ston1th/docstore
1
1
Fork 0
Code Issues 3 Releases Activity

improved logging

Browse source
This commit is contained in:
ston1th 2022-09-19 22:52:40 +02:00
commit 05b41cedce
81 changed files with 9802 additions and 279 deletions
Download patch file Download diff file Expand all files Collapse all files

18
pkg/scan/scanner.go
View file

@ -5,15 +5,16 @@ package scan
import (
"errors"
"fmt"
"git.giftfish.de/ston1th/docstore/pkg/core"
"git.giftfish.de/ston1th/docstore/pkg/fs"
"git.giftfish.de/ston1th/docstore/pkg/log"
"git.giftfish.de/ston1th/godrop/v2"
"os"
"os/exec"
"path/filepath"
"regexp"
"time"
"git.giftfish.de/ston1th/docstore/pkg/core"
"git.giftfish.de/ston1th/docstore/pkg/fs"
"git.giftfish.de/ston1th/godrop/v2"
"github.com/go-logr/logr"
)
const (
@ -44,13 +45,14 @@ var (
)
type Scanner struct {
log logr.Logger
base string
timeout time.Duration
t Tesseract
p PDF
}
func New(c *core.Config) (s *Scanner, err error) {
func New(log logr.Logger, c *core.Config) (s *Scanner, err error) {
var (
tcmd string
pcmd string
@ -92,6 +94,7 @@ func New(c *core.Config) (s *Scanner, err error) {
}
args := []string{"-l", lang, "--oem", fmt.Sprintf("%d", c.TesseractOEM)}
s = &Scanner{
log,
c.DataDir,
time.Second * time.Duration(c.CMDTimeout),
Tesseract{tcmd, args, []string{fmt.Sprintf("OMP_THREAD_LIMIT=%d", threads)}},
@ -101,6 +104,7 @@ func New(c *core.Config) (s *Scanner, err error) {
}
func (s *Scanner) Scan(file string) (filename, text string, err error) {
log := s.log
file = fs.Clean(file)
scanfile := filepath.Join(s.base, file)
var pdffile string
@ -117,7 +121,7 @@ func (s *Scanner) Scan(file string) (filename, text string, err error) {
p, _ := filepath.Split(file)
pdfCmd := exec.Command(s.t.cmd, append(s.t.args, []string{scanfile, "stdout", "pdf"}...)...)
pdfCmd.Env = s.t.env
log.Debugf("%v %s %v", pdfCmd.Env, pdfCmd.Path, pdfCmd.Args)
log.V(2).Info("tesseract command", "env", pdfCmd.Env, "path", pdfCmd.Path, "args", pdfCmd.Args)
pdf, err = timeout(pdfCmd, s.timeout)
if err != nil {
return
@ -140,7 +144,7 @@ func (s *Scanner) Scan(file string) (filename, text string, err error) {
filename = file
}
txtCmd := exec.Command(s.p.cmd, append(s.p.args, []string{pdffile, "-"}...)...)
log.Debugf("%s %v", txtCmd.Path, txtCmd.Args)
log.V(2).Info("text2pdf command", "path", txtCmd.Path, "args", txtCmd.Args)
txt, err := timeout(txtCmd, s.timeout)
if err != nil {
return