Fix values similar to booleans being incorrectly parsed to booleans

[MrGreenTea]

This included values that strat with t or f like test and friend as well as values that have true or false as prefix like truely and falsely.

All of them are now incorrectly rejected by tomlkit.value.

Closes #165

[frostming]

Sorry, I revisited the original issue and fix, and came up with a different fix approach.

The real problem here are 1) the error isn't thrown correctly and 2) the value() API doesn't check the EOF of the given string.

The parsing of boolean values is correct and doesn't need a fix. So can you kindly revert the change in parsing but ensure the EOF is reached after value parsing. This approach doesn't only fix the boolean parsing, but it also fixes other incomplete parsings like "som string"1.23.

Thanks for your patience.

[MrGreenTea]

I'm not sure I follow exactly. Can you provide me with some tests that reproduce the behaviour you expect? Or some examples?

[frostming]

tomlkit.value('"foo"1.2')
# should reject

the current behavior parses as 'foo' but it doesn't reach the end. 1.2 is remaining unparsed, we should check against this case. Then the original issue will also be fixed.

[MrGreenTea]

If I understand you correctly, we should check if EOF was reached when parsing a value?
Would a simple check like this suffice?

if not self.end():
    raise ParseError

[frostming]

lgtm, but it only applies to value api so it should be parser.end()

also a meaningful error message would be good

[frostming]

LGTM! Thanks

[MrGreenTea]

@frostming thanks for the help! Is there a chance to get a new release with this bugfix soon?