plexus-python-common 1.0.63__tar.gz → 1.0.64__tar.gz

{plexus_python_common-1.0.63 → plexus_python_common-1.0.64}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: plexus-python-common
-Version: 1.0.63
+Version: 1.0.64
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13

{plexus_python_common-1.0.63 → plexus_python_common-1.0.64}/src/plexus/common/utils/tagutils.py

@@ -1,6 +1,7 @@
 import contextlib
 import dataclasses
 import datetime
+import math
 import os
 import pathlib
 import textwrap
@@ -16,6 +17,7 @@ import sqlalchemy as sa
 import sqlalchemy.dialects.sqlite as sa_sqlite
 import sqlalchemy.orm as sa_orm
 from iker.common.utils.dbutils import ConnectionMaker
+from iker.common.utils.dtutils import dt_from_ts_us, dt_to_ts_us
 from iker.common.utils.funcutils import memorized, singleton
 from iker.common.utils.iterutils import batched, head_or_none
 from iker.common.utils.iterutils import dicttree
@@ -48,6 +50,8 @@ __all__ = [
     "tag_cache_file_path",
     "TagCache",
     "tag_cache",
+    "standard_clip_duration_us",
+    "populate_clip_ranges",
 ]
 
 
@@ -1093,3 +1097,52 @@ def tag_cache(*, identifier: str | None = None, file_path: str | None = None) ->
     if file_path is not None:
         return TagCache(file_path=file_path)
     return TagCache(file_path=tag_cache_file_path())
+
+
+
+@singleton
+def standard_clip_duration_us() -> int:
+    """Duration of clips to split samples into, in microseconds, which is fixed to 20 seconds for now."""
+    return 20 * 1_000_000  # 20 seconds
+
+
+
+def populate_clip_ranges(
+    data_begin_dt: datetime.datetime,
+    data_end_dt: datetime.datetime,
+    *,
+    clip_duration_us: int = None
+) -> Generator[tuple[datetime.datetime, datetime.datetime, datetime.datetime, datetime.datetime]]:
+    """
+    Generate clip begin datetime, clip data begin datetime, and clip data end datetime for each clip slot that overlaps
+    with the given data begin and end datetimes.
+    The clip begin datetime is the beginning of the clip slot, the clip data begin datetime is the maximum of the clip
+    begin datetime and the data begin datetime, and the clip data end datetime is the minimum of the clip end datetime
+    and the data end datetime.
+
+    :param data_begin_dt: The beginning datetime of the data.
+    :param data_end_dt: The end datetime of the data.
+    :param clip_duration_us: The duration of each clip in microseconds. If not provided,
+                             the standard clip duration will be used.
+    :return: A generator of tuples containing the clip begin datetime, clip data begin datetime,
+             and clip data end datetime
+    """
+    clip_duration_us = clip_duration_us or standard_clip_duration_us()
+
+    if data_begin_dt == data_end_dt:
+        clip_slot = math.floor(dt_to_ts_us(data_begin_dt) / clip_duration_us)
+        clip_begin_dt = dt_from_ts_us(clip_slot * clip_duration_us)
+        clip_end_dt = dt_from_ts_us((clip_slot + 1) * clip_duration_us)
+        yield clip_begin_dt, clip_end_dt, data_begin_dt, data_end_dt
+        return
+
+    begin_clip_slot = math.floor(dt_to_ts_us(data_begin_dt) / clip_duration_us)
+    end_clip_slot = math.ceil(dt_to_ts_us(data_end_dt)  / clip_duration_us)
+
+    for clip_slot in range(begin_clip_slot, end_clip_slot):
+        clip_begin_dt = dt_from_ts_us(clip_slot * clip_duration_us)
+        clip_end_dt = dt_from_ts_us((clip_slot + 1) * clip_duration_us)
+        clip_data_begin_dt = max(clip_begin_dt, data_begin_dt)
+        clip_data_end_dt = min(clip_end_dt, data_end_dt)
+
+        yield clip_begin_dt, clip_end_dt, clip_data_begin_dt, clip_data_end_dt

{plexus_python_common-1.0.63 → plexus_python_common-1.0.64}/src/plexus_python_common.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: plexus-python-common
-Version: 1.0.63
+Version: 1.0.64
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13

{plexus_python_common-1.0.63 → plexus_python_common-1.0.64}/test/plexus_tests/common/utils/tagutils_test.py

@@ -10,6 +10,7 @@ from iker.common.utils.iterutils import last
 from iker.common.utils.jsonutils import JsonObject
 
 from plexus.common.utils.tagutils import MutableTagset, Tag, TagCache, Tagset
+from plexus.common.utils.tagutils import populate_clip_ranges
 from plexus.common.utils.tagutils import predefined_tagsets, render_tagset_markdown_readme
 from plexus.common.utils.tagutils import tag_cache_file_path
 
@@ -497,3 +498,177 @@ class TagUtilsTest(unittest.TestCase):
                              tag_records_count * 2 * (tags_count - tagset_tags_count) // tags_count)
             self.assertEqual(len(list(dst_target_cache.iter_tags(tagsets=[tagset], tagset_inverted=False))),
                              tag_records_count * 2 * tagset_tags_count // tags_count)
+
+    data_populate_clip_ranges = [
+        (
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:50.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:40.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:40.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:40.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:01:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:40.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:50.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:15.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:15.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:30.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:40.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:30.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:05.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:05.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:05.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:05.000000+00:00")),
+            ],
+        )
+    ]
+
+    @ddt.idata(data_populate_clip_ranges)
+    @ddt.unpack
+    def test_populate_clip_ranges(self, data_begin_dt, data_end_dt, expected):
+        result = list(populate_clip_ranges(data_begin_dt, data_end_dt))
+        self.assertEqual(result, expected)
+
+    data_populate_clip_ranges__customized_duration = [
+        (
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:50.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:30.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:30.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:30.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:40.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:30.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:40.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:40.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:50.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:40.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:50.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:15.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:15.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:30.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00")),
+                (dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:30.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:20.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:30.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:00.000000+00:00")),
+            ],
+        ),
+        (
+            dt_parse_iso("2023-01-01T00:00:05.000000+00:00"),
+            dt_parse_iso("2023-01-01T00:00:05.000000+00:00"),
+            [
+                (dt_parse_iso("2023-01-01T00:00:00.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:10.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:05.000000+00:00"),
+                 dt_parse_iso("2023-01-01T00:00:05.000000+00:00")),
+            ],
+        )
+    ]
+
+    @ddt.idata(data_populate_clip_ranges__customized_duration)
+    @ddt.unpack
+    def test_populate_clip_ranges__customized_duration(self, data_begin_dt, data_end_dt, expected):
+        result = list(populate_clip_ranges(data_begin_dt, data_end_dt, clip_duration_us=10_000_000))
+        self.assertEqual(result, expected)