#!/usr/bin/env nu

# Linkify ADR mentions in outreach/web/src/*.html.
#
# Two passes per file:
#   1. Fix href format: absolute https://ontoref.dev/adr/adr-NNN → site-relative /adr/NNN
#   2. Wrap plain ADR-NNN text (not already inside <a ...>...</a>) with a link
#
# All hrefs use site-relative paths (/adr/NNN) so links work in both localhost
# and production without domain changes.
# Single-quotes are used for injected href/title/class attributes so the link
# is safe inside both text nodes and double-quoted data-en/data-es attributes.
#
# Usage (from outreach/site):
#   nu scripts/build/linkify-adrs.nu
#   nu scripts/build/linkify-adrs.nu --map site/public/r/adr-map.json --web ../../outreach/web --daemon https://ontoref.dev

def escape-attr [s: string] {
  $s
  | str replace --all '&' '&amp;'
  | str replace --all '<' '&lt;'
  | str replace --all '>' '&gt;'
  | str replace --all "'" '&apos;'
  | str replace --all '"' '&quot;'
}

def main [
  --map: string = "site/public/r/adr-map.json"
  --web: string = "../../outreach/web"
  --daemon: string = "https://ontoref.dev"   # used only to recognise existing absolute hrefs to fix
] {
  let adr_map = (open $map)
  let web_dir = ($web | path expand)
  let src_dir = ($web_dir | path join "src")

  if not ($src_dir | path exists) {
    error make { msg: $"src dir not found: ($src_dir)" }
  }

  let files = (glob $"($src_dir)/*.html" | sort)
  mut total_files = 0
  mut total_changes = 0

  for f in $files {
    mut content = (open --raw $f)
    let original = $content

    # Sort ADRs longest-id-first so adr-010 doesn't shadow adr-0100 (hypothetical),
    # and to process higher numbers first (safer for text replacement ordering).
    let adrs = ($adr_map | transpose key val | sort-by key --reverse)

    for entry in $adrs {
      let id = $entry.key       # "adr-009"
      let val = $entry.val
      let num = ($id | str replace 'adr-' '')    # "009"
      let adr_upper = $"ADR-($num)"              # "ADR-009"
      let path = $"/adr/($num)"                  # "/adr/009" — site-relative, works everywhere
      let safe_name = (escape-attr ($val.name? | default ""))

      # Pass 1: fix existing absolute hrefs to the canonical site-relative path.
      # Handles double-quoted (original manual links) and single-quoted (injected links).
      $content = ($content | str replace --all $"href=\"($daemon)/adr/($id)\"" $"href=\"($path)\"")
      $content = ($content | str replace --all $"href=\"($daemon)/adr/($num)\"" $"href=\"($path)\"")
      $content = ($content | str replace --all $"href='($daemon)/adr/($id)'" $"href='($path)'")
      $content = ($content | str replace --all $"href='($daemon)/adr/($num)'" $"href='($path)'")

      # Pass 2: linkify plain ADR-NNN not already followed by </a>
      # Uses single-quote attrs: safe inside double-quoted HTML attribute values.
      let link = $"<a href='($path)' title='($safe_name)' class='adr-ref' target='_blank' rel='noopener'>($adr_upper)</a>"
      let pattern = $"ADR-($num)\(?!</a>\)"
      $content = ($content | str replace --all --regex $pattern $link)
    }

    if $content != $original {
      $content | save -f $f
      let linked_after = ($content | find --regex "ADR-[0-9]{3}</a>" | length)
      print $"  ($f | path basename): updated"
      $total_files = ($total_files + 1)
      $total_changes = ($total_changes + 1)
    }
  }

  print $"linkify-adrs: ($total_files) files updated"
}