gsl-statique-litterateur/var/lib/gsl/scripts/gsl__post_checkers

#!/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 | Check Post from modules
#=======================================================================
gsl__post_all_checkers() {
clear
unset gsl_check_done
 
echo -ne "Searching for #1..."
gsl__post_check_h1 || return
echo -n "\r\033[2K: : Searching for NEEDED HEADERS..."
gsl__post_check_needed_headers
echo -n "\r\033[2K: : Searching for Post TYPE..."
gsl__post_check_type
echo -ne "\r\033[2K: Searching for PARAGRAPHS"
gsl__post_check_paragraphs
echo -ne "\r\033[2K: Searching for Content MARKERS..."
gsl__post_check_markers
echo -ne "\r\033[2K: Searching for BLOCKQUOTEs..."
gsl__post_check_blockquote
echo -ne "\r\033[2K: Searching for ABBRs..."
gsl__post_check_abbr
echo -ne "\r\033[2K: Searching for LINKs..."
gsl__post_check_links
echo -ne "\r\033[2K: Searching for FILES..."
gsl__post_check_files
echo -ne "\r\033[2K: Searching for IMAGES..."
gsl__post_check_images
echo -ne "\r\033[2K: Searching for BLOCK-CODES..."
gsl__post_check_bcodes
echo -ne "\r\033[2K: Searching for some STATS..."
gsl__post_check_stats

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

# End of checkers : show logs for war and err
gsl__db_line_post
}

#=======================================================================
# Check Post for MISSING & VALID Content HEADERS
#=======================================================================
gsl__post_check_needed_headers() {
#-----------------------------------------------------------------------
# Check if Missing NEEDED HEADERS
#-----------------------------------------------------------------------
# Title
gsl_header_title=`gsl__get_header "$gsl_marker_title" "$gsl_post"`
gsl__check_needed_headers \
    "$gsl_header_title" \
    "$gsl_marker_title" \
    "POST TITLE"

# Slug
gsl_header_slug=`gsl__get_header "$gsl_marker_slug" "$gsl_post"`
gsl__check_needed_headers \
    "$gsl_header_slug" \
    "$gsl_marker_slug" \
    "POST-TITLE" \
  && gsl_slug_err=true

# Author
gsl_header_author=`gsl__get_header "$gsl_marker_author" "$gsl_post"`
gsl__check_needed_headers \
    "$gsl_header_author" \
    "$gsl_marker_author" \
    "REGISTRED NAME" \
  && gsl_author_err=true

# Date
gsl_header_date=`gsl__get_header "$gsl_marker_date" "$gsl_post"`
gsl__check_needed_headers \
    "$gsl_header_date" \
    "$gsl_marker_date" \
    "YYYY-MM-DD" \
  && gsl_date_err=true

# Description
gsl_header_info=`gsl__get_header "$gsl_marker_info" "$gsl_post"`
gsl__check_needed_headers \
    "$gsl_header_info" \
    "$gsl_marker_info" \
    "DESCRIPTION"

# Tags
gsl_header_tags=`gsl__get_header "$gsl_marker_tags" "$gsl_post"`
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" \
      "Post" \
      "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" \
      "Post" \
      "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" \
      "Post" \
      "Header $gsl_marker_slug" \
      "Changed: $gsl_header_slug_test" \
      "${PWD}/$gsl_post"
    
    gsl_header_slug=$gsl_header_slug_test
    gsl_checker_war=true
  fi
fi
}

#=======================================================================
# Check Post for begining content
#=======================================================================
gsl__post_check_h1() {
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" \
    "Post" \
    "Content begin" \
    "Missing: #1 TITLE" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true
fi
}

#=======================================================================
# Check/Set for Post ID
#=======================================================================
gsl__post_check_type() {
if [[ `awk  -v m="$gsl_marker_type" -v l="$gsl_post_begin" \
            'NR < l && $0 ~ m' \
            "$gsl_post"` ]];then
  
  gsl_post_type=`gsl__get_header "$gsl_marker_type" "$gsl_post"`
  # No ID found
  case "$gsl_post_type" in
    page|post) true ;;
    *)
      sed -i  "s|$gsl_marker_type.*|${gsl_marker_type}post|" \
              "$gsl_post"
      ;;
  esac
else
  sed -i  "1i${gsl_marker_type}post" \
          "$gsl_post"
fi
}

#=======================================================================
# Check for paragraphs ( and )
#=======================================================================
gsl__post_check_paragraphs() {
gsl_post_p_open_nbr=`
awk -v line="$gsl_post_begin" \
    -v op="$gsl_mo_p" \
    'NR > line && $1 == op' \
    "$gsl_post" \
  | wc -l`
      
gsl_post_p_close_nbr=`
awk -v line="$gsl_post_begin" \
    -v cp="$gsl_mc_p" \
    'NR > line && $1 == cp' \
    "$gsl_post" \
  | wc -l`
   
if (( "$gsl_post_p_open_nbr" == 0 ));then
  gsl__logs_print \
    "$gsl_log_e" \
    "Post" \
    "Content paragraphs" \
    "Missing content: ( and ) at begining lines" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true
  return
fi

if (( "$gsl_post_p_open_nbr" == "$gsl_post_p_close_nbr" ));then
   gsl_post_stat_p="$gsl_post_p_open_nbr"
else
  gsl__logs_print \
    "$gsl_log_e" \
    "Post" \
    "Content paragraphs" \
    "Mismatch: (=$gsl_post_p_open_nbr ; )=$gsl_post_p_close_nbr" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true
  return
fi
}

#=======================================================================
# Check and count MARKERS content (Strong, links...)
#=======================================================================
gsl__post_check_markers() {
gsl_markers_nbr=$(( ${#gsl_o_markers[@]} - 1 ))
gsl_post_stat_m=(0 0 0 0 0 0)

while read -r "gsl_post_line"
do
  gsl_line_nbr=`awk -F":" '{print $1}' <<< $gsl_post_line`
  
  for gsl_i in `seq 0 $gsl_markers_nbr`
  do      
    if [[ "$gsl_post_line" =~ ${gsl_o_markers[gsl_i]} ]];then
      # Open Marker by Type
      gsl_o=`
      grep  -o "${gsl_o_markers[gsl_i]}" \
            <<< $gsl_post_line \
          | wc -l`
      # Close Marker by Type
      gsl_c=`
      grep  -o "${gsl_c_markers[gsl_i]}" \
            <<< $gsl_post_line \
          | wc -l`
       
      # Strong Vs Bold | 2= **_ / 3= *_ in array
      if (( $gsl_i == 2 ));then
        gsl_oS=$gsl_o # Open Strong marker nbr
        gsl_cS=$gsl_c # Close Strong marker nbr
      
      elif (( $gsl_i == 3 ));then
        # Count difference from Strong
        if (( $gsl_oS ));then
          gsl_o=$(( $gsl_o - $gsl_oS ))
          unset gsl_oS
        fi
        if (( $gsl_cS ));then
          gsl_c=$(( $gsl_c - $gsl_cS ))
          unset gsl_cS
        fi
      fi
    
      # Compare
      if ! (( $gsl_o == $gsl_c ));then
        gsl__logs_print \
          "$gsl_log_e" \
          "Post" \
          "${gsl_i_markers[gsl_i]}" \
          "Line:$gsl_line_nbr > Mismatch ${gsl_u_markers[gsl_i]}" \
          "${PWD}/$gsl_post"
        gsl_checker_err=true
      else
        gsl_post_stat_m[gsl_i]=$(( gsl_post_stat_m[gsl_i] + gsl_o ))
      fi
    fi
  done
done < <(awk  -v line="$gsl_post_begin" \
              'NR >= line {print NR":",$0}' \
              "$gsl_post")

gsl_post_stat_mdb=${gsl_post_stat_m[@]}

# Show Counts in logs
for gsl_i in `seq 0 $gsl_markers_nbr`
do
  if (( ${gsl_post_stat_m[gsl_i]} >= 1 ));then
  gsl__logs_print \
    "$gsl_log_i" \
    "Post" \
    "${gsl_i_markers[gsl_i]}" \
    "NBR=${gsl_post_stat_m[gsl_i]} > ${gsl_u_markers[gsl_i]}" \
    "${PWD}/$gsl_post"
  fi
done
}

#=======================================================================
# Check for blockquotes: ---
#=======================================================================
gsl__post_check_blockquote() {
gsl_post_blockquotes_nbr=`
awk -v line="$gsl_post_begin" \
    -v bq="$gsl_mark_blockquote" \
    'NR > line && $1 == bq' \
    "$gsl_post" \
  | wc -l`

gsl_post_blockquotes_nbr_r=$(( $gsl_post_blockquotes_nbr % 2 ))

if [[ "$gsl_post_blockquotes_nbr_r" -eq 0 ]];then
  gsl_post_stat_bq=$(( $gsl_post_blockquotes_nbr / 2 ))
  gsl__logs_print \
    "$gsl_log_i" \
    "Post" \
    "Content citations" \
    "NBR=$gsl_post_stat_bq > $gsl_mark_blockquote (2x)" \
    "${PWD}/$gsl_post"
fi
}

#=======================================================================
# Check Post for ABBR HEADER + Content
#=======================================================================
gsl__post_check_abbr() {
gsl_post_stat_abbr=0
while read -r "gsl_header_content_line"
do

  # Get & Check HEADER CONTENT
  gsl__get_header_fields "$gsl_marker_abbr"
  gsl__check_header_fields "Post" "Header $gsl_marker_abbr" "$gsl_log_act_abbr" \
    || continue

  # Check POST CONTENT
  [[ `gsl__get_content_line \
        "$gsl_post_header_field_1" \
        "$gsl_post"` ]] \
  && gsl_post_stat_abbr=$(( gsl_post_stat_abbr + 1 )) \
  && continue

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

done < <(gsl__get_header "$gsl_marker_abbr" "$gsl_post")

gsl__logs_print \
  "$gsl_log_i" \
  "Post" \
  "Header $gsl_marker_abbr" \
  "NBR=$gsl_post_stat_abbr" \
  "${PWD}/$gsl_post"
}

#=======================================================================
# Check Post for LINKs HEADER + Content
#=======================================================================
gsl__post_check_links() {
gsl_post_stat_link=0
while read -r "gsl_header_content_line"
do
  # Get & Check Header CONTENT
  gsl__get_header_fields "$gsl_marker_link"
  gsl__check_header_fields "Post" "Header $gsl_marker_link" "$gsl_log_act_link" \
    || continue
  
  # Not ALT TEXT ?
  gsl__check_header_field3 "Post" "Header $gsl_marker_link"

  # Find if ref is in CONTENT
  [[ `gsl__get_content_with_markers "$gsl_mo_link" "$gsl_mc_link"` ]] \
    && gsl_post_stat_link=$(( $gsl_post_stat_link + 1 )) \
    && continue
    
  # Check URL
  (( `curl  -o /dev/null --silent --head --write-out \
            '%{http_code}' \
            "$gsl_post_header_field_2"` == 200 )) \
    && continue \
    || gsl__logs_print \
        "$gsl_log_e" \
        "Post" \
        "Header $gsl_marker_link" \
        "URL offline: $gsl_post_header_field_2" \
        "${PWD}/$gsl_post"

  # Missing CONTENT
  gsl__logs_print \
    "$gsl_log_e" \
    "Post" \
    "$gsl_log_c_link" \
    "Missing: [_${gsl_post_header_field_1}_]" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true

done < <(gsl__get_header "$gsl_marker_link" "$gsl_post")
}

#=======================================================================
# Check Post for FILES HEADER + Content
#=======================================================================
gsl__post_check_files() {
gsl_post_stat_file=0
while read -r "gsl_header_content_line"
do

  # Get & Check Header CONTENT
  gsl__get_header_fields "$gsl_marker_file"
  gsl__check_header_fields "Post" "Header $gsl_marker_file" "$gsl_log_act_file" \
    || continue

  # Not ALT TEXT ?
  gsl__check_header_field3 "Post" "Header $gsl_marker_file"

  if ! [[ -f "$gsl_dir_domain_files/$gsl_post_header_field_2" ]];then
    gsl__logs_print \
      "$gsl_log_e" \
      "Post" \
      "Header $gsl_marker_file" \
      "Not found: $gsl_post_header_field_2" \
      "$gsl_dir_domain_files/$gsl_post_header_field_2"
    gsl_checker_err=true
  else
  # Register file to DB with hash
    gsl_file_csum=`
    cksum "$gsl_dir_domain_files/$gsl_post_header_field_2" \
        | awk '{print $1}'`
    gsl_db_post_files+="$gsl_post_header_field_2|"
    gsl__db_line_file
  fi
  
  # Find if ref is in CONTENT
  [[ `gsl__get_content_with_markers "$gsl_mo_file" "$gsl_mc_file"` ]] \
    && gsl_post_stat_file=$(( $gsl_post_stat_file + 1 )) \
    && continue

  # Missing CONTENT
  gsl__logs_print \
    "$gsl_log_e" \
    "Post" \
    "$gsl_log_c_file" \
    "Missing: <_${gsl_post_header_field_1}_>" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true

done < <(gsl__get_header "$gsl_marker_file" "$gsl_post")
}

#=======================================================================
# Check Post for CODES HEADER + Content
#=======================================================================
gsl__post_check_bcodes() {
gsl_post_stat_bcode=0
while read -r "gsl_header_content_line"
do

  # Get & Check Header CONTENT
  gsl__get_header_fields "$gsl_marker_code"
  gsl__check_header_fields "Post" "Header $gsl_marker_code" "$gsl_log_act_code" \
    || continue

  if ! [[ -f "$gsl_dir_domain_files/$gsl_post_header_field_2" ]];then
    gsl__logs_print \
      "$gsl_log_e" \
      "Post" \
      "Header $gsl_marker_code" \
      "Not found: $gsl_post_header_field_2" \
      "$gsl_dir_domain_files/$gsl_post_header_field_2"
    gsl_checker_err=true
  else
  # Register file to DB with hash
    gsl_file_csum=`
    cksum "$gsl_dir_domain_files/$gsl_post_header_field_2" \
        | awk '{print $1}'`
    gsl_db_post_files+="$gsl_post_header_field_2|"
    gsl__db_line_file
  fi

  # Find if ref is in CONTENT
  [[ `gsl__get_content_line \
        "$gsl_mark_code : $gsl_post_header_field_1" \
        "$gsl_post"` ]] \
    && gsl_post_stat_bcode=$(( $gsl_post_stat_bocde + 1 )) \
    && continue

  # Missing CONTENT
  gsl__logs_print \
    "$gsl_log_e" \
    "Post" \
    "$gsl_log_c_code" \
    "Missing: $gsl_mark_code : ${gsl_post_header_field_1}" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true

done < <(gsl__get_header "$gsl_marker_code" "$gsl_post")
}

#=======================================================================
# Check Post for IMAGES HEADER + Content
#=======================================================================
gsl__post_check_images() {
gsl_post_stat_image=0
while read -r "gsl_header_content_line"
do

  # Get & Check Header CONTENT
  gsl__get_header_fields "$gsl_marker_image"
  gsl__check_header_fields "Post" "Header $gsl_marker_image" "$gsl_log_act_image" \
    || continue

  # Not ALT TEXT ?
  gsl__check_header_field3 "Post" "Header $gsl_marker_image"

  if ! [[ -f "$gsl_dir_domain_files/$gsl_post_header_field_2" ]];then
    gsl__logs_print \
      "$gsl_log_e" \
      "Post" \
      "Header $gsl_marker_image" \
      "Not found: $gsl_post_header_field_2" \
      "$gsl_dir_domain_files/$gsl_post_header_field_2"
    gsl_checker_err=true
  else
  # Register file to DB with hash
    gsl_file_csum=`
    cksum "$gsl_dir_domain_files/$gsl_post_header_field_2" \
        | awk '{print $1}'`
    gsl_db_post_files+="$gsl_post_header_field_2|"
    
    gsl__db_exists "$gsl_file_db_files" "$gsl_post_header_field_2"
    gsl__db_line_file
  fi
  
  # Find if ref is in CONTENT
  [[ `gsl__get_content_line \
        "$gsl_mark_image : $gsl_post_header_field_1" \
        "$gsl_post"` ]] \
    && gsl_post_stat_image=$(( $gsl_post_stat_image + 1 )) \
    && continue

  # Missing CONTENT
  gsl__logs_print \
    "$gsl_log_e" \
    "Post" \
    "$gsl_log_c_image" \
    "Missing: $gsl_mark_image : ${gsl_post_header_field_1}" \
    "${PWD}/$gsl_post"
  gsl_checker_err=true

done < <(gsl__get_header "$gsl_marker_image" "$gsl_post")
}

#=======================================================================
# Check for more STATS
#=======================================================================
gsl__post_check_stats() {
  gsl_post_stat_w=`
  awk -v l="$gsl_post_begin" \
      'NR >= l' \
      "$gsl_post" \
    | wc -w`

  gsl_post_stat_w=$(( $gsl_post_stat_w - 2*$gsl_post_stat_p ))
}