#! /bin/bash

HUGO_POSTS="${HUGO_POSTS:-"./posts"}"
RCLONE_ALIAS="${RCLONE_ALIAS:-"hugo"}"
RCLONE_CONFIG="${RCLONE_CONFIG:-./rclone.conf}"
RCLONE_FOLDER="${RCLONE_FOLDER:-"shared_from_bbarl64"}"
RCLONE_FOLDER_POSTS="${RCLONE_FOLDER_POSTS:-"posts"}"
RCLONE_FOLDER_META="${RCLONE_FOLDER_META:-"meta"}"
RCLONE_OPTS="${RCLONE_OPTS:-"--drive-shared-with-me"}"
TMPDIR="${TMPDIR:-"/tmp"}"

main() {
   flags "$@"
   posts
   meta
}

flags() {
   set -e
   set -o pipefail
   cd "$(dirname "$BASH_SOURCE")"
}

posts() (
   list() {
      gd list_posts
   }

   up_to_date() {
      local metadatad="$(metadatad "$1")"
      if [ -f "$metadatad" ]; then
         if [ "$1" == "$(cat "$metadatad")" ]; then
            return 0
         fi
      fi
      return 1
   }

   pulled() {
      echo "$TMPDIR/$(echo "$1" | jq -r .Name)"
   }

   pull() {
      local filename="$(echo "$json" | jq -r .Name)"
      gd pull_posts "$filename"
   }

   extracted() {
      local pulled_file="$(pulled "$1")"
      local date="$(echo "$1" | jq -r .ModTime)"
      date="${date%%T*}"
      local pulled_file_safe_basename="$(basename "$pulled_file" | sed 's/.zip$//' | sed 's/[^a-zA-Z0-9]/_/g')"
      echo "$TMPDIR/${date}_${pulled_file_safe_basename}"
   }

   extract() {
      local extracted_dir="$(extracted "$1")"
      local index_html="$extracted_dir/.index.html"
      local index_md="$extracted_dir/index.md"
      local pulled_file="$(pulled "$1")"
      local u_date_iso="$(echo "$1" | jq -r .ModTime)"

      if [ -d "$extracted_dir" ]; then
         rm -rf "$extracted_dir"
      fi
      mkdir -p "$extracted_dir"
      7z x -o"$extracted_dir" "$pulled_file"

      cat "$extracted_dir"/*.html | html_only_body > "$index_html"
      local tags=($(cat "$index_html" | html_to_plaintext | grep -o '#[a-zA-Z0-9]*' | grep '[a-zA-Z]' | sed 's/^#//' | sort -u))
      local tags_csv="$(first=false; for tag in "${tags[@]}"; do true; if $first; then echo -n ", "; fi; first=true; echo -n "$tag"; done)"

      printf '
         ---
         title: "%s"
         date: %s
         draft: false
         tags: [%s]
         ---
         ' \
         "$(basename "$pulled_file" | sed 's/.zip$//' | sed 's/"/\\"/g')" \
         "$u_date_iso" \
         "$tags_csv" \
         | sed 's/^[ ]*//' \
         | grep . \
         > "$index_md"

      cat "$index_html" >> "$index_md"
      echo "" >> "$index_md"
      rm "$index_html" "$extracted_dir"/*.html

      for ext in png jpg jpeg gif JPG; do find "$extracted_dir" -name "*.$ext"; done | while read -r line; do
         convert "$line" -ordered-dither o8x8,8,8,4 "$line.2"
         mv "$line.2" "$line"
      done
   }

   imported() {
      local extracted_dir="$(extracted "$1")"
      echo "$HUGO_POSTS/$(basename "$extracted_dir")"
   }

   import() {
      local extracted_dir="$(extracted "$1")"
      local target_dir="$(imported "$1")"
      if [ -d "$target_dir" ]; then
         rm -rf "$target_dir"
      fi
      mv "$extracted_dir" "$target_dir"
   }

   metadatad() {
      local imported="$(imported "$1")"
      echo "$imported"/.metadata.json
   }

   metadata() {
      local metadatad="$(metadatad "$1")"
      echo "$1" > "$metadatad"
   }

   local want=()
   for b64_json in $(list); do
      local json="$(echo "$b64_json" | base64 --decode)"
      local filename="$(echo "$json" | jq -r .Name)"
      want+=("$(imported "$json")")
      if up_to_date "$json"; then
         log "$filename: up to date"
         continue
      fi
      log "$filename: pull"
      pull "$json"
      log "$filename: extract"
      extract "$json"
      log "$filename: import"
      import "$json"
      log "$filename: metadata"
      metadata "$json"
      rm "$(pulled "$json")"
   done
   for d in "$HUGO_POSTS"/*; do
      if  [[ ! " ${want[*]} " =~ " $d " ]]; then
         log "$d: stale, pruning"
         rm -rf "$d"
      fi
   done
)

meta() {
   list() {
      gd list_meta
   }

   pull() {
      local filename="$(echo "$json" | jq -r .Name)"
      gd pull_meta "$filename"
   }

   pulled() {
      echo "$TMPDIR/$(echo "$1" | jq -r .Name)"
   }

   meta_file() {
      echo "./meta/.overrides/$(echo "$1" | jq -r .Name)/.meta"
   }

   up_to_date() {
      test -f "$(meta_file "$1")" && test "$1" == "$(cat "$(meta_file "$1")")" && test -e "$(must "$1")"
   }

   extracted_dir() {
      echo "./meta/.overrides/$(echo "$1" | jq -r .Name)"
   }

   extracted() {
      echo "./meta/.overrides/$(echo "$1" | jq -r .Name | sed 's/.zip$//')"
   }

   must() {
      echo "./meta/$(basename "$(extracted "$1")")"
   }

   extract() {
      local extracted_dir="$(extracted_dir "$1")"
      rm -rf "$extracted_dir" "${extracted_dir%.zip}"
      mkdir -p "$extracted_dir"
      7z x -o"$extracted_dir" "$(pulled "$1")"
      local extracted_file="$(ls "$extracted_dir"/*.html)"
      if [ $(ls "$extracted_dir" | wc -l) == 1 ]; then
         cp "$extracted_file" "$(extracted "$1")"
         extracted_file="$(extracted "$1")"
      else
         mv "$extracted_file" "$(dirname "$extracted_file")"/index.md
         mv "$extracted_dir" "$(extracted "$1")"
      fi
      local extracted="$(extracted "$1")"
      case "${extracted##*.}" in
         yaml )   true     ;;
         * )      return   ;;
      esac
      cat "$extracted" | html_to_plaintext > "$extracted.2"; mv "$extracted.2" "$extracted"
   }

   merge() {
      local extracted="$(extracted "$1")"
      case "${extracted##*.}" in
         yaml ) merge_yaml "$1" ;;
         * )
      esac
   }

   merge_yaml() {
      local override="$(extracted "$1")"
      local default="$(echo "$override" | sed 's/\.overrides/\.default/')"
      if [ ! -f "$default" ]; then
         return
      fi
      cat "$override" | yq -j eval - | jq > "$override.json"
      cat "$default" | yq -j eval - | jq > "$default.json"
      jq -s '.[0] * .[1]' "$default.json" "$override.json" | yq -P eval - > ./meta/"$(basename "$override")"
   }

   import() {
      local must="$(must "$1")"
      if [ -f "$must" ]; then
         rm "$must"
      fi
      merge "$1"
      if [ -f "$must" ]; then
         return
      fi
      log cp -r "${must%/*}/.overrides/${must##*/}" "$must"
      cp -r "${must%/*}/.overrides/${must##*/}" "$must"
   }

   for b64_json in $(list); do
      local json="$(echo "$b64_json" | base64 --decode | jq .)"
      local filename="$(echo "$json" | jq -r .Name)"
      if up_to_date "$json"; then
         log "$filename: up to date"
         continue
      fi
      log "$filename: pulling"
      pull "$json"
      log "$filename: extracting"
      extract "$json"
      import "$json"
      echo "$json" > "$(meta_file "$json")"
   done
}

fatal() {
   log "$@"
   exit 2
}

log() {
   echo "$(date +%H:%M:%S) > $*" >&2
}

gd() (
   list_posts() {
      gd list "$RCLONE_FOLDER_POSTS"
   }
   list_meta() {
      gd list "$RCLONE_FOLDER_META"
   }
   list() {
      rc lsjson "$RCLONE_ALIAS:$RCLONE_FOLDER/$1" \
         | jq -c .[] \
         | grep -E 'zip"' \
         | while read -r line; do
         echo "$line" | base64
      done
   }
   pull_posts() {
      pull "$RCLONE_FOLDER_POSTS/$1"
   }
   pull_meta() {
      pull "$RCLONE_FOLDER_META/$1"
   }
   pull() {
      rc copy "$RCLONE_ALIAS:$RCLONE_FOLDER/$1" "$TMPDIR"/
   }
   rc() {
      rclone \
         --config "$RCLONE_CONFIG" \
         --fast-list \
         $RCLONE_OPTS \
         "$@"
   }

   "$@"
)

html_to_plaintext() {
   sed 's/<style.*<\/style>//' | sed 's/&nbsp;/ /g' | sed -e 's/<[^>]*>/\n/g' | grep .
   #sed 's/<[^>]*>//g'
}

html_only_body() {
   sed 's/.*<body/<body/' \
   | sed 's/<\/body>.*/<\/body>/'
}

if [ "$0" == "$BASH_SOURCE" ]; then
   main "$@"
fi