Fix (doc-)line-too-long start location

[MichaReiser]

I noticed in the byte offset refactor that the (doc-)line-too-long start range can point past the end of the line if the line contains characters with a width greater or equal to two. This is because we set the start location's column to limit instead of computing the character at which the line width is exceeded.

The current range can cause panics when printing the diagnostics because location can be greater than end_location which panics when slicing into the text string (the byte offset refactor will catch this earlier because it uses TextRange that enforces this invariant).

[MichaReiser]

Current dependencies on/for this PR:

• main
  • PR Fix (doc-)line-too-long start location #4006 👈

This comment was auto-generated by Graphite.

[MichaReiser]

We could use this as a chance to improve the diagnostic. Any recommendations for a replacement term for characters that captures that we compare the width and not characters count?

[github-actions]

PR Check Results

Ecosystem

ℹ️ ecosystem check detected changes. (+6, -6, 0 error(s))

zulip (+6, -6)

- zerver/tests/test_message_fetch.py:2275:101: E501 Line too long (123 > 100 characters)
+ zerver/tests/test_message_fetch.py:2275:92: E501 Line too long (123 > 100 characters)
- zerver/tests/test_message_fetch.py:2443:101: E501 Line too long (110 > 100 characters)
+ zerver/tests/test_message_fetch.py:2443:89: E501 Line too long (110 > 100 characters)
- zerver/webhooks/slack_incoming/tests.py:149:101: E501 Line too long (191 > 100 characters)
+ zerver/webhooks/slack_incoming/tests.py:149:103: E501 Line too long (191 > 100 characters)
- zerver/webhooks/slack_incoming/tests.py:151:101: E501 Line too long (217 > 100 characters)
+ zerver/webhooks/slack_incoming/tests.py:151:103: E501 Line too long (217 > 100 characters)
+ zerver/webhooks/slack_incoming/tests.py:155:100: E501 Line too long (211 > 100 characters)
- zerver/webhooks/slack_incoming/tests.py:155:101: E501 Line too long (211 > 100 characters)
- zilencer/management/commands/populate_db.py:872:101: E501 Line too long (106 > 100 characters)
+ zilencer/management/commands/populate_db.py:872:95: E501 Line too long (106 > 100 characters)

Benchmark

Linux

group                                      main                                    pr
-----                                      ----                                    --
linter/all-rules/large/dataset.py          1.00     17.4±1.07ms     2.3 MB/sec     1.06     18.5±1.04ms     2.2 MB/sec
linter/all-rules/numpy/ctypeslib.py        1.00      4.2±0.23ms     4.0 MB/sec     1.05      4.4±0.27ms     3.8 MB/sec
linter/all-rules/numpy/globals.py          1.03   574.4±38.69µs     5.1 MB/sec     1.00   555.3±33.52µs     5.3 MB/sec
linter/all-rules/pydantic/types.py         1.00      7.1±0.38ms     3.6 MB/sec     1.06      7.6±0.39ms     3.4 MB/sec
linter/default-rules/large/dataset.py      1.00      8.7±0.43ms     4.7 MB/sec     1.15     10.0±0.62ms     4.1 MB/sec
linter/default-rules/numpy/ctypeslib.py    1.00  1990.1±137.15µs     8.4 MB/sec    1.05      2.1±0.12ms     7.9 MB/sec
linter/default-rules/numpy/globals.py      1.00   242.5±13.53µs    12.2 MB/sec     1.02   246.8±17.48µs    12.0 MB/sec
linter/default-rules/pydantic/types.py     1.00      4.1±0.28ms     6.2 MB/sec     1.08      4.5±0.29ms     5.7 MB/sec
parser/large/dataset.py                    1.00      6.2±0.32ms     6.6 MB/sec     1.06      6.6±0.41ms     6.2 MB/sec
parser/numpy/ctypeslib.py                  1.04  1381.8±103.98µs    12.1 MB/sec    1.00  1322.7±85.90µs    12.6 MB/sec
parser/numpy/globals.py                    1.09    143.3±7.18µs    20.6 MB/sec     1.00   131.7±11.99µs    22.4 MB/sec
parser/pydantic/types.py                   1.12      3.1±0.25ms     8.1 MB/sec     1.00      2.8±0.18ms     9.1 MB/sec

Windows

group                                      main                                   pr
-----                                      ----                                   --
linter/all-rules/large/dataset.py          1.00     16.6±0.13ms     2.5 MB/sec    1.01     16.7±0.16ms     2.4 MB/sec
linter/all-rules/numpy/ctypeslib.py        1.00      4.3±0.06ms     3.8 MB/sec    1.01      4.4±0.07ms     3.8 MB/sec
linter/all-rules/numpy/globals.py          1.00   450.8±11.22µs     6.5 MB/sec    1.01   454.7±28.86µs     6.5 MB/sec
linter/all-rules/pydantic/types.py         1.00      7.1±0.09ms     3.6 MB/sec    1.01      7.2±0.12ms     3.5 MB/sec
linter/default-rules/large/dataset.py      1.00      8.4±0.07ms     4.8 MB/sec    1.01      8.5±0.09ms     4.8 MB/sec
linter/default-rules/numpy/ctypeslib.py    1.00  1819.8±16.32µs     9.2 MB/sec    1.02  1857.6±24.53µs     9.0 MB/sec
linter/default-rules/numpy/globals.py      1.00    190.0±3.38µs    15.5 MB/sec    1.02    194.4±5.79µs    15.2 MB/sec
linter/default-rules/pydantic/types.py     1.00      3.8±0.05ms     6.7 MB/sec    1.02      3.9±0.06ms     6.5 MB/sec
parser/large/dataset.py                    1.00      6.7±0.05ms     6.1 MB/sec    1.00      6.7±0.06ms     6.1 MB/sec
parser/numpy/ctypeslib.py                  1.00  1286.4±15.52µs    12.9 MB/sec    1.00  1292.2±18.08µs    12.9 MB/sec
parser/numpy/globals.py                    1.00    128.2±1.98µs    23.0 MB/sec    1.00    128.5±1.75µs    23.0 MB/sec
parser/pydantic/types.py                   1.00      2.9±0.03ms     8.9 MB/sec    1.01      2.9±0.04ms     8.9 MB/sec

[charliermarsh]

I wonder if we should move this computation into is_overlong somehow, so that it's guaranteed to yield a consistent result with line_width?

[MichaReiser]

Done. is_overlong now does all the calculations and returns an Option<Overlong> to give the caller access to the computed range and width.