Extracting text from a specific page of the document

[bm777]

How can I load a specific page and extract text?

And, identify and retrieve the paragraph with the highest occurrence of the target word from a multi-paragraph page.

[nmammeri]

There are 2 parts to this question:

1. Loading a specific pdf page: The short answer: it is not possible with the current version. However, I'm thinking how can this be handled in future versions.

• One way would be to split the pdf file using a different library then pass the byte[] stream to Extractous.
• The second way would be to add a specific Extractor function to return a list of pages.

2. Retrieving a paragraph with the highest occurrence of the target word: I think this out of scope of Extractous and is something that can be performed post extraction. The extracted text is delimited by spaces, to get the paragraphs I would split the string then perform a search on all paragraphs and select the one with the highest occurrence of the the target word.

[bm777]

Okay, understood. Future release -<

For the second question, no way to use Regex there? in the preprocessor.

Giving this task to a different library will increase the processing time, no?

• I was just suggesting since you guys are using the speed of Rust, why not include it as an option? But by default, it returns just text.
• And lastly, people are lazy which is why they use unstructured.

[nmammeri]

If I understood correctly, the use cases you want are kind of:

• Extract text for specific page
• Search for a target word and extract paragraph or text around it.

Interesting use case we'll definitely look into this but I'm wondering if unstructured has such functionality.

[bm777]

• Extracting text from specific page is available in Unstructured [here]. (I'm not using it, I'm using a different tool)
• Search for a target word, and return the max occurrence in paragraphs. (Currently, I use regex it is just a naive way)

Looking forward to play with Extractous