#!/bin/bash
# Extract function fingerprints from ELF binaries
# Outputs JSON array with function name, offset, size, and hashes

set -euo pipefail

DIR="${1:-.}"

extract_functions_from_binary() {
    local binary="$1"
    
    # Skip non-ELF files
    file "$binary" 2>/dev/null | grep -q "ELF" || return 0
    
    # Get function symbols with objdump
    objdump -t "$binary" 2>/dev/null | \
    awk '/\.text.*[0-9a-f]+.*F/ {
        gsub(/\*.*\*/, "", $1)
        if ($5 != "" && length($4) > 0) {
            size = strtonum("0x" $4)
            if (size >= 16) {
                print $1, $4, $NF
            }
        }
    }' | while read -r offset size name; do
        # Skip compiler-generated symbols
        case "$name" in
            __*|_GLOBAL_*|.plt*|.text*|frame_dummy|register_tm_clones|deregister_tm_clones|_start|_init|_fini)
                continue
                ;;
        esac
        
        # Convert hex size
        dec_size=$((16#$size))
        
        # Compute hash of function bytes
        local hash=$(objdump -d --start-address="0x$offset" --stop-address="$((16#$offset + dec_size))" "$binary" 2>/dev/null | \
            grep -E "^[[:space:]]*[0-9a-f]+:" | \
            awk '{for(i=2;i<=NF;i++){if($i~/^[0-9a-f]{2}$/){printf "%s", $i}}}' | \
            sha256sum | cut -d' ' -f1)
        
        [ -n "$hash" ] || hash="unknown"
        
        printf '{"name":"%s","offset":"0x%s","size":%d,"hash":"%s"}\n' \
            "$name" "$offset" "$dec_size" "$hash"
    done
}

# Output JSON array
echo "["
first=true

find "$DIR" -type f \( -executable -o -name "*.so" -o -name "*.so.*" \) 2>/dev/null | while read -r binary; do
    file "$binary" 2>/dev/null | grep -q "ELF" || continue
    
    extract_functions_from_binary "$binary" | while read -r json; do
        [ -z "$json" ] && continue
        if [ "$first" = "true" ]; then
            first=false
            echo "$json"
        else
            echo ",$json"
        fi
    done
done

echo "]"