User Tools

Site Tools


driving_jobs_scanner

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
driving_jobs_scanner [2024/11/23 10:55] – [Research phase] kalaschnikowdriving_jobs_scanner [2024/12/20 12:03] (current) – [Research phase] kalaschnikow
Line 1: Line 1:
 +====== Driving jobs scanner ======
 +
 +Rolf wants to implement a tool which scans driving jobs (Fahraufträge) PDFs and creates DB table entries from them.
 +
 +We spoke about it on 22.11.2024.
 +
 +
 +===== Research phase =====
 +
 +  * OCR necessary
 +    * because PDF do not contain text
 +
 +  * free OCR tools
 +    * PDF24 tools 
 +      * online (not locally)
 +      * freeware, but not open source
 +      * license unclear (commercial use allowed?)
 +    * Tesseract
 +    * ocrmypdf 
 +      * "Tesseract for PDFs"
 +      * Windows + Linux
 +      * use ''--sidecar'' option to create a text file from PDF
 +        * Example:
 +          * ''ocrmypdf -l deu --sidecar output.txt --output-type pdf input.pdf output.pdf''
 +      * Misc
 +        * Useful link for Windows:
 +          * https://ub-mannheim.github.io/Tesseract_Dokumentation/OCRmyPDF_Windows_und_Linux.html
 +
 +    * extracting text from PDF
 +      * [[https://github.com/itext|iText]]
 +        * but the PDF file should already contain text
 +        * see also:
 +          * https://stackoverflow.com/questions/8821107/pdf-text-extraction-using-itext
 +      * Aspose.PDF for Java
 +        * https://products.aspose.com/pdf/de/java/editor/
 +      * Misc
 +        * https://edrawmind.wondershare.com/productivity-improvement/7-best-ocr-tools-for-linux.html
 +        * https://www.baeldung.com/linux/ocr-tools
 +
 +===== My Notes =====
 +
 +  * OCR notwendig
 +  * Andere Formate als PDF möglich?
 +  * Text nach OCR "zerschnipselt"
 +  * Auswahl mit Firefox noch OK, aber auch nicht 100%
 +  * mit OCR
 +    * "gefährlich", da die Fahraufträge verschiedenartig sein können
 +    * definieren wie der Text zu interpretieren ist
 +    * definieren was tun, wenn der Text zugeordnet sein kann
 +    * definieren was passiert, wenn etwas falsch ist
 +
 +  * 200 PDFs
 +  * Power Apps 
 +  * mit Flow
 +  * Email
 +    * 1. Ladeliste PDF => Auftrag
 +    * 2. Email
 +    * 3. 14 Tage später Gutschrift
 +  * Flow
 +    * langsam
 +    * Lizenzkosten
 +  * Desktop App
 +    * File reinziehen