qsgen2/qsgen2

#!/usr/bin/zsh

set -x

# Quick Site Generator 2 is a static website generator inspired by Nikola.
# It is written for the Z shell (zsh) because that's what I use and also because I like it better than Bash.
#
# This script is an almost complete rewrite of my old script because it became overly complicated and
# had way too many bugs, even though it worked on simple sites.
#
# qsgen2 uses different templates and its own formattings tags to generate the static HTML pages.
#
# The file structure in the project directory should look like this:
#
# .blog_cache
# .pages_cache
# config
# index.tpl
# templates/
# templates/<theme>/pages.tpl
# templates/<theme>/blogs.tpl
# templates/<theme>/blog_index.tpl
# blog/
# blog/2024-01-26-1.blog
#
# Explanation of the file structure.
# The file 'config' contains the settings for this specific project.
# The file 'index.tpl' is what becomes index.html that is served to the public.
# The 'templates/' directory contains the templates used to generate the pages and blog posts.
# The 'blog/' directory contains the blog posts. The date format of the files are used to create a blog/index.html with the newest post first.
#
# Contents of the file 'config' --> CHANGE THESE <--
# export site_name="The name of the website"
# - This is the project directory
# export project_dir=${HOME}/www/vikingo
# - This is where the generated files will be put
# export www_root=${HOME}/www_root/smelror.com
# #################################################################################################
# I don't think these need to be here as they'll always be in the same place with the same names
# export pages=${project_dir}/themes/<theme>/pages.tpl
# export blogs=${project_dir}/themes/<theme>/blogs.tpl
# export blog_list=${project_dir}/themes/<theme>/blog_list.tpl
# export blog_index=${project_dir}/themes/<theme>/blog_index.tpl
###################################################################################################
# export blog_in_index=false

VERSION="0.0.2 alpha" # Tue-2024-01-30
QSGEN="Quick Site Generator 2"

# Set to true or false
# This will show debug information from almost every function in this script
debug=false

function include () {
    
    # This function is used to include other functions that will normally be in
    # ${HOME}/bin/include/

        # Edit this path to reflect your installation
        local inc_file=${HOME}/bin/include/${1}.inc
        if [[ ! -f ${inc_file} ]]; then
                local inc_opt=$( echo ${1} | cut -d\/ -f2 )
                echo "Supplied option \"${inc_opt}\" is not a valid include."
        else
                builtin source ${inc_file} ${2}
        fi

}

# Including some colors to the script
include common/colors

echo "${bold_magenta}${QSGEN} ${VERSION}${end}"

function _version() {
    echo "${yellow}- Created by kekePower - 2018-$(date +%Y)${end}"
    echo "${yellow}- https://github.com/kekePower/qsgen2/${end}"
    echo "${yellow}- See '${1} help' for more information.${end}"
    exit
}

function _help() {
    echo "This is where I'll write the Help documentation."
	exit
}

if [[ "$1" == "version" || "$1" == "-v" || "$1" == "--version" ]]; then
  _version ${0:t}
elif [[ "$1" == "help" || "$1" == "-h" || "$1" == "--help" ]]; then
  _help ${0:t}
fi

# Loading Zsh modules
zmodload zsh/files

# Check for, an source, the config file for this specific website
if [[ -f $(pwd)/config ]]; then
    if (${debug}); then echo "${red}Config file found and sourced${end}\n- ${yellow} - $(pwd)/config${end}"; fi
	# CONFIG=$(pwd)/config
	builtin source $(pwd)/config
else
	echo "${red}Cannot find configuration file.${end}"
	echo "${yellow} - Please create the file 'config' in your project directory.${end}"
	exit
fi
debug=true
if (${debug}); then
    qsconfig=$( cat $(pwd)/config | grep -v \# | awk '{print substr($0, index($0, " ") + 1)}' )
    # qsconfig=$(<${qsconfig})
	echo "${red}Contents of Config file:${end}"
    for lines in ${qsconfig}
    do
        echo "${yellow}${lines}${end}"
    done
    #echo "${yellow} - site_name=${site_name}${end}"
    #echo "${yellow} - site_tagline=${site_tagline}${end}"
    #echo "${yellow} - theme=${theme}${end}"
    #echo "${yellow} - project_dir=${project_dir}${end}"
    #echo "${yellow} - www_root=${www_root}${end}"
    #echo "${yellow} - blog_in_index=${blog_in_index}${end}"
    #echo "${yellow} - generator=${generator}${end}"
    #echo "${yellow} - sitemap=${sitemap}${end}"
fi

function _msg() {
    # Use this function to write out messages based on their type
    # Types are: std=green - info=yellow - debug=red other=bold_yellow
    # Usage: _msg <type> "<message>"
    
    case ${1} in
        std)
            echo "${green}${2}${end}"
        ;;
        info)
            echo "${yellow}${2}${end}"
        ;;
        debug)
            echo "${red}${2}${end}"
        ;;
        other)
            echo "${bold_yellow}${2}${end}"
        ;;
        *)
            echo "${2}"
        ;;
}

# Let's check if qsgen2 can generate this site by checking if 'generator' is available
if [[ ! ${generator} ]] || [[ -d $(pwd)/.git ]]; then
    _msg other "${0:t} cannot parse this site. Exiting."
    exit
fi

# We define the variable 'engine' based on what's in the 'config' file.
if [[ ${generator} == "native" ]]; then
        # Usage: ${engine} ${1} - Where 1 is the file you want to convert
        engine=_html
elif [[ ${generator} == "markdown" ]]; then
    if [[ ! -f /usr/bin/markdown ]]; then
        _msg other "Please install the 'discount' package to use Markdown."
        exit
    else
        # Usage: ${engine} ${1} - Where 1 is the file you want parsed
        engine=$( /usr/bin/markdown ${1} -d )
    fi
fi
if (${debug}) _msg debug "Using the ${generator} engine"

# Define cache files for blogs and pages
blog_cache_file="${project_dir}/.blog_cache"
pages_cache_file="${project_dir}/.pages_cache"

builtin cd ${project_dir}

# Let's put these here for now.
export today=$( date "+%Y-%m-%d - %T" )
export blogdate=$( date +%a-%Y-%b-%d )

# Let's create arrays of all the files we'll be working on

function _list_blog_idx() {
    ls -har blog/*.idx | while read -r file; do
    blog_idx_array+=($file)
    done
}
function _list_blog_tmp_idx() {
    ls -har blog/*.tmp.html | while read -r file; do
    blog_tmp_idx_array+=($file)
    done
}
function _list_blog() {
    ls -1btar blog/*.blog | while read -r file; do
    blogs_file_array+=($file)
    done
}

# BLOG CACHE
function _blog_cache() {
    local debug=false
    # Create an associative array for the blog cache
    typeset -A blog_cache

    # Load the existing blog cache
    if [[ -f $blog_cache_file ]]; then
        while IFS=':' read -r name hash; do
            blog_cache[$name]=$hash
    		if (${debug}) _msg debug "_blog_cache: HASH VALUE: ${blog_cache[${name}]}"
        done < "$blog_cache_file"
    fi

    # Initialize the array for storing blog files to process
    make_blog_array=()

    # Process blog files
    for blog_file in $(ls -har blog/*.blog); do
        # Compute the current blog file hash
        current_hash=$(md5sum "$blog_file" | awk '{print $1}')
	
    	if (${debug}) _msg debug "1. _blog_cache: ${blog_file}"
    	if (${debug}) _msg debug "2. _blog_cache: current_cache: ${current_hash}"

        # Check if the blog file is new or has changed
        if [[ ${blog_cache[$blog_file]} != "$current_hash" ]]; then
    		if (${debug}) _msg debug "3. _blog_cache: new_cache_file: ${blog_file}$"
    		if (${debug}) _msg debug "4. _blog_cache: new_current_cache: ${current_hash}"
            # Blog file is new or has changed; add it to the processing array
            make_blog_array+=("$blog_file")

            # Update the blog cache with the new hash
            blog_cache[$blog_file]=$current_hash
        fi
    done

    # Rebuild the blog cache file from scratch
    : >| "$blog_cache_file"  # Truncate the file before writing
    for name in "${(@k)blog_cache}"; do
        echo "$name:${blog_cache[$name]}" >> "$blog_cache_file"
    done
}


# PAGES CACHE
# Returns the array pages_array()
function _pages_cache() {
    local debug=false
    # Create an associative array for the pages cache
    typeset -A pages_cache

    # Load the existing pages cache
    if [[ -f $pages_cache_file ]]; then
        while IFS=':' read -r name hash; do
            pages_cache[$name]=$hash
    		if (${debug}) _msg debug "PAGES HASH VALUE: ${pages_cache[${name}]}"
        done < "$pages_cache_file"
    fi

    # Initialize the array for storing pages files to process
    pages_array=()

    # Process pages files
    for file in $(ls -1bt *tpl); do
        # Compute the current blog file hash
        current_hash=$(md5sum "$file" | awk '{print $1}')
    	
    	if (${debug}) _msg debug "1. pages_cache: ${pages_cache[$file]}"
    	if (${debug}) _msg debug "1. c_pages_cache: urrent_cache: ${current_hash}"

        # Check if the pages file is new or has changed
        if [[ ${pages_cache[$file]} != "$current_hash" ]]; then
    		if (${debug}) _msg debug "2. _pages_cache: pages_file: ${pages_cache[$file]}"
    		if (${debug}) _msg debug "2. _pages_cache: current_cache: ${current_hash}"
            # Pages file is new or has changed; add it to the processing array
            pages_array+=("$file")
    
            # Update the pages cache with the new hash
            pages_cache[$file]=$current_hash
        fi
    done

    # Rebuild the pages cache file from scratch
    : >| "$pages_cache_file"  # Truncate the file before writing
    for name in "${(@k)pages_cache}"; do
        echo "$name:${pages_cache[$name]}" >> "$pages_cache_file"
    done
}

function _last_updated() {
    # This function updates #updated and #version tags in the provided string for buffers

    local content="${1}"
    local debug=false

    if (${debug}) _msg debug "_last_updated: Setting date and version in footer"

    # Perform the replacements
    local updated_content=$(echo "${content}" | sed \
        -e "s|#updated|${TODAY}|" \
        -e "s|#version|${QSGEN} ${VERSION}|")

    # Return the updated content
    echo "${updated_content}"
}

function _f_last_updated() {
    # This function updates #updated and #version tags in the provided string in files

    local content="${1}"
    local debug=false

    if (${debug}) _msg debug "_f_last_updated: Setting date and version in footer of file ${1}"

    # Perform the replacements
    tee < ${content} | sed \
        -e "s|#updated|${TODAY}|" \
        -e "s|#version|${QSGEN} ${VERSION}|" \
        > ${content}

}

function _file_to_lower() {
    
    local filename="${1}"

    # Replace spaces with dashes
    filename="${filename// /-}"

    # Convert to lowercase and remove invalid characters
    filename=$(echo "${filename}" | sed -e 's/^[^a-zA-Z0-9_.]+//g' -e 's/[^a-zA-Z0-9_-]+/-/g')

    echo ${filename}

}


function _pages() {
    # This function generates all the new and updated Pages
    
    local debug=false
    
    # Load the cache for Pages
    if (${debug}) _msg debug "_pages: Running function _pages_cache"
    _pages_cache
    
    if (( ${#pages_array[@]} > 0 )); then
    
      # If pages_array is not empty, we do work
      if (${debug}) _msg debug "_pages: pages_array is not empty"
      for pages_in_array in ${pages_array[@]}
      do
          
        if (${debug}) _msg debug "_pages: Setting Pages template"
        local pages=${project_dir}/themes/${theme}/pages.tpl
        
        # Let's check if we can access the pages.tpl file.
        # It not, exit script.
        if [[ ! -f ${pages} ]]; then
            _msg info "Unable to find the Pages template: ${pages}"
            exit
        else
            # Read template once
            if (${debug}) _msg debug "_pages: Reading Pages template into pages_tpl"
            pages_tpl="$(<${pages})"
        fi
        
        _msg std "Generating Page: ${pages_in_array}"
        # Read the file once
        if (${debug}) _msg debug "_pages: Loading page_content once - ${pages_in_array}"
        local page_content="$(<${pages_in_array})"
              
        # Grab the title from the Page
        if (${debug}) _msg debug "_pages: Grepping for page_title"
        local page_title=$( echo ${page_content} | head -2 | grep \#title | cut -d= -f2 )
        if (${debug}) _msg debug "_pages: ${page_title}"
        
        # Remove the #title line from the buffer. No longer needed.
        if (${debug}) _msg debug "_pages: Removing #title line from page_content"
        local page_content=$( echo ${page_content} | grep -v \#title )
        
        # HTML'ify the page content
        if (${debug}) _msg debug "_pages: Running engine on ${pages_in_array}"
        local page_content=$( ${engine} ${page_content} )
        
        # Look for links, images and videos and convert them if present.
        if (${debug}) _msg debug "_pages: Checking for #link, #showimg and #ytvideo in page_content"
        if [[ $( echo ${page_content} | grep \#link ) ]]; then
            if (${debug}) _msg debug "_pages: #link is present, run _link: page_content"
            local page_content=$( _link "${page_content}" )
        fi
        if [[ $( echo ${page_content} | grep \#showimg ) ]]; then
            if (${debug}) _msg debug "_pages: #showimg is present, run _image: page_content"
            local page_content=$( _image "${page_content}" )
        fi
        if [[ $( echo ${page_content} | grep \#ytvideo ) ]]; then
            if (${debug}) _msg debug "_pages: #ytvideo is present, run _youtube: page_content"
            local page_content=$( _youtube "${page_content}" )
        fi
        
        # Let's find the file 'index.tpl' and add the blog if blog_in_index is true
        if [[ ${pages_in_array} == "index.tpl" ]] && [[ ${blog_in_index} == "true" ]]; then
            if (${debug}) _msg debug "_pages: Inserting blog list to index.html"
            local blog_index_list=$(<${project_dir}/blog/index.tmp.html)
            echo ${blog_index_list} >> ${page_content}
        fi
        
        # Insert page_content into pages_tpl by replacing the BODY tag present there
        if (${debug}) _msg debug "_pages: Replacing BODY with page_content in pages_tpl using Perl"
        # Use Perl for multi-line and special character handling
        local pages_tpl=$( echo "${pages_tpl}" | perl -pe "s|BODY|${page_content}|gs" )
        
        # Replace every #pagetitle in pages_tpl
        if (${debug}) _msg debug "_pages: Replacing #pagetitle in pages_tpl"
        local pages_tpl=$( echo ${pages_tpl} | perl -pe "s|#pagetitle|${page_title}|gs" )
        
        # Replace every #tagline in pages_tpl
        if (${debug}) _msg debug "_pages: Replacing tagline"
        local pages_tpl=$( echo ${pages_tpl} | perl -pe "s|#tagline|${site_tagline}|gs" )
        
        # Replace #updated with today's date and #version with Name and Version to footer
        if (${debug}) _msg debug "_pages: _last_updated in pages_tpl"
        local pages_tpl=$( _last_updated ${pages_tpl} )
        
        # Clean up unused tags, if any
        if (${debug}) _msg debug "_pages: Running _cleanup"
        local pages_tpl=$( _cleanup "${pages_tpl}" )
        
        # Always use lowercase for file names
        if (${debug}) _msg debug "_pages: Lowercase filnames, always"
        local pages_title_lower=$( _file_to_lower "${pages_in_array}" )
        
        # Write pages_tpl to disk
        _msg std "Writing ${www_root}/${pages_title_lower%.*}.html to disk."
        echo "${pages_tpl}" > ${www_root}/${pages_title_lower%.*}.html
        
      done
      
    else

        _msg info "No new or updated Pages"

    fi
    
}

function _blogs() {
    # Rewritten _blogs function
    # This function either generates blog files or exports metadata based on the argument

    local debug=false

    # Running function _list_blog
    if (${debug}) _msg debug "_blogs: Running function _list_blog"
    _list_blog
    # Running function _blog_cache
    if (${debug}) _msg debug "_blogs: Running function _blog_cache"
    _blog_cache

    if (( ${#make_blog_array[@]} > 0 )); then
        
        # Declare the array to hold metadata strings for each blog
        BLOG_META_STR_ARRAY=()

            # Regular blog creation process
            #if [[ ! ${mode} ]]; then
                
                if [[ -f ${project_dir}/themes/${theme}/blogs.tpl ]]; then
                    local blog_tpl=$(<"${project_dir}/themes/${theme}/blogs.tpl")
                else
                    _msg info "Unable to find theme template for Blogs."
                    exit
                fi
                
                for blog in "${make_blog_array[@]}"; do
                    
                    if (${debug}) _msg debug "_blogs: Processing pre-data for ${blog}"
                
                    local content="$(<"${blog}")"
                    local sdate btitle ingress body blog_index blog_dir blog_url

                    # Extract blog information
                    sed -i "s/GETDATE/${blogdate}/" ${blog}
                    # Array sdate = Name day=1, Year=2, Month=3, Number day=4
                    sdate=( $( echo ${content} | grep DATE | sed "s|DATE\ ||" | sed "s|\-|\ |g" ) )
                    btitle=$( echo ${content} | grep BLOG_TITLE | cut -d' ' -f2- )
                    ingress=$( echo ${content} | sed "s/'/\\\'/g" | xargs | grep -Po "#INGRESS_START\K(.*?)#INGRESS_STOP" | sed "s|\ \#INGRESS_STOP||" | sed "s|^\ ||" )
                    body=$( echo ${content} | sed "s/'/\\\'/g" | xargs | grep -Po "#BODY_START\K(.*?)#BODY_STOP" | sed "s|\ \#BODY_STOP||" | sed "s|^\ ||" )

                    blog_index=$(echo "${btitle:l}" | sed 's/ /_/g; s/,//g; s/\.//g; s/://g; s/[()]//g')

                    blog_dir="/blog/${sdate[2]}/${sdate[3]:l}/${sdate[4]}"
                    blog_url="${blog_dir}/${blog_index}.html"
                    
                    if (${debug}) _msg debug "_blogs: Adding data for ${blog} to array to export"

                    # Concatenate all metadata into a single string for the current blog
                    local metadata_str="SDATE: ${sdate[@]}||BTITLE: ${btitle}||INGRESS: ${ingress}||URL: ${blog_url}"
                    # Append this metadata string to the array
                    BLOG_META_STR_ARRAY+=("${metadata_str}")
                    
                   if (${debug}) _msg debug "_blogs: Processing ${blog}"
                       
                   _msg std "Generating blog ${blog_index}.html"

                   # Prepare the blog template
                   if (${debug}) _msg debug "_blogs: Processing substitutes in ${blog}"
                   local blog_content=$(echo "${blog_tpl}" | perl -pe "s|BLOGTITLE|${btitle}|g")
                   blog_content=$(echo "${blog_content}" | perl -pe "s|CALADAY|${sdate[1]}|g")
                   blog_content=$(echo "${blog_content}" | perl -pe "s|CALNDAY|${sdate[4]}|g")
                   blog_content=$(echo "${blog_content}" | perl -pe "s|CALMONTH|${sdate[3]}|g")
                   blog_content=$(echo "${blog_content}" | perl -pe "s|CALYEAR|${sdate[2]}|g")
                   blog_content=$(echo "${blog_content}" | perl -pe "s|BLOGURL|${blog_url}|g")
                   blog_content=$(echo "${blog_content}" | perl -pe "s|\QINGRESS\E|${ingress}|g")
                   blog_content=$(echo "${blog_content}" | perl -pe "s|\QBODY\E|${body}|g")


                   if (${debug}) _msg debug "_blogs: Running function _html for ${blog}"
                   # Apply transformations
                   blog_content="$(_html "${blog_content}")"
                   # Look for links, images and videos and convert them if present.
                   if (${debug}) _msg debug "_blogs: Checking for #link, #showimg and #ytvideo in blog_content"
                   if [[ $( echo ${blog_content} | grep \#link ) ]]; then
                       if (${debug}) _msg debug "_blogs: If #link is present, run _link: blog_content"
                       blog_content="$(_link "${blog_content}")"
                   fi
                   if [[ $( echo ${blog_content} | grep \#showimg ) ]]; then
                       if (${debug}) _msg debug "_blogs: If #showimg is present, run _link: blog_content"
                       blog_content="$(_image "${blog_content}")"
                   fi
                   if [[ $( echo ${blog_content} | grep \#ytvideo ) ]]; then
                       if (${debug}) _msg debug "_blogs: If #ytvideo is present, run _link: blog_content"
                       blog_content="$(_youtube "${blog_content}")"
                   fi                   
                   blog_content="$(_cleanup "${blog_content}")"
                   blog_content="$(_last_updated "${blog_content}")"
                
                   # Create directory if it doesn't exist
                   if (${debug}) _msg debug "_blogs: Creating directoty ${www_root}/${blog_dir}"
                   [[ ! -d "${www_root}/${blog_dir}" ]] && mkdir -p "${www_root}/${blog_dir}"

                   # Write to file
                   if (${debug}) _msg debug "_blogs: Writing ${blog} to disk: ${www_root}${blog_url}"
                   echo "${blog_content}" > "${www_root}${blog_url}"
                
                done
            # Now BLOG_META_STR_ARRAY contains the metadata string for each blog post
            export BLOG_META_STR_ARRAY

    else
        _msg info "No new or updated Blogs detected."
    fi
}

function _blog_idx_for_index() {
    # This function generates the file blog/index.tmp.html
    
    local debug=false
    
    if (${debug}) _msg debug "_blog_idx_for_index: Initiating function"
    
    local blog_list_tpl=$(<${project_dir}/themes/${theme}/blog_list.tpl)
    
    # Truncate file before writing new one
    : >| "${project_dir}/blog/index.tmp.html"
    
    if (${debug}) _msg debug ":: _blog_idx_for_index: BLOG_META_STR_ARRAY: ${BLOG_META_STR_ARRAY[@]}"
    
    for meta_str in ${BLOG_META_STR_ARRAY[@]}
    do
        if (${debug}) _msg debug "_blog_idx_for_index: meta_str from BLOG_META_STR_ARRAY from _blogs"
        if (${debug}) _msg debug ":: _blog_idx_for_index: ${meta_str}"
        local debug=false
        # Split meta_str into individual metadata components
        local -a meta_array=("${(@s/||/)meta_str}")

        # Initialize variables to store each component
        local sdate=""
        local btitle=""
        local ingress=""
        local url=""

        # Iterate over each component and extract information
        if (${debug}) _msg debug "_blog_idx_for_index: Iterate over each component and extract information"
        for component in "${meta_array[@]}"; do
            local debug=false
            case "${component}" in
                SDATE:*) sdate=${component#SDATE: } ;;
                BTITLE:*) btitle=${component#BTITLE: } ;;
                INGRESS:*) ingress=${component#INGRESS: } ;;
                URL:*) url=${component#URL: } ;;
            esac

            local adate=( $( echo ${sdate} ) )
            local bdate="${adate[1]} - ${adate[4]}/${adate[3]}/${adate[2]}"
            blog_list_content=$(echo "${blog_list_tpl}" | perl -pe "s|BLOGURL|${url}|g")
            blog_list_content=$(echo "${blog_list_content}" | perl -pe "s|BLOGTITLE|${btitle}|g")
            blog_list_content=$(echo "${blog_list_content}" | perl -pe "s|INGRESS|${ingress}|g")
            blog_list_content=$(echo "${blog_list_content}" | perl -pe "s|BLOGDATE|${bdate}|g")    
        
        done
    
            if (${debug}) _msg debug "_blog_idx_for_index: Writing _blog_idx_for_index to file: ${project_dir}/blog/index.tmp.html"
            if (${debug}) _msg debug "_blog_idx_for_index: blog_list_content = ${blog_list_content}"
            echo ${blog_list_content} >> ${project_dir}/blog/index.tmp.html

    done

}

function _blog_index() {

    # This function generates the /blog/index.html file that gets its data from _blog_list_for_index()
    if [[ ${blog_in_index} == "false" ]]; then
        
        _msg info "Running function _blog_index"
        if (${debug}) _msg debug "_blog_index: blog_in_index = ${blog_in_index}"
        
        local debug=false
        local blog_index=$(<${project_dir}/themes/${theme}/blog_index.tpl)
        local blog_index_list=$(<${project_dir}/blog/index.tmp.html)

        _msg std "Updating the Blog Index file"
        # blog_index_content=$(echo "${blog_index}" | perl -pe "s|BODY|${blog_index_list}|g")
        echo "${blog_index}" | perl -pe "s|BODY|${blog_index_list}|g" > ${www_root}/blog/index.html
        _f_last_updated ${www_root}/blog/index.html
    elif [[ ${blog_in_index} == "true" ]]; then
        # Write blog list to /index.html and /blog/index.html
    fi
    
}

function _m_sitemap() {
    
if [[ ${sitemap} == "true" ]]; then
    
    _msg info "Generating sitemap.xml"

    local sitemap_file="${www_root}/sitemap.xml"

    # Find all HTML files and store them in an array
    local -a html_files=("${(@f)$(find "${www_root}" -type f -name "*.html")}")

    # Start of the XML file
    echo '<?xml version="1.0" encoding="UTF-8"?>' > ${sitemap_file}
    echo "<!-- Sitemap generated by ${QSGEN} ${VERSION} - https://github.com/kekePower/qsgen2 -->" >> ${sitemap_file}
    echo "<?xml-stylesheet type=\"text/xsl\" href=\"${site_url}/css/default-sitemap.xsl?sitemap=page\"?>" >> ${sitemap_file}
    echo '<urlset' >> ${sitemap_file}
    echo '  xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"' >> ${sitemap_file}
    echo '  xmlns:xhtml="http://www.w3.org/1999/xhtml"' >> ${sitemap_file}
    echo '  xmlns:image="http://www.google.com/schemas/sitemap-image/1.1"' >> ${sitemap_file}
    echo '>' >> ${sitemap_file}

    # Add each URL to the sitemap
    for file in "${html_files[@]}"; do
        # Remove www_root from the path and prepend site_url
        local url="${site_url}${file#$www_root}"
        local lastmod=$(stat -c %y "${file}" 2>/dev/null | cut -d' ' -f1,2 | sed 's/ /T/' | sed 's/\..*$//')

        echo "  <url>" >> ${sitemap_file}
        echo "    <loc>${url}</loc>" >> ${sitemap_file}
        echo "    <lastmod><![CDATA[${lastmod}+01:00]]></lastmod>" >> "${sitemap_file}"
        echo "    <changefreq><![CDATA[always]]></changefreq>" >> ${sitemap_file}
        echo "    <priority><![CDATA[1]]></priority>" >> ${sitemap_file}
        echo "  </url>" >> ${sitemap_file}
    done

    # End of the XML file
    echo '</urlset>' >> "${sitemap_file}"

    if (${debug}) _msg debug "Sitemap generated at ${sitemap_file}"

fi

}

function _link() {
    # This converts #link tags to actual clickable links in a provided string

    local content="${1}"
    local modified_content=""
    local debug=false

    # Process the content line by line
    echo "${content}" | while IFS= read -r line; do
        if [[ ${line} == *"#link"* ]]; then
            if (${debug}) _msg debug "_link: URL_MAIN(line): ${line}"

            # Extract the URL and the link text
            local url_full=$(echo "${line}" | awk -F'#link ' '{print $2}' | awk -F'¤' '{print $1 "¤" $2}')
            local url_dest=$(echo "${url_full}" | awk -F'¤' '{print $1}')
            local url_txt=$(echo "${url_full}" | awk -F'¤' '{print $2}')
            
            if (${debug}) _msg debug "_link_ URL: ${url_dest}"
            if (${debug}) _msg debug "_link: Text: ${url_txt}"

            # Form the replacement HTML link
            local modified_link="<a href=\"${url_dest}\">${url_txt}"
            if [[ ${url_dest} =~ ^https?:// ]]; then
                # Add external link icon for external URLs
                modified_link+="<img class=\"exticon\" alt=\"External site icon\" src=\"/images/ext-url.png\" width=\"16\" />"
            fi
            modified_link+="</a>"
            line=${line//"#link ${url_full}"/${modified_link}}
        fi
        modified_content+="${line}\n"
    done

    # Return the modified content
    echo -e "${modified_content}"
}

function _image() {
    # This replaces #showimg tags with actual HTML img tags in a provided string

    local content="${1}"
    local modified_content=""
    local debug=false

    # Process the content line by line
    echo "${content}" | while IFS= read -r line; do
        if [[ ${line} == *"#showimg"* ]]; then
            if (${debug}) _msg debug "_image: Processing line: ${line}"

            # Extract image link and alt text
            local img_link=$(echo "${line}" | awk -F'#showimg ' '{print $2}')
            local image=$(echo "${img_link}" | awk -F'¤' '{print $1}')
            local img_alt=$(echo "${img_link}" | awk -F'¤' '{print $2}')

            # Determine the source of the image
            local real_image=""
            if [[ ${image} =~ ^https?:// ]]; then
                real_image=${image}
            elif [[ ${image} =~ ^\/ ]]; then
                real_image=${image}
            else
                real_image="/images/${image}"
            fi

            # Form the replacement HTML image tag
            local img_tag="<img src=\"${real_image}\" alt=\"${img_alt}\" width=\"500\" />"
            line=${line//"#showimg ${img_link}"/${img_tag}}
        fi
        modified_content+="${line}\n"
    done

    # Return the modified content
    echo -e "${modified_content}"
}

function _youtube() {
    # This embeds a YouTube video in a provided string

    local content="${1}"
    local modified_content=""
    local debug=false

    # Process the content line by line
    echo "${content}" | while IFS= read -r line; do
        if [[ ${line} == *"#ytvideo"* ]]; then
            if (${debug}) _msg debug "_youtube: Processing line: ${line}"

            # Extract YouTube video ID
            local yt_id=$(echo "${line}" | awk -F'#ytvideo ' '{print $2}')

            # Form the replacement YouTube iframe embed
            local yt_iframe="<iframe width=\"560\" height=\"315\" src=\"https://www.youtube.com/embed/${yt_id}\" title=\"YouTube video player\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" allowfullscreen></iframe>"
            line=${line//"#ytvideo ${yt_id}"/${yt_iframe}}
        fi
        modified_content+="${line}\n"
    done

    # Return the modified content
    echo -e "${modified_content}"
}

function _cleanup() {
    # This removes tags used in the templates that may be left over for some reason

    local content="${1}"
    local debug=false

    if (${debug}) _msg debug "_cleanup: Cleaning up tags in content"

    # Perform the cleanup
    local cleaned_content=$(echo "${content}" | sed \
        -e "s|¤||g" \
        -e "s|#showimg\ ||g" \
        -e "s|#ytvideo\ ||g" \
        -e "s|#link\ ||g")

    # Return the cleaned content
    echo "${cleaned_content}"
}

function _html() {
    # This function converts formatting tags into their HTML equivalents in a provided string

    local content="${1}"
    local debug=false

    if (${debug}) _msg debug "_html: Generating HTML from content"

    # Perform HTML tag substitutions
    local html_content=$(echo "${content}" | sed \
        -e "s|#BR|<br/>|g" \
        -e "s|#BD|<b>|g" \
        -e "s|#EBD|</b>|g" \
        -e "s|#UN|<u>|g" \
        -e "s|#EUN|</u>|g" \
        -e "s|#P|<p>|g" \
        -e "s|#EP|</p>|g" \
        -e "s|#Q|<blockquote>|g" \
        -e "s|#EQ|</blockquote>|g" \
        -e "s|#STRONG|<strong>|g" \
        -e "s|#ESTRONG|</strong>|g" \
        -e "s|#I|<i>|g" \
        -e "s|#EI|</i>|g" \
        -e "s|#C|<code>|g" \
        -e "s|#EC|</code>|g" \
        -e "s|#EM|<em>|g" \
        -e "s|#SEM|</em>|g" \
        -e "s|#OT|&quot;|g" \
        -e "s|#UL|<ul>|g" \
        -e "s|#OL|<ol>|g" \
        -e "s|#LI|<li class=\"libody\">|g" \
        -e "s|#ELI|</li>|g" \
        -e "s|#EUL|</ul>|g" \
        -e "s|#EOL|</ol>|g" \
        -e "s|#H1|<h1>|g" \
        -e "s|#H2|<h2>|g" \
        -e "s|#H3|<h3>|g" \
        -e "s|#H4|<h4>|g" \
        -e "s|#H5|<h5>|g" \
        -e "s|#H6|<h6>|g" \
        -e "s|#EH1|</h1>|g" \
        -e "s|#EH2|</h2>|g" \
        -e "s|#EH3|</h3>|g" \
        -e "s|#EH4|</h4>|g" \
        -e "s|#EH5|</h5>|g" \
        -e "s|#EH6|</h6>|g" \
        -e "s|#LT|&lt;|g" \
        -e "s|#GT|&gt;|g" \
        -e "s|#NUM|&num;|g")

    # Return the HTML content
    echo "${html_content}"
}

# Time to test the first function
_msg std "Running function _blogs"
_blogs
_msg std "Running function _blog_idx_for_index"
_blog_idx_for_index
_blog_index
_msg std "Running function _pages"
_pages
_msg std "Running function _sitemap"
_m_sitemap