I get the identical issue for my purchase order from suppliers.

suppliers send their delivery orders or invoices either as a mail content (one ligne per product) or pdf

Importing these data into odoo is a nightmare.

if you just create a csv file from text and then import Odoo will will create new products thus creating duplicates, unles you provide the product id if the product already exist in odoo. but the PO import menu do not identify the PO order line product name with the product id!!!!

I use the follwing tools: for OCR Tabula, Tesseract, 

for data cleaning :Google OpenRefine

Same question, any update for the topic? 

I tried to install the OCR from webkul, but it is not for sales order directly, I need to do more coding to make it. And, then it will have Automation in action.

Update :

Odoo 12 introduced Documents module. which allows to import vendor bills from PDF ( with embedded xml) or OCR  into Odoo automatically. So it is possible theoretically to convert Customer's orders into Odoo's SO. 

why odoo didn't include this and only limited to vendor bills?

on the other hand, there are several tools to extract tables from PDF files, which can be parsed somehow and imported to odoo. 

My questions aims to see if someone did that?   what are suggested solutions ? 

I think u can't do this.

Odoo works with internal references for each product and vendor. Its not like it relates text, it works with externa references in a Database.