format StmtBreak

[davidszotten]

Summary

format StmtBreak

trying to learn how to help out with the formatter. starting simple

Test Plan

new snapshot test

[konstin]

thanks

[github-actions]

PR Check Results

Ecosystem

✅ ecosystem check detected no changes.

Benchmark

Linux

group                                      main                                   pr
-----                                      ----                                   --
formatter/large/dataset.py                 1.00      6.8±0.01ms     6.0 MB/sec    1.00      6.8±0.01ms     6.0 MB/sec
formatter/numpy/ctypeslib.py               1.00   1397.9±1.97µs    11.9 MB/sec    1.00   1400.6±2.58µs    11.9 MB/sec
formatter/numpy/globals.py                 1.00    136.8±0.33µs    21.6 MB/sec    1.01    138.5±0.32µs    21.3 MB/sec
formatter/pydantic/types.py                1.00      2.8±0.00ms     9.2 MB/sec    1.01      2.8±0.01ms     9.1 MB/sec
linter/all-rules/large/dataset.py          1.02     14.2±0.06ms     2.9 MB/sec    1.00     14.0±0.04ms     2.9 MB/sec
linter/all-rules/numpy/ctypeslib.py        1.01      3.5±0.00ms     4.7 MB/sec    1.00      3.5±0.00ms     4.8 MB/sec
linter/all-rules/numpy/globals.py          1.02    369.0±1.19µs     8.0 MB/sec    1.00    363.2±0.82µs     8.1 MB/sec
linter/all-rules/pydantic/types.py         1.01      6.2±0.01ms     4.1 MB/sec    1.00      6.1±0.01ms     4.2 MB/sec
linter/default-rules/large/dataset.py      1.03      7.5±0.06ms     5.4 MB/sec    1.00      7.3±0.01ms     5.5 MB/sec
linter/default-rules/numpy/ctypeslib.py    1.02   1570.1±3.32µs    10.6 MB/sec    1.00   1533.2±3.95µs    10.9 MB/sec
linter/default-rules/numpy/globals.py      1.02    168.5±0.16µs    17.5 MB/sec    1.00    165.7±0.36µs    17.8 MB/sec
linter/default-rules/pydantic/types.py     1.02      3.4±0.01ms     7.6 MB/sec    1.00      3.3±0.00ms     7.7 MB/sec

Windows

group                                      main                                   pr
-----                                      ----                                   --
formatter/large/dataset.py                 1.00      7.8±0.02ms     5.2 MB/sec    1.01      7.9±0.04ms     5.2 MB/sec
formatter/numpy/ctypeslib.py               1.00  1574.9±10.08µs    10.6 MB/sec    1.01  1583.4±10.25µs    10.5 MB/sec
formatter/numpy/globals.py                 1.00    155.8±1.46µs    18.9 MB/sec    1.01    156.8±2.23µs    18.8 MB/sec
formatter/pydantic/types.py                1.00      3.2±0.02ms     8.0 MB/sec    1.01      3.2±0.02ms     7.9 MB/sec
linter/all-rules/large/dataset.py          1.00     15.9±0.07ms     2.6 MB/sec    1.01     16.0±0.11ms     2.5 MB/sec
linter/all-rules/numpy/ctypeslib.py        1.00      4.2±0.02ms     4.0 MB/sec    1.01      4.2±0.03ms     4.0 MB/sec
linter/all-rules/numpy/globals.py          1.00    428.5±8.35µs     6.9 MB/sec    1.00    429.4±5.24µs     6.9 MB/sec
linter/all-rules/pydantic/types.py         1.00      7.0±0.03ms     3.6 MB/sec    1.02      7.1±0.05ms     3.6 MB/sec
linter/default-rules/large/dataset.py      1.00      8.3±0.05ms     4.9 MB/sec    1.01      8.4±0.02ms     4.8 MB/sec
linter/default-rules/numpy/ctypeslib.py    1.01  1742.1±12.50µs     9.6 MB/sec    1.00   1732.9±9.27µs     9.6 MB/sec
linter/default-rules/numpy/globals.py      1.00    186.6±1.34µs    15.8 MB/sec    1.00    186.6±1.24µs    15.8 MB/sec
linter/default-rules/pydantic/types.py     1.00      3.8±0.02ms     6.7 MB/sec    1.00      3.8±0.02ms     6.7 MB/sec

[davidszotten]

do i need any custom comment handling like in

ruff/crates/ruff_python_formatter/src/statement/stmt_while.rs

Line 21 in 30574e1

let comments = f.context().comments().clone();

? if not, how come it's needed for while?

[konstin]

FormatNodeRule has default implementations for fmt_leading_comments, fmt_trailing_comments and fmt_dangling_comments which do the right thing for break (a comment can be on its own line before the break or trailing the break either on the same line or on its own line line after).

For while it's more complicated, e.g. you can have

while x: # trailing condition comment, do not indent
    body # trailing body comment, keep with body
    # trailing body comment, keep with body
# leading else comment, keep with else
else:  # trailing else comment, do not indent either
    print("else")

Since our AST is often lacking nodes for parts of the syntax, we often have to set comments as "dangling" in placements.rs and handle them manually in the node formatting. For the while case there is e.g. no node for the else, we get the or body as a Vec<Stmt>. That means we get comments once for the entire while-else statement and have to parse out the leading else comments. Similarly the : doesn't have any representation, so we mark trailing comments after a colon here.

[davidszotten]

thank you, i sort of followed that. will hopefully become more clear as i dig in more

[charliermarsh]

Thanks @davidszotten!