}}
                 {{if not $item.Flag}}
                 <a href="/index{{$item.Abs}}" class="btn btn-sm btn-success">Index</a>
@@ -109,7 +109,7 @@ const (
     </ol>
     <div class="btn-group btn-breadcrumb">
         {{if .Data.Scan}}
-        <a href="#" class="btn btn-sm btn-warning">Scanning..</a>
+        <a href="#" class="btn btn-sm btn-warning">Scanning</a>
         {{else}}
         {{if not .Data.Flag}}
         <a href="/index{{.Data.Name}}" class="btn btn-sm btn-success">Index</a>