Resp 3 add make sample data

src/RESPFlow/access_files.py

@@ -1,4 +1,5 @@
 import os, re
+import pandas as pd
 
 #
 # =============================================================================
@@ -130,4 +131,44 @@ def map_files(in_path: str,
         elif (directory.endswith(file_ext)) and ((expression is None) or (re.match(expression, fileName)!=None)):
             file_dirs[fileName] = new_path
 
-    return file_dirs
+    return file_dirs
+
+# 
+# =============================================================================
+# 
+
+from pathlib import Path
+from importlib import resources as ir
+import shutil
+
+def make_sample_data(path_names: dict[str, str]) -> None:
+    """
+    Copies sample data files from the RESPFlow package to the raw data folder
+    specified in `path_names`.
+
+    Parameters
+    ----------
+    path_names : dict[str, str]
+        A dictionary of file locations with keys for stage in the processing
+        pipeline. Must contain a 'raw' key specifying the destination folder
+        for the sample data.
+
+    Raises
+    ------
+    ValueError
+        If 'raw' key is not found in `path_names`.
+
+    Returns
+    -------
+    None
+        The function performs file operations only; it does not return a value.
+    """
+    if "raw" not in path_names:
+        raise ValueError("Raw path not detected in path_names.")
+
+    dest = Path(path_names["raw"])
+    dest.mkdir(parents=True, exist_ok=True)
+
+    with ir.as_file(ir.files("RESPFlow").joinpath("data")) as src:
+        shutil.copytree(src, dest, dirs_exist_ok=True)
+

tests/test_access_files.py

@@ -1,6 +1,21 @@
+# ============================================================================
+# Imports
+# ============================================================================
+
+# --- Standard library ---
 import os
+from pathlib import Path
+from importlib import resources as ir
+
+# --- Third-party packages ---
 import pytest
-import RESPFlow
+
+# --- Local application / package imports ---
+from RESPFlow.access_files import make_paths, map_files, make_sample_data
+
+# ============================================================================
+# Tests
+# ===========================================================================
 
 # ============================================================================
 # Test make_paths function
@@ -23,7 +38,7 @@ def fake_makedirs(path, exist_ok=True):
 
 
 def test_make_paths_defaults(mock_filesystem):
-    paths = RESPFlow.make_paths()
+    paths = make_paths()
 
     expected_keys = {
         'raw', 'notch', 'bandpass', 'fwr',
@@ -45,7 +60,7 @@ def test_make_paths_custom_root_raw(mock_filesystem):
     custom_root = "my_root"
     custom_raw = "my_raw"
 
-    paths = RESPFlow.make_paths(root=custom_root, raw=custom_raw)
+    paths = make_paths(root=custom_root, raw=custom_raw)
 
     # Assert custom raw should be absolute and point to my_raw
     assert paths['raw'] == "/abs/my_raw"
@@ -61,10 +76,114 @@ def test_make_paths_custom_root_raw(mock_filesystem):
 # Test map_files function
 # ============================================================================
 
+def _touch(p: str, text: str = "") -> None:
+    os.makedirs(os.path.dirname(p), exist_ok=True)
+    with open(p, "w", encoding="utf-8") as f:
+        f.write(text)
+
+
+def test_map_files_basic_recursive(tmp_path):
+    # data/
+    #   10/file1.csv
+    #   10/nested/info.txt
+    #   23/file2.csv
+    root = tmp_path / "data"
+    f1 = root / "10" / "file1.csv"
+    f_txt = root / "10" / "nested" / "info.txt"
+    f2 = root / "23" / "file2.csv"
+    _touch(str(f1), "a,b\n1,2\n")
+    _touch(str(f_txt), "hello\n")
+    _touch(str(f2), "x,y\n3,4\n")
+
+    result = map_files(str(root))  # default file_ext='csv'
+
+    # Only CSVs should appear
+    keys = set(result.keys())
+    assert os.path.join("10", "file1.csv") in keys
+    assert os.path.join("23", "file2.csv") in keys
+
+    # Non-csv shouldn't be included
+    assert not any("info.txt" in k for k in keys)
+
+    # Values should be absolute paths pointing to the files
+    assert os.path.isabs(result[os.path.join("10", "file1.csv")])
+    assert os.path.exists(result[os.path.join("23", "file2.csv")])
+
+
+def test_map_files_extension_filter(tmp_path):
+    root = tmp_path / "data"
+    _touch(str(root / "a.csv"), "csv\n")
+    _touch(str(root / "b.txt"), "txt\n")
+
+    only_txt = map_files(str(root), file_ext="txt")
+
+    assert set(only_txt.keys()) == {"b.txt"}
+
+    only_csv = map_files(str(root), file_ext="csv")
+
+    assert set(only_csv.keys()) == {"a.csv"}
+
+
+def test_map_files_regex_filter_on_filename(tmp_path):
+    root = tmp_path / "data"
+    _touch(str(root / "10" / "keep_this.csv"), "ok\n")
+    _touch(str(root / "10" / "skip_this.csv"), "no\n")
+    _touch(str(root / "23" / "also_skip.csv"), "no\n")
+
+    # Match only files whose name ends with 'keep_this.csv'
+    expr = r".*keep_this\.csv$"
+
+    result = map_files(str(root), expression=expr)
+
+    keys = set(result.keys())
+    assert any(k.endswith(os.path.join("10", "keep_this.csv")) for k in keys)
+    assert not any(k.endswith("skip_this.csv") for k in keys)
+
+def test_map_files_invalid_regex_raises(tmp_path):
+    root = tmp_path / "data"
+    _touch(str(root / "a.csv"), "a\n")
+
+    with pytest.raises(Exception, match="Invalid regex expression"):
+        map_files(str(root), expression="(")  # invalid regex
+
 # ============================================================================
-# Tests for map_files
+# Tests for make_sample_data function
 # ============================================================================
 
+def _fake_pkg_with_data(tmp_path: Path) -> Path:
+    """Create a fake RESPFlow/data tree with a few files."""
+    root = tmp_path / "RESPFlow_fake"
+    (root / "data" / "10").mkdir(parents=True, exist_ok=True)
+    (root / "data" / "10" / "file1.csv").write_text("a,b\n1,2\n", encoding="utf-8")
+    (root / "data" / "10" / "nested").mkdir(parents=True, exist_ok=True)
+    (root / "data" / "10" / "nested" / "info.txt").write_text("hello\n", encoding="utf-8")
+    (root / "data" / "23").mkdir(parents=True, exist_ok=True)
+    (root / "data" / "23" / "file2.csv").write_text("x,y\n3,4\n", encoding="utf-8")
+    return root
+
+
+def test_copies_all_and_preserves_structure(tmp_path: Path, monkeypatch):
+    fake_pkg = _fake_pkg_with_data(tmp_path)
+    # Make ir.files("RESPFlow") point to our fake package root
+    monkeypatch.setattr(ir, "files", lambda _pkg: fake_pkg)
+
+    dest = tmp_path / "raw"
+    make_sample_data({"raw": str(dest)})
+
+    assert (dest / "10" / "file1.csv").is_file()
+    assert (dest / "10" / "nested" / "info.txt").is_file()
+    assert (dest / "23" / "file2.csv").is_file()
+    # spot-check contents
+    assert (dest / "10" / "file1.csv").read_text(encoding="utf-8").startswith("a,b\n")
+
+
+def test_missing_raw_raises(monkeypatch, tmp_path: Path):
+    fake_pkg = _fake_pkg_with_data(tmp_path)
+    monkeypatch.setattr(ir, "files", lambda _pkg: fake_pkg)
+
+    with pytest.raises(ValueError, match="Raw path not detected"):
+        make_sample_data({})
+
 # ============================================================================
 # End of tests
 # ============================================================================