orchestrator/agent-team/tests/test_confluence_client.py

"""Unit tests for agent_team.confluence.client (OAuth 2LO + basic auth + dry-run updater).

Fully hermetic: the HTTP transport is dependency-injected with an in-memory
fake :class:`HttpClient`, so no network call, OAuth token, or live Confluence is
exercised. Credentials are injected via ``monkeypatch.setenv``/``delenv`` and
never read from a real environment.
"""

from __future__ import annotations

import base64
import json
from typing import Any
from urllib import parse as _urlparse

import pytest

from agent_team.nodes.confluence_writer_llm import (
    ConfluenceDraftError,
    parse_confluence_reply,
)
from agent_team.confluence.client import (
    ACCESSIBLE_RESOURCES_URL,
    ATLASSIAN_API_BASE,
    DEFAULT_OAUTH_TOKEN_URL,
    ConfluenceAuth,
    ConfluenceClient,
    ConfluenceError,
    PlannedPageUpdate,
    body_diff,
    init_auth,
)

# --------------------------------------------------------------------------- #
# Fakes / fixtures
# --------------------------------------------------------------------------- #

_CONFLUENCE_ENV_KEYS = (
    "CONFLUENCE_OAUTH_CLIENT_ID",
    "CONFLUENCE_OAUTH_CLIENT_SECRET",
    "CONFLUENCE_OAUTH_TOKEN_URL",
    "CONFLUENCE_CLOUD_ID",
    "CONFLUENCE_BASE_URL",
    "CONFLUENCE_EMAIL",
    "CONFLUENCE_API_TOKEN",
)


class FakeHttp:
    """In-memory :class:`HttpClient` recording calls and returning scripted
    ``(status, body)`` per verb.

    ``responses`` maps a verb to a list of scripted ``(status, body)`` tuples,
    consumed in order; a single tuple is reused for every call to that verb.
    """

    def __init__(
        self,
        *,
        get: Any = None,
        post: Any = None,
        put: Any = None,
    ) -> None:
        self._scripts: dict[str, Any] = {"get": get, "post": post, "put": put}
        self.calls: list[dict[str, Any]] = []

    def _next(self, verb: str) -> tuple[int, Any]:
        script = self._scripts[verb]
        if script is None:
            raise AssertionError(f"unexpected {verb.upper()} call")
        if isinstance(script, list):
            return script.pop(0)
        return script

    def get(
        self, url: str, *, headers: dict[str, str]
    ) -> tuple[int, dict[str, Any] | bytes]:
        self.calls.append({"verb": "get", "url": url, "headers": headers})
        return self._next("get")

    def post(
        self, url: str, *, headers: dict[str, str], data: bytes
    ) -> tuple[int, dict[str, Any] | bytes]:
        self.calls.append(
            {"verb": "post", "url": url, "headers": headers, "data": data}
        )
        return self._next("post")

    def put(
        self, url: str, *, headers: dict[str, str], data: bytes
    ) -> tuple[int, dict[str, Any] | bytes]:
        self.calls.append({"verb": "put", "url": url, "headers": headers, "data": data})
        return self._next("put")


def _clear_confluence_env(monkeypatch: pytest.MonkeyPatch) -> None:
    for key in _CONFLUENCE_ENV_KEYS:
        monkeypatch.delenv(key, raising=False)


def _storage_page(page_id: str, version: int, value: str) -> dict[str, Any]:
    return {
        "id": page_id,
        "title": "Existing Title",
        "version": {"number": version},
        "body": {"storage": {"representation": "storage", "value": value}},
    }


# --------------------------------------------------------------------------- #
# OAuth 2LO path
# --------------------------------------------------------------------------- #


def test_oauth_token_post_and_cloud_id_from_env(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_ID", "cid")
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_SECRET", "secret")
    monkeypatch.setenv("CONFLUENCE_CLOUD_ID", "cloud-123")

    http = FakeHttp(post=(200, {"access_token": "tok-abc"}))
    auth = init_auth(http)

    assert auth.mode == "oauth"
    assert auth.base == f"{ATLASSIAN_API_BASE}/cloud-123"
    assert auth.base == "https://api.atlassian.com/ex/confluence/cloud-123"
    assert auth.headers()["Authorization"] == "Bearer tok-abc"

    # Exactly one POST (the token request) — cloudId came from env, no GET.
    (post_call,) = [c for c in http.calls if c["verb"] == "post"]
    assert post_call["url"] == DEFAULT_OAUTH_TOKEN_URL
    form = dict(_urlparse.parse_qsl(post_call["data"].decode("utf-8")))
    assert form["grant_type"] == "client_credentials"
    assert form["client_id"] == "cid"
    assert form["client_secret"] == "secret"
    assert not [c for c in http.calls if c["verb"] == "get"]


def test_oauth_cloud_id_resolved_from_accessible_resources(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_ID", "cid")
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_SECRET", "secret")
    monkeypatch.setenv("CONFLUENCE_BASE_URL", "https://seahaven.atlassian.net")

    resources = [
        {"id": "other-cloud", "url": "https://other.atlassian.net"},
        {"id": "match-cloud", "url": "https://seahaven.atlassian.net"},
    ]
    http = FakeHttp(
        post=(200, {"access_token": "tok-xyz"}),
        get=(200, resources),
    )
    auth = init_auth(http)

    # Site URL matches the second resource, so its id wins.
    assert auth.base == f"{ATLASSIAN_API_BASE}/match-cloud"
    get_call = next(c for c in http.calls if c["verb"] == "get")
    assert get_call["url"] == ACCESSIBLE_RESOURCES_URL
    assert get_call["headers"]["Authorization"] == "Bearer tok-xyz"


def test_oauth_cloud_id_falls_back_to_first_resource(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_ID", "cid")
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_SECRET", "secret")
    # No CONFLUENCE_BASE_URL -> no site match -> first resource id wins.

    resources = [{"id": "first-cloud", "url": "https://a.atlassian.net"}]
    http = FakeHttp(post=(200, {"access_token": "t"}), get=(200, resources))
    auth = init_auth(http)
    assert auth.base == f"{ATLASSIAN_API_BASE}/first-cloud"


def test_oauth_token_request_failure_raises_skip(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_ID", "cid")
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_SECRET", "secret")

    http = FakeHttp(post=(401, {"error": "invalid_client"}))
    with pytest.raises(ConfluenceError) as exc:
        init_auth(http)
    assert exc.value.status == 0  # conf_api_init-style "skip, no false alarm".


def test_oauth_unresolvable_cloud_id_raises_skip(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_ID", "cid")
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_SECRET", "secret")

    http = FakeHttp(post=(200, {"access_token": "t"}), get=(200, []))
    with pytest.raises(ConfluenceError) as exc:
        init_auth(http)
    assert exc.value.status == 0


def test_oauth_token_url_override(monkeypatch: pytest.MonkeyPatch) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_ID", "cid")
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_SECRET", "secret")
    monkeypatch.setenv("CONFLUENCE_CLOUD_ID", "cloud-1")
    monkeypatch.setenv("CONFLUENCE_OAUTH_TOKEN_URL", "https://custom.example/oauth")

    http = FakeHttp(post=(200, {"access_token": "t"}))
    init_auth(http)
    post_call = next(c for c in http.calls if c["verb"] == "post")
    assert post_call["url"] == "https://custom.example/oauth"


# --------------------------------------------------------------------------- #
# Basic-auth fallback
# --------------------------------------------------------------------------- #


def test_basic_auth_fallback_when_oauth_absent(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_BASE_URL", "https://seahaven.atlassian.net/")
    monkeypatch.setenv("CONFLUENCE_EMAIL", "adam@seahavenind.com")
    monkeypatch.setenv("CONFLUENCE_API_TOKEN", "api-tok")

    http = FakeHttp()  # No HTTP call expected for basic-auth resolution.
    auth = init_auth(http)

    assert auth.mode == "basic"
    # Trailing slash stripped.
    assert auth.base == "https://seahaven.atlassian.net"
    expected = base64.b64encode(b"adam@seahavenind.com:api-tok").decode("ascii")
    assert auth.headers()["Authorization"] == f"Basic {expected}"
    assert http.calls == []


def test_oauth_wins_over_basic_when_both_present(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_ID", "cid")
    monkeypatch.setenv("CONFLUENCE_OAUTH_CLIENT_SECRET", "secret")
    monkeypatch.setenv("CONFLUENCE_CLOUD_ID", "cloud-1")
    monkeypatch.setenv("CONFLUENCE_BASE_URL", "https://seahaven.atlassian.net")
    monkeypatch.setenv("CONFLUENCE_EMAIL", "adam@seahavenind.com")
    monkeypatch.setenv("CONFLUENCE_API_TOKEN", "api-tok")

    http = FakeHttp(post=(200, {"access_token": "tok"}))
    auth = init_auth(http)
    assert auth.mode == "oauth"


# --------------------------------------------------------------------------- #
# Missing-creds path (detectable skip, no crash)
# --------------------------------------------------------------------------- #


def test_missing_creds_raises_skip_status_zero(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    http = FakeHttp()
    with pytest.raises(ConfluenceError) as exc:
        init_auth(http)
    assert exc.value.status == 0  # mirrors conf_api_init returning non-zero.
    assert http.calls == []


def test_incomplete_basic_creds_raises_skip(
    monkeypatch: pytest.MonkeyPatch,
) -> None:
    _clear_confluence_env(monkeypatch)
    # Base URL + email but NO api token -> not fully configured -> skip.
    monkeypatch.setenv("CONFLUENCE_BASE_URL", "https://seahaven.atlassian.net")
    monkeypatch.setenv("CONFLUENCE_EMAIL", "adam@seahavenind.com")
    with pytest.raises(ConfluenceError) as exc:
        init_auth(FakeHttp())
    assert exc.value.status == 0


def test_client_uses_injected_env_mapping() -> None:
    # No monkeypatch: the explicit env mapping is read at call time, not import.
    env = {
        "CONFLUENCE_BASE_URL": "https://seahaven.atlassian.net",
        "CONFLUENCE_EMAIL": "adam@seahavenind.com",
        "CONFLUENCE_API_TOKEN": "tok",
    }
    page = _storage_page("100", 4, "<p>hi</p>")
    http = FakeHttp(get=(200, page))
    client = ConfluenceClient(http=http, env=env)
    fetched = client.get_page("100")
    assert fetched["version"]["number"] == 4
    assert http.calls[0]["url"].startswith("https://seahaven.atlassian.net")


# --------------------------------------------------------------------------- #
# get_page
# --------------------------------------------------------------------------- #


def test_get_page_parses_storage_format() -> None:
    page = _storage_page("12345", 7, "<p>Current body</p>")
    http = FakeHttp(get=(200, page))
    auth = ConfluenceAuth(
        mode="oauth",
        base="https://api.atlassian.com/ex/confluence/cloud-1",
        _headers={"Authorization": "Bearer t", "Accept": "application/json"},
    )
    client = ConfluenceClient(http=http, auth=auth)

    result = client.get_page("12345")
    assert result["version"]["number"] == 7
    assert result["body"]["storage"]["value"] == "<p>Current body</p>"
    (call,) = http.calls
    assert call["url"] == (
        "https://api.atlassian.com/ex/confluence/cloud-1"
        "/wiki/api/v2/pages/12345?body-format=storage"
    )


def test_get_page_encodes_injected_id_in_request_path() -> None:
    # Defense-in-depth: a malformed (injected) id must not appear verbatim in the
    # request path — it would otherwise rewrite the authenticated request
    # (request-path injection carrying the service-account credential).
    page = _storage_page("123", 1, "<p>x</p>")
    http = FakeHttp(get=(200, page))
    auth = ConfluenceAuth(mode="basic", base="https://x.atlassian.net")
    client = ConfluenceClient(http=http, auth=auth)

    injected = "123?body-format=atlas_doc_format"
    client.get_page(injected)
    (call,) = http.calls
    # The raw injected query must NOT survive as a literal path/query segment.
    assert injected not in call["url"]
    # The whole id is percent-encoded into a single path segment; the "?" that
    # would start a new query string is escaped to %3F and cannot inject.
    assert (
        "/pages/123%3Fbody-format%3Datlas_doc_format?body-format=storage" in call["url"]
    )


def test_get_page_non_2xx_raises_with_status() -> None:
    http = FakeHttp(get=(404, {"message": "Not Found"}))
    auth = ConfluenceAuth(mode="basic", base="https://x.atlassian.net")
    client = ConfluenceClient(http=http, auth=auth)
    with pytest.raises(ConfluenceError) as exc:
        client.get_page("999")
    assert exc.value.status == 404
    assert "Not Found" in exc.value.body


# --------------------------------------------------------------------------- #
# update_page — dry-run (default) issues NO PUT
# --------------------------------------------------------------------------- #


def _basic_client(http: FakeHttp) -> ConfluenceClient:
    auth = ConfluenceAuth(
        mode="basic",
        base="https://seahaven.atlassian.net",
        _headers={"Authorization": "Basic x", "Accept": "application/json"},
    )
    return ConfluenceClient(http=http, auth=auth)


def test_update_page_dry_run_default_issues_no_put() -> None:
    current = _storage_page("777", 3, "<p>old</p>")
    http = FakeHttp(get=(200, current))  # PUT script omitted -> would assert.
    client = _basic_client(http)

    planned = client.update_page("777", "New Title", "<p>new</p>", 3)

    assert isinstance(planned, PlannedPageUpdate)
    assert planned.applied is False
    assert planned.new_version == 4  # current + 1
    assert planned.current_version == 3
    assert planned.body_storage == "<p>new</p>"
    # A unified diff against the current body shows the change.
    assert "<p>old</p>" in planned.body_delta
    assert "<p>new</p>" in planned.body_delta
    # Only the read (GET) happened — NO PUT.
    assert [c["verb"] for c in http.calls] == ["get"]


def test_update_page_dry_run_diff_against_empty_when_read_fails() -> None:
    # get_page returns 404 -> client swallows and diffs against an empty baseline.
    http = FakeHttp(get=(404, {"message": "gone"}))
    client = _basic_client(http)
    planned = client.update_page("404id", "T", "<p>brand new</p>", 0)
    assert planned.applied is False
    assert planned.new_version == 1
    assert "<p>brand new</p>" in planned.body_delta


# --------------------------------------------------------------------------- #
# update_page — apply=True issues the PUT
# --------------------------------------------------------------------------- #


def test_update_page_apply_true_issues_put() -> None:
    current = _storage_page("555", 9, "<p>old</p>")
    updated = _storage_page("555", 10, "<p>updated</p>")
    http = FakeHttp(get=(200, current), put=(200, updated))
    client = _basic_client(http)

    result = client.update_page("555", "Title", "<p>updated</p>", 9, apply=True)

    assert result.applied is True
    assert result.new_version == 10

    put_call = next(c for c in http.calls if c["verb"] == "put")
    assert put_call["url"] == "https://seahaven.atlassian.net/wiki/api/v2/pages/555"
    payload = json.loads(put_call["data"].decode("utf-8"))
    assert payload["id"] == "555"
    assert payload["status"] == "current"
    assert payload["title"] == "Title"
    assert payload["version"]["number"] == 10
    assert payload["body"]["representation"] == "storage"
    assert payload["body"]["value"] == "<p>updated</p>"
    assert put_call["headers"]["Content-Type"] == "application/json"


def test_update_page_apply_true_non_2xx_put_raises() -> None:
    current = _storage_page("321", 1, "<p>x</p>")
    http = FakeHttp(get=(200, current), put=(409, {"message": "version conflict"}))
    client = _basic_client(http)
    with pytest.raises(ConfluenceError) as exc:
        client.update_page("321", "T", "<p>y</p>", 1, apply=True)
    assert exc.value.status == 409
    assert "version conflict" in exc.value.body


# --------------------------------------------------------------------------- #
# Pure planning helper
# --------------------------------------------------------------------------- #


def test_parse_confluence_reply_rejects_non_numeric_page_id() -> None:
    # Layer-1 guard: an LLM-supplied page_id that is not a bare numeric content
    # id (here carrying a request-path-injection payload) must fail the draft.
    reply = json.dumps(
        {
            "title": "T",
            "body_storage": "<p>b</p>",
            "page_id": "123?body-format=atlas_doc_format",
        }
    )
    with pytest.raises(ConfluenceDraftError):
        parse_confluence_reply(reply)


def test_parse_confluence_reply_accepts_numeric_page_id() -> None:
    reply = json.dumps({"title": "T", "body_storage": "<p>b</p>", "page_id": "123456"})
    draft = parse_confluence_reply(reply)
    assert draft["page_id"] == "123456"


def test_body_diff_empty_when_identical() -> None:
    assert body_diff("<p>same</p>", "<p>same</p>", page_id="1") == ""


def test_body_diff_shows_added_and_removed() -> None:
    delta = body_diff("line a\nline b\n", "line a\nline c\n", page_id="99")
    assert "page/99@current" in delta
    assert "page/99@planned" in delta
    assert "-line b" in delta
    assert "+line c" in delta