fix: improve logging, add extra template

Dockerfile

@@ -8,4 +8,5 @@ FROM debian:bookworm
 RUN apt update && apt install -y libtesseract5
 COPY --from=build /usr/bin/rok-server /usr/bin/rok-server
 EXPOSE 8080
+ENV GOMEMLIMIT=100MiB
 ENTRYPOINT [ "/usr/bin/rok-server" ]

cmd/rok-scanner/main.go

@@ -2,11 +2,12 @@ package main
 
 import (
 	"fmt"
-	"github.com/rokmonster/ocr/internal/pkg/rokocr/tesseractutils"
 	"os"
 	"strings"
 	"time"
 
+	"github.com/rokmonster/ocr/internal/pkg/rokocr/tesseractutils"
+
 	"github.com/olekukonko/tablewriter"
 	config "github.com/rokmonster/ocr/internal/pkg/config/scannerconfig"
 	schema "github.com/rokmonster/ocr/internal/pkg/ocrschema"
@@ -50,6 +51,7 @@ func writeCSV(data []schema.OCRResult, template schema.OCRTemplate) {
 }
 
 func main() {
+
 	rokocr.Prepare(flags.CommonConfiguration)
 	rokocr.DownloadTesseractData(flags.CommonConfiguration)
 	rokocr.PreloadTemplates(flags.CommonConfiguration)

docs/install/golang_mac.md

@@ -23,8 +23,8 @@ curl -LO https://go.dev/dl/go1.19.2.darwin-amd64.pkg && open ./go1.19.2.darwin-a
 # install tesseract & opencv
 brew install tesseract opencv
 # You might need these
-# export CPATH="/opt/homebrew/include"
-# export LIBRARY_PATH="/opt/homebrew/lib"
+export CPATH="/opt/homebrew/include"
+export LIBRARY_PATH="/opt/homebrew/lib"
 ```
 
 ### PLEASE CHECK FOR ERRORS!!!

internal/pkg/ocrschema/result-schema.go

@@ -1,6 +1,9 @@
 package ocrschema
 
+import "time"
+
 type OCRResult struct {
 	Filename string                 `json:"filename"`
 	Data     map[string]interface{} `json:"data"`
+	Took     time.Duration          `json:"duration"`
 }

internal/pkg/rokocr/tesseractutils/parser.go

@@ -1,11 +1,13 @@
 package tesseractutils
 
 import (
-	imgutils2 "github.com/rokmonster/ocr/internal/pkg/utils/imgutils"
-	"github.com/rokmonster/ocr/internal/pkg/utils/stringutils"
 	"image"
 	"os"
 	"path/filepath"
+	"time"
+
+	imgutils2 "github.com/rokmonster/ocr/internal/pkg/utils/imgutils"
+	"github.com/rokmonster/ocr/internal/pkg/utils/stringutils"
 
 	log "github.com/sirupsen/logrus"
 
@@ -14,6 +16,7 @@ import (
 
 func ParseImage(name string, img image.Image, template schema.OCRTemplate, tmpdir, tessdata string) schema.OCRResult {
 	log.Debugf("[%s] Processing with template: %s", filepath.Base(name), template.Title)
+	start := time.Now()
 
 	results := make(map[string]interface{})
 
@@ -35,5 +38,6 @@ func ParseImage(name string, img image.Image, template schema.OCRTemplate, tmpdi
 	return schema.OCRResult{
 		Filename: filepath.Base(name),
 		Data:     results,
+		Took:     time.Since(start),
 	}
 }

internal/pkg/rokocr/tesseractutils/recognition.go

@@ -4,7 +4,6 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
-	"time"
 
 	"github.com/rokmonster/ocr/internal/pkg/utils/fileutils"
 	"github.com/rokmonster/ocr/internal/pkg/utils/imgutils"
@@ -13,17 +12,6 @@ import (
 	schema "github.com/rokmonster/ocr/internal/pkg/ocrschema"
 )
 
-// func processSingleFile(index, total int, f string) *schema.OCRResult {
-// 	start := time.Now()
-// 	result, err := ParseSingleFile(f, tessData, template, force)
-// 	if err != nil {
-// 		logrus.Printf("[%04d/%04d] %v - %v", index, total, filepath.Base(f), err)
-// 		return
-// 	}
-// 	logrus.Printf("[%04d/%04d] %v Took: %v ms", index, total, filepath.Base(f), time.Since(start).Milliseconds())
-// 	return result
-// }
-
 func RunRecognitionChan(mediaDir, tessData string, template schema.OCRTemplate, force bool) <-chan schema.OCRResult {
 
 	out := make(chan schema.OCRResult)
@@ -33,13 +21,11 @@ func RunRecognitionChan(mediaDir, tessData string, template schema.OCRTemplate,
 		total := len(files)
 
 		for index, f := range files {
-			start := time.Now()
 			result, err := ParseSingleFile(f, tessData, template, force)
 			if err != nil {
-				logrus.Printf("[%04d/%04d] %v - %v", index, total, filepath.Base(f), err)
+				logrus.Errorf("[%04d/%04d] %v - %v", index, total, filepath.Base(f), err)
 				continue
 			}
-			logrus.Printf("[%04d/%04d] %v Took: %v ms", index, total, filepath.Base(f), time.Since(start).Milliseconds())
 			out <- *result
 		}
 		close(out)
@@ -69,5 +55,5 @@ func ParseSingleFile(f, tessData string, template schema.OCRTemplate, force bool
 		return &result, nil
 	}
 
-	return nil, fmt.Errorf("image doesn't match the template")
+	return nil, fmt.Errorf("image doesn't match the template: Template: %s @ %s", template.Title, template.Version)
 }

internal/pkg/www/jobs_controller.go

@@ -5,20 +5,19 @@ import (
 	"encoding/binary"
 	"encoding/json"
 	"fmt"
-	"github.com/rokmonster/ocr/internal/pkg/www/middlewares"
 	"net/http"
 	"os"
 	"path/filepath"
 	"strconv"
 	"strings"
 	"time"
 
-	"github.com/rokmonster/ocr/internal/pkg/rokocr/tesseractutils"
-	"github.com/rokmonster/ocr/internal/pkg/utils/fileutils"
-
 	"github.com/gin-gonic/gin"
 	"github.com/rokmonster/ocr/internal/pkg/ocrschema"
 	"github.com/rokmonster/ocr/internal/pkg/rokocr"
+	"github.com/rokmonster/ocr/internal/pkg/rokocr/tesseractutils"
+	"github.com/rokmonster/ocr/internal/pkg/utils/fileutils"
+	"github.com/rokmonster/ocr/internal/pkg/www/middlewares"
 	log "github.com/sirupsen/logrus"
 	bolt "go.etcd.io/bbolt"
 )
@@ -218,11 +217,13 @@ func (controller *JobsController) StartJobByID(c *gin.Context) {
 		if len(templates) > 0 {
 			log.Debugf("Loaded %v templates", len(templates))
 			template := ocrschema.FindTemplate(mediaDir, templates)
+			log.Infof("[Job: %04d] Picked template: %s by %s", job.ID, template.Title, template.Author)
 			_ = controller.updateJobTemplate(job.ID, template)
 
 			var data []ocrschema.OCRResult
-			for elem := range tesseractutils.RunRecognitionChan(mediaDir, controller.tessdataDir, template, false) {
+			for elem := range tesseractutils.RunRecognitionChan(mediaDir, controller.tessdataDir, template, true) {
 				data = append(data, elem)
+				log.Printf("[Job: %04d][%04d/%04d] %v Took: %v ms", job.ID, index, fileCount, elem.Filename, elem.Took.Milliseconds())
 				index = index + 1
 				_ = controller.updateJobStatus(job.ID, fmt.Sprintf("Processing: %v/%v", index, fileCount))
 				_ = controller.updateJobResults(job.ID, data)