Kebab-case tags are not considered. Unicode tags are not considered.

[endusol]

Kebab-cased tags are not considered by current regexps. Tag for test: #spam-spam.

Unicode tags are not considered by current regexps. Tag for test: #спам_спам.

Tags that start with digit are not considered. Of course full-digits tags are not considered by Obsidian as well, but seems that tags that just start with digits, should be. Tag for test: #123a.

[endusol]

Please, see the note:

#спам
#spam-spom
`#inblock`
#123a #a123 #123

Note, what tags are considered by Obsidian itself:

[endusol]

This note seems good for testing:

# Valid tags

#- #_ - Single dash/underscore.
#kebab-notation #snake_notation - Kebab/Snake.
#спам - Unicode
#123a #123- #123_ - Digits at the start.
#parent/child - Nested.
\<#non_html - Not a part of an HTML tag.

# Non-tags

#123 - Fully digital.
/#slashed - Slashed.
\#backslahed - Backslashed.
<span tag='#html'>spam</span> - Part of HTML tag.
(#braced) - Braced.
`#blocked` - Code-block.

Please, see the below code:

import pathlib
import pandas as pd
import obsidiantools.api as otools

path = pathlib.Path('..').resolve()
vault = otools.Vault(path).connect(attachments=True, show_nested_tags=True).gather()

_tags = lambda n: {*vault.tags_index[n], *vault.front_matter_index[n].get('tags', [])}
_row = lambda n, p: {'path': p, 'tags': _tags(n)}
df = pd.DataFrame([_row(n, p) for n, p in vault.md_file_index.items()])

for _ in df.iloc[0].tags:
    print(_)

This is the output:

parent/child
_\-
kebab-notation
non_html
#frontmatter
snake_notation
slashed

As you can see, not all the tags are considered.

Now, try to disable the nested tags flag and run the code again.

[endusol]

Please, see the following RegExps:

# Fixes for kebab-case and unicode tags:
CHAR_TM = r'(?:[\/-]|[^\W\d])'  # Tag-Making charachters: slash, hyphen, or any unicode letter.
CHAR = r'[\/\w_-]'              # Slash, hyphen, underscore, or any alphanumeric.
NAME_IT = r'(?<!\\\(\/)'        # Как это обозвать?
_constants.TAG_INCLUDE_NESTED_REGEX = fr'{NAME_IT}#(?:{CHAR_TM}+{CHAR}*|{CHAR}*{CHAR_TM}+)'
_constants.TAG_MAIN_ONLY_REGEX = fr'{NAME_IT}#(?:{CHAR_TM}+{CHAR}*|{CHAR}*{CHAR_TM}+)'

Please, see the output:

#snake_notation
#frontmatter
#-
#123_
#/
#123-
#/noroot
#123a
#slashed
#braced
#parent/child
#_
#спам
#non_html
#kebab-notation

[endusol]

Current workaround:

from obsidiantools import _constants, md_utils

# Fixes for kebab-case and unicode tags:
CHAR_TM = r'(?:[\/-]|[^\W\d])'     # Tag-Making charachters: slash, hyphen, or any unicode letter.
CHAR = r'[\/\w_-]'                 # Slash, hyphen, underscore, or any alphanumeric.
FORBIDDEN_PREFIX = r'(?<!\\\(\/)'
TAG_REGEX = fr'{FORBIDDEN_PREFIX}#(?:{CHAR_TM}+{CHAR}*|{CHAR}*{CHAR_TM}+)'
setattr(_constants, 'TAG_INCLUDE_NESTED_REGEX', TAG_REGEX)
setattr(_constants, 'TAG_MAIN_ONLY_REGEX', TAG_REGEX)
setattr(md_utils, 'TAG_INCLUDE_NESTED_REGEX', TAG_REGEX)
setattr(md_utils, 'TAG_MAIN_ONLY_REGEX', TAG_REGEX)

# Proceed as always:
pass

mb its better to just store one regex and let users filter tags afterwards instead of having the show_nested_tags parameter?