Regular expression typo false positive

[pdostal]

Hello,

I've this case:

error: `ot` should be `to`, `of`, `or`
  --> ./lib/qam.pm:71:44
   |
71 |     if ($patch_status =~ /Status\s*:\s+[nN]ot\s[nN]eeded/) {
   |                                            ^^

Would that be possible to filter those out?

[epage]

Similar to #643, our main two routes are

• Heuristics.
  • We need to distinguish from paths when doing so
  • Not all regex engines have leading, trailing /
• Mark sections for ignoring like in Exclude specific line #316

[epage]

FYI #695 provides a new workaround for false positives

[kdeldycke]

To illustrate this issue with more test cases, here is another false positive of a regular expression encountered in a markdown document (as produced by typos-cli 1.20.8):

error: `ba` should be `by`, `be`
  --> ./content/2011/postgresql-commands.md:47:95
   |
47 |   $ psql --tuples-only --no-align -d database_id -c "SELECT id FROM res_users;" | sed ':a;N;$!ba;s/\n/ /g'
   |                                                                                               ^^
   |

In my case, the fix consisted in adding the following configuration to my pyproject.toml:

[tool.typos]
default.extend-ignore-identifiers-re = [
    "ba",
]

[epage]

Personally, I would recommend using default.extend-ignore-re to look for the pattern of your style of regexes and avoid checking them completely, rather than playing whack-a-mole with specific identifiers within a regex.

When handling identifiers, I would recommend to instead use

[default.extend-identifiers]
ba = "ba"

[kdeldycke]

Personally, I would recommend using default.extend-ignore-re

Ah yes, thanks @epage for the precision. It's better indeed. It just took me while to understand the different stages of tokenization process of typos and the influence of its parameters on that.

In the end I fine-tuned typos with the following config:

[tool.typos]
default.extend-ignore-re = [
    "!ba;",
]