fix unused import, fix type, make Newline removal configurable

config.go

@@ -36,15 +36,17 @@ type TesConfig struct {
 	NatsConnectRetries int `env:"TES_NATS_CONNECT_RETRIES" default:"10"`
 	// if true, disable HTTP Server in favor of NATS Microservice interface
 	NoHttp bool `env:"TES_NO_HTTP" default:"false"`
+	// if true, extracted text will be compacted by replacing newlines with whitespace
+	RemoveNewlines bool `env:"TES_REMOVE_NEWLINES" default:"true"`
 	// How many replicas of the bucket to create. Default: 1
 	Replicas int `env:"TES_REPLICAS" default:"1"`
 	// HTTP listen address and/or port. Default: ':8080'
 	SrvAddr string `env:"TES_HOST_PORT" default:":8080"`
 }
 
-// NewTesConigFromEnv returns a service config object
+// NewTesConfigFromEnv returns a service config object
 // populated with defaults and values from environment vars
-func NewTesConigFromEnv() TesConfig {
+func NewTesConfigFromEnv() TesConfig {
 	var cfg TesConfig
 	if err := env.Load(&cfg, nil); err != nil {
 		logger.Error("Loading config failed", "err", err)

main.go

@@ -8,6 +8,7 @@ import (
 
 	"github.com/gin-contrib/expvar"
 	"github.com/gin-gonic/gin"
+	"github.com/johbar/text-extraction-service/v2/pkg/dehyphenator"
 	"github.com/johbar/text-extraction-service/v2/pkg/docparser"
 	sloggin "github.com/samber/slog-gin"
 )
@@ -24,13 +25,14 @@ var (
 )
 
 func main() {
+	tesConfig = NewTesConfigFromEnv()
+	dehyphenator.RemoveNewlines = tesConfig.RemoveNewlines
 	args := os.Args
 	// one shot mode: don't start a server, just process a single file provided on the command line
 	if len(args) > 1 {
 		PrintMetadataAndTextToStdout(args[1])
 		return
 	}
-	tesConfig = NewTesConigFromEnv()
 	closeDocChan = make(chan Document, 100)
 	saveExtractedDocChan = make(chan *ExtractedDocument, 100)
 	go saveAndCloseExtracedDocs()
@@ -70,7 +72,7 @@ func main() {
 		logger.Info("Service started with no HTTP endpoints. Waiting for interrupt.")
 		<-wait
 	}
-	logger.Info("Using PDF implementation", "lib", pdfImplementation)
+	logger.Info("PDF implementation", "lib", pdfImplementation)
 	if !docparser.Initialized {
 		logger.Warn("wvWare is not in PATH! We will not be able to extract legacy MS Word documents.")
 	}

poppler.go

@@ -5,7 +5,6 @@ package main
 import (
 	"errors"
 	"io"
-	"log/slog"
 	"strconv"
 	"strings"
 	"time"