ontoref-outreach/site/scripts/build/linkify-site.nu

#!/usr/bin/env nu

# Linkify site conventions in blog markdown source:
#
#   ontoref / Ontoref →  <a href="/about" class="brand-site">…</a>   (brand mention, case-preserving)
#   ADR-NNN   →  <a href="/adr/NNN" class="adr-ref">ADR-NNN</a>     (decision deep-link)
#
# Operates on the MARKDOWN SOURCE (not the generated HTML) so the change survives
# every `just content` regeneration. Both rewrites are idempotent via lookarounds:
# an already-wrapped mention (preceded by `>` / followed by `</a>`) is left as-is.
#
# Skipped, so code and metadata never get a stray link:
#   - YAML frontmatter (between the leading --- delimiters)
#   - fenced code blocks (``` … ```)
#   - inline code spans (`…`)
#   - headings (lines beginning with #)
#   - compound identifiers: ontoref-daemon, ontoref/dist, .ontoref, /adr/038 (lookarounds)
#
# Usage (from outreach/site):
#   nu scripts/build/linkify-site.nu                       # default glob
#   nu scripts/build/linkify-site.nu site/content/blog/**/*.md
#   nu scripts/build/linkify-site.nu --check               # CI gate, no writes

const LINK = '<a href="/about" class="brand-site">${n}</a>'
const WORD = '(?<![>\w./-])(?<n>[Oo]ntoref)(?![<\w./-])'
const ADR_LINK = '<a href="/adr/${num}" class="adr-ref">ADR-${num}</a>'
const ADR_WORD = '(?<![>/\w-])ADR-(?<num>\d+)(?!</a>)'

# Apply both site rewrites outside inline-code spans on a single line.
def linkify-line [line: string]: nothing -> string {
  $line
  | split row '`'
  | enumerate
  | each {|it|
      if ($it.index mod 2) == 0 {
        $it.item
        | str replace --all --regex $WORD $LINK
        | str replace --all --regex $ADR_WORD $ADR_LINK
      } else {
        $it.item
      }
    }
  | str join '`'
}

def main [
  pattern: string = "site/content/blog/**/*.md"
  --check   # report posts that WOULD change and exit non-zero; never writes
] {
  let files = (glob $pattern | sort)
  mut changed = 0

  for f in $files {
    let lines = (open --raw $f | lines)
    mut out = []
    mut in_front = false
    mut in_fence = false
    mut idx = 0

    for line in $lines {
      let trimmed = ($line | str trim)

      # Frontmatter: a leading --- opens it, the next --- closes it.
      if $trimmed == "---" and not $in_fence {
        if $idx == 0 {
          $in_front = true
        } else if $in_front {
          $in_front = false
        }
        $out = ($out | append $line)
        $idx = ($idx + 1)
        continue
      }

      if ($trimmed | str starts-with '```') {
        $in_fence = (not $in_fence)
        $out = ($out | append $line)
        $idx = ($idx + 1)
        continue
      }

      if $in_front or $in_fence or ($trimmed | str starts-with '#') {
        $out = ($out | append $line)
        $idx = ($idx + 1)
        continue
      }

      $out = ($out | append (linkify-line $line))
      $idx = ($idx + 1)
    }

    let result = ($out | str join "\n")
    let original = (open --raw $f)
    # Preserve a trailing newline if the source had one.
    let final = if ($original | str ends-with "\n") { $result + "\n" } else { $result }

    if $final != $original {
      if $check {
        print $"  ($f | path basename): unlinked site mentions pending"
      } else {
        $final | save -f $f
        print $"  ($f | path basename): linked"
      }
      $changed = ($changed + 1)
    }
  }

  if $check {
    if $changed > 0 {
      error make { msg: $"linkify-site --check: ($changed) post\(s\) have unlinked site mentions — run `just content`" }
    }
    print "linkify-site --check: all posts clean"
  } else {
    print $"linkify-site: ($changed) file\(s\) updated"
  }
}