chore: Bing up to speed with dev

src/transport_performance/gtfs/multi_validation.py

@@ -68,6 +68,8 @@ class MultiGtfsInstance:
         Plot each of the stops from all GtfsInstance's on a folium Map object.
     validate_empty_feeds()
         Check if there are empty feeds within the MultiGtfsInstance.
+    get_dates()
+        Get the range of dates that the gtfs(s) span.
 
     Raises
     ------
@@ -641,3 +643,35 @@ def viz_stops(
         if return_viz:
             return map
         return None
+
+    def get_dates(self, return_range: bool = True) -> list:
+        """Get all available dates from calendar.txt (or calendar_dates.txt).
+
+        Parameters
+        ----------
+        return_range : bool, optional
+           Whether to return the raw dates, or the min/max range, by default
+           True
+
+        Returns
+        -------
+        list
+            Either the full set of dates, or the range that the dates span
+            between
+
+        """
+        _type_defence(return_range, "return_range", bool)
+        available_dates = set()
+        for inst in self.instances:
+            # gtfs-kit makes calendar None if it isn't present
+            if isinstance(inst.feed.calendar, type(None)):
+                available_dates.update(
+                    inst.feed.calendar_dates["date"].unique()
+                )
+            else:
+                available_dates.update(inst.feed.calendar["start_date"])
+                available_dates.update(inst.feed.calendar["end_date"])
+        sorted_dates = sorted(available_dates)
+        if return_range:
+            return [min(sorted_dates), max(sorted_dates)]
+        return sorted_dates

src/transport_performance/osm/osm_utils.py

@@ -53,6 +53,9 @@ def filter_osm(
 
     """
     # defence
+    _type_defence(pbf_pth, "pbf_pth", (pathlib.Path, str))
+    # if pbf_pth is str, convert to Path in order to access as_posix method
+    pbf_pth = pathlib.Path(pbf_pth)
     _is_expected_filetype(pbf_pth, param_nm="pbf_pth", exp_ext=".pbf")
     _enforce_file_extension(out_pth, ".pbf", ".pbf", "out_pth")
     for nm, val in {

tests/gtfs/test_multi_validation.py

@@ -4,6 +4,7 @@
 import glob
 import pathlib
 import shutil
+from copy import deepcopy
 
 import numpy as np
 import pandas as pd
@@ -32,6 +33,23 @@ def multi_gtfs_fixture(multi_gtfs_paths):
     return m_gtfs
 
 
+@pytest.fixture(scope="function")
+def multi_gtfs_altered_fixture(multi_gtfs_fixture):
+    """Test fixture with calendar_dates.txt in GTFS instance 0."""
+    # deepcopy otherwise it overwrites multi_gtfs_fixture
+    mgtfs = deepcopy(multi_gtfs_fixture)
+    dummy_cdates = pd.DataFrame(
+        {
+            "service_id": ["000001", "000001", "000002", "000003"],
+            "date": ["20231211", "20240601", "20240613", "20220517"],
+            "exception_type": [1, 1, 1, 1],
+        }
+    )
+    mgtfs.instances[0].feed.calendar_dates = dummy_cdates
+    mgtfs.instances[0].feed.calendar = None
+    return mgtfs
+
+
 class TestMultiGtfsInstance(object):
     """Tests for the MultiGtfsInstance class."""
 
@@ -462,3 +480,44 @@ def test_viz_stops(self, multi_gtfs_fixture, tmp_path):
         assert isinstance(
             no_stop_code_map, folium.Map
         ), "Map not plotted without stop_code present"
+
+    def test_get_dates_defence(self, multi_gtfs_fixture):
+        """Defensive tests for .get_dates()."""
+        with pytest.raises(
+            TypeError, match=".*return_range.*expected.*bool.*Got.*str.*"
+        ):
+            multi_gtfs_fixture.get_dates(return_range="test")
+
+    def test_get_dates(self, multi_gtfs_fixture, multi_gtfs_altered_fixture):
+        """General tests for .get_dates()."""
+        # multi gtfs with only calendar.txt
+        # return_range=True
+        assert (
+            len(multi_gtfs_fixture.get_dates()) == 2
+        ), "Unexpected number of dates returned"
+        assert (
+            multi_gtfs_fixture.get_dates()[0] == "20230605"
+        ), "min not as expected"
+        assert (
+            multi_gtfs_fixture.get_dates()[1] == "20240426"
+        ), "max not as expected"
+        # return_range=False
+        assert (
+            len(multi_gtfs_fixture.get_dates(return_range=False)) == 5
+        ), "Unexpected number of dates"
+        # multi_gtfs with both calendar.txt and calendar_dates.txt
+        # return_range=True
+        assert (
+            len(multi_gtfs_altered_fixture.get_dates()) == 2
+        ), "Unexpected number of dates"
+        assert (
+            multi_gtfs_altered_fixture.get_dates()[0] == "20220517"
+        ), "min not as expected"
+        assert (
+            multi_gtfs_altered_fixture.get_dates()[1] == "20240613"
+        ), "max not as expected"
+        # return_range=False
+        assert (
+            len(multi_gtfs_altered_fixture.get_dates(return_range=False)) == 6
+        ), "Unexpected number of dates"
+        pass

tests/osm/test_osm_utils.py

@@ -3,6 +3,9 @@
 import os
 from unittest.mock import patch, call
 import re
+import glob
+
+from pyprojroot import here
 
 from transport_performance.osm.osm_utils import filter_osm
 
@@ -118,3 +121,21 @@ def test_filter_osm_with_osmosis(self, mock_print, tmpdir):
             .startswith("call('Filter completed. Written to ")
         )
         assert out is None
+
+    @pytest.mark.runinteg
+    @patch("builtins.print")
+    def test_filter_osm_takes_globstring(self, mock_print, tmpdir):
+        """Assert filter operation executes when user passes a globstring."""
+        found_pbf = sorted(glob.glob(str(here("tests/data/*.pbf"))))[0]
+        target_pth = os.path.join(tmpdir, "test_globstring_input.osm.pbf")
+        filter_osm(
+            pbf_pth=found_pbf, out_pth=target_pth, install_osmosis=False
+        )
+        # check the file exists
+        assert os.path.exists(
+            target_pth
+        ), f"Filtered pbf file not found: {target_pth}"
+        func_out = mock_print.mock_calls
+        assert (
+            func_out[-1].__str__().endswith("test_globstring_input.osm.pbf')")
+        )