[Python-Dev] PEP 498: Literal String Interpolation is ready for pronouncement

[Eric V. Smith]

On 9/5/2015 1:28 PM, Nathaniel Smith wrote:
> On Sep 5, 2015 9:20 AM, "Guido van Rossum" <guido at python.org
> <mailto:guido at python.org>> wrote:
>>
>> The processing of f-strings described by the PEP uses several phases:
>>
>> - find the end of the string (the final quote[s]) using the same
> algorithm used for all string literals
>> - expand \ escapes (e.g. \uXXXX)
> 
> It might be worth calling out explicitly in the text how this works for
> what I suspect will be the most common case: embedded quoting embedded
> quotes, like
> 
>     f"{ table[\"foo\"] }"
> 
>> - look for single {, then scan ahead to a matching } -- this skips
> matching () [] {}
> 
> I assume this part of the algorithm uses the lexer and scans for
> matching tokens rather than just scanning for characters? I tried
> looking at the PEP to double check but couldn't find the relevant part.
> It's important for things like
> 
>      f"{ '}' }"

Actually, my current implementation doesn't use the lexer, although I
suppose it could. I'm currently manually scanning the string, keeping
track of strings and parens. To find the end of an expression, it looks
for a '!', ':', or non-doubled '}', not inside of a string or (), [], or
{}. There's a special case for '!=' so the bang isn't seen as ending the
expression.

So it does work on this f-string:

>>> f"{ '}' }"
'}'

Although by using the lexer, I'd be worried about multiple levels of
escaping. I'd have to give it some thought.


>> - look for optional !a,!r,!s and !<spec> inside each {...} pair
>> - take what's left, enclose it in (), parse as expression
> 
> In fact maybe this whole list that Guido wrote should be copied into the
> PEP; it's a good summary :-).

I think the PEP does say these things, but maybe it could be tightened up.

Eric.