Multiline f-strings and prefixing all with f

[PatrickJordanNutanix]

We have soft rule in our code base that means we should prefix all multiline f-strings with an f even if no interpolation is happening on part of the string. For example, instead of doing this:

long_string = (
  "This is a long string to demonstrate that when it wraps on to the next "
  f"line, all lines should be prefixed with an {f_prefix} to improve "
  "readability and effectively treat them along the same lines as a trailing "
  "comma, even if interpolation only happens on a subset of the lines."
)

We should be doing this:

long_string = (
  f"This is a long string to demonstrate that when it wraps on to the next "
  f"line, all lines should be prefixed with an {f_prefix} to improve "
  f"readability and effectively treat them along the same lines as a trailing "
  f"comma, even if interpolation only happens on a subset of the lines."
)

Is there an existing rule that can catch this linting error? If not, can I request one be added please?

[MichaReiser]

Does this rule of yours only apply if one of the string literals contain an expression or for all implicit concatenated strings?

E.g. what about

long_string = (
  "This is a long string to demonstrate that when it wraps on to the next "
  "line, all lines should be prefixed with an ... to improve "
  "readability and effectively treat them along the same lines as a trailing "
  "comma, even if interpolation only happens on a subset of the lines."
)

[PatrickJordanNutanix]

Does this rule of yours only apply if one of the string literals contain an expression or for all implicit concatenated strings?

Yes, only if at some point a variable is being interpolated into it. If it's a multiline string with no interpolation, then we do not prefix with an f.

[MichaReiser]

Thanks. I now understand your use case. We could phrase it in a more general way and say that you want a common prefix for all string literals in implicit concatenated strings. The rule would then also enforce that all strings are raw-strings

a = (
	r"flag"
	"this"
)

# instead write
a = (
	r"flag"
	r"this"
)

# or
a = (
	"flag"
	"this"
)

Considering that this is an opinionated stylistic rule, we should wait for #1774 to avoid enabling the rule by default for everyone using select=ALL.