website-htmx-rustelo-code/scripts/sh/content/content-manager.sh

#!/bin/bash

# Unified Content Management Script
# Replaces: generate-index.sh, validate-id-consistency.sh, validate-content.sh
# Usage: ./content-manager.sh [command] [options]

set -e

# Get script directory and load shared library
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
source "$SCRIPT_DIR/lib.sh"

# =============================================================================
# COMMAND FUNCTIONS
# =============================================================================

# Generate JSON indices from markdown frontmatter
cmd_generate_indices() {
    log_step "Generating content indices..."
    
    local content_types=($(get_enabled_content_types))
    local languages=($(get_supported_languages))
    local total_errors=0
    
    for content_type in "${content_types[@]}"; do
        for language in "${languages[@]}"; do
            if ! content_dir_exists "$content_type" "$language"; then
                log_warning "Content directory not found: ${CONTENT_DIR}/${content_type}/${language}"
                continue
            fi
            
            log_step "Generating $content_type index for language: $language"
            
            local content_dir="${CONTENT_DIR}/${content_type}/${language}"
            local index_file="$content_dir/index.json"
            local temp_entries="/tmp/index_entries_$$"
            local entry_count=0
            
            # Clear temp file
            > "$temp_entries"
            
            # Process each markdown file
            for md_file in $(get_markdown_files "$content_type" "$language"); do
                if process_markdown_file "$md_file" "$content_type" "$language" > /tmp/json_entry_$$; then
                    # Add comma if not first entry
                    if [ $entry_count -gt 0 ]; then
                        echo "," >> "$temp_entries"
                    fi
                    cat /tmp/json_entry_$$ >> "$temp_entries"
                    entry_count=$((entry_count + 1))
                    rm -f /tmp/json_entry_$$
                else
                    log_warning "Failed to process: $md_file"
                fi
            done
            
            # Determine array name (keep legacy naming for recipes)
            local array_name="${content_type}_posts"
            if [ "$content_type" = "recipes" ]; then
                array_name="prescriptions"  # Keep legacy array name for recipes
            fi
            
            # Build final index.json
            cat > "$index_file" << EOF
{
  "$array_name": [
$(cat "$temp_entries")
  ],
  "language": "$language"
}
EOF
            
            # Validate JSON
            if validate_json "$index_file"; then
                log_success "Generated valid $content_type index ($language): $entry_count entries"
            else
                log_error "Generated invalid JSON for $content_type ($language)"
                total_errors=$((total_errors + 1))
            fi
            
            # Cleanup
            rm -f "$temp_entries"
        done
    done
    
    if [ $total_errors -eq 0 ]; then
        log_success "🎉 Index generation completed successfully!"
    else
        log_error "Index generation completed with $total_errors errors"
        exit 1
    fi
}

# Validate ID consistency across languages
cmd_validate_ids() {
    log_step "Validating ID consistency across languages..."
    
    local content_types=($(get_enabled_content_types))
    local languages=($(get_supported_languages))
    local total_errors=0
    local primary_lang="en"
    
    for content_type in "${content_types[@]}"; do
        log_step "Validating $content_type ID consistency..."
        
        local primary_dir="${CONTENT_DIR}/${content_type}/${primary_lang}"
        if [ ! -d "$primary_dir" ]; then
            log_error "Primary language directory not found: $primary_dir"
            total_errors=$((total_errors + 1))
            continue
        fi
        
        local temp_ids="/tmp/id_validation_${content_type}_$$"
        > "$temp_ids"
        
        # Collect all IDs from primary language
        for primary_file in $(get_markdown_files "$content_type" "$primary_lang"); do
            local filename=$(basename "$primary_file" .md)
            local primary_id=$(extract_id "$primary_file")
            
            if [ -z "$primary_id" ]; then
                log_warning "No ID found in primary file: $primary_file"
                continue
            fi
            
            echo "$filename:$primary_id" >> "$temp_ids"
            
            # Check corresponding files in other languages
            for lang in "${languages[@]}"; do
                if [ "$lang" = "$primary_lang" ]; then
                    continue
                fi
                
                local lang_file="${CONTENT_DIR}/${content_type}/${lang}/${filename}.md"
                
                if [ ! -f "$lang_file" ]; then
                    log_warning "Missing $lang translation for $filename"
                    continue
                fi
                
                local lang_id=$(extract_id "$lang_file")
                
                if [ -z "$lang_id" ]; then
                    log_error "No ID found in $lang file: $lang_file"
                    total_errors=$((total_errors + 1))
                    continue
                fi
                
                if [ "$primary_id" != "$lang_id" ]; then
                    log_error "ID mismatch for $filename: $primary_lang='$primary_id' vs $lang='$lang_id'"
                    total_errors=$((total_errors + 1))
                fi
            done
        done
        
        # Check for duplicate IDs within each language
        for lang in "${languages[@]}"; do
            if ! content_dir_exists "$content_type" "$lang"; then
                continue
            fi
            
            local temp_lang_ids="/tmp/lang_ids_${content_type}_${lang}_$$"
            > "$temp_lang_ids"
            
            for md_file in $(get_markdown_files "$content_type" "$lang"); do
                local file_id=$(extract_id "$md_file")
                if [ -n "$file_id" ]; then
                    echo "$file_id" >> "$temp_lang_ids"
                fi
            done
            
            # Check for duplicates
            local duplicates=$(sort "$temp_lang_ids" | uniq -d)
            if [ -n "$duplicates" ]; then
                echo "$duplicates" | while read -r duplicate_id; do
                    log_error "Duplicate ID '$duplicate_id' found in $content_type ($lang)"
                    total_errors=$((total_errors + 1))
                done
            fi
            
            rm -f "$temp_lang_ids"
        done
        
        rm -f "$temp_ids"
    done
    
    if [ $total_errors -eq 0 ]; then
        log_success "🎉 All content IDs are consistent across languages!"
    else
        log_error "❌ Content validation failed with $total_errors issue(s)"
        exit 1
    fi
}

# Validate content structure and quality
cmd_validate_content() {
    log_step "Validating content structure and quality..."
    
    local content_types=($(get_enabled_content_types))
    local languages=($(get_supported_languages))
    local total_issues=0
    
    for content_type in "${content_types[@]}"; do
        for language in "${languages[@]}"; do
            if ! content_dir_exists "$content_type" "$language"; then
                log_warning "Content directory not found: ${CONTENT_DIR}/${content_type}/${language}"
                continue
            fi
            
            log_step "Validating $content_type content for language: $language"
            local issues_found=0
            
            # Check for index.json
            local index_file="${CONTENT_DIR}/${content_type}/${language}/index.json"
            if [ -f "$index_file" ]; then
                if validate_json "$index_file"; then
                    log_success "Valid index.json for $content_type ($language)"
                else
                    log_error "Invalid index.json for $content_type ($language)"
                    issues_found=$((issues_found + 1))
                fi
            else
                log_warning "No index.json found for $content_type ($language)"
                issues_found=$((issues_found + 1))
            fi
            
            # Validate markdown files
            local md_files=($(get_markdown_files "$content_type" "$language"))
            local valid_files=0
            
            for md_file in "${md_files[@]}"; do
                if has_valid_frontmatter "$md_file"; then
                    valid_files=$((valid_files + 1))
                else
                    log_error "Invalid or missing frontmatter: $md_file"
                    issues_found=$((issues_found + 1))
                fi
            done
            
            log_info "$content_type ($language): $valid_files/${#md_files[@]} files with valid frontmatter"
            total_issues=$((total_issues + issues_found))
        done
    done
    
    if [ $total_issues -eq 0 ]; then
        log_success "🎉 All content validation passed!"
    else
        log_error "❌ Content validation found $total_issues issue(s)"
        exit 1
    fi
}

# Show content statistics
cmd_stats() {
    log_step "Generating content statistics..."
    
    local content_types=($(get_enabled_content_types))
    local languages=($(get_supported_languages))
    
    echo ""
    echo "📊 Content Statistics:"
    echo "======================"
    
    for content_type in "${content_types[@]}"; do
        echo ""
        echo "📝 Content Type: $content_type"
        echo "--------------------------------"
        
        local total_files=0
        for language in "${languages[@]}"; do
            if content_dir_exists "$content_type" "$language"; then
                local md_files=($(get_markdown_files "$content_type" "$language"))
                local count=${#md_files[@]}
                echo "  $language: $count files"
                total_files=$((total_files + count))
            else
                echo "  $language: 0 files (directory not found)"
            fi
        done
        echo "  Total: $total_files files"
    done
    
    echo ""
    echo "🌍 Languages: ${languages[*]}"
    echo "📂 Content Types: ${content_types[*]}"
    echo "📍 Content Directory: $CONTENT_DIR"
}

# =============================================================================
# MAIN COMMAND DISPATCHER
# =============================================================================

show_help() {
    cat << EOF
Unified Content Management Script

Usage: $0 [command] [options]

Commands:
  generate-indices     Generate index.json files from markdown frontmatter
  validate-ids        Validate ID consistency across languages  
  validate-content    Validate content structure and quality
  stats              Show content statistics
  all                Run generate-indices, validate-ids, and validate-content
  help               Show this help message

Options:
  --content-dir DIR   Content directory (default: content)
  --verbose          Enable verbose output

Examples:
  $0 generate-indices
  $0 validate-ids
  $0 validate-content  
  $0 all
  $0 stats

This script replaces the previous separate scripts:
  - generate-index.sh
  - validate-id-consistency.sh  
  - validate-content.sh

The script automatically reads content types from content-kinds.toml
and supports all enabled content types and languages.
EOF
}

# Parse command line arguments
VERBOSE=false
CONTENT_DIR_OVERRIDE=""

while [[ $# -gt 0 ]]; do
    case $1 in
        --content-dir)
            CONTENT_DIR_OVERRIDE="$2"
            shift 2
            ;;
        --verbose)
            VERBOSE=true
            shift
            ;;
        --help|-h|help)
            show_help
            exit 0
            ;;
        *)
            COMMAND="$1"
            shift
            ;;
    esac
done

# Override content directory if specified
if [ -n "$CONTENT_DIR_OVERRIDE" ]; then
    CONTENT_DIR="$CONTENT_DIR_OVERRIDE"
    CONTENT_KINDS_FILE="${CONTENT_DIR}/content-kinds.toml"
fi

# Initialize library
init_content_lib "content-manager"

# Execute command
case "${COMMAND:-help}" in
    generate-indices|indices|index)
        cmd_generate_indices
        ;;
    validate-ids|validate-id|ids)
        cmd_validate_ids
        ;;
    validate-content|validate|content)
        cmd_validate_content
        ;;
    stats|statistics|info)
        cmd_stats
        ;;
    all|full)
        cmd_generate_indices
        echo ""
        cmd_validate_ids  
        echo ""
        cmd_validate_content
        ;;
    help|--help|-h)
        show_help
        ;;
    *)
        log_error "Unknown command: $COMMAND"
        echo ""
        show_help
        exit 1
        ;;
esac