Abstract

PEP 498 introduced Literal String Interpolation (or “f-strings”). The expression portions of those literals however are subject to certain restrictions. This PEP proposes a formal grammar lifting those restrictions, promoting “f-strings” to “f expressions” or f-literals.

This PEP expands upon the f-strings introduced by PEP 498, so this text requires familiarity with PEP 498.

PEP Status

This PEP is deferred until an implementation is available.

Terminology

This text will refer to the existing grammar as “f-strings”, and the proposed one as “f-literals”.

Furthermore, it will refer to the {}-delimited expressions in f-literals/f-strings as “expression portions” and the static string content around them as “string portions”.

Motivation

The current implementation of f-strings in CPython relies on the existing string parsing machinery and a post processing of its tokens. This results in several restrictions to the possible expressions usable within f-strings:

1. It is impossible to use the quote character delimiting the f-string within the expression portion:

   >>> f'Magic wand: { bag['wand'] }'
                                ^
   SyntaxError: invalid syntax
   

2. A previously considered way around it would lead to escape sequences in executed code and is prohibited in f-strings:

   >>> f'Magic wand { bag[\'wand\'] } string'
   SyntaxError: f-string expression portion cannot include a backslash
   

3. Comments are forbidden even in multi-line f-strings:

   >>> f'''A complex trick: {
   ... bag['bag']  # recursive bags!
   ... }'''
   SyntaxError: f-string expression part cannot include '#'
   

4. Expression portions need to wrap ':' and '!' in braces:

   >>> f'Useless use of lambdas: { lambda x: x*2 }'
   SyntaxError: unexpected EOF while parsing
   

These limitations serve no purpose from a language user perspective and can be lifted by giving f-literals a regular grammar without exceptions and implementing it using dedicated parse code.

Rationale

The restrictions mentioned in Motivation are non-obvious and counter-intuitive unless the user is familiar with the f-literals’ implementation details.

As mentioned, a previous version of PEP 498 allowed escape sequences anywhere in f-strings, including as ways to encode the braces delimiting the expression portions and in their code. They would be expanded before the code is parsed, which would have had several important ramifications:

#. It would not be clear to human readers which portions are Expressions and which are strings. Great material for an “obfuscated/underhanded Python challenge” #. Syntax highlighters are good in parsing nested grammar, but not in recognizing escape sequences. ECMAScript 2016 (JavaScript) allows escape sequences in its identifiers [1] and the author knows of no syntax highlighter able to correctly highlight code making use of this.

As a consequence, the expression portions would be harder to recognize with and without the aid of syntax highlighting. With the new grammar, it is easy to extend syntax highlighters to correctly parse and display f-literals:

f'Magic wand: {bag['wand']:^10}'

Highlighting expression portions with possible escape sequences would mean to create a modified copy of all rules of the complete expression grammar, accounting for the possibility of escape sequences in key words, delimiters, and all other language syntax. One such duplication would yield one level of escaping depth and have to be repeated for a deeper escaping in a recursive f-literal. This is the case since no highlighting engine known to the author supports expanding escape sequences before applying rules to a certain context. Nesting contexts however is a standard feature of all highlighting engines.

Familiarity also plays a role: Arbitrary nesting of expressions without expansion of escape sequences is available in every single other language employing a string interpolation method that uses expressions instead of just variable names. [2]

Specification

PEP 498 specified f-strings as the following, but places restrictions on it:

f ' <text> { <expression> <optional !s, !r, or !a> <optional : format specifier> } <text> ... '

All restrictions mentioned in the PEP are lifted from f-literals, as explained below:

1. Expression portions may now contain strings delimited with the same kind of quote that is used to delimit the f-literal.
2. Backslashes may now appear within expressions just like anywhere else in Python code. In case of strings nested within f-literals, escape sequences are expanded when the innermost string is evaluated.
3. Comments, using the '#' character, are possible only in multi-line f-literals, since comments are terminated by the end of the line (which makes closing a single-line f-literal impossible).
4. Expression portions may contain ':' or '!' wherever syntactically valid. The first ':' or '!' that is not part of an expression has to be followed a valid coercion or format specifier.

A remaining restriction not explicitly mentioned by PEP 498 is line breaks in expression portions. Since strings delimited by single ' or " characters are expected to be single line, line breaks remain illegal in expression portions of single line strings.

Backwards Compatibility

f-literals are fully backwards compatible to f-strings, and expands the syntax considered legal.

Reference Implementation

TBD

References

[1]

ECMAScript IdentifierName specification ( http://ecma-international.org/ecma-262/6.0/#sec-names-and-keywords )

Yes, const cthulhu = { H̹̙̦̮͉̩̗̗ͧ̇̏̊̾Eͨ͆͒̆ͮ̃͏̷̮̣̫̤̣Cͯ̂͐͏̨̛͔̦̟͈̻O̜͎͍͙͚̬̝̣̽ͮ͐͗̀ͤ̍̀͢M̴̡̲̭͍͇̼̟̯̦̉̒͠Ḛ̛̙̞̪̗ͥͤͩ̾͑̔͐ͅṮ̴̷̷̗̼͍̿̿̓̽͐H̙̙̔̄͜\u0042: 42 } is valid ECMAScript 2016

[2]

Wikipedia article on string interpolation ( https://en.wikipedia.org/wiki/String_interpolation )

Copyright

This document has been placed in the public domain.