Files
pdf-xrechnung/lib/extract.go
T
2026-02-06 21:09:58 +01:00

58 lines
1.2 KiB
Go
Executable File

package lib
import (
"errors"
"fmt"
"os"
"path/filepath"
"strings"
attachment2 "gitlab.com/romalor/pdflib/pkg/components/attachment"
"gitlab.com/romalor/pdflib/pkg/pdf"
)
func ExtractFromPdf(inputFilePath string, outputFilePath string) error {
logger := GetDebugLogger()
logger.Log(fmt.Sprintf("enter ExtractFromPdf(%s, %s)", inputFilePath, outputFilePath))
defer func() {
logger.Log("leave extractFromPdf")
}()
if inputFilePath == "" {
return errors.New("no input file specified")
}
if outputFilePath == "" {
outputFilePath = strings.TrimSuffix(inputFilePath, filepath.Ext(inputFilePath)) + ".xml"
}
logger.Log(fmt.Sprintf("pdfInput: '%s' output: '%s'", inputFilePath, outputFilePath))
inputFile, err := os.Open(inputFilePath)
if err != nil {
return err
}
defer inputFile.Close()
var attachments []attachment2.Attachment
var doc *pdf.Document
doc, err = pdf.OpenReader(inputFile)
if err != nil {
return err
}
attachments, err = doc.Attachments()
if err != nil {
return err
}
if len(attachments) == 0 {
return errors.New("no attachments found")
}
err = doc.ExtractAttachment(attachments[0].Name, outputFilePath)
return err
}