Getting relevant data

[sohamvg]

1. The json files contain a text object depending on the text format but it is not uniform. e.g. "theorem 2.1" can be converted into one text object {theorem%202.1} (where %20 denotes space) or two text objects {theorem} & {2.1} even when "theorem" & "2.1" are of the same style (font size etc.).
2. How to detect newline using x,y coordinates of the text objects?