odoo-expertos.com/generate_sitemap.sh

#!/bin/bash

# Multi-Language Sitemap Generator for Odoo Expertos
# Generates sitemaps with hreflang annotations for ES (default), EN, DE, FR

DOMAIN="https://www.odoo-expertos.com"
TODAY=$(date +%Y-%m-%d)
LANGUAGES=("" "en" "de" "fr" "pt" "ar")
HREFLANGS=("es" "en" "de" "fr" "pt" "ar")

# Generate main sitemap index
cat << 'SITEMAP_INDEX_START' > sitemap.xml
<?xml version="1.0" encoding="UTF-8"?>
<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
    <sitemap>
        <loc>https://www.odoo-expertos.com/sitemap-es.xml</loc>
    </sitemap>
    <sitemap>
        <loc>https://www.odoo-expertos.com/sitemap-en.xml</loc>
    </sitemap>
    <sitemap>
        <loc>https://www.odoo-expertos.com/sitemap-de.xml</loc>
    </sitemap>
    <sitemap>
        <loc>https://www.odoo-expertos.com/sitemap-fr.xml</loc>
    </sitemap>
    <sitemap>
        <loc>https://www.odoo-expertos.com/sitemap-pt.xml</loc>
    </sitemap>
    <sitemap>
        <loc>https://www.odoo-expertos.com/sitemap-ar.xml</loc>
    </sitemap>
</sitemapindex>
SITEMAP_INDEX_START

# Function to generate hreflang entries for a given path
generate_hreflang() {
    local base_path="$1"
    echo "        <xhtml:link rel=\"alternate\" hreflang=\"es\" href=\"${DOMAIN}${base_path}\" />"
    echo "        <xhtml:link rel=\"alternate\" hreflang=\"en\" href=\"${DOMAIN}/en${base_path}\" />"
    echo "        <xhtml:link rel=\"alternate\" hreflang=\"de\" href=\"${DOMAIN}/de${base_path}\" />"
    echo "        <xhtml:link rel=\"alternate\" hreflang=\"fr\" href=\"${DOMAIN}/fr${base_path}\" />"
    echo "        <xhtml:link rel=\"alternate\" hreflang=\"pt\" href=\"${DOMAIN}/pt${base_path}\" />"
    echo "        <xhtml:link rel=\"alternate\" hreflang=\"ar\" href=\"${DOMAIN}/ar${base_path}\" />"
    echo "        <xhtml:link rel=\"alternate\" hreflang=\"x-default\" href=\"${DOMAIN}${base_path}\" />"
}

# Function to generate sitemap for a specific language
generate_lang_sitemap() {
    local lang="$1"
    local lang_prefix="$2"
    local sitemap_file="sitemap-${lang}.xml"

    cat << SITEMAP_HEADER > "$sitemap_file"
<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:xhtml="http://www.w3.org/1999/xhtml">
SITEMAP_HEADER

    # Homepage
    cat << HOMEPAGE >> "$sitemap_file"
    <url>
        <loc>${DOMAIN}${lang_prefix}/</loc>
        <lastmod>${TODAY}</lastmod>
        <changefreq>daily</changefreq>
        <priority>1.0</priority>
$(generate_hreflang "/")
    </url>
HOMEPAGE

    # Main category pages
    for category in "odoo" "odoo-hosting" "odoo-ia"; do
        cat << CATEGORY >> "$sitemap_file"
    <url>
        <loc>${DOMAIN}${lang_prefix}/${category}/</loc>
        <lastmod>${TODAY}</lastmod>
        <changefreq>weekly</changefreq>
        <priority>0.9</priority>
$(generate_hreflang "/${category}/")
    </url>
CATEGORY
    done

    # Content pages for this language
    if [ -z "$lang_prefix" ]; then
        # Spanish (root) - scan existing content
        find . -name "*.html" -type f | grep -E "^./odoo|^./odoo-hosting|^./odoo-ia" | grep -v test-files | grep -v "^./en" | grep -v "^./de" | grep -v "^./fr" | while read file; do
            path=$(echo "$file" | sed 's|^\./||' | sed 's|/index\.html$|/|' | sed 's|\.html$|.html|')

            # Skip category index pages (already added above)
            if [[ "$path" == "odoo/" || "$path" == "odoo-hosting/" || "$path" == "odoo-ia/" ]]; then
                continue
            fi

            depth=$(echo "$path" | tr -cd '/' | wc -c)
            if [ $depth -le 2 ]; then
                priority="0.8"
            else
                priority="0.7"
            fi

            cat << URL >> "$sitemap_file"
    <url>
        <loc>${DOMAIN}/${path}</loc>
        <lastmod>${TODAY}</lastmod>
        <changefreq>monthly</changefreq>
        <priority>${priority}</priority>
$(generate_hreflang "/${path}")
    </url>
URL
        done
    else
        # Other languages - scan their directories
        if [ -d ".${lang_prefix}" ]; then
            find ".${lang_prefix}" -name "*.html" -type f | while read file; do
                # Get path relative to language directory
                rel_path=$(echo "$file" | sed "s|^\.${lang_prefix}/||" | sed 's|/index\.html$|/|' | sed 's|\.html$|.html|')
                base_path="/${rel_path}"

                depth=$(echo "$rel_path" | tr -cd '/' | wc -c)
                if [ $depth -le 2 ]; then
                    priority="0.8"
                else
                    priority="0.7"
                fi

                cat << URL >> "$sitemap_file"
    <url>
        <loc>${DOMAIN}${lang_prefix}/${rel_path}</loc>
        <lastmod>${TODAY}</lastmod>
        <changefreq>monthly</changefreq>
        <priority>${priority}</priority>
$(generate_hreflang "${base_path}")
    </url>
URL
            done
        fi
    fi

    # Legal pages (only for main Spanish sitemap)
    if [ -z "$lang_prefix" ]; then
        for page in "cookie-policy" "legal-notice" "privacy-policy" "terms-of-use"; do
            cat << LEGAL >> "$sitemap_file"
    <url>
        <loc>${DOMAIN}/${page}.html</loc>
        <lastmod>${TODAY}</lastmod>
        <changefreq>yearly</changefreq>
        <priority>0.3</priority>
    </url>
LEGAL
        done
    fi

    # Close sitemap
    echo "</urlset>" >> "$sitemap_file"

    url_count=$(grep -c '<url>' "$sitemap_file")
    echo "Generated $sitemap_file with $url_count URLs"
}

# Generate sitemaps for each language
generate_lang_sitemap "es" ""
generate_lang_sitemap "en" "/en"
generate_lang_sitemap "de" "/de"
generate_lang_sitemap "fr" "/fr"
generate_lang_sitemap "pt" "/pt"
generate_lang_sitemap "ar" "/ar"

# Summary
echo ""
echo "=== Sitemap Generation Complete ==="
echo "Main sitemap index: sitemap.xml"
for lang in "es" "en" "de" "fr" "pt" "ar"; do
    echo "  - sitemap-${lang}.xml: $(grep -c '<url>' sitemap-${lang}.xml) URLs"
done