From c335c0c9d8673247dd7141cc9c075a5846467e28 Mon Sep 17 00:00:00 2001
From: karlicoss <karlicoss@gmail.com>
Date: Sat, 21 Oct 2023 23:57:01 +0100
Subject: [PATCH] core/stats: report datetime of first item in addition to last

quite useful for quickly determining time span of a data source
---
 my/core/common.py | 20 ++++++++++++++------
 my/core/stats.py  |  1 +
 2 files changed, 15 insertions(+), 6 deletions(-)

diff --git a/my/core/common.py b/my/core/common.py
index b692730..b34d6d2 100644
--- a/my/core/common.py
+++ b/my/core/common.py
@@ -431,22 +431,25 @@ def stat(
     }
 
 
-def _stat_iterable(it: Iterable[C], quick: bool=False) -> Any:
+def _stat_iterable(it: Iterable[C], quick: bool = False) -> Any:
     from more_itertools import ilen, take, first
 
     # todo not sure if there is something in more_itertools to compute this?
     total = 0
     errors = 0
-    last = None
+    first_item = None
+    last_item = None
 
     def funcit():
-        nonlocal errors, last, total
+        nonlocal errors, first_item, last_item, total
         for x in it:
             total += 1
             if isinstance(x, Exception):
                 errors += 1
             else:
-                last = x
+                last_item = x
+                if first_item is None:
+                    first_item = x
             yield x
 
     eit = funcit()
@@ -471,8 +474,13 @@ def _stat_iterable(it: Iterable[C], quick: bool=False) -> Any:
     if errors > 0:
         res['errors'] = errors
 
-    if last is not None:
-        dt = guess_datetime(last)
+    if first_item is not None:
+        dt = guess_datetime(first_item)
+        if dt is not None:
+            res['first'] = dt
+
+    if last_item is not None:
+        dt = guess_datetime(last_item)
         if dt is not None:
             res['last'] = dt
     return res
diff --git a/my/core/stats.py b/my/core/stats.py
index 9dfaa04..1818b63 100644
--- a/my/core/stats.py
+++ b/my/core/stats.py
@@ -37,6 +37,7 @@ def test_guess_stats() -> None:
     r = res['iter_data']
     assert r == {
         'count': 9,
+        'first': datetime(2020, 1, 1, 1, 1, 1),
         'last': datetime(2020, 1, 3, 1, 1, 1),
     }