Confluence HTML export - Find all macros used per page

find-confluence-macros-by-page.sh

grep -I -oH '<[^>]*macro[^>]*>' ./*.html \
| grep -Fv 'src=confmacro' \
| awk -F: '
{
  file=$1
  tag=$0
  sub(/^[^:]*:/, "", tag)

  out=tag

  if (match(tag, /class="[^"]*macro[^"]*"/)) {
    out=substr(tag, RSTART+7, RLENGTH-8)
  } else if (match(tag, /class='\''[^'\'']*macro[^'\'']*'\''/)) {
    out=substr(tag, RSTART+7, RLENGTH-8)
  } else if (match(tag, /aria-label="[^"]*"/)) {
    out=substr(tag, RSTART+12, RLENGTH-13)
  } else if (match(tag, /[[:alnum:]_-]*macro[[:alnum:]_-]*/)) {
    out=substr(tag, RSTART, RLENGTH)
  }

  gsub(/rbtoc[0-9]+/, "rbtoc", out)

  key=file SUBSEP out
  if (seen[key]++) next

  if (file != last_file) {
    if (last_file != "") print ""
    print file
    last_file=file
  }

  print "  - " out
}'