From 76d91ea8f65c6ad52efb24ac2c94b22c2908bc5c Mon Sep 17 00:00:00 2001 From: Nick White Date: Mon, 21 Mar 2022 13:51:51 +0000 Subject: Only generate full-size PDF if requested This avoids the issue that large PDFs require a lot of RAM, so there are chances of running out of memory. Plus it's a waste of space and time. --- cmd/rescribe/main.go | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'cmd/rescribe/main.go') diff --git a/cmd/rescribe/main.go b/cmd/rescribe/main.go index 96f6162..eba8e84 100644 --- a/cmd/rescribe/main.go +++ b/cmd/rescribe/main.go @@ -484,7 +484,7 @@ func startProcess(ctx context.Context, logger log.Logger, tessCommand string, bo } fmt.Printf("Processing book\n") - err = processbook(ctx, trainingName, tessCommand, conn) + err = processbook(ctx, trainingName, tessCommand, conn, fullpdf) if err != nil { _ = os.RemoveAll(tempdir) return fmt.Errorf("Error processing book: %v", err) @@ -648,7 +648,7 @@ func downloadbook(dir string, name string, conn Pipeliner) error { return nil } -func processbook(ctx context.Context, training string, tesscmd string, conn Pipeliner) error { +func processbook(ctx context.Context, training string, tesscmd string, conn Pipeliner, fullpdf bool) error { origPattern := regexp.MustCompile(`[0-9]{4}.(jpg|png)$`) wipePattern := regexp.MustCompile(`[0-9]{4,6}(.bin)?.(jpg|png)$`) ocredPattern := regexp.MustCompile(`.hocr$`) @@ -764,7 +764,7 @@ func processbook(ctx context.Context, training string, tesscmd string, conn Pipe stopTimer(stopIfQuiet) conn.Log("Message received on analyse queue, processing", msg.Body) fmt.Printf("\n Analysing OCR and compiling PDFs\n") - err = pipeline.ProcessBook(ctx, msg, conn, pipeline.Analyse(conn), ocredPattern, conn.AnalyseQueueId(), "") + err = pipeline.ProcessBook(ctx, msg, conn, pipeline.Analyse(conn, fullpdf), ocredPattern, conn.AnalyseQueueId(), "") resetTimer(stopIfQuiet, quietTime) if err != nil { return fmt.Errorf("Error during analysis: %v", err) -- cgit v1.2.1-24-ge1ad