bk2md.awk

awk

#!/usr/bin/awk -f

# Usage: ./bk2md bookmarks.html

# Usa arquivo bookmarks.html por padrão
BEGIN {
  if (ARGC == 1) {
    ARGV[1] = "bookmarks.html"
    ARGC = 2
  }
}

# Título
/<TITLE/ {
  h1 = gensub(/<TITLE>([^<]+).*$/, "\\1", 1)
  print "# " h1
}

# Subtítulo
/^[[:space:]]{4}<DT><H3/ {
  h2 = gensub(/[[:space:]]+<DT><H3.+">([^<]+).*$/, "\\1", "g")
  printf "\n## %s \n\n", h2
}

# Links
/DT><A / {
  title= gensub(/^[ \t]*<DT><A HREF=".+>([^<]+).*$/, "\\1", "g")
  link = gensub(/^.*<DT><A HREF="([^"]+)".*$/, "\\1", "g")
  printf "- [%s] (%s)\n", title, link
}

# ft=awk