Properly generate epoch structure.

eeglabio/epochs.py

@@ -92,7 +92,7 @@ def export_set(fname, data, sfreq, events, tmin, tmax, ch_names, event_id=None,
     ev_dur = np.zeros((trials,), dtype=np.int64)
 
     # indices of epochs each event belongs to
-    ev_epoch = np.arange(1, trials + 1)
+    ev_epoch = ev_lat // data.shape[1] + 1
 
     # merge annotations into events array
     if annotations is not None:
@@ -123,13 +123,31 @@ def export_set(fname, data, sfreq, events, tmin, tmax, ch_names, event_id=None,
                         names=["type", "latency", "duration", "epoch"])
 
     # construct epochs array
+    # true epochs array, one subarray per events in epoch
+    # make sure epoch count is increasing (it should be)
+    # splitting code from https://stackoverflow.com/a/43094244/8170714
+    epoch_start_idx = np.unique(all_epoch, return_index=True)[1][1:]  # skip 0
+    ep_event = np.split(np.arange(1, len(all_epoch)+1, dtype=np.double),
+                        epoch_start_idx)
+    # starting latency for each epoch in seconds
+    ep_lat_offset = (all_epoch - 1) * data.shape[1] / sfreq
+    all_lat_shifted = all_lat / sfreq - ep_lat_offset  # shifted rel to ep onset
+    # convert lat, pos, type to cell arrays by converting to object arrays
+    ep_lat = np.split(all_lat_shifted.astype(dtype=object) * 1000,
+                      epoch_start_idx)
+    ep_pos = np.split(all_epoch.astype(dtype=object), epoch_start_idx)
+    ep_types = np.split(all_types.astype(dtype=object), epoch_start_idx)
+
+    # regular one event per epoch
     # same as the indices for event epoch, except use array
-    ep_event = [np.array(n) for n in ev_epoch]
-    ep_lat = [np.array(n) for n in ev_lat]
-    ep_types = [np.array(n) for n in ev_types]
-
-    epochs = fromarrays([ep_event, ep_lat, ep_types],
-                        names=["event", "eventlatency", "eventtype"])
+    # ep_event = [np.array(n) for n in ev_epoch]
+    # ep_lat = [np.array(n) for n in ev_lat]
+    # ep_types = [np.array(n) for n in ev_types]
+
+    field_names = ["event", "eventlatency", "eventposition", "eventtype"]
+    epochs = fromarrays([np.array(arr, dtype=object) for arr in
+                         [ep_event, ep_lat, ep_pos, ep_types]],
+                        names=field_names)
 
     if isinstance(ref_channels, list):
         ref_channels = " ".join(ref_channels)