From 64be0e9706af84a3d3fa2297c7396439a525c8e4 Mon Sep 17 00:00:00 2001
From: Dima Gerasimov <karlicoss@gmail.com>
Date: Thu, 30 Aug 2018 17:31:11 +0300
Subject: [PATCH] grouping

---
 location/__init__.py | 48 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)

diff --git a/location/__init__.py b/location/__init__.py
index f521011..11bc401 100644
--- a/location/__init__.py
+++ b/location/__init__.py
@@ -32,6 +32,8 @@ def tagger(dt: datetime, lat: float, lon: float) -> Tag:
     else:
         return "other"
 
+# TODO hope they are sorted...
+# TODO that could also serve as basis for timezone provider.
 def iter_locations() -> Iterator[Location]:
     with open(PATH) as fo:
         reader = csv.reader(fo)
@@ -59,6 +61,52 @@ def get_locations(cached: bool=False) -> Iterable[Location]:
     else:
         return list(iter_locations())
 
+class LocInterval(NamedTuple):
+    from_: Location
+    to: Location
+
+def get_groups(cached: bool=False) -> List[LocInterval]:
+    locs = get_locations(cached=cached)
+    i = 0
+    groups: List[LocInterval] = []
+    curg: List[Location] = []
+
+    def add_to_group(x):
+        nonlocal curg
+        if len(curg) < 2:
+            curg.append(x)
+        else:
+            curg[-1] = x
+
+    def dump_group():
+        nonlocal curg
+        if len(curg) > 0:
+            groups.append(LocInterval(from_=curg[0], to=curg[-1]))
+            curg = []
+
+    while i < len(locs):
+        last = None if len(curg) == 0 else curg[-1]
+        cur = locs[i]
+        j = i
+        match = False
+        while not match and j < len(locs) and j < i + 10: # TODO FIXME time distance here... e.g. half an hour?
+            cur = locs[j]
+            if last is None or cur.tag == last.tag:
+                # ok
+                add_to_group(cur)
+                i = j + 1
+                match = True
+            else:
+                j += 1
+        # if we made here without advancing
+        if not match:
+            dump_group()
+            i += 1
+        else:
+            pass
+    dump_group()
+    return groups
+
 def update_cache():
     import dill # type: ignore
     datas = get_locations(cached=False)