archie-mckenzie / ltt Goto Github PK

Add a cost optimizer mode (not for code or text that includes code).

In this mode:
[ ] Remove all new lines and line breaks characters. Everything on a single line.
[ ] Remove all empy lines containing only space characters.
[ ] Remove all repeated spaces.
[ ] Replace repeated points (more than 2 times) with … (1 char)
[ ] Replace repeated punctuations used for q»uot»e and unquote with a single char version ('' becomes ", >> becomes »).
[ ] Replace the single quote character with double quote character (Example: 'Welcome' becomes "Welcome").
[ ] Remove all repeated punctuations (more than one time) (Example: "Yes!!!" becomes "Yes!").
[ ] Convert dates in numerical format (Example: "Fifteen December 1789" becomes "15/10/1789").
[ ] Convert numbers in numerical format (Example: "One Hundred Two Thousands" becomes "102000").
[ ] Convert times in numerical format when convenient (Example: "twelve past five" becomes "05:12").
[ ] Abbreviate titles (Example: "Mister, Missus and Missis" becomes "Mr, Mrs and Ms").
[ ] Replace 'and' with '&'
[ ] Remove all symbols except for punctuation, mathematical or economical (like asterisks, #, symbols, borders, lines, arrows, etc.).
[ ] Use the character - for bullet lists and remove all other symbols.
[ ] Replace the accent character after a letter with the corresponding accented letter (example: e' becomes è or é)
[ ] Remove all emojies, both literal and graphic (Example: "Hello! :-DDD" becomes "Hello!")

The user can enable or disable each one of those options.

The same options can be used for reducing the GPT answer tokens simply adding the above list in the Prompt (example:"You must always use the following rules when writing the answer: ..." )

When the program receive the translated text, it can reverse some of those shortening modifications. For example restoring '&' with 'and', adding line breaks and new lines to ease the reading, making the numbers more readable (example: 102000 becomes 102,000, etc.)

This can provide significants costs reduction when translating a long novel.