init: resume 작성 하네스 설계

build/lib/resume_harness/io.py

@@ -0,0 +1,72 @@
+"""Safe, small input helpers for harness contracts."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any, TypeVar
+
+import yaml
+from pydantic import BaseModel
+
+
+MAX_INPUT_BYTES = 2 * 1024 * 1024
+ModelT = TypeVar("ModelT", bound=BaseModel)
+
+
+class InputError(ValueError):
+    """Raised when an input file cannot be safely decoded as a model payload."""
+
+
+def load_mapping(path: str | Path, *, max_bytes: int = MAX_INPUT_BYTES) -> dict[str, Any]:
+    """Load one JSON/YAML mapping without constructing arbitrary Python objects."""
+
+    input_path = Path(path)
+    try:
+        size = input_path.stat().st_size
+    except OSError as exc:
+        raise InputError(f"입력 파일을 읽을 수 없습니다: {input_path}") from exc
+
+    if size > max_bytes:
+        raise InputError(f"입력 파일이 {max_bytes}바이트 제한을 초과했습니다: {input_path}")
+
+    try:
+        text = input_path.read_text(encoding="utf-8")
+    except (OSError, UnicodeDecodeError) as exc:
+        raise InputError(f"입력 파일은 UTF-8 텍스트여야 합니다: {input_path}") from exc
+
+    suffix = input_path.suffix.casefold()
+    try:
+        if suffix == ".json":
+            value = json.loads(text)
+        elif suffix in {".yaml", ".yml"}:
+            value = yaml.safe_load(text)
+        else:
+            raise InputError("지원 형식은 .json, .yaml, .yml입니다.")
+    except (json.JSONDecodeError, yaml.YAMLError) as exc:
+        raise InputError(f"JSON/YAML 구문이 올바르지 않습니다: {input_path}") from exc
+
+    if not isinstance(value, dict):
+        raise InputError(f"입력 최상위 값은 객체(mapping)여야 합니다: {input_path}")
+    return value
+
+
+def load_model(path: str | Path, model_type: type[ModelT]) -> ModelT:
+    """Load and validate a Pydantic contract from JSON/YAML."""
+
+    return model_type.model_validate(load_mapping(path))
+
+
+def dump_json(model: BaseModel) -> str:
+    """Serialize a contract deterministically for audit-friendly output."""
+
+    return json.dumps(
+        model.model_dump(mode="json"),
+        ensure_ascii=False,
+        indent=2,
+        sort_keys=True,
+    )
+
+
+__all__ = ["InputError", "MAX_INPUT_BYTES", "dump_json", "load_mapping", "load_model"]
+