I performed a small experiment where I found a newspaper article form the 1920s - it was a Christmas type story with a lot of dialogue and I asked it to convert it to json format for my text 2 speech (demo). It proved a 1st draft TV script could be written from a book in a few minutes and keeping the spoken lines and probably getting the speaker right 'most of the time'. The story started with a monologue interspersed with movement and so it broke it apart. In a script actions/soundfx could be added even between spoken parts in a monologue. I could parse Wuthering Heights and have the dialogue verbatim.