Messed order of parsed texts

Hi Guys,

Found something interesting not in a pleasant manner about the texts parsed. The bulk paragraphs are wrongly placed at the bottom of the parsed texts. Could anyone help correct this wrong order?

Thanks so much. Luke

Input Pdf file left and output text right shown in screenshot below:

Below is the code: if pdf_file.endswith('.pdf') or pdf_file.endswith('.PDF'): parsed_curr_pg = parser.from_file(pdf_file, 'http://localhost:9998/tika') curr_pg_text = parsed_curr_pg['content'] with open('%s.txt' %pdf_file_nm, 'a', encoding='utf-8') as curr_pg: curr_pg.write(curr_pg_text) curr_pg.write('\page_break')

[jfrfonseca]

I managed to get it to work! I used a Docker container to run Tika as a server. Following the instructions in the official apache tika-docker repo (https://github.com/apache/tika-docker#custom-config) I created a custom config file setting the sortByPosition propriety in the PDF parser, and run the container mapping a volume to a local config file. Results as expected!

[paconius]

Sorry, I’m in the same boat as you WRT having no concrete idea of how to fix this. I am simply passing along new leads as I find them. AFAIK, you need to create a separate config file to provide to parser. The format of the file is described here: https://tika.apache.org/1.18/configuring.html. However, I’m not sure what string should go in the config file to set the sort preference. Also, I’m not 100% sure changing the sort order will fix the problem – however, it appears to be the best option based on what I’ve read in other threads.