website-htmx-rustelo-code/scripts/sh/content/validate-content-consistency.sh

#!/bin/bash

# Content Consistency Validation Script
# Validates that content structure is consistent and server can process all content

# set -e removed - handle errors explicitly to avoid early exits on warnings

# Configuration
CONTENT_DIR="${SITE_CONTENT_PATH:-site/content}"
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"

# Colors for output
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
BLUE='\033[0;34m'
NC='\033[0m' # No Color

# Counters
TOTAL_CHECKS=0
PASSED_CHECKS=0
FAILED_CHECKS=0

# Function to print colored output
print_status() {
    local status=$1
    local message=$2
    case $status in
        "PASS")
            echo -e "${GREEN}✅ PASS${NC}: $message"
            ((PASSED_CHECKS++))
            ;;
        "FAIL")
            echo -e "${RED}❌ FAIL${NC}: $message"
            ((FAILED_CHECKS++))
            ;;
        "WARN")
            echo -e "${YELLOW}⚠️  WARN${NC}: $message"
            ;;
        "INFO")
            echo -e "${BLUE}ℹ️  INFO${NC}: $message"
            ;;
    esac
    ((TOTAL_CHECKS++))
}

# Function to check if directory exists and is not empty
check_directory() {
    local dir=$1
    local description=$2
    
    if [ ! -d "$dir" ]; then
        print_status "FAIL" "$description directory missing: $dir"
        return 1
    fi
    
    local file_count
    file_count=$(ls -A "$dir" 2>/dev/null | wc -l)
    if [ "$file_count" -eq 0 ]; then
        print_status "WARN" "$description directory is empty: $dir"
        return 0  # Return 0 for warnings to not exit with set -e
    fi
    
    print_status "PASS" "$description directory exists and has content: $dir"
    return 0
}

# Function to validate markdown files
validate_markdown_files() {
    local base_dir=$1
    local content_type=$2
    
    print_status "INFO" "Validating $content_type markdown files in $base_dir"
    
    local md_count=0
    local invalid_count=0
    
    while IFS= read -r -d '' file; do
        ((md_count++))
        
        # Check if file has frontmatter
        if ! head -n1 "$file" | grep -q "^---$"; then
            print_status "FAIL" "Missing frontmatter in: ${file#$PROJECT_ROOT/}"
            ((invalid_count++))
            continue
        fi
        
        # Check if file has closing frontmatter
        if ! sed -n '2,$p' "$file" | grep -q "^---$"; then
            print_status "FAIL" "Invalid frontmatter (no closing ---) in: ${file#$PROJECT_ROOT/}"
            ((invalid_count++))
            continue
        fi
        
        # Check for required frontmatter fields
        local frontmatter=$(sed -n '/^---$/,/^---$/p' "$file")
        
        if ! echo "$frontmatter" | grep -q "^title:"; then
            print_status "FAIL" "Missing 'title' field in: ${file#$PROJECT_ROOT/}"
            ((invalid_count++))
        fi
        
        if ! echo "$frontmatter" | grep -q "^slug:"; then
            print_status "FAIL" "Missing 'slug' field in: ${file#$PROJECT_ROOT/}"
            ((invalid_count++))
        fi
        
    done < <(find "$base_dir" -name "*.md" -type f -print0)
    
    if [ $md_count -eq 0 ]; then
        print_status "WARN" "No markdown files found in $base_dir"
    else
        if [ $invalid_count -eq 0 ]; then
            print_status "PASS" "All $md_count $content_type markdown files are valid"
        else
            print_status "FAIL" "$invalid_count of $md_count $content_type markdown files have issues"
        fi
    fi
}

# Function to validate content structure
validate_content_structure() {
    print_status "INFO" "=== Content Structure Validation ==="
    
    # Check main content directory
    check_directory "$CONTENT_DIR" "Main content"
    
    # Check content-kinds.toml exists
    local content_kinds_file="$CONTENT_DIR/content-kinds.toml"
    if [ ! -f "$content_kinds_file" ]; then
        print_status "FAIL" "content-kinds.toml missing: $content_kinds_file"
        return 1
    fi
    print_status "PASS" "content-kinds.toml exists"
    
    # Check routes configuration
    if [ -f "$CONTENT_DIR/routes.toml" ]; then
        print_status "PASS" "routes.toml exists"
    elif [ -d "$CONTENT_DIR/routes/" ]; then
        print_status "PASS" "routes/ directory exists"
    else
        print_status "FAIL" "No routes configuration found (routes.toml or routes/ directory)"
    fi
    
    # Validate each enabled content type
    while IFS= read -r content_type; do
        local content_dir="$CONTENT_DIR/$content_type"
        if check_directory "$content_dir" "Content type '$content_type'"; then
            # Check for language directories
            local lang_count=0
            for lang_dir in "$content_dir"/*/; do
                if [ -d "$lang_dir" ]; then
                    local lang=$(basename "$lang_dir")
                    check_directory "$lang_dir" "Language '$lang' in '$content_type'"
                    validate_markdown_files "$lang_dir" "$content_type/$lang"
                    ((lang_count++))
                fi
            done
            
            if [ $lang_count -eq 0 ]; then
                print_status "WARN" "No language directories found in $content_type"
            fi
        fi
    done < <(get_enabled_content_types)
}

# Function to get enabled content types from content-kinds.toml
get_enabled_content_types() {
    local content_kinds_file="$CONTENT_DIR/content-kinds.toml"
    
    if [ ! -f "$content_kinds_file" ]; then
        echo "blog"
        echo "recipes"
        return
    fi
    
    # Simplified parsing - look for directory entries that aren't commented out
    grep -E '^\s*directory\s*=' "$content_kinds_file" | \
    sed 's/.*directory\s*=\s*"\([^"]*\)".*/\1/' | \
    while read -r dir; do
        if [ -d "$CONTENT_DIR/$dir" ]; then
            echo "$dir"
        fi
    done
}

# Function to check server can process content
validate_server_processing() {
    print_status "INFO" "=== Server Content Processing Validation ==="
    
    # Check if server binary exists
    if [ ! -f "$PROJECT_ROOT/target/debug/server" ] && [ ! -f "$PROJECT_ROOT/target/release/server" ]; then
        print_status "WARN" "Server binary not found - run 'cargo build --features content-static' first"
        return
    fi
    
    print_status "PASS" "Server binary is available for content processing"
}

# Function to validate i18n structure
validate_i18n_structure() {
    print_status "INFO" "=== Internationalization Structure Validation ==="
    
    local locales_dir="$CONTENT_DIR/locales"
    if check_directory "$locales_dir" "Locales"; then
        # Check for language directories in locales
        for lang_dir in "$locales_dir"/*/; do
            if [ -d "$lang_dir" ]; then
                local lang=$(basename "$lang_dir")
                check_directory "$lang_dir" "Locale '$lang'"
                
                # Check for common.ftl
                if [ -f "$lang_dir/common.ftl" ]; then
                    print_status "PASS" "common.ftl exists for language '$lang'"
                else
                    print_status "WARN" "common.ftl missing for language '$lang'"
                fi
            fi
        done
    fi
    
    # Check menus
    local menus_dir="$CONTENT_DIR/menus"
    if check_directory "$menus_dir" "Menus"; then
        for menu_file in "$menus_dir"/*.toml; do
            if [ -f "$menu_file" ]; then
                local lang=$(basename "$menu_file" .toml)
                print_status "PASS" "Menu configuration exists for language '$lang'"
            fi
        done
    fi
}

# Function to remove deprecated files/directories
cleanup_deprecated() {
    print_status "INFO" "=== Cleanup Deprecated Build Outputs ==="
    
    # The /public directory is now deprecated since server processes content dynamically
    if [ -d "$PROJECT_ROOT/public" ]; then
        local item_count=$(find "$PROJECT_ROOT/public" -type f | wc -l)
        if [ "$item_count" -gt 5 ]; then  # Allow for a few static files
            print_status "WARN" "Found $item_count files in /public - consider removing pre-built HTML"
            print_status "INFO" "Server now processes content dynamically from /content"
        else
            print_status "PASS" "/public directory has minimal static files"
        fi
    fi
}

# Main validation function
main() {
    echo "🔍 Content Consistency Validation"
    echo "=================================="
    echo "Content directory: $CONTENT_DIR"
    echo "Project root: $PROJECT_ROOT"
    echo ""
    
    validate_content_structure
    echo ""
    
    validate_i18n_structure
    echo ""
    
    validate_server_processing
    echo ""
    
    cleanup_deprecated
    echo ""
    
    # Summary
    echo "🏁 Validation Summary"
    echo "===================="
    echo "Total checks: $TOTAL_CHECKS"
    echo "Passed: $PASSED_CHECKS"
    echo "Failed: $FAILED_CHECKS"
    
    if [ $FAILED_CHECKS -eq 0 ]; then
        print_status "PASS" "All content validation checks passed!"
        echo ""
        echo "✅ Content structure is consistent and ready for server processing"
        echo "🚀 You can run: cargo run --features content-static"
        exit 0
    else
        echo ""
        echo "❌ Content validation found $FAILED_CHECKS issues"
        echo "🔧 Please fix the issues above before building"
        exit 1
    fi
}

# Run main function
main "$@"