feat: added union_no_overlap from aw-research

Author: ErikBjare

aw_transform/__init__.py

@@ -14,6 +14,7 @@
 from .simplify import simplify_string
 from .flood import flood
 from .classify import categorize, tag, Rule
+from .union_no_overlap import union_no_overlap
 
 __all__ = [
     "flood",
@@ -24,6 +25,7 @@
     "period_union",
     "filter_period_intersect",
     "union",
+    "union_no_overlap",
     "concat",
     "sum_durations",
     "sort_by_timestamp",

aw_transform/union_no_overlap.py

@@ -0,0 +1,122 @@
+"""
+Originally from aw-research
+"""
+
+from copy import deepcopy
+from typing import List, Tuple, Optional
+from datetime import datetime, timedelta, timezone
+
+from timeslot import Timeslot
+
+from aw_core import Event
+
+
+def _split_event(e: Event, dt: datetime) -> Tuple[Event, Optional[Event]]:
+    if e.timestamp < dt < e.timestamp + e.duration:
+        e1 = deepcopy(e)
+        e2 = deepcopy(e)
+        e1.duration = dt - e.timestamp
+        e2.timestamp = dt
+        e2.duration = (e.timestamp + e.duration) - dt
+        return (e1, e2)
+    else:
+        return (e, None)
+
+
+def test_split_event():
+    now = datetime(2018, 1, 1, 0, 0).astimezone(timezone.utc)
+    td1h = timedelta(hours=1)
+    e = Event(timestamp=now, duration=2 * td1h, data={})
+    e1, e2 = _split_event(e, now + td1h)
+    assert e1.timestamp == now
+    assert e1.duration == td1h
+    assert e2.timestamp == now + td1h
+    assert e2.duration == td1h
+
+
+def union_no_overlap(events1: List[Event], events2: List[Event]) -> List[Event]:
+    """Merges two eventlists and removes overlap, the first eventlist will have precedence
+
+    Example:
+      events1  | xxx    xx     xxx     |
+      events1  |  ----     ------   -- |
+      result   | xxx--  xx ----xxx  -- |
+    """
+    events1 = deepcopy(events1)
+    events2 = deepcopy(events2)
+
+    # I looked a lot at aw_transform.union when I wrote this
+    events_union = []
+    e1_i = 0
+    e2_i = 0
+    while e1_i < len(events1) and e2_i < len(events2):
+        e1 = events1[e1_i]
+        e2 = events2[e2_i]
+        e1_p = Timeslot(e1.timestamp, e1.timestamp + e1.duration)
+        e2_p = Timeslot(e2.timestamp, e2.timestamp + e2.duration)
+
+        if e1_p.intersects(e2_p):
+            if e1.timestamp <= e2.timestamp:
+                events_union.append(e1)
+                e1_i += 1
+
+                # If e2 continues after e1, we need to split up the event so we only get the part that comes after
+                _, e2_next = _split_event(e2, e1.timestamp + e1.duration)
+                if e2_next:
+                    events2[e2_i] = e2_next
+                else:
+                    e2_i += 1
+            else:
+                e2_next, e2_next2 = _split_event(e2, e1.timestamp)
+                events_union.append(e2_next)
+                e2_i += 1
+                if e2_next2:
+                    events2.insert(e2_i, e2_next2)
+        else:
+            if e1.timestamp <= e2.timestamp:
+                events_union.append(e1)
+                e1_i += 1
+            else:
+                events_union.append(e2)
+                e2_i += 1
+    events_union += events1[e1_i:]
+    events_union += events2[e2_i:]
+    return events_union
+
+
+def test_union_no_overlap():
+    from pprint import pprint
+
+    now = datetime(2018, 1, 1, 0, 0)
+    td1h = timedelta(hours=1)
+    events1 = [
+        Event(timestamp=now + 2 * i * td1h, duration=td1h, data={"test": 1})
+        for i in range(3)
+    ]
+    events2 = [
+        Event(timestamp=now + (2 * i + 0.5) * td1h, duration=td1h, data={"test": 2})
+        for i in range(3)
+    ]
+
+    events_union = _union_no_overlap(events1, events2)
+    # pprint(events_union)
+    dur = sum((e.duration for e in events_union), timedelta(0))
+    assert dur == timedelta(hours=4, minutes=30)
+    assert sorted(events_union, key=lambda e: e.timestamp)
+
+    events_union = _union_no_overlap(events2, events1)
+    # pprint(events_union)
+    dur = sum((e.duration for e in events_union), timedelta(0))
+    assert dur == timedelta(hours=4, minutes=30)
+    assert sorted(events_union, key=lambda e: e.timestamp)
+
+    events1 = [
+        Event(timestamp=now + (2 * i) * td1h, duration=td1h, data={"test": 1})
+        for i in range(3)
+    ]
+    events2 = [Event(timestamp=now, duration=5 * td1h, data={"test": 2})]
+    events_union = _union_no_overlap(events1, events2)
+    pprint(events_union)
+    dur = sum((e.duration for e in events_union), timedelta(0))
+    assert dur == timedelta(hours=5, minutes=0)
+    assert sorted(events_union, key=lambda e: e.timestamp)