From 518522e6850ebf91254334ba5bad48266ea51e80 Mon Sep 17 00:00:00 2001
From: Gerrit Holl <gerrit.holl@dwd.de>
Date: Wed, 20 Apr 2022 16:06:48 +0200
Subject: [PATCH] Add method to find coverage gaps

---
 typhon/files/fileset.py            | 29 +++++++++++++++++++++++++++
 typhon/tests/files/test_fileset.py | 32 ++++++++++++++++++++++++++++++
 2 files changed, 61 insertions(+)

diff --git a/typhon/files/fileset.py b/typhon/files/fileset.py
index be1f199d..fa80d6dc 100644
--- a/typhon/files/fileset.py
+++ b/typhon/files/fileset.py
@@ -2998,6 +2998,35 @@ def write(self, data, file_info, in_background=False, **write_args):
         else:
             self.handler.write(data, file_info, **write_args)
 
+    def find_coverage_gaps(self, start_time, end_time):
+        """Find coverage gaps between timestamps.
+
+        Between ``start_time`` and ``end_time``, yield pairs of times for which
+        the fileset does _not_ have data.
+
+        Args:
+            start_time: Start time in any form understood by
+                :class:`~pandas.Timestamp`.
+            end_time: Ending time in the same form.
+
+        Yields:
+            Yields zero or more instances of :class:`~pandas.Interval`
+            corresponding to each time segment in the overall interval that is
+            _not_ covered by data files.
+        """
+
+        last = pd.Timestamp(start_time)
+        for fi in self.find(start_time, end_time, no_files_error=False):
+            if fi.times[0] > last:
+                yield pd.Interval(
+                    pd.Timestamp(last),
+                    pd.Timestamp(fi.times[0]))
+            last = fi.times[1]
+        if last < end_time:
+            yield pd.Interval(
+                    pd.Timestamp(last),
+                    pd.Timestamp(end_time))
+
 
 class FileSetManager(dict):
     def __init__(self, *args, **kwargs):
diff --git a/typhon/tests/files/test_fileset.py b/typhon/tests/files/test_fileset.py
index 81e432b3..7568e121 100644
--- a/typhon/tests/files/test_fileset.py
+++ b/typhon/tests/files/test_fileset.py
@@ -8,6 +8,8 @@
 import shutil
 import logging
 
+import pandas as pd
+
 from typhon.files import FileHandler, FileInfo, FileSet, FileSetManager
 from typhon.files.utils import get_testfiles_directory
 
@@ -823,3 +825,33 @@ def test_compare_fileinfo(self):
         assert f2 != f4
         assert f3 != f4
         assert f1 != "fake/path"
+
+    def test_coverage_gaps(self, tmp_path):
+        """Test that coverage gaps are correctly found."""
+
+        (tmp_path / "abisko").mkdir(parents=True)
+        fs = FileSet(
+                tmp_path / "abisko" /
+                "y{year}-m{month}-d{day}-h{hour}-m{minute}-he{end_hour}-me{end_minute}")
+
+        pI = pd.Interval
+        pT = pd.Timestamp
+        gaps = list(fs.find_coverage_gaps(
+                datetime.datetime(1900, 1, 1, 0, 0),
+                datetime.datetime(1900, 1, 1, 0, 10)))
+        assert gaps == [
+                pI(pT("1900-01-01T00:00:00"), pT("1900-01-01T00:10:00"))]
+        (tmp_path / "abisko" / "y1900-m01-d01-h00-m02-he00-me04").touch()
+        (tmp_path / "abisko" / "y1900-m01-d01-h00-m06-he00-me07").touch()
+        gaps = list(fs.find_coverage_gaps(
+                datetime.datetime(1900, 1, 1, 0, 0),
+                datetime.datetime(1900, 1, 1, 0, 15)))
+        assert gaps == [
+                pI(pT("1900-01-01T00:00:00"), pT("1900-01-01T00:02:00")),
+                pI(pT("1900-01-01T00:04:00"), pT("1900-01-01T00:06:00")),
+                pI(pT("1900-01-01T00:07:00"), pT("1900-01-01T00:15:00"))]
+        (tmp_path / "abisko" / "y1900-m01-d01-h00-m07-he00-me12").touch()
+        gaps = list(fs.find_coverage_gaps(
+                datetime.datetime(1900, 1, 1, 0, 7),
+                datetime.datetime(1900, 1, 1, 0, 12)))
+        assert gaps == []