Refactor regular text extraction into dedicated module

[stefan6419846]

At the moment, the regular (non-layout mode) text extraction is mostly implemented in a method inside the PageObject class directly. This makes it hard to keep the complexity low, requires nonglobal variables and bloats the PageObject more and more on every new error handler.

For this reason, we should look into moving the actual internal implementation to a dedicated module/class to allow for easier refactoring and reduce file/module complexity.

[MartinThoma]

Maybe creating a TextExtraction class would help:

• We could replace the nonlocal keywords by object attributes
• We could get rid of the nested functions
• We could introduce methods that handle operations (_handle_operation_save_graphics_state, _handle_operations_restore_graphics_state, _handle_operation_horizontal_text_scaling, ...)
  • That would allow us to have a dictionary mapping the operator to the operation which could reduce the complexity of process_operation quite a bit

What do you think?

[stefan6419846]

Yes, this is similar to what I have been thinking. Regarding the different operations, this probably depends on whether their API signatures can be unified well enough, but the easier to read and test, the better.