- When TextractDocument is created from blocks, the operation can tak…

[vucetica]

…e a lot of time (for the example PDFs that we've had, it took hours on decent machines).

Cause

• blocks.Find is extremely slow and it is all over the place in the code.

Solution

• Instead of having blocks as a list, I turned into dictionary and having lookups which perform much faster.

Results

• Our document took about 2h to process. This time is reduced now to approx 1s (the same CPU).

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[sahays]

@vucetica your changes looks good to me. If you could change the namespace and commit again, I will be able to merge the changes.

[vucetica]

Thanks, @sahays . I've just changed the namespace. Please let me know if you find anything else.