Regex order/options to allow beginning delimiter

[keeganjk]

Hello, my current goal for how files should look is number-track_name, that is, there is an initial number for the year or track, a hyphen as a delimiter, and then the title for the album or track. Is there any way I can perform a replace only on e.g. $title and $album, or is there a regular expression solution?

This is what my replace section currently looks like:

replace:
    '[\\/]': _
    '[\x00-\x1f]': _
    '[<>:\!\?\*\|_]': _
    '-+': _
    '^(\d+)\s+': '\1-'
    '\s+': _
    '[:,]': ''
    '[^a-z0-9_-]': ''

'^(\d+)\s+': '\1-' is supposed to preserve the initial number place a hyphen, but it looks like these hyphens are being replaced with underscores by '-+': _. What could I possibly do about this?

[sampsyo]

Hmm… that's an interesting problem; it's not exactly a use case that we've built the replace feature for. That is, its original purpose is to "legalize" filenames, in the sense that it matches illegal characters that the filesystem can't support well and replaces them with other stuff. Having one rule that produces characters that another rule classifies as illegal characters is not something we anticipated. The code for filename sanitation repeatedly applies the rules for exactly this kind of situation: where some rule might produce an invalid filename that needs to be further cleaned up by other rules.

I wonder if a custom plugin might be the best way to do this? Not entirely sure what a good route would be, to be honest, unless it's possible to make that -+ rule more specific.

[keeganjk]

Interesting, that makes sense. I am trying to "legalize" this to strict regulations (maybe unnecessarily), that is, allowing only [a-z0-9_.-] where hyphen separates "items" (e.g. year or track number) and the dot is reserved for the file extension, so that leaves the underscore specifically for spaces within titles. Maybe doing this is more complicated than I thought, and/or unnecessary.

Though yes, if the regex I am currently trying is not possible, making a more detailed regex for -+ (to fit the case where it is not preceeded by ^\d+, which mostly works except for + within a reverse-lookback like (?<!\d+)) would be an ideal solution.

[wisp3rwind]

Hello, my current goal for how files should look is number-track_name, that is, there is an initial number for the year or track, a hyphen as a delimiter, and then the title for the album or track. Is there any way I can perform a replace only on e.g. $title and $album, or is there a regular expression solution?

This is what my replace section currently looks like:

replace:
    '[\\/]': _
    '[\x00-\x1f]': _
    '[<>:\!\?\*\|_]': _
    '-+': _
    '^(\d+)\s+': '\1-'
    '\s+': _
    '[:,]': ''
    '[^a-z0-9_-]': ''

'^(\d+)\s+': '\1-' is supposed to preserve the initial number place a hyphen, but it looks like these hyphens are being replaced with underscores by '-+': _. What could I possibly do about this?

I'm having a hard time following what exactly you try to do here, since this is missing your configuration for context (in particular the paths section) and you don't really describe your actual goal.

That being said, this sounds like the inline plugin should allow you to define sanitized fields, using any python code you like.

[keeganjk]

The rest of my configuration is quite simple; the only other field in use is paths. Here is the full scope:

directory: /media/hdd2/aud/
plugins: inline convert

paths: 
    default: %lower{%asciify{$albumartist/$year $album/$track $title}}
    comp: %lower{%asciify{compilations/$year $album%aunique{}/$track $title}}

replace:
    '[\\/]': _
    '[\x00-\x1f]': _
    '[<>:\!\?\*\|_]': _
    '-+': _
    '^(\d+)\s+': '\1-'
    '\s+': _
    '[:,]': ''
    '[^a-z0-9_-]': ''

I left spaces in paths as a "placeholder" that would separately be replaced with a hyphen.

My goal is to restrict filenames to simple ASCII characters: lowercase alphanumeric, the underscore for spaces (which, not being a delimiter, make sense for use within an item like title (for organizational reasons, and out of confidence that this will lead to no problems in compatibility or strange characters)), the dot being reserved for file extensions (not making much sense elsewhere), and the hyphen, a delimiter, being used to separate individual items (like track and title). Though maybe it does not make sense to do this, or it is useless (one should be able to simply tell that the first number indicates the track, though it seems to logically make sense to me to format this way).

inline seems promising for what I am trying to do; I will give this a shot.

[keeganjk]

I have included inline and the following seems to bring forth a solution for the problem I described:

directory: /media/hdd2/aud/
plugins: inline convert

# _f: _formatted
item_fields:
  album_f: album.replace('-', "_")
  title_f: title.replace('-', "_")
  albumartist_f: albumartist.replace('-', "_")

paths: 
  default: %lower{%asciify{$albumartist_f/$year-$album_f/$track-$title_f}}
  comp: %lower{%asciify{compilations/$year-$album_f%aunique{}/$track-$title_f}}

replace:
  '[\\/]': _
  '[\x00-\x1f]': _
  '[<>:\!\?\*\|_]': _
  '\s+': _
  '_+': _
  '[^a-z0-9_-]': ''