analyze-project-optimized.sh

#!/bin/bash

# Optimized Project Analysis Script
# Provides comprehensive statistics about project files and structure
# Optimized for performance while maintaining accuracy

set -e

PROJECT_ROOT="$(pwd)"
TIMESTAMP=$(date +"%Y-%m-%d_%H-%M-%S")
OUTPUT_FILE="project-analysis-${TIMESTAMP}.txt"

echo "🔍 ModularHomeInventory Project Analysis (Optimized)"
echo "===================================================="
echo "Generated: $(date)"
echo "Project Root: $PROJECT_ROOT"
echo ""

# Function to format file sizes
format_size() {
    local size=$1
    if [ $size -gt 1073741824 ]; then
        printf "%.1fGB" $(echo "scale=1; $size / 1073741824" | bc -l)
    elif [ $size -gt 1048576 ]; then
        printf "%.1fMB" $(echo "scale=1; $size / 1048576" | bc -l)
    elif [ $size -gt 1024 ]; then
        printf "%.1fKB" $(echo "scale=1; $size / 1024" | bc -l)
    else
        echo "${size}B"
    fi
}

# Function to count lines efficiently
count_lines_safe() {
    local file="$1"
    if [ -f "$file" ] && [ -r "$file" ]; then
        wc -l < "$file" 2>/dev/null | tr -d ' ' || echo "0"
    else
        echo "0"
    fi
}

# Create temporary files for efficiency
TEMP_DIR=$(mktemp -d)
trap "rm -rf $TEMP_DIR" EXIT

echo "📊 Collecting file information (this may take a moment)..."

# Efficiently collect all file information at once
find . -type f -exec stat -f "%z %N" {} \; 2>/dev/null > "$TEMP_DIR/all_files_with_sizes.txt" &
find . -type d > "$TEMP_DIR/all_dirs.txt" &
find . -name "*.swift" > "$TEMP_DIR/swift_files.txt" &
wait

{
    echo "🔍 ModularHomeInventory Project Analysis (Optimized)"
    echo "===================================================="
    echo "Generated: $(date)"
    echo "Project Root: $PROJECT_ROOT"
    echo ""

    # Overall Statistics
    echo "📊 OVERALL PROJECT STATISTICS"
    echo "=============================="
    
    total_files=$(wc -l < "$TEMP_DIR/all_files_with_sizes.txt" | tr -d ' ')
    total_dirs=$(wc -l < "$TEMP_DIR/all_dirs.txt" | tr -d ' ')
    total_size=$(awk '{sum+=$1} END {print sum+0}' "$TEMP_DIR/all_files_with_sizes.txt")
    
    echo "Total Files: $total_files"
    echo "Total Directories: $total_dirs"
    echo "Total Size: $(format_size $total_size)"
    echo ""

    # File Extension Analysis - Optimized
    echo "📁 FILE EXTENSION ANALYSIS"
    echo "=========================="
    echo "Extension | Count | Total Size | Avg Size"
    echo "----------|-------|------------|----------"
    
    # Extract extensions and calculate stats
    awk '{
        file = $2;
        size = $1;
        if (match(file, /\.([^.\/]+)$/, arr)) {
            ext = arr[1];
            count[ext]++;
            total_size[ext] += size;
        }
    }
    END {
        for (ext in count) {
            avg = total_size[ext] / count[ext];
            printf "%s %d %d %.0f\n", ext, count[ext], total_size[ext], avg;
        }
    }' "$TEMP_DIR/all_files_with_sizes.txt" | sort -k2,2nr | head -25 | \
    while read ext count size avg; do
        printf "%-9s | %-5s | %-10s | %s\n" "$ext" "$count" "$(format_size $size)" "$(format_size $avg)"
    done
    echo ""

    # Swift File Analysis - Optimized
    echo "🐦 SWIFT FILE ANALYSIS"
    echo "======================"
    
    swift_count=$(wc -l < "$TEMP_DIR/swift_files.txt" | tr -d ' ')
    
    if [ $swift_count -gt 0 ]; then
        # Calculate Swift file sizes
        grep "\.swift$" "$TEMP_DIR/all_files_with_sizes.txt" | awk '{sum+=$1} END {print sum+0}' > "$TEMP_DIR/swift_total_size.txt"
        swift_size=$(cat "$TEMP_DIR/swift_total_size.txt")
        
        echo "Total Swift Files: $swift_count"
        echo "Total Swift Size: $(format_size $swift_size)"
        
        # Count lines in Swift files efficiently using parallel processing
        echo "Calculating Swift lines (parallel processing)..."
        cat "$TEMP_DIR/swift_files.txt" | xargs -n 50 -P 8 wc -l 2>/dev/null | awk '{sum+=$1} END {print sum+0}' > "$TEMP_DIR/swift_lines.txt" &
        
        # Continue with other analysis while lines are being counted
        echo "Average Swift File Size: $(format_size $((swift_size / swift_count)))"
        echo ""

        # Longest Swift Files by Line Count - More Efficient
        echo "📏 LONGEST SWIFT FILES (Top 20)"
        echo "==============================="
        echo "Lines | File"
        echo "------|----"
        
        # Use parallel processing for line counting
        cat "$TEMP_DIR/swift_files.txt" | head -200 | xargs -n 10 -P 8 -I {} sh -c '
            for file in "$@"; do
                if [ -f "$file" ] && [ -r "$file" ]; then
                    lines=$(wc -l < "$file" 2>/dev/null | tr -d " " || echo "0")
                    echo "$lines|$file"
                fi
            done
        ' _ {} | sort -t'|' -k1,1nr | head -20 | \
        while IFS='|' read -r lines file; do
            printf "%-5s | %s\n" "$lines" "$file"
        done
        
        # Wait for total lines calculation to finish
        wait
        swift_total_lines=$(cat "$TEMP_DIR/swift_lines.txt")
        echo ""
        echo "Total Swift Lines: $swift_total_lines"
        echo ""
    else
        echo "No Swift files found."
        echo ""
    fi

    # Largest Files by Size - Optimized
    echo "📦 LARGEST FILES (Top 20)"
    echo "========================="
    echo "Size | File"
    echo "-----|----"
    
    sort -k1,1nr "$TEMP_DIR/all_files_with_sizes.txt" | head -20 | \
    while read size file; do
        printf "%-10s | %s\n" "$(format_size $size)" "$file"
    done
    echo ""

    # Largest Directories - More Efficient
    echo "📂 LARGEST DIRECTORIES (Top 20)"
    echo "==============================="
    echo "Size | Directory"
    echo "-----|----------"
    
    # Calculate directory sizes more efficiently
    awk '{
        file = $2;
        size = $1;
        # Extract directory path
        if (match(file, /^(.*)\/[^\/]+$/, arr)) {
            dir = arr[1];
            if (dir == "") dir = ".";
            dir_size[dir] += size;
        }
    }
    END {
        for (dir in dir_size) {
            printf "%d|%s\n", dir_size[dir], dir;
        }
    }' "$TEMP_DIR/all_files_with_sizes.txt" | sort -t'|' -k1,1nr | head -20 | \
    while IFS='|' read -r size dir; do
        printf "%-10s | %s\n" "$(format_size $size)" "$dir"
    done
    echo ""

    # Module Analysis (Swift Package Modules) - Optimized
    echo "🏗️ MODULE ANALYSIS"
    echo "=================="
    echo "Module | Swift Files | Lines | Size"
    echo "-------|-------------|-------|-----"
    
    find . -name "Package.swift" -type f | while read package; do
        module_dir=$(dirname "$package")
        module_name=$(basename "$module_dir")
        
        # Count Swift files in this module
        swift_count=$(find "$module_dir" -name "*.swift" -type f | wc -l | tr -d ' ')
        
        if [ $swift_count -gt 0 ]; then
            # Calculate lines and size for this module
            swift_lines=$(find "$module_dir" -name "*.swift" -type f -exec wc -l {} \; 2>/dev/null | awk '{sum+=$1} END {print sum+0}')
            swift_size=$(find "$module_dir" -name "*.swift" -type f -exec stat -f%z {} \; 2>/dev/null | awk '{sum+=$1} END {print sum+0}')
            
            printf "%s|%s|%s|%s\n" "$module_name" "$swift_count" "$swift_lines" "$swift_size"
        fi
    done | sort -t'|' -k3,3nr | while IFS='|' read -r name count lines size; do
        printf "%-25s | %-11s | %-5s | %s\n" "$name" "$count" "$lines" "$(format_size $size)"
    done
    echo ""

    # Code Complexity Indicators - Optimized
    echo "🧮 CODE COMPLEXITY INDICATORS"
    echo "============================="
    
    if [ $swift_count -gt 0 ]; then
        # Use parallel grep for better performance
        echo "Analyzing code constructs..."
        
        cat "$TEMP_DIR/swift_files.txt" | xargs -P 8 grep -h "^class \|^struct \|^protocol \|^extension \|^enum \|func " 2>/dev/null | \
        awk '
        /^class / { classes++ }
        /^struct / { structs++ } 
        /^protocol / { protocols++ }
        /^extension / { extensions++ }
        /^enum / { enums++ }
        /func / { functions++ }
        END {
            print "Classes: " (classes+0)
            print "Structs: " (structs+0) 
            print "Protocols: " (protocols+0)
            print "Extensions: " (extensions+0)
            print "Enums: " (enums+0)
            print "Functions: " (functions+0)
        }'
    else
        echo "No Swift files to analyze for complexity."
    fi
    echo ""

    # Test File Analysis - Optimized
    echo "🧪 TEST FILE ANALYSIS"
    echo "====================="
    
    test_files=$(grep -E "(Test|Tests)\.swift$" "$TEMP_DIR/swift_files.txt" | wc -l | tr -d ' ')
    
    if [ $test_files -gt 0 ]; then
        test_lines=$(grep -E "(Test|Tests)\.swift$" "$TEMP_DIR/swift_files.txt" | xargs wc -l 2>/dev/null | tail -1 | awk '{print $1}')
        test_size=$(grep -E "(Test|Tests)\.swift$" "$TEMP_DIR/swift_files.txt" | xargs -I {} stat -f%z {} 2>/dev/null | awk '{sum+=$1} END {print sum+0}')
        
        echo "Test Files: $test_files"
        echo "Test Lines: $test_lines"
        echo "Test Size: $(format_size $test_size)"
        
        if [ $swift_count -gt 0 ]; then
            test_coverage_ratio=$(echo "scale=1; $test_files * 100 / $swift_count" | bc -l 2>/dev/null || echo "0")
            echo "Test Coverage Ratio: ${test_coverage_ratio}% (test files / total swift files)"
        fi
    else
        echo "No test files found."
    fi
    echo ""

    # Configuration Files - Optimized
    echo "⚙️ CONFIGURATION FILES"
    echo "======================"
    
    config_patterns=("Package.swift" "Makefile" "project.yml" "Podfile" "Cartfile" "Gemfile" "Fastfile" ".swiftlint.yml" ".swiftformat" "Info.plist")
    
    for pattern in "${config_patterns[@]}"; do
        count=$(grep -c "$pattern$" "$TEMP_DIR/all_files_with_sizes.txt" 2>/dev/null || echo "0")
        if [ $count -gt 0 ]; then
            echo "$pattern: $count files"
            grep "$pattern$" "$TEMP_DIR/all_files_with_sizes.txt" | head -5 | awk '{print "  - " $2}'
        fi
    done
    echo ""

    # File Age Analysis (if git is available) - Optimized
    if command -v git >/dev/null 2>&1 && [ -d .git ]; then
        echo "📅 FILE AGE ANALYSIS (Git History)"
        echo "=================================="
        
        echo "Recently Modified Files (Last 7 days):"
        git log --since="7 days ago" --name-only --pretty=format: | grep -v "^$" | sort | uniq -c | sort -nr | head -10 | \
        while read count file; do
            if [ -n "$file" ] && [ "$file" != " " ]; then
                echo "  $count changes: $file"
            fi
        done
        echo ""
        
        echo "Most Active Files (All Time - Top 10):"
        git log --name-only --pretty=format: | grep -v "^$" | sort | uniq -c | sort -nr | head -10 | \
        while read count file; do
            if [ -n "$file" ] && [ "$file" != " " ]; then
                echo "  $count changes: $file"
            fi
        done
        echo ""
    fi

    # Archive Analysis
    echo "🗄️ ARCHIVE ANALYSIS"
    echo "==================="
    
    zip_count=$(grep "\.zip$" "$TEMP_DIR/all_files_with_sizes.txt" | wc -l | tr -d ' ')
    if [ $zip_count -gt 0 ]; then
        echo "Archive Files Found:"
        grep "\.zip$" "$TEMP_DIR/all_files_with_sizes.txt" | while read size file; do
            echo "  $(basename "$file") - $(format_size $size)"
        done
    else
        echo "No archive files found."
    fi
    echo ""

    # Build System Analysis
    echo "🔧 BUILD SYSTEM ANALYSIS"
    echo "========================"
    
    packages=$(grep "Package.swift$" "$TEMP_DIR/all_files_with_sizes.txt" | wc -l | tr -d ' ')
    makefiles=$(grep "Makefile$" "$TEMP_DIR/all_files_with_sizes.txt" | wc -l | tr -d ' ')
    xcodeproj=$(grep "\.xcodeproj/" "$TEMP_DIR/all_files_with_sizes.txt" | wc -l | tr -d ' ')
    
    echo "Swift Packages: $packages"
    echo "Makefiles: $makefiles" 
    echo "Xcode Project Files: $xcodeproj"
    echo ""

    # Summary
    echo "📋 SUMMARY"
    echo "=========="
    project_type="medium-sized"
    if [ $total_files -gt 5000 ]; then project_type="very large"
    elif [ $total_files -gt 2000 ]; then project_type="large"
    elif [ $total_files -gt 500 ]; then project_type="medium-sized"
    else project_type="small"; fi
    
    echo "This is a $project_type Swift project with:"
    echo "• $swift_count Swift source files"
    if [ -f "$TEMP_DIR/swift_lines.txt" ]; then
        echo "• $(cat "$TEMP_DIR/swift_lines.txt") total lines of Swift code"
    fi
    echo "• $test_files test files"
    echo "• $packages Swift Package modules"
    echo "• Total project size: $(format_size $total_size)"
    echo "• Most common file type: $(awk '{file=$2; if(match(file,/\.([^.\/]+)$/,arr)) count[arr[1]]++} END {max=0; for(ext in count) if(count[ext]>max) {max=count[ext]; maxext=ext}} END {print maxext}' "$TEMP_DIR/all_files_with_sizes.txt")"
    echo ""
    echo "Generated: $(date)"
    echo "Analysis completed in optimized mode."
    
} | tee "$OUTPUT_FILE"

echo ""
echo "✅ Analysis complete! Results saved to: $OUTPUT_FILE"
echo "📊 Quick Summary:"
echo "   • Total files: $(wc -l < "$TEMP_DIR/all_files_with_sizes.txt" | tr -d ' ')"
echo "   • Swift files: $(wc -l < "$TEMP_DIR/swift_files.txt" | tr -d ' ')"
echo "   • Project size: $(format_size $(awk '{sum+=$1} END {print sum+0}' "$TEMP_DIR/all_files_with_sizes.txt"))"