"""Toolchain preflight (architecture D3; Development Plan M4.1).

A model works out what a project needs and sets it up on the build VM; code
makes sure that happens, bounds it and checks the result.

- **Ready or not** is the harness's: before a driver run it checks, in one
  round trip, the baseline it relies on itself, the workspace clone, and a
  marker per project (``~/.nisse/<project>.ready``) that a finished preflight
  leaves. No marker means a new project, VM or wipe: preflight runs first.
- **The preflight run** is the item's driver model with its own prompt and
  the execution tools (screened as ever), ending with ``toolchain_ready``.
- **Versions are the harness's facts:** it runs each tool's version command
  itself (version-shaped commands only) and records what they print; it
  re-checks them before later runs, and a change is a fact for the driver.
"""

import base64
import json
import shlex
from dataclasses import dataclass, field
from typing import Any

from nisse_core.actions import split_shell, unwrap
from nisse_dev.execution import SSHHost

# What the harness itself runs on a build VM (L13, D4): part of the
# template, never of a project's toolchain.
BASELINE = ("git", "base64", "realpath", "mkdir", "ssh-agent", "ssh-add", "timeout")

_VERSION_FLAGS = ("--version", "-V", "-v", "version", "-version")


@dataclass(frozen=True)
class Readiness:
    missing: tuple[str, ...]  # baseline commands the VM lacks
    cloned: bool  # the workspace clone is there
    marker: dict[str, Any] | None  # what the last preflight left, if any


@dataclass(frozen=True)
class Tool:
    name: str
    command: str  # version-shaped: prints the tool's version
    version: str = ""  # what it printed, when the harness last ran it

    def to_data(self) -> list[str]:
        return [self.name, self.command, self.version]


@dataclass
class Report:
    summary: str
    tools: list[Tool] = field(default_factory=list)
    missing: list[tuple[str, str]] = field(default_factory=list)  # (name, why)


MARKERS = "$HOME/.nisse"  # on the build VM


def marker_path(project: str, markers: str = MARKERS) -> str:
    return f"{markers}/{project}.ready"


def check(
    host: SSHHost, project: str, workspace: str, markers: str = MARKERS
) -> Readiness:
    """The baseline, the clone and the marker, in one round trip."""
    probe = "; ".join(
        [
            *(
                f"command -v {shlex.quote(c)} >/dev/null 2>&1 || echo missing:{c}"
                for c in BASELINE
            ),
            f"test -d {shlex.quote(workspace)}/.git && echo cloned",
            (
                f'[ -f "{marker_path(project, markers)}" ] && '
                f'echo marker:$(base64 -w0 < "{marker_path(project, markers)}")'
            ),
            "true",
        ]
    )
    result = host.execute(probe, timeout_s=30)
    missing: list[str] = []
    cloned = False
    marker: dict[str, Any] | None = None
    for line in result.stdout.splitlines():
        if line.startswith("missing:"):
            missing.append(line.removeprefix("missing:"))
        elif line == "cloned":
            cloned = True
        elif line.startswith("marker:"):
            try:
                marker = json.loads(base64.b64decode(line.removeprefix("marker:")))
            except ValueError:
                marker = None  # unreadable: as good as none
    return Readiness(tuple(missing), cloned, marker)


def write_marker(
    host: SSHHost, project: str, data: dict[str, Any], markers: str = MARKERS
) -> None:
    text = base64.b64encode(json.dumps(data, sort_keys=True).encode()).decode()
    host.execute(
        f'mkdir -p "{markers}" && base64 -d > "{marker_path(project, markers)}"',
        stdin=text,
        timeout_s=30,
    )


def version_shaped(command: str) -> bool:
    """One plain command that only prints a version: ``rustc --version``,
    ``go version``, ``python3 -V``. Nothing else is run by the harness."""
    commands = split_shell(command)
    if commands is None or len(commands) != 1:
        return False
    argv, written = commands[0]
    if written or unwrap(list(argv)) != argv:
        return False  # no redirection, no sudo or env in front
    return (
        2 <= len(argv) <= 3
        and argv[-1] in _VERSION_FLAGS
        and not argv[0].startswith(("-", "/", "."))
        and argv[0] not in ("sh", "bash", "zsh", "dash", "eval", "exec")
    )


def versions(host: SSHHost, tools: list[Tool]) -> list[Tool]:
    """Each tool's version, as its command prints it now (first line, from
    stdout or stderr): run by the harness, in one round trip."""
    # The marker lives on the VM, where the model's commands could change
    # it: only version-shaped commands are ever run, wherever they came from.
    tools = [t for t in tools if version_shaped(t.command)]
    if not tools:
        return []
    script = "; ".join(
        f"echo @@{i}; if command -v {shlex.quote(t.command.split()[0])} "
        f">/dev/null 2>&1; then {t.command} 2>&1 | head -n 1; fi"
        for i, t in enumerate(tools)
    )
    result = host.execute(script, timeout_s=60)
    found: dict[int, str] = {}
    current = -1
    for line in result.stdout.splitlines():
        if line.startswith("@@") and line[2:].isdigit():
            current = int(line[2:])
            found[current] = ""
        elif current >= 0 and not found[current]:
            found[current] = line.strip()
    return [
        Tool(t.name, t.command, found.get(i, "") or "not found")
        for i, t in enumerate(tools)
    ]


def drift(before: list[Tool], now: list[Tool]) -> list[str]:
    """What changed, as "name: old → new"."""
    was = {t.name: t.version for t in before}
    return [
        f"{t.name}: {was[t.name]} → {t.version}"
        for t in now
        if t.name in was and was[t.name] != t.version
    ]


def tools_of(data: Any) -> list[Tool]:
    """Tools from stored data (a marker's or an item's)."""
    out: list[Tool] = []
    for entry in data or []:
        if isinstance(entry, list | tuple) and len(entry) == 3:
            out.append(Tool(str(entry[0]), str(entry[1]), str(entry[2])))
    return out


def said(tools: list[Tool]) -> str:
    return ", ".join(f"{t.name} {t.version}" for t in tools) or "nothing recorded"


TOOLCHAIN_READY = {
    "type": "function",
    "function": {
        "name": "toolchain_ready",
        "description": (
            "The project's toolchain is set up on this build VM (or as far as "
            "it can be): report it. Ends the preflight."
        ),
        "parameters": {
            "type": "object",
            "properties": {
                "summary": {
                    "type": "string",
                    "description": "What you found and did, in a sentence or two.",
                },
                "toolchain": {
                    "type": "array",
                    "description": (
                        "Each tool the project needs whose version matters, "
                        "with the command that prints its version (e.g. "
                        "'python3 --version', 'rustc -V', 'go version')."
                    ),
                    "items": {
                        "type": "object",
                        "properties": {
                            "name": {"type": "string"},
                            "version_command": {"type": "string"},
                        },
                        "required": ["name", "version_command"],
                    },
                },
                "installed": {
                    "type": "array",
                    "items": {"type": "string"},
                    "description": "What you installed, if anything.",
                },
                "missing": {
                    "type": "array",
                    "description": "What the project needs that you couldn't set up.",
                    "items": {
                        "type": "object",
                        "properties": {
                            "name": {"type": "string"},
                            "why": {"type": "string"},
                        },
                        "required": ["name", "why"],
                    },
                },
            },
            "required": ["summary", "toolchain"],
        },
    },
}

SYSTEM = """You are nisse-dev, running a toolchain step for {repo}: making its
build VM and the project's own files agree. Its clone is the workspace.
{workspace}

Work out from the project's own files what it needs to build, test and run: its
language runtimes and compilers, package and build tools, and anything else it
calls (system libraries, CLIs, a database for its tests). Honour its pins
(.python-version, rust-toolchain.toml, requires-python, engines, lockfiles).

{mode}
Check what's there with version commands; install what's missing. The build
VM is the project's and disposable: installing on it is fine. Don't install or
change anything for other hosts, and never install, store or ask for
credentials: the harness hands those out when they're needed.

Leave out the basics every machine has (ls, cat, git, curl, ssh, tar): those
are the VM template's. Leave out tools that act on other systems (gh, tea,
docker push, cloud CLIs): they aren't how work is done here.

When you're done, or can't get further, call toolchain_ready with what you
found and did: each tool whose version matters with a command that prints its
version, what you installed, and what's still missing and why. Keep it short:
this is setup, not the work."""

SETUP = """This build VM is new to the project, or was rebuilt. Set it up from what
the repository says: its pins, setup docs and CI configuration. Don't change
the repository's files: if they're out of step with what the project needs,
say so in your summary, and the work can put them right.
{last}"""

REQUESTED = """The work needs a toolchain change: {reason}
Install what it needs, then bring the project's own files in step with it: its
pins (.python-version, rust-toolchain.toml, requires-python, engines,
lockfiles), its setup docs (the README, or wherever it documents setup) and its
CI configuration. Those edits are part of the current work and are reviewed
with it."""


def report_of(arguments: dict[str, Any]) -> Report:
    tools = [
        Tool(str(t.get("name", "")).strip(), str(t.get("version_command", "")).strip())
        for t in arguments.get("toolchain") or []
        if isinstance(t, dict)
    ]
    missing = [
        (str(m.get("name", "")).strip(), str(m.get("why", "")).strip())
        for m in arguments.get("missing") or []
        if isinstance(m, dict)
    ]
    return Report(str(arguments.get("summary", "")).strip(), tools, missing)


def report_problem(arguments: dict[str, Any]) -> str | None:
    """A toolchain_ready the harness can't use: said back to the model."""
    report = report_of(arguments)
    bad = [t.command for t in report.tools if not version_shaped(t.command)]
    if bad:
        return (
            "these aren't plain version commands, which is all I'll run: "
            + "; ".join(bad)
            + ". Use the tool's own --version (or -V, version)."
        )
    if any(not t.name for t in report.tools):
        return "every toolchain entry needs a name"
    return None