procurement-ingest/lambdas/wo/email_processor/tests/test_comment_id.py

"""Issue #23: comment_id (WorkOrderComments range key) must be unique per source
email AND byte-identical across Lambda async retries of the same S3 object, with
wall-clock now() kept OUT of the key.

Advisory A1: on the ai_fallback path the parsed comment_time is model output and
not retry-stable, so it must never enter the key -- the time segment derives
from the (deterministic) email Date header instead.
"""

import handler

DATE_HEADER = "Mon, 27 Apr 2026 23:57:49 +0000 (UTC)"


def _parsed(wo="11144580730", comment_time="2026-04-27T23:51:48"):
    return {
        "work_order_id": wo,
        "email_type": "comment",
        "commenter": "jdoe",
        "comment_text": "a comment",
        "comment_time": comment_time,
    }


def _comment_ids(table):
    return [item["comment_id"] for item in table.puts]


def test_same_object_key_retry_is_idempotent(fake_dynamo):
    parsed = _parsed()
    handler.save_event(
        parsed, "s3://bucket/inbound/obj-a", "inbound/obj-a", "template", DATE_HEADER
    )
    handler.save_event(
        parsed, "s3://bucket/inbound/obj-a", "inbound/obj-a", "template", DATE_HEADER
    )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    ids = _comment_ids(table)
    # Two put_item calls with an identical key -> one logical row (overwrite).
    assert ids[0] == ids[1]
    assert len(table.store) == 1


def test_distinct_emails_same_wo_and_time_are_distinct_rows(fake_dynamo):
    parsed = _parsed()
    handler.save_event(
        parsed, "s3://bucket/inbound/obj-a", "inbound/obj-a", "template", DATE_HEADER
    )
    handler.save_event(
        parsed, "s3://bucket/inbound/obj-b", "inbound/obj-b", "template", DATE_HEADER
    )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    ids = _comment_ids(table)
    assert ids[0] != ids[1]
    assert len(table.store) == 2


def test_absent_comment_time_is_stable_across_retries(fake_dynamo):
    parsed = _parsed(comment_time=None)
    handler.save_event(
        parsed, "s3://bucket/inbound/obj-c", "inbound/obj-c", "template", DATE_HEADER
    )
    handler.save_event(
        parsed, "s3://bucket/inbound/obj-c", "inbound/obj-c", "template", DATE_HEADER
    )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    ids = _comment_ids(table)
    assert ids[0] == ids[1]  # 'nocomment' segment, not now()
    assert "#nocomment#" in ids[0]
    assert len(table.store) == 1


def test_event_id_independent_of_now(fake_dynamo, monkeypatch):
    """Freezing / advancing the clock must not change the range key."""
    parsed = _parsed(comment_time=None)

    class FrozenDT:
        offset = 0

        @classmethod
        def now(cls, tz=None):
            import datetime as _dt

            return _dt.datetime(2020, 1, 1, tzinfo=tz) + _dt.timedelta(
                seconds=cls.offset
            )

    monkeypatch.setattr(handler, "datetime", FrozenDT)
    handler.save_event(
        parsed, "s3://b/inbound/obj-d", "inbound/obj-d", "template", DATE_HEADER
    )
    FrozenDT.offset = 999999  # simulate a later retry
    handler.save_event(
        parsed, "s3://b/inbound/obj-d", "inbound/obj-d", "template", DATE_HEADER
    )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    ids = _comment_ids(table)
    assert ids[0] == ids[1]
    # created_at (display) may differ, but the KEY must not.
    assert len(table.store) == 1


def test_comment_id_format_shape(fake_dynamo):
    parsed = _parsed()
    handler.save_event(
        parsed, "s3://bucket/inbound/obj-e", "inbound/obj-e", "template", DATE_HEADER
    )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    cid = _comment_ids(table)[0]
    wo, time_part, suffix = cid.split("#")
    assert wo == "11144580730"
    assert time_part == "2026-04-27T23:51:48"
    assert len(suffix) == 12 and all(c in "0123456789abcdef" for c in suffix)


# --- Advisory A1: ai_fallback path ---


def test_ai_path_retry_with_drifted_comment_time_is_idempotent(fake_dynamo):
    """A retry where the model returns a DIFFERENT comment_time must still map
    to the same range key (the model output never enters the key)."""
    handler.save_event(
        _parsed(comment_time="2026-04-27T23:51:48"),
        "s3://bucket/inbound/obj-f",
        "inbound/obj-f",
        "ai_fallback",
        DATE_HEADER,
    )
    handler.save_event(
        _parsed(comment_time="2026-04-27T23:52:03"),  # model drifted on retry
        "s3://bucket/inbound/obj-f",
        "inbound/obj-f",
        "ai_fallback",
        DATE_HEADER,
    )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    ids = _comment_ids(table)
    assert ids[0] == ids[1]
    assert len(table.store) == 1


def test_ai_path_key_uses_date_header_not_model_output(fake_dynamo):
    handler.save_event(
        _parsed(),
        "s3://bucket/inbound/obj-g",
        "inbound/obj-g",
        "ai_fallback",
        DATE_HEADER,
    )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    cid = _comment_ids(table)[0]
    _, time_part, _ = cid.split("#")
    assert time_part == "2026-04-27T23:57:49+00:00"  # Date header, UTC ISO
    assert "23:51:48" not in cid  # parsed comment_time kept out of the key


def test_ai_path_missing_date_header_falls_back_to_nocomment(fake_dynamo):
    # 'Fri, 31 Dec 9999' parses but overflows on UTC conversion (OverflowError,
    # not ValueError) -- _header_date_iso must swallow it, never raise.
    for header in (None, "", "not a date", "Fri, 31 Dec 9999 23:59:59 -1400"):
        handler.save_event(
            _parsed(),
            "s3://bucket/inbound/obj-h",
            "inbound/obj-h",
            "ai_fallback",
            header,
        )
    table = fake_dynamo.tables[handler.COMMENTS_TABLE]
    ids = _comment_ids(table)
    assert all("#nocomment#" in cid for cid in ids)
    assert len(set(ids)) == 1  # stable regardless of header garbage