govoplan/tools/devkit/govoplan_devkit/package_tests.py

"""Bounded package metadata and conservative direct-Node test discovery."""

from __future__ import annotations

from pathlib import Path
import re
import shlex

from .common import digest, read_json, redact_argv, reject_symlinks

MAX_PACKAGE_BYTES = 1024 * 1024
MAX_SCRIPTS = 512
CORE_COMPONENT_SUITES = (
    "data-grid-actions",
    "dialog-focus",
    "explorer-tree",
    "icon-button",
    "layout-primitives",
    "mail-components",
    "metric-card",
    "page-layout",
    "workspace-layout",
    "people-picker",
    "password-field",
    "resource-access",
    "action-blocker",
    "documentation-help",
    "selection-list",
    "wysiwyg-editor",
)
CORE_RUNNER = "scripts/run-component-tests.mjs"


def read_package(path: Path) -> dict:
    try:
        package = read_json(path, max_bytes=MAX_PACKAGE_BYTES)
    except (OSError, ValueError) as exc:
        raise ValueError(f"Cannot safely read package metadata: {path}") from exc
    if not isinstance(package, dict):
        raise ValueError(f"Package metadata must be an object: {path}")
    scripts = package.get("scripts", {})
    if not isinstance(scripts, dict) or len(scripts) > MAX_SCRIPTS:
        raise ValueError(f"Package scripts must be a bounded object: {path}")
    for name, command in scripts.items():
        if (
            not isinstance(name, str)
            or not 1 <= len(name) <= 128
            or any(char in name for char in "\0\r\n")
        ):
            raise ValueError(f"Package script names must be bounded strings: {path}")
        if (
            not isinstance(command, str)
            or not 1 <= len(command) <= 8192
            or "\0" in command
        ):
            raise ValueError(
                f"Package script commands must be bounded nonempty strings: {path}"
            )
    return package


def core_component_alias(
    repo_name: str, name: str, command: str, package_path: Path
) -> str | None:
    if repo_name != "govoplan-core" or package_path.parent.name != "webui":
        return None
    if name == "test:components" and command == f"node {CORE_RUNNER}":
        return "all"
    for suite in CORE_COMPONENT_SUITES:
        if name == "test:" + suite and command == f"node {CORE_RUNNER} {suite}":
            return suite
    return None


def direct_node(command: str, package_path: Path) -> tuple[list[str] | None, str]:
    try:
        parts = shlex.split(command)
    except ValueError:
        return None, "Malformed command quoting; no command was guessed or executed."
    if not parts or parts[0] != "node":
        return (
            None,
            "Not a direct Node source test; use its explicitly reviewed owning workflow.",
        )
    offset = 2 if len(parts) > 1 and parts[1] == "--test" else 1
    if len(parts) != offset + 1 or not re.fullmatch(
        r"(?:scripts|tests)/[A-Za-z0-9_.-]+\.mjs", parts[offset]
    ):
        return (
            None,
            "Compound command, flags or arguments are unsupported by scoped source discovery.",
        )
    if parts[offset] == CORE_RUNNER:
        return (
            None,
            "Only exact known Core component aliases belong to the shared UI batch.",
        )
    target = package_path.parent / parts[offset]
    try:
        reject_symlinks(target)
        if not target.is_file() or not target.resolve().is_relative_to(
            package_path.parent.resolve()
        ):
            return None, "Declared test target is missing or escapes its package."
    except (OSError, ValueError):
        return None, "Declared test target is not a safe regular package file."
    return [
        "{node}",
        *parts[1:offset],
        str(target),
    ], "Direct package-owned source test."


def declared_tests(repo, package_path: Path) -> list[dict]:
    package = read_package(package_path)
    result = []
    for name, command in sorted(package.get("scripts", {}).items()):
        if name != "test" and not name.startswith("test:"):
            continue
        component = core_component_alias(repo.name, name, command, package_path)
        argv, reason = (
            direct_node(command, package_path)
            if component is None
            else (None, "Exact known Core component alias.")
        )
        result.append(
            {
                "repo": repo.name,
                "package_path": str(package_path),
                "name": name,
                "command_sha256": digest(command),
                "argv": redact_argv(argv) if argv else None,
                "component_suite": component,
                "reason": reason,
                "_argv": argv,
            }
        )
    return result


def discovered_sources(repo) -> list[dict]:
    webui = repo.path / "webui"
    found = {}
    for folder in ("scripts", "tests"):
        for pattern in ("test-interface-pattern*.mjs", "*structure*.mjs"):
            for target in (webui / folder).glob(pattern):
                command = (
                    "node "
                    + ("--test " if target.name.endswith(".test.mjs") else "")
                    + target.relative_to(webui).as_posix()
                )
                argv, reason = direct_node(command, webui / "package.json")
                found[str(target)] = {
                    "repo": repo.name,
                    "package_path": str(webui / "package.json"),
                    "name": "file:" + target.relative_to(webui).as_posix(),
                    "discovered": True,
                    "command_sha256": digest(command),
                    "argv": redact_argv(argv) if argv else None,
                    "component_suite": None,
                    "reason": reason,
                    "_argv": argv,
                }
                if len(found) > MAX_SCRIPTS:
                    raise ValueError(
                        f"Too many discovered source checks in {repo.name}"
                    )
    return [found[key] for key in sorted(found)]


def source_stage_id(repo_name: str, argv: list[str]) -> str:
    stem = Path(argv[-1]).stem
    return f"{repo_name}.{stem}"[:110] + "." + digest(argv)[:12]