#!/bin/bash
function fetch_blogs () {
  cd -- "/var/www/www.cozynet.org/html/"
  id='blog_cats'
  cd blogs/
  parse_blogs && echo "Completed."
}
function parse_blogs () {
  for file in *_blog.html;
  do
    [[ -f "$file" ]] || continue
    blog_cats=$(cat "${file}" | head -n13 | grep "\<meta " | grep categories | cut -d= -f3 | tr -d '"' | tr -d '>' | sed 's/, /,/g')
    blog_title=$(cat "${file}" | head -n13 | grep \<title\> | sed 's|  <title>||g' | sed 's|</title>||g')
    # Using a pipe "|" for a delimiter since colons and semi-colons are already being used. The pipe will be converted to a new line later.
    blog_index+=$(echo "${file}:${blog_cats}:${blog_title}|")
  done
  parse_cats
}
function parse_cats {
  contents=("${blog_index[0]}")
  all_cats=$(echo "${contents}" | tr '|' '\n' | cut -d':' -f'2' | tr ',' '\n' | sed '/^$/d' | sort)
  uniq_cats=$(echo "${all_cats}" | uniq)
  declare -a catsArray
  while read -r line
  do
    catsArray+=("${line}")
  done <<< "${uniq_cats}"
  for line in "${catsArray[@]}"
  do
    cat_count=$(echo "${all_cats}" | grep -w "${line}" | wc -l)
    cat_stat+=("${line} (${cat_count});")
    links=$(echo "${contents}" | tr '|' '\n' | cut -d':' -f'1,2' | grep -w "${line[@]}" | cut -d':' -f'1' | tr '\n' ',')
    for link_line in `echo "${links}"`
    do
      parse_links=$(echo "${link_line}" | tr ',' '\n')
      for index_link in `echo "${parse_links}"`
      do
        raw_link+=$(echo "${contents}" | tr '|' '\n' | cut -d':' -f'1,3' | grep -w "${index_link}" | sed 's/&amp;/\&/g' | sed 's/$/†/')
      done
      raw_links+=$(echo "${raw_link};" | sed 's/†;/;/g')
      raw_link=''
    done
    cat_links+=$(echo "${links};" | sed 's/,;/;/g')
  done
  update_db
}
function update_db () {
  touch "/var/www/www.cozynet.org/db/${id}_index.db"
  # Clear the file
  :> "/var/www/www.cozynet.org/db/${id}_index.db"
  echo "${cat_stat[*]}" | sed 's/;$//g' >> "/var/www/www.cozynet.org/db/${id}_index.db"
  echo "${cat_links[*]}" | sed 's/;$//g' >> "/var/www/www.cozynet.org/db/${id}_index.db"
  echo "${raw_links[*]}" >> "/var/www/www.cozynet.org/db/${id}_index.db"
}
fetch_blogs