NeuralEnsemble / python-neo / 1134

49%

master: 86%

Pull #550

travis-ci

Fixes for AxographIO's pairing of channel name/units with data, for signal start time, and for units detection (#2)

* Fix for AxographIO's pairing of channel name/units with data

When np.unique is used to find unique channel names, the names are automatically sorted alphabetically. Channel units are extracted from names and so have the same reordering. However, as AnalogSignals are added to Segments, data is iterated over according to the ordering in the original AxoGraph file. Consequently, when channels appear in non-alphabetical order in an AxoGraph file, name/units and data are paired incorrectly by AxographIO.

This change fixes the pairing mistake by preserving the original ordering of channels when finding unique channel names. Preserving the original ordering of channels, rather than reordering the data so that the AnalogSignals are sorted alphabetically within each Segment, is preferable since users will likely expect ordering to match their AxoGraph data files; if not they can sort the channels by name themselves. Sorting during reading of the file is not reversible for users who need to preserve the original ordering.

* Correction for AxographIO signal start time

The start time in AxoGraph files is not 0. Rather, it is one sample period after 0 (see spreadsheet view in AxoGraph or the contents of an exported text file). Both sample period and start time are made accessible directly by the axographio package, so this change uses them instead.

* Make AxographIO save file_origin when possible

* Fix units detection by AxographIO when parentheses exist in name

When an AxoGraph file contains a channel name that contains parentheses in addition to the units, e.g., "My channel (it's really great) (V)", the units are extracted incorrectly as the first parenthetical phrase. This change fixes this bug.

* Strip units from channel names

Units are stored already in AnalogSignals. If the signal is rescaled by converting units, e... (continued)

82 of 82 new or added lines in 3 files covered. (100.0%)

12082 of 24346 relevant lines covered (49.63%)

1.98 hits per line