API: Store Eyelink "messages" (i.e. events) as unicode strings, not byte strings

[scott-huberty]

Currently, eyelinkIO will read stimulus events ("messages" in EyeLink terminology) from the EDF file and store them as byte strings (see example below). This was inherited from the original implementation in the pyeparse package.

After a chat with @teonbrooks we've concluded that we should probably just store these as regular ol' Python strings, because 1) It will make working with these events easier for user, 2) We don't need to be Python 2 compatible, and 3) I don't think memory is an issue that warrants the use of byte strings.

---

I think that changing this line:

eyelinkio/src/eyelinkio/edf/read.py

Lines 197 to 198 in c3afb9a

	dtype = [("stime", np.float64), ("msg", "|S%s" % _MAX_MSG_LEN)]
	res["discrete"]["messages"] = np.empty((n_samps["messages"]), dtype=dtype)

to this should work:

dtype = [("stime", np.float64), ("msg", f"U{_MAX_MSG_LEN}")]
res["discrete"]["messages"] = np.empty((n_samps["messages"]), dtype=dtype)

Which will specify a Unicode string type while preserving the _MAX_MSG_LEN constraint that currently exists.

---

Example of current behavior

import eyelinkio

fname = eyelinkio.utils._get_test_fnames()[0]
edf = eyelinkio.read_edf(fname)
edf["discrete"]["messages"]

array([(  0.   , b'RECCFG CR 1000 2 1 L'), (  0.   , b'ELCLCFG MTABLER'),
       (  0.   , b'GAZE_COORDS 0.00 0.00 1919.00 1079.00'),
       (  0.   , b'THRESHOLDS L 120 255'),
       (  0.   , b'ELCL_PROC CENTROID (3)'),
       (  0.   , b'ELCL_PCR_PARAM 5 3.0'),
       (  0.   , b'!MODE RECORD CR 1000 2 1 L\n'),
       (  3.314, b'TRIALID 1'), (  6.329, b'TRIALID 2'),
       (  9.339, b'TRIALID 3'), ( 12.348, b'TRIALID 4'),
       ( 15.356, b'TRIALID 5'), ( 18.364, b'TRIALID 6'),
       ( 21.373, b'TRIALID 7'), ( 24.382, b'TRIALID 8'),
       ( 27.39 , b'TRIALID 9'), ( 30.399, b'TRIALID 10'),
       ( 33.409, b'TRIALID 11'), ( 36.418, b'TRIALID 12'),
       ( 39.426, b'TRIALID 13'), ( 42.434, b'TRIALID 14'),
       ( 45.444, b'TRIALID 15'), ( 48.453, b'TRIALID 16'),
       ( 51.461, b'TRIALID 17'), ( 54.469, b'TRIALID 18'),
       ( 57.478, b'TRIALID 19'), ( 60.485, b'TRIALID 20'),
       ( 63.493, b'TRIALID 21'), ( 66.501, b'TRIALID 22'),
       ( 69.51 , b'TRIALID 23'), ( 72.518, b'TRIALID 24'),
       ( 75.526, b'TRIALID 25'), ( 78.534, b'TRIALID 26'),
       ( 81.543, b'TRIALID 27'), ( 84.55 , b'TRIALID 28'),
       ( 87.559, b'TRIALID 29'), ( 90.567, b'TRIALID 30'),
       ( 93.575, b'TRIALID 31'), ( 96.583, b'TRIALID 32'),
       ( 99.591, b'TRIALID 33'), (102.599, b'TRIALID 34'),
       (105.608, b'TRIALID 35'), (108.615, b'TRIALID 36'),
       (111.623, b'TRIALID 37'), (114.63 , b'TRIALID 38'),
       (117.639, b'TRIALID 39'), (120.647, b'TRIALID 40'),
       (124.739, b'Ending')], dtype=[('stime', '<f8'), ('msg', 'S260')])