#!/usr/bin/env bash
# CWE-407 scan — SpiderMonkey JS engine (IonMonkey compiler, CFG, SCC)
set -euo pipefail
echo "# scan=spidermonkey host=$(hostname) date=$(date -u +%Y-%m-%dT%H:%M:%SZ)"

# Sparse clone of gecko-dev — only js/src
cd /tmp
git clone --depth 1 --filter=blob:none --sparse \
  https://github.com/mozilla/gecko-dev geckosrc 2>&1 | tail -1
cd geckosrc
git sparse-checkout set js/src 2>&1 | tail -1
SMROOT=/tmp/geckosrc/js/src

GRAPH_KW="topolog|scc\b|strongly.connected|cycle|dfs\b|postorder|dominator|liveness|reachab|tarjan|register.alloc|ion\b"
C_MEM='std::find[(]|\.Contains[(]|IsInVector\b|vector.*find\b|linear_search|Vector.*Contains|NodeVector.*find'

[ -d "$SMROOT" ] || { echo "# ERROR: SpiderMonkey source not found"; exit 0; }
echo "# roots: $SMROOT"
find "$SMROOT" -name "*.cpp" -o -name "*.h" 2>/dev/null | sort | while IFS= read -r f; do
  gl=$(grep -inE "$GRAPH_KW" "$f" 2>/dev/null | cut -d: -f1 | head -5 || true)
  [ -z "$gl" ] && continue
  while IFS= read -r ln; do
    s=$(( ln > 20 ? ln - 20 : 1 )); e=$(( ln + 20 ))
    hit=$(sed -n "${s},${e}p" "$f" 2>/dev/null | grep -iE "$C_MEM" | head -1 | sed 's/^\s*//' || true)
    [ -n "$hit" ] && echo "CANDIDATE	${f}:${ln}	${hit}"
  done <<< "$gl"
done
echo "# scan complete"