a text parser

[Mirage]

Anyhow... maybe when I get the competition site sorted we could have a regular... 'see who can do it fastest/smallest competition'... just a thought

Seems that's a good idea.

[arthurprs]

Anyhow... maybe when I get the competition site sorted we could have a regular... 'see who can do it fastest/smallest competition'... just a thought

Seems that's a good idea.

very good

[JernejL]

As for writing a JSON parser... I don't want to discourage anyone from learning, but I also wouldn't encourage re-inventing the wheel (unless of course the wheel you were looking at was square). There is already a JSON parser for Delphi, so if it were me, I'd look at their code, see if could be optimised... and if not... work on something else

it must work in freepascal too, the code there uses variants and is "not lightweight enough".

[AthenaOfDelphi]

it must work in freepascal too, the code there uses variants and is "not lightweight enough".

Then that sounds like a case of re-inventing the wheel... where the original is square

Good luck with that when you get round to it.

[AthenaOfDelphi]

Given the usefulness of this little session, I've added it to the library in a newly created code snippets section here

[JernejL]

AthenaOfDelphi's and consequently Arthurps's code for some reason separate also by the "." character, which makes it impossible to parse floating point values (think "123.45" ) and also "0D0A" newlines which isnt neccesary but they dont parse the tab #09.
This is easy to fix tho just by removing extra IF condition.

Arthurps's code also doesnt parse properly if you use several separators to separate values (think grid space / tab aligned values): it produces empty values in the array.

So correct characters for separating should be:
(aChar = (' ')) or (aChar = (';')) or
(aChar = (',')) or (aChar = (#09))

I didn't test User137 and Brainer's code since they use classes.

[AthenaOfDelphi]

AthenaOfDelphi's and consequently Arthurps's code for some reason separate also by the "." character, which makes it impossible to parse floating point values (think "123.45" ).
This is easy to fix tho just by removing extra IF condition.

I didn't test User137 and Brainer's code since they use classes.

I just banged in a bunch of separators without really thinking about it too much. I normally use things like INI files, type files or streamed classes for data storage... depends on the application.

Of course, we perhaps shouldn't have included , either since that works as the decimal separator depending on location... just a thought

[JernejL]

http://en.wikipedia.org/wiki/Decimal_separator

The list is faulty, i've been always taught to use decimal point and the list shows slovenia as using a decimal colon.. practically everyone uses decimal point these days.

[arthurprs]

yes "dot" should be removed, but it's very easy to modify the tokens separators "ifs"

[JernejL]

yes "dot" should be removed, but it's very easy to modify the tokens separators "ifs"

That's true and i changed it, and after that Athena's code works but your code has another issue - i quote myself:

Arthurps's code also doesnt parse properly if you use several separators to separate values (think grid space / tab aligned values): it produces empty values in the array.