Adding and testing TimeStampedGazePositions.from_dataframe classmethod.

2 files changed, 94 insertions, 7 deletions

diff --git a/src/argaze.test/GazeFeatures.py b/src/argaze.test/GazeFeatures.py
index f73eefe..8cd2e56 100644
--- a/src/argaze.test/GazeFeatures.py
+++ b/src/argaze.test/GazeFeatures.py
@@ -13,6 +13,7 @@ from dataclasses import dataclass
 from argaze import GazeFeatures
 
 import numpy
+import pandas
 
 def random_gaze_positions(size, screen_dimension: tuple[float, float] = (1, 1)):
     """ Generate random TimeStampedGazePsoitions for testing purpose.
@@ -209,6 +210,52 @@ class TestTimeStampedGazePositionsClass(unittest.TestCase):
 
         self.assertEqual(repr(ts_gaze_positions), "{\"0\": {\"message\": null, \"value\": [null, null], \"precision\": null}}")
 
+    def test_from_dataframe(self):
+        """Test from_dataframe classmethod."""
+
+        data = {'Specific timestamp label': [0, 1, 2, 3, 4], 
+                'Specific gaze position x label': [0, 10, numpy.nan, 30, 40], 
+                'Specific gaze position y label': [0, 100, numpy.nan, 300, 400]
+                }
+        dataframe = pandas.DataFrame().from_dict(data)
+
+        ts_gaze_positions = GazeFeatures.TimeStampedGazePositions.from_dataframe(dataframe, timestamp='Specific timestamp label', x='Specific gaze position x label', y='Specific gaze position y label')
+
+        # Check buffer length
+        self.assertEqual(len(ts_gaze_positions), 5)
+
+        # Check first gaze position is correctly stored and accessible as a GazePosition
+        self.assertIsInstance(ts_gaze_positions[0], GazeFeatures.GazePosition)
+        self.assertEqual(ts_gaze_positions[0].precision, 0)
+        self.assertEqual(ts_gaze_positions[0].valid, True)
+
+        # Check third gaze position is correctly stored and accessible as a UnvalidGazePosition
+        self.assertIsInstance(ts_gaze_positions[2], GazeFeatures.UnvalidGazePosition)
+        self.assertEqual(ts_gaze_positions[2].precision, None)
+        self.assertEqual(ts_gaze_positions[2].valid, False)
+
+        data = {'Specific timestamp label': [0, 1, 2, 3, 4], 
+                'Specific gaze position x label': [0, 10, numpy.nan, 30, 40], 
+                'Specific gaze position y label': [0, 100, numpy.nan, 300, 400],
+                'Specific precisison label': [15, 15, numpy.nan, 15, 15]
+                }
+        dataframe = pandas.DataFrame().from_dict(data)
+
+        ts_gaze_positions = GazeFeatures.TimeStampedGazePositions.from_dataframe(dataframe, timestamp='Specific timestamp label', x='Specific gaze position x label', y='Specific gaze position y label', precision='Specific precisison label')
+
+        # Check buffer length
+        self.assertEqual(len(ts_gaze_positions), 5)
+
+        # Check first gaze position is correctly stored and accessible as a GazePosition
+        self.assertIsInstance(ts_gaze_positions[0], GazeFeatures.GazePosition)
+        self.assertEqual(ts_gaze_positions[0].precision, 15)
+        self.assertEqual(ts_gaze_positions[0].valid, True)
+
+        # Check third gaze position is correctly stored and accessible as a UnvalidGazePosition
+        self.assertIsInstance(ts_gaze_positions[2], GazeFeatures.UnvalidGazePosition)
+        self.assertEqual(ts_gaze_positions[2].precision, None)
+        self.assertEqual(ts_gaze_positions[2].valid, False)
+
     def test_as_dataframe(self):
         """Test inherited as_dataframe method."""
 
diff --git a/src/argaze/GazeFeatures.py b/src/argaze/GazeFeatures.py
index 53bec25..870cd3c 100644
--- a/src/argaze/GazeFeatures.py
+++ b/src/argaze/GazeFeatures.py
@@ -125,9 +125,15 @@ class TimeStampedGazePositions(DataStructures.TimeStampedBuffer):
 
             assert(set(['value', 'precision']).issubset(value.keys()))
 
-            if 'message' in value.keys():
+            if math.isnan(value['precision']):
 
-                value = UnvalidGazePosition(value['message'])
+                if 'message' in value.keys():
+
+                    value = UnvalidGazePosition(value['message'])
+
+                else :
+
+                    value = UnvalidGazePosition()
 
             else:
 
@@ -148,14 +154,48 @@ class TimeStampedGazePositions(DataStructures.TimeStampedBuffer):
             return TimeStampedGazePositions({ast.literal_eval(ts_str): json_buffer[ts_str] for ts_str in json_buffer})
 
     @classmethod
-    def from_dataframe(self, dataframe: pandas.DataFrame, exclude=[]) -> TimeStampedGazePositionsType:
-        """Create a TimeStampedGazePositions from [Pandas DataFrame](https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.html)."""
+    def from_dataframe(self, dataframe: pandas.DataFrame, timestamp: str, x: str, y: str, precision: str = None) -> TimeStampedGazePositionsType:
+        """Create a TimeStampedGazePositions from [Pandas DataFrame](https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.html).
+
+        Parameters:
+            timestamp: specific timestamp column label.
+            x: specific x column label.
+            y: specific y column label.
+            precision: specific precision column label if exist.
+        """
+
+        # Select columns
+        if precision:
+
+            df = dataframe[[timestamp, x, y, precision]]
+
+        else:
 
-        dataframe.drop(exclude, inplace=True, axis=True)
+            df = dataframe[[timestamp, x, y]]
 
-        assert(dataframe.index.name == 'timestamp')
+        # Merge x and y columns into one 'value' column
+        df['value'] = tuple(zip(df[x], df[y]))
+        df.drop(columns= [x, y], inplace=True, axis=1)
+
+        # Handle precision data
+        if precision:
+
+             # Rename precision column into 'precision' column
+            df.rename(columns={precision: 'precision'}, inplace=True)
+
+        else:
+
+            # Append a precision column where precision is NaN if value is a tuple of NaN else 0
+            df['precision'] = df.apply(lambda row: numpy.nan if math.isnan(row.value[0]) or math.isnan(row.value[1]) else 0, axis=True)
+
+        # Rename timestamp column into 'timestamp' column then use it as index
+        df.rename(columns={timestamp: 'timestamp'}, inplace=True)
+        df.set_index('timestamp', inplace=True)
+        
+        # Filter duplicate timestamps
+        df = df[df.index.duplicated() == False]
 
-        return TimeStampedGazePositions(dataframe.to_dict('index'))
+        return TimeStampedGazePositions(df.to_dict('index'))
 
 GazeMovementType = TypeVar('GazeMovement', bound="GazeMovement")
 # Type definition for type annotation convenience