diff options
| author | Nick White <git@njw.name> | 2022-03-21 13:51:51 +0000 |
|---|---|---|
| committer | Nick White <git@njw.name> | 2022-03-21 13:51:51 +0000 |
| commit | 76d91ea8f65c6ad52efb24ac2c94b22c2908bc5c (patch) | |
| tree | 35bc0e895befc1c686a64565f8bdff470a8db1c2 /cmd/rescribe/main.go | |
| parent | af8650c074bc111200b132b0918d44cacd423b6e (diff) | |
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.
Diffstat (limited to 'cmd/rescribe/main.go')
| -rw-r--r-- | cmd/rescribe/main.go | 6 |
1 files changed, 3 insertions, 3 deletions
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) |