hakorune/tools/selfhost/selfhost_build.sh

#!/usr/bin/env bash
# selfhost_build.sh — Hybrid selfhost build (Stage‑B → Program(JSON v0) → optional run via Core‑Direct)
#
# Goals (80/20):
# - Take a Hako source (.hako), compile with Hakorune Stage‑B to Program(JSON v0).
# - Optionally run via Gate‑C/Core Direct (in‑proc) to verify exit code.
# - (Future) Optionally convert to MIR and build an executable via ny-llvmc.
#
# Usage:
#   tools/selfhost/selfhost_build.sh --in source.hako [--json out.json] [--run]
#   Env:
#     NYASH_BIN: path to hakorune/nyash binary (auto-detected if omitted)
#     NYASH_ROOT: repo root (auto-detected)
#
set -euo pipefail

ROOT="${NYASH_ROOT:-$(cd "$(dirname "$0")/../.." && pwd)}"
BIN="${NYASH_BIN:-}"
if [ -z "${BIN}" ]; then
  if [ -x "$ROOT/target/release/hakorune" ]; then BIN="$ROOT/target/release/hakorune";
  elif [ -x "$ROOT/target/release/nyash" ]; then BIN="$ROOT/target/release/nyash";
  else echo "[selfhost] error: NYASH_BIN not set and no binary found under target/release" >&2; exit 2; fi
fi

IN=""
JSON_OUT=""
MIR_OUT=""
EXE_OUT=""
DO_RUN=0

while [ $# -gt 0 ]; do
  case "$1" in
    --in) IN="$2"; shift 2;;
    --json) JSON_OUT="$2"; shift 2;;
    --run) DO_RUN=1; shift;;
    --mir) MIR_OUT="$2"; shift 2;;
    --exe) EXE_OUT="$2"; shift 2;;
    *) echo "[selfhost] unknown arg: $1" >&2; exit 2;;
  esac
done

if [ -z "$IN" ]; then echo "[selfhost] --in <file.hako> is required" >&2; exit 2; fi
if [ ! -f "$IN" ]; then echo "[selfhost] input not found: $IN" >&2; exit 2; fi

tmp_json="${JSON_OUT:-/tmp/hako_stageb_$$.json}"

# Emit Program(JSON v0; prefer BuildBox for emit-only when HAKO_USE_BUILDBOX=1)
RAW="/tmp/hako_stageb_raw_$$.txt"
SRC_CONTENT="$(cat "$IN")"
if [ "${HAKO_USE_BUILDBOX:-0}" = "1" ] && [ "$DO_RUN" = "0" ] && [ -z "$EXE_OUT" ]; then
  WRAP="/tmp/hako_buildbox_wrap_$$.hako"
  cat > "$WRAP" <<'HAKO'
include "lang/src/compiler/build/build_box.hako"
static box Main { method main(args) {
  local src = env.get("HAKO_SRC");
  local j = BuildBox.emit_program_json_v0(src, null);
  print(j);
  return 0;
} }
HAKO
  (
    export HAKO_SRC="$SRC_CONTENT"
    export NYASH_QUIET=0 HAKO_QUIET=0 NYASH_CLI_VERBOSE=0
    cd "$ROOT" && "$BIN" --backend vm "$WRAP"
  ) > "$RAW" 2>&1 || true
  rm -f "$WRAP" 2>/dev/null || true
else
  (
    export NYASH_PARSER_ALLOW_SEMICOLON=1
    export NYASH_ALLOW_USING_FILE=0
    export HAKO_ALLOW_USING_FILE=0
    export NYASH_USING_AST=1
    export HAKO_PARSER_STAGE3=1
    export NYASH_PARSER_STAGE3=1
    export NYASH_VARMAP_GUARD_STRICT=0
    export NYASH_BLOCK_SCHEDULE_VERIFY=0
    export NYASH_QUIET=0 HAKO_QUIET=0 NYASH_CLI_VERBOSE=0
    cd "$ROOT" && \
      "$BIN" --backend vm \
        "$ROOT/lang/src/compiler/entry/compiler_stageb.hako" -- \
        --source "$SRC_CONTENT"
  ) > "$RAW" 2>&1 || true
fi

if ! awk '(/"version":0/ && /"kind":"Program"/){print;found=1;exit} END{exit(found?0:1)}' "$RAW" > "$tmp_json"; then
  echo "[selfhost] Stage‑B emit failed" >&2
  tail -n 120 "$RAW" >&2 || true
  rm -f "$RAW" 2>/dev/null || true
  exit 1
fi
rm -f "$RAW" 2>/dev/null || true

if [ -n "$JSON_OUT" ]; then
  echo "[selfhost] JSON v0 written: $tmp_json" >&2
fi

# Optional: emit MIR(JSON) from source (runner compiles .hako directly; Stage‑B JSON is for reference)
if [ -n "$MIR_OUT" ]; then
  echo "[selfhost] emitting MIR JSON → $MIR_OUT" >&2
  "$BIN" --backend mir --emit-mir-json "$MIR_OUT" "$IN" >/dev/null
fi

# Optional: build native EXE via ny-llvmc harness (fallback path; parses original source)
if [ -n "$EXE_OUT" ]; then
  # Requirements: ny-llvmc present and harness envs
  NYLL="${NYASH_NY_LLVM_COMPILER:-$ROOT/target/release/ny-llvmc}"
  if [ ! -x "$NYLL" ] && [ ! -f "$NYLL" ]; then
    echo "[selfhost] ny-llvmc not found: $NYLL (skip EXE). Set NYASH_NY_LLVM_COMPILER or build ny-llvmc" >&2
    exit 2
  fi
  NYRT_DIR="${NYASH_EMIT_EXE_NYRT:-$ROOT/target/release}"
  export NYASH_LLVM_USE_HARNESS=1
  export NYASH_NY_LLVM_COMPILER="$NYLL"
  export NYASH_EMIT_EXE_NYRT="$NYRT_DIR"
  echo "[selfhost] emitting EXE via ny-llvmc → $EXE_OUT" >&2
  # Note: This path compiles the original Hako source via LLVM backend. Stage‑B JSON is emitted above for reference.
  "$BIN" --backend llvm --emit-exe "$EXE_OUT" "$IN"
  # Done
  if [ -z "$JSON_OUT" ]; then rm -f "$tmp_json" 2>/dev/null || true; fi
  exit 0
fi

if [ "$DO_RUN" = "1" ]; then
  # Run via Core‑Direct (in‑proc), quiet
  set +e
  NYASH_GATE_C_CORE=1 HAKO_GATE_C_CORE=1 HAKO_CORE_DIRECT=1 HAKO_CORE_DIRECT_INPROC=1 \
    NYASH_QUIET=1 HAKO_QUIET=1 NYASH_CLI_VERBOSE=0 NYASH_NYRT_SILENT_RESULT=1 \
    "$BIN" --json-file "$tmp_json" >/dev/null 2>&1
  rc=$?
  set -e
  if [ -z "$JSON_OUT" ]; then rm -f "$tmp_json" 2>/dev/null || true; fi
  exit $rc
else
  # Emit-only
  echo "$tmp_json"
fi