#!/bin/bash
# DBIS Cross-Reference Verification Script (Simplified)
# Automated link verification and cross-reference checking

set -e

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

# Configuration
PROJECT_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
REPORT_FILE="${PROJECT_ROOT}/CROSS_REFERENCE_VERIFICATION_REPORT.md"
BROKEN_LINKS_FILE="${PROJECT_ROOT}/BROKEN_LINKS.txt"
TOTAL_LINKS=0
VALID_LINKS=0
BROKEN_COUNT=0

echo "=========================================="
echo "DBIS Cross-Reference Verification"
echo "=========================================="
echo ""
echo -e "${BLUE}Project Root:${NC} $PROJECT_ROOT"
echo ""

# Initialize report
cat > "$REPORT_FILE" << 'EOF'
# CROSS-REFERENCE VERIFICATION REPORT
## Automated Link Verification Results

**Generated:** 
**Project Root:** 

---

## SUMMARY

- **Total Links Scanned:** 0
- **Valid Links:** 0
- **Broken Links:** 0
- **Success Rate:** 0.00%

---

## BROKEN LINKS

EOF

# Function to check if file exists
check_file_exists() {
    local file_path="$1"
    local source_file="$2"
    local source_dir="$(dirname "$source_file")"
    
    # Handle different path types
    if [[ "$file_path" == http://* ]] || [[ "$file_path" == https://* ]] || [[ "$file_path" == mailto:* ]]; then
        return 0  # External links - skip for now
    fi
    
    # Handle relative paths
    if [[ "$file_path" == /* ]]; then
        # Absolute path from project root
        full_path="${PROJECT_ROOT}${file_path}"
    elif [[ "$file_path" == ../* ]]; then
        # Relative path going up
        resolved_path="$(cd "$source_dir" && cd "$(dirname "$file_path")" 2>/dev/null && pwd)/$(basename "$file_path")"
        if [[ -f "$resolved_path" ]]; then
            return 0
        fi
    else
        # Relative path in same directory or subdirectory
        full_path="${source_dir}/${file_path}"
    fi
    
    # Remove anchor if present
    file_only="${full_path%%#*}"
    
    if [[ -f "$file_only" ]]; then
        return 0
    else
        return 1
    fi
}

# Main verification process
echo -e "${BLUE}Scanning markdown files...${NC}"
echo ""

find "$PROJECT_ROOT" -name "*.md" -type f | while IFS= read -r file; do
    # Skip certain directories
    if [[ "$file" == *"/node_modules/"* ]] || [[ "$file" == *"/.git/"* ]] || [[ "$file" == *"/scripts/"* ]]; then
        continue
    fi
    
    relative_file="${file#$PROJECT_ROOT/}"
    
    # Extract markdown links using grep
    grep -oE '\[([^\]]+)\]\(([^)]+)\)' "$file" 2>/dev/null | while IFS= read -r link_match; do
        # Extract the path part (between parentheses)
        link_path=$(echo "$link_match" | sed -n 's/.*(\(.*\))/\1/p')
        
        if [[ -z "$link_path" ]]; then
            continue
        fi
        
        TOTAL_LINKS=$((TOTAL_LINKS + 1))
        
        # Skip external links
        if [[ "$link_path" == http://* ]] || [[ "$link_path" == https://* ]] || [[ "$link_path" == mailto:* ]]; then
            VALID_LINKS=$((VALID_LINKS + 1))
            continue
        fi
        
        # Check if file exists
        if check_file_exists "$link_path" "$file"; then
            VALID_LINKS=$((VALID_LINKS + 1))
        else
            BROKEN_COUNT=$((BROKEN_COUNT + 1))
            echo -e "${RED}✗${NC} $relative_file -> $link_path" | tee -a "$BROKEN_LINKS_FILE"
        fi
    done
done

# Update report with summary
{
    echo "**Generated:** $(date -u +"%Y-%m-%d %H:%M:%S UTC")"
    echo "**Project Root:** $PROJECT_ROOT"
    echo ""
    echo "---"
    echo ""
    echo "## SUMMARY"
    echo ""
    echo "- **Total Links Scanned:** $TOTAL_LINKS"
    echo "- **Valid Links:** $VALID_LINKS"
    echo "- **Broken Links:** $BROKEN_COUNT"
    if [[ $TOTAL_LINKS -gt 0 ]]; then
        SUCCESS_RATE=$(awk "BEGIN {printf \"%.2f\", ($VALID_LINKS/$TOTAL_LINKS)*100}")
        echo "- **Success Rate:** ${SUCCESS_RATE}%"
    else
        echo "- **Success Rate:** N/A"
    fi
    echo ""
    echo "---"
    echo ""
    echo "## BROKEN LINKS"
    echo ""
    if [[ -f "$BROKEN_LINKS_FILE" ]]; then
        cat "$BROKEN_LINKS_FILE"
    else
        echo "✅ No broken links found!"
    fi
    echo ""
    echo "---"
    echo ""
    echo "## RECOMMENDATIONS"
    echo ""
    echo "1. Fix all broken links identified above"
    echo "2. Verify and correct invalid paths"
    echo "3. Update cross-references in affected documents"
    echo "4. Re-run verification after fixes"
    echo ""
    echo "---"
    echo ""
    echo "**END OF VERIFICATION REPORT**"
} > "$REPORT_FILE"

echo ""
echo "=========================================="
echo "Verification Summary"
echo "=========================================="
echo ""
echo -e "${GREEN}Total Links Scanned:${NC} $TOTAL_LINKS"
echo -e "${GREEN}Valid Links:${NC} $VALID_LINKS"
echo -e "${RED}Broken Links:${NC} $BROKEN_COUNT"
if [[ $TOTAL_LINKS -gt 0 ]]; then
    SUCCESS_RATE=$(awk "BEGIN {printf \"%.2f\", ($VALID_LINKS/$TOTAL_LINKS)*100}")
    echo -e "${BLUE}Success Rate:${NC} ${SUCCESS_RATE}%"
fi
echo ""
echo -e "${BLUE}Report generated:${NC} $REPORT_FILE"
if [[ -f "$BROKEN_LINKS_FILE" ]]; then
    echo -e "${BLUE}Broken links list:${NC} $BROKEN_LINKS_FILE"
fi
echo ""
echo -e "${GREEN}✓ Verification complete!${NC}"