Convert durations to ISO8601 format (podcast episodes)

learning_resources/etl/podcast.py

@@ -12,6 +12,7 @@
 
 from learning_resources.constants import LearningResourceType
 from learning_resources.etl.constants import ETLSource
+from learning_resources.etl.utils import iso8601_duration
 from learning_resources.models import PodcastEpisode
 from main.utils import clean_data, frontend_absolute_url, now_in_utc
 
@@ -174,7 +175,7 @@ def transform_episode(rss_data, offered_by, topics, parent_image):
         "podcast_episode": {
             "episode_link": rss_data.link.text if rss_data.link else None,
             "duration": (
-                rss_data.find("itunes:duration").text
+                iso8601_duration(rss_data.find("itunes:duration").text)
                 if rss_data.find("itunes:duration")
                 else None
             ),

learning_resources/etl/podcast_test.py

@@ -162,7 +162,7 @@ def test_transform(mock_github_client, title, topics, offered_by):
                     "published": True,
                     "podcast_episode": {
                         "episode_link": "https://soundcloud.com/podcast/episode1",
-                        "duration": "00:17:16",
+                        "duration": "PT17M16S",
                         "rss": episodes_rss[0].prettify(),
                     },
                     "resource_type": LearningResourceType.podcast_episode.name,
@@ -185,7 +185,7 @@ def test_transform(mock_github_client, title, topics, offered_by):
                     "published": True,
                     "podcast_episode": {
                         "episode_link": "https://soundcloud.com/podcast/episode2",
-                        "duration": "00:17:16",
+                        "duration": "PT17M16S",
                         "rss": episodes_rss[1].prettify(),
                     },
                     "resource_type": LearningResourceType.podcast_episode.name,

learning_resources/etl/utils.py

@@ -21,6 +21,7 @@
 import rapidjson
 import requests
 from django.conf import settings
+from django.utils.dateparse import parse_duration
 from django.utils.functional import SimpleLazyObject
 from django.utils.text import slugify
 from tika import parser as tika_parser
@@ -703,3 +704,31 @@ def parse_certification(offeror, runs_data):
             if (availability and availability != AvailabilityType.archived.value)
         ]
     )
+
+
+def iso8601_duration(duration_str: str) -> str or None:
+    """
+    Parse the duration from a string and return it in ISO-8601 format
+
+    Args:
+        duration_str (str): The duration as a string in one of various formats
+
+    Returns:
+        str: the duration in ISO-8601 format
+    """
+    if not duration_str:
+        return None
+    delta = parse_duration(duration_str)
+    if delta is None:
+        log.warning("Could not parse duration string %s", duration_str)
+        return None
+
+    hours, remainder = divmod(delta.total_seconds(), 3600)
+    minutes, seconds = divmod(remainder, 60)
+
+    if hours or minutes or seconds:
+        hour_duration = f"{int(hours)}H" if hours else ""
+        minute_duration = f"{int(minutes)}M" if minutes else ""
+        second_duration = f"{int(seconds)}S" if seconds else ""
+        return f"PT{hour_duration}{minute_duration}{second_duration}"
+    return "PT0S"

learning_resources/etl/utils_test.py

@@ -460,3 +460,31 @@ def test_calc_checksum(previous_archive, identical):
 def test_get_department_id_by_name(dept_name, dept_id):
     """Test that the correct department ID (if any) is returned"""
     assert utils.get_department_id_by_name(dept_name) == dept_id
+
+
+@pytest.mark.parametrize(
+    ("duration_str", "expected"),
+    [
+        ("1:00:00", "PT1H"),
+        ("1:30:04", "PT1H30M4S"),
+        ("00:00", "PT0S"),
+        ("00:00:00", "PT0S"),
+        ("00:01:00", "PT1M"),
+        ("01:00:00", "PT1H"),
+        ("00:00:01", "PT1S"),
+        ("02:59", "PT2M59S"),
+        ("72:59", "PT1H12M59S"),
+        ("3675", "PT1H1M15S"),
+        ("5", "PT5S"),
+        ("PT1H30M4S", "PT1H30M4S"),
+        ("", None),
+        (None, None),
+        ("bad_duration", None),
+        ("PTBarnum", None),
+    ],
+)
+def test_parse_duration(mocker, duration_str, expected):
+    """Test that parse_duration returns the expected duration"""
+    mock_warn = mocker.patch("learning_resources.etl.utils.log.warning")
+    assert utils.iso8601_duration(duration_str) == expected
+    assert mock_warn.call_count == (1 if duration_str and expected is None else 0)