I am not sure if that is available via Tesseract C++ API. However it would be nice to have a possibility to iterate over the page segments and get rect coordinates together with recognized text within that rect.
For example, I want to disregard headers/footers during OCR process. I can estimate header/footer location of the page (plus analysing the text which is typical), but for that I need rect coordinates. When headers/footers are incorporated onto output string it is very difficult to cut them away.
Another scenario: I have some text images with line number column (i.e. all text lines are followed by a number / line-counter). Visually it is very easy to cut this column away, but again hardly possible to remove line numbers from resulting text without danger to remove numbers from actual text.
Using ResultIterator, you can have the block/para/word/symbol(character) coordinates. See the unit test cases for example.