slic/tests/test_core_runname.py

import pytest
from pathlib import Path
from slic.core.scanner.runname import (
    extract_runnumber, 
    extract_runnumbers, 
    RunFilenameGenerator,
    EVERYTHING
)


@pytest.fixture
def tmpdir_runs(tmp_path):
    d = tmp_path / "runs"
    d.mkdir()
    return d


@pytest.mark.parametrize(
    "fname,prefix,separator,expected",
    [
        ("scan0004_test.json", "scan", "_", 4),
        ("scan0001_alpha.txt", "scan", "_", 1),
        ("run-0042_demo.csv", "run-", "_", 42),
        ("data9999_final.txt", "data", "_", 9999),
        ("scan5_test.json", "scan", "_", 5),
        (str(Path("/tmp/data/scan0007_exp.json")), "scan", "_", 7),
        (str(Path("/Users/yasmine_tligui/test_pv/runs/scan0123_test.json")), "scan", "_", 123),
        ("scan0010_report.csv", "scan", "_", 10),
        ("scan0300_analysis.yaml", "scan", "_", 300),
        ("experimentscan0070_trial.txt", "experimentscan", "_", 70),
        ("meas:001_test.txt", "meas:", "_", 1),
        (str(Path("a/b/c/scan0042_demo.json")), "scan", "_", 42),
        ("scan1234_extra_part.json", "scan", "_", 1234),

        # error
        ("scan_0042_test.json", "scan_", "_", 42),
    ],
)

def test_extract_runnumber_valid_cases(fname, prefix, separator, expected):
    assert extract_runnumber(fname, prefix, separator) == expected


@pytest.mark.parametrize(
    "fname,prefix,separator",
    [
        ("scan00A3_test.json", "scan", "_"),
        ("file_test.json", "scan", "_"),
        ("scan_test.json", "scan", "_"),
        ("run#_info.txt", "run#", "#"),
        ("1234_scan_test.json", "scan", "_"),
    ],
)
def test_extract_runnumber_invalid_cases(fname, prefix, separator):
    with pytest.raises(ValueError):
        extract_runnumber(fname, prefix, separator)


@pytest.mark.parametrize(
    "fnames,prefix,separator,expected",
    [
        (["scan0001_a", "scan0002_b", "scan0030_c"], "scan", "_", [1, 2, 30]),
        (["scan1_a", "scan02_b", "scan300_c"], "scan", "_", [1, 2, 300]),
        (["run-0004_x", "run-0005_y", "run-0010_z"], "run-", "_", [4, 5, 10]),
        (["data9_file", "data10_file", "data11_file"], "data", "_", [9, 10, 11]),
        (["scan001--a", "scan002--b", "scan003--c"], "scan", "--", [1, 2, 3]),
    ],
)
def test_extract_runnumbers_valid(fnames, prefix, separator, expected):
    result = extract_runnumbers(fnames, prefix, separator)
    assert result == expected


@pytest.mark.parametrize(
    "prefix,run_index,separator,name,suffix,expected",
    [
        ("pfx", "7", "-", "abc", ".txt", "pfx7-abc"),
        ("scan", "001", "_", "data", "_scan_info.json", "scan001_data"),
        ("run", "42", "-", "test", ".meta", "run42-test"),
        ("exp", "003", "--", "trial", ".log", "exp003--trial"),
        ("data", "99", "::", "final", ".json", "data99::final"),
        ("experiment_", "005", "_", "measure", ".csv", "experiment_005_measure"),
    ],
)
def test_fill_filename_pattern(prefix, run_index, separator, name, suffix, expected):
    gen = RunFilenameGenerator(
        base_dir=".",
        prefix=prefix,
        separator=separator,
        suffix=suffix
    )
    formatted = gen._fill_filename_pattern(run_index, name)
    assert formatted == expected


@pytest.mark.parametrize(
    "prefix,n_digits,separator,suffix,expected",
    [
        ("scan", 3, "_", "_scan_info.json", "scan" + "[0-9]"*3 + "_" + EVERYTHING + "_scan_info.json"),
        ("run-", 4, "-", ".json", "run-" + "[0-9]"*4 + "-" + EVERYTHING + ".json"),
        ("data", 2, "#", ".meta", "data" + "[0-9]"*2 + "#" + EVERYTHING + ".meta"),
        ("exp", 1, ".", "_info.txt", "exp" + "[0-9]"*1 + "." + EVERYTHING + "_info.txt"),
    ],
)
def test_pattern_exact_match(tmpdir_runs, prefix, n_digits, separator, suffix, expected):
    gen = RunFilenameGenerator(
        base_dir=tmpdir_runs,
        prefix=prefix,
        n_digits=n_digits,
        separator=separator,
        suffix=suffix,
    )
    pattern = gen.pattern
    assert pattern == expected


def test_get_existing_runnumbers_empty(monkeypatch, tmpdir_runs):
    gen = RunFilenameGenerator(tmpdir_runs)
    monkeypatch.setattr("slic.core.scanner.runname.glob_files", lambda b, p: [])
    assert gen.get_existing_runnumbers() == []


@pytest.mark.parametrize(
    "file_structure,search_base_dir,prefix,separator,suffix,n_digits,expected_next",
    [
        (
            {
                "experiments": ["scan0001_test_scan_info.json", "scan0002_test_scan_info.json"],
                "other_folder": ["scan0003_test_scan_info.json", "scan0004_test_scan_info.json"]
            },
            "experiments",
            "scan", "_", "_scan_info.json", 4,
            "scan0003_test"
        ),
        (
            {
                "project/data": ["scan0001_test_scan_info.json"],
                "project/data/raw": ["scan0002_test_scan_info.json"],
                "project/backup": ["scan0003_test_scan_info.json"],
                "other_project": ["scan0004_test_scan_info.json"]
            },
            "project/data",
            "scan", "_", "_scan_info.json", 4,
            "scan0002_test"
        ),
        (
            {
                "empty_dir": [],
                "full_dir": ["scan0001_test_scan_info.json", "scan0002_test_scan_info.json"],
            },
            "empty_dir",
            "scan", "_", "_scan_info.json", 4,
            "scan0000_test"
        ),
    ],
)
def test_get_next_run_filename(tmpdir, file_structure, search_base_dir, prefix, separator, suffix, n_digits, expected_next):
    for dir_path, filenames in file_structure.items():
        current_dir = tmpdir
        for part in dir_path.split("/"):
            current_dir = current_dir / part
            if not current_dir.exists():
                current_dir.mkdir()
        
        for filename in filenames:
            (current_dir / filename).write("test")
    
    actual_base_dir = tmpdir / search_base_dir
    
    gen = RunFilenameGenerator(
        base_dir=str(actual_base_dir),
        prefix=prefix,
        n_digits=n_digits,
        separator=separator,
        suffix=suffix,
    )
    next_file = gen.get_next_run_filename("test")
    assert next_file == expected_next