"""Tests for FEED_FILE_TIMESTAMP behavior across all splitter functions.

These tests cover the regression introduced in INT-2478 and fixed in INT-2649:
FEED_FILE_TIMESTAMP is appended to output filenames after reading the Header
row. All H/T validation must compare raw (pre-mutation) Snowflake filenames so
that the timestamp suffix does not trigger a spurious ValueError.

Covered code paths:
  - split_rows_to_files        (US / GB / EX_US modes)
  - split_res_to_files         (AGGREGATE / SAP_SETTLEMENT — STREAMING mode)

FEED_FILE_TIMESTAMP absent: existing behaviour must be unchanged (no regression).
FEED_FILE_TIMESTAMP present:
  - output filename is timestamped
  - H/T validation passes when raw names match
  - H/T validation still raises ValueError when raw names genuinely differ
"""

import os

import pytest

import feed_file_exporter as ffe
from constants import database as db_consts

TIMESTAMP = '20260224_023117'


def _make_rows(raw_name):
    """Return a minimal H/B/T row list for the given raw filename."""
    return [
        {
            db_consts.RECORD_TYPE: 'H',
            db_consts.BODY_CONTENT: 'HEADER',
            db_consts.FILE_NAME: raw_name,
        },
        {
            db_consts.RECORD_TYPE: 'N',
            db_consts.BODY_CONTENT: 'BODY',
            db_consts.FILE_NAME: raw_name,
        },
        {
            db_consts.RECORD_TYPE: 'T',
            db_consts.BODY_CONTENT: 'TAIL',
            db_consts.FILE_NAME: raw_name,
        },
    ]


class _Row:
    """Minimal stand-in for a SQLAlchemy RowProxy."""

    def __init__(self, d):
        self._d = d

    def keys(self):
        return list(self._d.keys())

    def __iter__(self):
        return iter(self._d.values())


class _RP:
    """Minimal stand-in for a SQLAlchemy ResultProxy."""

    def __init__(self, rows):
        self._rows = rows

    def fetchone(self):
        return _Row(self._rows[0])

    def __iter__(self):
        for r in self._rows[1:]:
            yield _Row(r)


# ── split_rows_to_files (US / GB / EX_US) ────────────────────────────────────

def test_split_rows_timestamp_applied_to_filename_with_extension(
    tmp_path, monkeypatch
):
    """Timestamp is inserted before the file extension (.txt → _TS.txt)."""
    monkeypatch.setenv('FEED_FILE_TIMESTAMP', TIMESTAMP)
    rows = _make_rows('sony_stars_US_001.txt')
    count, upload_list, gen_list = ffe.split_rows_to_files(rows, str(tmp_path))
    expected_name = f'sony_stars_US_001_{TIMESTAMP}.txt'
    assert count == 3
    assert upload_list == [expected_name]
    assert len(gen_list) == 1
    assert os.path.basename(gen_list[0]) == expected_name
    assert os.path.exists(gen_list[0])


def test_split_rows_timestamp_applied_to_filename_no_extension(
    tmp_path, monkeypatch
):
    """Timestamp is appended at the end when the filename has no extension."""
    monkeypatch.setenv('FEED_FILE_TIMESTAMP', TIMESTAMP)
    rows = _make_rows('sony_stars_US_001')
    _, upload_list, gen_list = ffe.split_rows_to_files(rows, str(tmp_path))
    expected_name = f'sony_stars_US_001_{TIMESTAMP}'
    assert upload_list == [expected_name]
    assert os.path.basename(gen_list[0]) == expected_name


def test_split_rows_ht_validation_passes_with_timestamp(tmp_path, monkeypatch):
    """H/T validation uses raw names — must NOT raise when timestamp is set.

    Regression test for INT-2649: the original bug raised ValueError on every
    production run because the mutated (timestamped) filename was compared
    against the raw Snowflake Tail filename.
    """
    monkeypatch.setenv('FEED_FILE_TIMESTAMP', TIMESTAMP)
    rows = _make_rows('sony_stars_US_001.txt')
    # Must not raise
    count, _, _ = ffe.split_rows_to_files(rows, str(tmp_path))
    assert count == 3


def test_split_rows_ht_mismatch_still_raises_with_timestamp(
    tmp_path, monkeypatch
):
    """A genuine raw H/T name mismatch still raises ValueError with timestamp.

    Confirms the H/T guard still works correctly after the INT-2649 fix.
    """
    monkeypatch.setenv('FEED_FILE_TIMESTAMP', TIMESTAMP)
    rows = _make_rows('A.txt')
    rows[-1][db_consts.FILE_NAME] = 'B.txt'  # raw tail differs from raw header
    with pytest.raises(ValueError, match='[Mm]ismatch'):
        ffe.split_rows_to_files(rows, str(tmp_path))


def test_split_rows_timestamp_absent_filename_unchanged(tmp_path, monkeypatch):
    """Without FEED_FILE_TIMESTAMP the filename is unchanged (no regression)."""
    monkeypatch.delenv('FEED_FILE_TIMESTAMP', raising=False)
    rows = _make_rows('sony_stars_US_001.txt')
    _, upload_list, gen_list = ffe.split_rows_to_files(rows, str(tmp_path))
    assert upload_list == ['sony_stars_US_001.txt']
    assert os.path.basename(gen_list[0]) == 'sony_stars_US_001.txt'


# ── split_res_to_files (AGGREGATE / SAP_SETTLEMENT — STREAMING mode) ─────────

def test_split_res_timestamp_applied_to_filename(tmp_path, monkeypatch):
    """STREAMING mode applies FEED_FILE_TIMESTAMP to the output filename."""
    monkeypatch.setenv('FEED_FILE_TIMESTAMP', TIMESTAMP)
    rows = _make_rows('aggregate_feed.txt')
    rp = _RP(rows)
    count, gen_list = ffe.split_res_to_files(rp, str(tmp_path))
    expected_name = f'aggregate_feed_{TIMESTAMP}.txt'
    assert count == 3
    assert len(gen_list) == 1
    assert os.path.basename(gen_list[0]) == expected_name
    assert os.path.exists(gen_list[0])


def test_split_res_ht_validation_passes_with_timestamp(tmp_path, monkeypatch):
    """STREAMING mode H/T validation uses raw names — must NOT raise.

    Same root-cause regression as INT-2649 applied to the STREAMING path.
    """
    monkeypatch.setenv('FEED_FILE_TIMESTAMP', TIMESTAMP)
    rows = _make_rows('aggregate_feed.txt')
    rp = _RP(rows)
    count, _ = ffe.split_res_to_files(rp, str(tmp_path))
    assert count == 3


def test_split_res_ht_mismatch_still_raises_with_timestamp(
    tmp_path, monkeypatch
):
    """STREAMING mode genuine raw H/T mismatch still raises ValueError."""
    monkeypatch.setenv('FEED_FILE_TIMESTAMP', TIMESTAMP)
    rows = _make_rows('A.txt')
    rows[-1][db_consts.FILE_NAME] = 'B.txt'
    rp = _RP(rows)
    with pytest.raises(ValueError, match='[Mm]ismatch'):
        ffe.split_res_to_files(rp, str(tmp_path))


def test_split_res_timestamp_absent_filename_unchanged(tmp_path, monkeypatch):
    """Without FEED_FILE_TIMESTAMP STREAMING mode filename is unchanged."""
    monkeypatch.delenv('FEED_FILE_TIMESTAMP', raising=False)
    rows = _make_rows('aggregate_feed.txt')
    rp = _RP(rows)
    _, gen_list = ffe.split_res_to_files(rp, str(tmp_path))
    assert os.path.basename(gen_list[0]) == 'aggregate_feed.txt'