From a977bf5cb5457f23e7b984b27fc4afffafa41991 Mon Sep 17 00:00:00 2001
From: Alina Voilova <alina.voilova@lichtblick.de>
Date: Mon, 15 Apr 2024 10:23:26 +0200
Subject: [PATCH] added valuerror for sod with less than daily freq

---
 dev_scripts/intersect_test copy 2.py | 79 ++++++++++++++++++++++++++++
 portfolyo/tools/intersect.py         | 24 ++++-----
 tests/tools/test_intersect_flex.py   |  8 ++-
 3 files changed, 96 insertions(+), 15 deletions(-)
 create mode 100644 dev_scripts/intersect_test copy 2.py

diff --git a/dev_scripts/intersect_test copy 2.py b/dev_scripts/intersect_test copy 2.py
new file mode 100644
index 0000000..8210189
--- /dev/null
+++ b/dev_scripts/intersect_test copy 2.py	
@@ -0,0 +1,79 @@
+import pandas as pd
+
+from portfolyo import tools
+
+
+def freq():
+    # ignore freq
+    a = pd.date_range(
+        "2022-04-01",
+        "2024-07-01",
+        freq="QS",
+        inclusive="left",
+    )
+    b = pd.date_range(
+        "2021-01-01",
+        "2024-01-01",
+        freq="D",
+        inclusive="left",
+    )
+    intersect = tools.intersect.indices_flex(a, b, ignore_freq=True)
+    print(intersect)
+
+
+def start_day():
+    a = pd.date_range(
+        "2022-04-01 00:00", "2022-05-10 00:00", freq="15T", inclusive="left"
+    )
+    b = pd.date_range(
+        "2022-04-01 06:00", "2022-07-15 06:00", freq="15T", inclusive="left"
+    )
+    intersect = tools.intersect.indices_flex(a, b, ignore_start_of_day=True)
+    print(intersect)
+
+
+def tz():
+    a = pd.date_range(
+        "2022-04-01 00:00",
+        "2022-05-10 00:00",
+        freq="H",
+        tz=None,
+        inclusive="left",
+    )
+    b = pd.date_range(
+        "2022-04-25 00:00",
+        "2022-05-15 00:00",
+        freq="H",
+        tz="Europe/Berlin",
+        inclusive="left",
+    )
+    intersect = tools.intersect.indices_flex(a, b, ignore_tz=True)
+    print(intersect)
+
+
+def all():
+    a = pd.date_range(
+        "2022-01-01 00:00",
+        "2023-01-01 00:00",
+        freq="15T",
+        tz="Asia/Kolkata",
+        inclusive="left",
+    )
+    b = pd.date_range(
+        "2022-01-20 06:00",
+        "2023-01-01 06:00",
+        freq="H",
+        tz=None,
+        inclusive="left",
+    )
+    intersect = tools.intersect.indices_flex(
+        a, b, ignore_freq=True, ignore_tz=True, ignore_start_of_day=True
+    )
+    # print(a)
+    print(intersect)
+
+
+# tz()
+start_day()
+# freq()
+# all()
diff --git a/portfolyo/tools/intersect.py b/portfolyo/tools/intersect.py
index 5e2df66..15d0f17 100644
--- a/portfolyo/tools/intersect.py
+++ b/portfolyo/tools/intersect.py
@@ -3,7 +3,8 @@
 import pandas as pd
 from portfolyo import tools
 
-from portfolyo.tools.freq import longest
+from portfolyo.tools.right import stamp
+from portfolyo.tools.freq import longest, longer_or_shorter
 from datetime import datetime
 
 
@@ -119,6 +120,11 @@ def indices_flex(
     distinct_sod = set([i[0].time() for i in idxs])
     if len(distinct_sod) != 1 and ignore_start_of_day is False:
         raise ValueError(f"Indices must have equal start-of-day; got {distinct_sod}.")
+    for i in range(len(idxs)):
+        if len(distinct_sod) != 1 and longer_or_shorter(idxs[i].freq, "D") == -1:
+            raise ValueError(
+                "Downsample all indices to daily-or-longer, or trim them so they have the same start-of-day, before attempting to calculate the intersection"
+            )
 
     freq, name, tz = [], [], []
     for i in range(len(idxs)):
@@ -126,16 +132,7 @@ def indices_flex(
         name.append(idxs[i].name)
         tz.append(idxs[i].tz)
 
-    # add one interval of the respective freq to each index (this way, a given date-range from A-B that was exclusive
-    # of B is now inclusive of B - this helps when we need to convert frequencies or times-of-day without loosing
-    # data. At the end, we exclude the end-date of the final result again.)
-    # idxs = [
-    #     idx.append(
-    #         pd.DatetimeIndex([idx[-1] + pd.tseries.frequencies.to_offset(idx.freq)])
-    #     )
-    #     for idx in idxs
-    # ]
-
+    longest_freq = freq[0]
     if ignore_freq is True and len(distinct_freqs) != 1:
         # Find the longest frequency
         longest_freq = longest(*freq)
@@ -169,8 +166,9 @@ def indices_flex(
     idxs_out = []
     for i in range(len(idxs)):
         start = min(values)
+        # end = stamp(start, longest_freq._prefix)
         end = max(values)
-        inclusive = "both"
+        end = stamp(end, longest_freq)
 
         if ignore_start_of_day is True:
             start = datetime.combine(pd.to_datetime(start).date(), start_of_day[i])
@@ -184,7 +182,7 @@ def indices_flex(
                 freq=freq[i],
                 name=name[i],
                 tz=tz[i],
-                inclusive=inclusive,
+                inclusive="left",
             )
         )
 
diff --git a/tests/tools/test_intersect_flex.py b/tests/tools/test_intersect_flex.py
index f480d89..29e1281 100644
--- a/tests/tools/test_intersect_flex.py
+++ b/tests/tools/test_intersect_flex.py
@@ -95,7 +95,7 @@ def test_intersect_flex_ignore_start_of_day(
     do_test_intersect(
         "idx",
         idxs,
-        expected_startdate,
+        ValueError if freq == "15T" or freq == "H" else expected_startdate,
         expected_tz=tz,
         expected_freq=freq,
         expected_starttime=starttime,
@@ -190,7 +190,11 @@ def test_ignore_all(  # indexorframe: str,
     do_test_intersect(
         "idx",
         idxs,
-        expected_startdate,
+        (
+            ValueError
+            if freq[0] == "15T" or freq[0] == "H" or freq[1] == "15T" or freq[1] == "H"
+            else expected_startdate
+        ),
         expected_tz=tz,
         expected_freq=freq[0],
         expected_starttime=starttime,