[Bug]: Case insensitive attribute selector is escaped incorrectly

[Merri]

Describe the bug

Given this selector: [data--~="is½" i] {}

Used with HTML like this: <span data--="myThing is½"> or <span data--="MYTHING IS½">.

The processed output selector [data--~=is\BD i] {} will not match due to lacking whitespace.

That output is understood as equivalent of [data--~="is½i"]. Notice the letter i has moved from outside the quotes to inside.

Expected behaviour

The selector should match even after processing.

So either don't escape the ½ and other similar characters, or add one extra whitespace to the output after the escape sequence so that the insensitivity indicator doesn't get understood as part of the string: [data--~=is\BD i] {}

Just in case it isn't obvious; two space characters are required to keep the insensitivity indicator separate.

Steps to reproduce

Run minify with any CSS file that has ½ (or other special character) as the last character of a case insensitive attribute selector.

Version

6.0.1

Preset

(no preset)

Environment

System:
    OS: Windows 10 10.0.19045
    CPU: (16) x64 AMD Ryzen 7 4800HS with Radeon Graphics
    Memory: 6.94 GB / 15.42 GB

Package details

+-- cssnano@6.0.1
| +-- cssnano-preset-default@6.0.1
| | +-- css-declaration-sorter@6.4.1
| | | `-- postcss@8.4.29 deduped
| | +-- cssnano-utils@4.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-calc@9.0.1
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-colormin@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-convert-values@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-discard-comments@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-discard-duplicates@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-discard-empty@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-discard-overridden@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-merge-longhand@6.0.0
| | | +-- postcss@8.4.29 deduped
| | | `-- stylehacks@6.0.0
| | |   `-- postcss@8.4.29 deduped
| | +-- postcss-merge-rules@6.0.1
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-minify-font-values@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-minify-gradients@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-minify-params@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-minify-selectors@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-charset@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-display-values@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-positions@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-repeat-style@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-string@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-timing-functions@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-unicode@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-url@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-normalize-whitespace@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-ordered-values@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-reduce-initial@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-reduce-transforms@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-svgo@6.0.0
| | | `-- postcss@8.4.29 deduped
| | +-- postcss-unique-selectors@6.0.0
| | | `-- postcss@8.4.29 deduped
| | `-- postcss@8.4.29 deduped
| `-- postcss@8.4.29 deduped
+-- postcss-cli@10.1.0
| +-- postcss-load-config@4.0.1
| | `-- postcss@8.4.29 deduped
| +-- postcss-reporter@7.0.5
| | `-- postcss@8.4.29 deduped
| `-- postcss@8.4.29 deduped
`-- postcss@8.4.29

Additional context

No response

[ludofischer]

Definitely a bug here, but the selector is hopefully fairly unusual.