Get page information from extracted text

[atahanuz]

Hello, I'm using megaparse vision to extract text from a PDF. After the extraction, how can find which part of the extracted text belonged to which page (i.e. page 7 of the PDF) ?

[linear]

MEG-42 Get page information from extracted text

[dosubot]

Hey @atahanuz! I'm here to help you with bug fixes, answer questions, and become a contributor. Let's squash those bugs together! 🤖

To determine which part of the extracted text from a PDF belongs to a specific page, such as page 7, you can use the page_range attribute in the Block objects. Each Block object has a page_range attribute that indicates the page number(s) the text belongs to. You can filter the blocks by checking if the page_range matches the desired page number.

Here's a code snippet to illustrate how you can filter the blocks for page 7:

# Assuming `mp_document` is the MPDocument object returned by the convert method
page_number = 7
page_7_blocks = [
    block for block in mp_document.content
    if block.page_range and block.page_range[0] == page_number
]

# Now `page_7_blocks` contains all blocks from page 7

This code iterates over the content of the MPDocument and selects blocks where the page_range matches page 7 ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}

[atahanuz]

@dosu
page_range attribute gives (0,48), which is the entire document. I haven't been able to access specific pages

[dosubot]

If the page_range attribute is giving you (0, 48) for the entire document, it seems like the page_range is not being set correctly for individual pages. This might be due to how the process_file and __to_elements_list__ methods are handling the page information.

In MegaParse, the process_file method converts each page of a PDF into a base64 encoded image, and the number of pages (n_pages) is used to set the page_range in the __to_elements_list__ method. If the page_range is not reflecting individual pages, you might want to check if the metadata.page_number is being correctly assigned during the conversion process.

The __convert_element_to_block method is responsible for setting the page_range using the metadata.page_number from each Element. Ensure that this metadata is correctly populated for each page:

page_range=(metadata.page_number, metadata.page_number)
if metadata.page_number
else None

If the metadata.page_number is not available or not set correctly, the page_range will default to None or cover the entire document. You may need to verify that the metadata.page_number is being correctly extracted and assigned during the parsing process ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}

[atahanuz]

Dosubot weren't be able to help, can someone do? 🥲