#!/bin/bash
# file: gsl__post_checkers
# Folder: /var/lib/gsl/scripts
# By echolib
# License: GNU AFFERO GENERAL PUBLIC LICENSE Version 3, 19 November 2007

#----------------------------------------------------------------------
# --------------------------------------------------
# From: gsl__loop_posts | MAIN checkers from modules
# --------------------------------------------------
#----------------------------------------------------------------------
gsl__all_checkers() {
# Init Stats
gsl_stat_abbr=0
gsl_stat_abbrs=0
gsl_stat_link=0
gsl_stat_links=0
gsl_stat_file=0
gsl_count_files=0
gsl_stat_image=0
gsl_stat_fcode=0

gsl_db_tmp=`mktemp`

gsl__check_h1 || return
gsl__post_content_only

gsl__check_type
gsl__check_headers
gsl__check_paragraphs "$gsl_tmp_post"
gsl__check_blockquote "$gsl_tmp_post"

gsl__do_header check "$gsl_marker_link" "$gsl_post"
gsl__do_header check "$gsl_marker_abbr" "$gsl_post"
gsl__do_header check "$gsl_marker_file" "$gsl_post"
gsl__do_header check "$gsl_marker_image" "$gsl_post"
gsl__do_header check "$gsl_marker_code" "$gsl_post"
gsl__stat_words "$gsl_tmp_post"
gsl__do_strongs_bolds stats "$gsl_tmp_post"
gsl__do_italics stats "$gsl_tmp_post"
gsl__do_icode stats "$gsl_tmp_post"

rm -f "$gsl_tmp_post"

echo -ne "\r\033[2K"

gsl__db_set_post_status chk
rm -f "$gsl_db_tmp"
}

#----------------------------------------------------------------------
# -------------------------------------------------
# Modules checkers
# -------------------------------------------------
#----------------------------------------------------------------------
#======================================================================
# Check Post for begining content
#======================================================================
gsl__check_h1() {
echo -ne ": Searching for begining #1..."

gsl_post_begin=`
grep -n "#1" $gsl_post \
    | head -1 \
    | awk -F: '{print $1}'`

if ! [[ "$gsl_post_begin" ]];then
  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_h1" \
    "Missing: #1 TITLE" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true
fi
}

#======================================================================
# Check/Set Post Type
#======================================================================
gsl__check_type() {
echo -ne "\r\033[2K: Searching for Post TYPE..."

case "$gsl_post_type" in
  page|post)
    gsl__logs_print \
      "$gsl_log_i" \
      "Content" \
      "Set as" \
      "$gsl_post_type: $gsl_post" \
      "${PWD}/$gsl_post"
    ;;
  *)
    gsl_post_type="post"
    sed -i  "1i$gsl_marker_type$gsl_post_type" \
            "$gsl_post" && \
    gsl__logs_print \
      "$gsl_log_w" \
      "Content" \
      "Set as" \
      "Default: $gsl_post_type for $gsl_post" \
      "${PWD}/$gsl_post"
    gsl__get_sum "$gsl_post"
    ;;
esac
}

#======================================================================
# Check for paragraphs ( and ) | $1: File
#======================================================================
gsl__check_paragraphs() {
echo -ne "\r\033[2K: Searching for Paragraphs..."

gsl_post_p_open=`
awk -v op="$gsl_mo_p" \
    '$1 == op' \
    "$1" \
  | wc -l`
      
gsl_post_p_close=`
awk -v cp="$gsl_mc_p" \
    '$1 == cp' \
    "$1" \
  | wc -l`

if (( "$gsl_post_p_open" == 0 ));then

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_loc_c_p" \
    "Missing: ( and ) at begining lines" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true
  return

fi

if (( "$gsl_post_p_open" == "$gsl_post_p_close" ));then

   gsl_stat_p="$gsl_post_p_open"
  
else

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_p" \
    "Mismatch: (=$gsl_post_p_open ; )=$gsl_post_p_close" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true
  return

fi
}

#======================================================================
# Check for blockquotes: --- | $1: File
#======================================================================
gsl__check_blockquote() {
echo -ne "\r\033[2K: Searching for BLOCKQUOTEs..."

gsl_post_blockquotes_nbr=`
awk -v bq="$gsl_mark_blockquote" \
    '$1 == bq' \
    "$1" \
  | wc -l`

gsl_post_blockquotes_nbr_r=$(( $gsl_post_blockquotes_nbr % 2 ))

if [[ "$gsl_post_blockquotes_nbr_r" -eq 0 ]];then

  gsl_stat_bq=$(( $gsl_post_blockquotes_nbr / 2 ))

else

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_cite" \
    "Mismatch: --- not paired" \
    "${PWD}/$gsl_post"
    gsl_checker_err=true
fi
}

#======================================================================
# From gsl__do_header: Check Link
#======================================================================
gsl__check_link() {
# Get & Check Header CONTENT
gsl__check_header_fields \
  "Content" "$gsl_log_h_link" "$gsl_log_act_link" \
|| return

echo -ne "\r\033[2K: Searching for Links... $gsl_post_hf_1"

# No ALT TEXT ?
gsl__check_header_field3 "Content" "$gsl_log_h_link" "$gsl_marker_link"

gsl_count_links=`
gsl__get_content_line "_$gsl_post_hf_1" "$gsl_post" | wc -l`

if (( "$gsl_count_links" > 0 ));then

  # Check URL
  gsl_url_status=`
  timeout 2 curl -o /dev/null --silent --head --write-out \
        '%{http_code}' \
        "$gsl_post_hf_2" 2>/dev/null || gsl_url_status=408`

  case "$gsl_url_status" in
    1*|2*|3*)
      gsl__logs_print \
        "$gsl_log_i" \
        "Content" \
        "$gsl_log_h_link" \
        "Online:$gsl_url_status ${gsl_post_hf_2:0:40}..." \
        "${PWD}/$gsl_post"
      ;;
    ""|*)
      gsl__logs_print \
        "$gsl_log_w" \
        "Content" \
        "$gsl_log_h_link" \
        "Offline:$gsl_url_status ${gsl_post_hf_2:0:40}..." \
        "${PWD}/$gsl_post"
      
      # Decomment below if you want the offline link avoiding convertion
      #gsl_checker_err=true
      gsl_checker_war=true
      return
      ;;
  esac

  # Stats
  ((gsl_stat_link++))
  gsl_stat_links=$(( gsl_stat_links + gsl_count_links ))

else

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_link" \
    "Unused: _$gsl_post_hf_1" \
    "${PWD}/$gsl_post"
    gsl_checker_err=true

fi
}

#======================================================================
# From gsl__do_header: Check ABBR
#======================================================================
gsl__check_abbr() {
gsl__check_header_fields \
  "Content" "$gsl_log_h_abbr" "$gsl_log_act_abbr" \
|| return

echo -ne "\r\033[2K: Searching for ABBRs... $gsl_post_hf_1"

gsl_count_abbrs=`
gsl__get_content_line "$gsl_post_hf_1" "$gsl_post" | wc -l`

if (( "$gsl_count_abbrs" > 0 ));then
  
  ((gsl_stat_abbr++))
  gsl_stat_abbrs=$(( gsl_stat_abbrs + gsl_count_abbrs ))

else

  # Missing CONTENT
  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_abbr" \
    "Missing: $gsl_post_header_field_1" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true

fi
}

#======================================================================
# From gsl__do_header: Check Files
#======================================================================
gsl__check_file() {
gsl__check_header_fields \
  "Content" "$gsl_log_h_file" "$gsl_log_act_file" \
|| return

echo -ne "\r\033[2K: Searching for Content Files... $gsl_post_hf_1"

gsl__check_header_field3 "Content" "$gsl_log_h_file" "$gsl_marker_file"

# File exists in folder
if ! [[ -f "$gsl_dir_domain_files/$gsl_post_hf_2" ]];then

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "Header $gsl_marker_file" \
    "Not found: $gsl_post_hf_2" \
    "$gsl_dir_domain_files/$gsl_post_hf_2"
  gsl_checker_err=true
  
else

# Register file to DB with hash
  gsl_file_csum=`
  cksum "$gsl_dir_domain_files/$gsl_post_hf_2" \
      | awk '{print $1}'`

  printf '%s\n' \
         "File:$gsl_dir_domain_images/$gsl_post_hf_2:$gsl_file_csum" \
      >> "$gsl_db_tmp"

fi

# Exists in Content
gsl_count_files=`
gsl__get_content_line "__$gsl_post_hf_1" "$gsl_post" | wc -l`

if (( $gsl_count_files > 0 ));then

  ((gsl_stat_file++))

else

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_file" \
    "Missing: __$gsl_post_hf_1" \
    "${PWD}/$gsl_post"
    gsl_checker_err=true
  
fi
}

#======================================================================
# From gsl__do_header: Check Files
#======================================================================
gsl__check_image() {
gsl__check_header_fields \
  "Content" "$gsl_log_h_image" "$gsl_log_act_image" \
|| return

echo -ne "\r\033[2K: Searching for Images... $gsl_post_hf_1"

gsl__check_header_field3 "Content" "$gsl_log_h_image" "$gsl_marker_image" \
|| return

# File exists in folder
if ! [[ -f "$gsl_dir_domain_images/$gsl_post_hf_2" ]];then

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_h_image" \
    "Not found: $gsl_post_hf_2" \
    "$gsl_dir_domain_images/$gsl_post_hf_2"
  gsl_checker_err=true
  
else

# Register file to DB with hash
  gsl_file_csum=`
  cksum "$gsl_dir_domain_images/$gsl_post_hf_2" \
      | awk '{print $1}'`
  
  printf '%s\n' \
         "Image:$gsl_dir_domain_images/$gsl_post_hf_2:$gsl_file_csum" \
      >> "$gsl_db_tmp"

fi

# Exists in Content
gsl_count_images=`
gsl__get_content_line "_image:$gsl_post_hf_1" "$gsl_post" | wc -l`

if (( $gsl_count_images > 0 ));then

  ((gsl_stat_image++))

else

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_image" \
    "Missing: _image:$gsl_post_hf_1" \
    "${PWD}/$gsl_post"
    gsl_checker_err=true
  
fi
}

#======================================================================
# From gsl__do_header: Check Files
#======================================================================
gsl__check_fcode() {

gsl__check_header_fields \
  "Content" "$gsl_log_h_code" "$gsl_log_act_code" \
|| return

echo -ne "\r\033[2K: Searching for File Codes... $gsl_post_hf_1"

gsl__check_header_field3 "Content" "$gsl_log_h_code" "$gsl_marker_code" \

# File exists in folder
if ! [[ -f "$gsl_dir_domain_files/$gsl_post_hf_2" ]];then

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_h_code" \
    "Not found: $gsl_post_hf_2" \
    "$gsl_dir_domain_files/$gsl_post_hf_2"
  gsl_checker_err=true
  
else

# Register file to DB with hash
  gsl_file_csum=`
  cksum "$gsl_dir_domain_files/$gsl_post_hf_2" \
      | awk '{print $1}'`
  
  printf '%s\n' \
         "FCode:$gsl_dir_domain_filess/$gsl_post_hf_2:$gsl_file_csum" \
      >> "$gsl_db_tmp"

fi

# Exists in Content
gsl_count_fcodes=`
gsl__get_content_line "_code:$gsl_post_hf_1" "$gsl_post" | wc -l`

if (( $gsl_count_fcodes > 0 ));then

  ((gsl_stat_fcode++))

else

  gsl__logs_print \
    "$gsl_log_e" \
    "Content" \
    "$gsl_log_c_code" \
    "Missing: _code:$gsl_post_hf_1" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true
  
fi
}

#======================================================================
# Stat: Count words
#======================================================================
gsl__stat_words() {
echo -ne "\r\033[2K: Counting Words..."
gsl_stat_words=`cat "$1" | wc -w`

gsl_stat_words=$(( $gsl_stat_words - 2*$gsl_stat_p ))
}

#======================================================================
# Check Post for MISSING & VALID Content HEADERS
#======================================================================
gsl__check_headers() {
echo -ne "\r\033[2K: Searching HEADERS..."
#--------------------------------
# Check if Missing NEEDED HEADERS
#--------------------------------
# Title
echo -ne "\r\033[2K: Searching HEADERS... $gsl_header_title"
gsl__check_needed_headers \
  "$gsl_header_title" \
  "$gsl_marker_title" \
  "POST TITLE"

# Slug
echo -ne "\r\033[2K: Searching HEADERS... $gsl_header_slug"
gsl__check_needed_headers \
  "$gsl_header_slug" \
  "$gsl_marker_slug" \
  "POST-TITLE" \
&& gsl_slug_err=true

# Author
echo -ne "\r\033[2K: Searching HEADERS... $gsl_header_author"
gsl__check_needed_headers \
  "$gsl_header_author" \
  "$gsl_marker_author" \
  "REGISTRED NAME" \
&& gsl_author_err=true

# Date
echo -ne "\r\033[2K: Searching HEADERS... $gsl_header_date"
gsl__check_needed_headers \
  "$gsl_header_date" \
  "$gsl_marker_date" \
  "YYYY-MM-DD" \
&& gsl_date_err=true

# Description
echo -ne "\r\033[2K: Searching HEADERS... ${gsl_header_info:0:40}..."
gsl__check_needed_headers \
  "$gsl_header_info" \
  "$gsl_marker_info" \
  "DESCRIPTION"

# Tags
echo -ne "\r\033[2K: Searching HEADERS... ${gsl_header_tags:0:40}..."
gsl__check_needed_headers \
  "$gsl_header_tags" \
  "$gsl_marker_tags" \
  "TAG1,OTHER TAG2,TAG3"

#--------------------------------
# Check for VALID Content HEADERS
#--------------------------------
# Author registred
if  ! [[ "$gsl_author_err" ]] && \
    ! [[ `grep "$gsl_header_author" "$gsl_file_auth_ndd"` ]];then

    gsl__logs_print \
      "$gsl_log_e" \
      "Content" \
      "Header $gsl_marker_author" \
      "$gsl_header_author not registred for domain $gsl_find_domain" \
      "${PWD}/$gsl_post"
    gsl_checker_err=true

fi

# Date Format YYYY-MM-DD
if  ! [[ "$gsl_date_err" ]] && \
    ! [[ "$gsl_header_date" =~ $gsl_test_date ]];then

    gsl__logs_print \
      "$gsl_log_e" \
      "Content" \
      "Header $gsl_marker_date" \
      "$gsl_header_date not YYYY-MM-DD" \
      "${PWD}/$gsl_post"
    gsl_checker_err=true

fi

# Slug format title-post
if ! [[ "$gsl_slug_err" ]];then
  gsl_header_slug_test=${gsl_header_slug// /-}
  
  if ! [[ "$gsl_header_slug" == "$gsl_header_slug_test" ]];then

    gsl_new_header_slug="$gsl_marker_slug$gsl_header_slug_test"
    sed -i \
        "s|$gsl_marker_slug$gsl_header_slug|$gsl_new_header_slug|" \
        "$gsl_post" && \
    gsl__logs_print \
      "$gsl_log_w" \
      "Content" \
      "Header $gsl_marker_slug" \
      "Changed: $gsl_header_slug_test" \
      "${PWD}/$gsl_post"
    
    gsl__get_sum "$gsl_post"
    gsl_header_slug=$gsl_header_slug_test
    gsl_checker_war=true

  fi

fi
}