Added date filter

eyra · Nov 26, 2023 · 70f4ee5 · 70f4ee5
1 parent cc9733c
commit 70f4ee5
Show file tree

Hide file tree

Showing 9 changed files with 22 additions and 10 deletions.
diff --git a/public/port-0.0.0-py3-none-any.whl b/public/port-0.0.0-py3-none-any.whl
diff --git a/src/framework/processing/py/dist/port-0.0.0-py3-none-any.whl b/src/framework/processing/py/dist/port-0.0.0-py3-none-any.whl
diff --git a/src/framework/processing/py/numpy-1.25.2-cp311-cp311-macosx_11_0_arm64.whl b/src/framework/processing/py/numpy-1.25.2-cp311-cp311-macosx_11_0_arm64.whl
diff --git a/src/framework/processing/py/port/__pycache__/script.cpython-311.pyc b/src/framework/processing/py/port/__pycache__/script.cpython-311.pyc
diff --git a/src/framework/processing/py/port/api/__pycache__/props.cpython-311.pyc b/src/framework/processing/py/port/api/__pycache__/props.cpython-311.pyc
diff --git a/src/framework/processing/py/port/script.py b/src/framework/processing/py/port/script.py
@@ -11,6 +11,8 @@
 
 ExtractionResult = namedtuple("ExtractionResult", ["id", "title", "data_frame"])
 
+filter_start_date = datetime(2017, 1, 1)
+
 
 class FileInZipNotFoundError(Exception):
     """Raised when a specific file is not found within the ZIP archive."""
@@ -25,19 +27,13 @@ class InvalidXMLError(Exception):
 class HealthDataHandler(xml.sax.ContentHandler):
     def __init__(self, callback):
         self.callback = callback
-        self.in_step_count = False
 
     def startElement(self, tag, attributes):
         if tag == "Record" and attributes["type"] == "HKQuantityTypeIdentifierStepCount":
-            self.in_step_count = True
             value = int(attributes["value"])
-            startDate = self.parse_naive_datetime(attributes["startDate"])
-            self.callback(value, startDate)
+            start_date = self.parse_naive_datetime(attributes["startDate"])
+            self.callback(value, start_date)
 
-    def endElement(self, tag):
-        if tag == "Record" and self.in_step_count:
-            self.in_step_count = False
-
     def parse_naive_datetime(self, date_str):
         dt = datetime.strptime(date_str, '%Y-%m-%d %H:%M:%S %z')
         return dt.replace(tzinfo=None)
@@ -47,9 +43,11 @@ def __init__(self):
         self.start_times = []
         self.steps = []
 
-    def __call__(self, value, startDate):
+    def __call__(self, value, start_date):
+        if start_date < filter_start_date:
+            return
         self.steps.append(value)
-        self.start_times.append(startDate)
+        self.start_times.append(start_date)
 
     def to_dataframe(self):
         return pd.DataFrame({'Start Time': self.start_times, 'Steps': self.steps})

diff --git a/src/framework/processing/py/tests/__pycache__/__init__.cpython-311.pyc b/src/framework/processing/py/tests/__pycache__/__init__.cpython-311.pyc
diff --git a/src/framework/processing/py/tests/__pycache__/script_test.cpython-311-pytest-7.4.2.pyc b/src/framework/processing/py/tests/__pycache__/script_test.cpython-311-pytest-7.4.2.pyc
diff --git a/src/framework/processing/py/tests/script_test.py b/src/framework/processing/py/tests/script_test.py
@@ -53,6 +53,20 @@ def test_aggregate_same_day():
     assert df.iloc[0]['Date'] == '2023-06-25'
     assert df.iloc[0]['Steps'] == 40
 
+
+def test_filters_out_data_based_on_date():
+    # see the script for the earliest data which will be included
+    xml_data = """
+    <HealthData locale="en_NL">
+     <Record type="HKQuantityTypeIdentifierStepCount" startDate="2016-06-24 23:10:45 +0100" value="2"/>
+     <Record type="HKQuantityTypeIdentifierStepCount" startDate="2017-01-1 23:10:45 +0100" value="3"/>
+    </HealthData>
+    """
+    df = aggregate_daily_steps(io.StringIO(xml_data))
+    assert len(df) == 1
+    assert df.iloc[0]['Date'] == '2017-01-01'
+    assert df.iloc[0]['Steps'] == 3
+
 def test_no_records():
     xml_data = """
     <HealthData locale="en_NL">