Module thunderfish.hopkinsloader
Load EODs from Hopkins files.
Carl Hopkins and John Sullivan stored only a few cut-out EOD waveforms of Mormyrid EODs in a specific mat file. These recordings are available at the Macaulay library.
Functions
load_hopkins(): load a Hopkins file containing a few EOD pulses.analyse_hopkins(): analyze the content of Hopkins files.
Expand source code
"""Load EODs from Hopkins files.
Carl Hopkins and John Sullivan stored only a few cut-out EOD waveforms
of Mormyrid EODs in a specific mat file. These recordings are
available at the Macaulay library.
## Functions
- `load_hopkins()`: load a Hopkins file containing a few EOD pulses.
- `analyse_hopkins()`: analyze the content of Hopkins files.
"""
import sys
import numpy as np
import matplotlib.pyplot as plt
from scipy.io import loadmat
from audioio import print_metadata
def load_hopkins(file_path):
""" Load a Hopkins file containing a few EOD pulses.
Parameters
----------
file_path: str
The mat file with the data.
Returns
-------
data: list of 2-D ndarrays
A list of single EOD pulses.
First column is time in seconds, second column EOD waveform.
md: nested dict
Metadata.
Raises
------
FileNotFoundError
`file_path` does not exist.
ValueError
`file_path` is not a valid mat file.
"""
x = loadmat(file_path, squeeze_me=True)
if not 'eod' in x or not hasattr(x['eod'], 'size'):
return [], {}
y = x['eod'].reshape(x['eod'].size)
if not 'wave' in y.dtype.names or not 'time' in y.dtype.names:
return [], {}
# assemble data:
data = []
for k in range(len(y['wave'])):
eod = np.zeros((len(y['wave'][k]), 2))
eod[:, 0] = y['time'][k]
eod[:, 1] = (y['wave'][k]).astype(float)
data.append(eod)
# assemble metadata:
md = {}
eod_md = []
for n in y.dtype.names:
t = type(y[n][0])
if not t is np.ndarray:
# some metadata may or may not differ between EODs:
for k in range(len(y[n])):
if y[n][k] and n != 'eodnum' and y[n][k] != y[n][0]:
while len(eod_md) < len(y[n]):
eod_md.append({})
for k in range(len(y[n])):
v = y[n][k]
if isinstance(v, str):
v = v.replace('Date:', '')
v = v.replace('Time:', '')
v = v.replace('Time', '')
v = v.strip()
eod_md[k][n] = v
break
else:
v = y[n][0]
if isinstance(v, str):
v = v.replace('Date:', '')
v = v.replace('Time:', '')
v = v.replace('Time', '')
v = v.replace(' T: ', 'T')
v = v.strip()
md[n] = v
for k in range(len(eod_md)):
md[f'EOD{k}'] = eod_md[k]
return data, md
def analyse_hopkins(pathes):
""" Analyze the content of Hopkins files.
Prints out some statistics about the field names and types.
Parameters
----------
pathes: list of str
Files to be analyzed.
"""
keys = {}
types = {}
data_types = {}
for file_path in pathes:
x = loadmat(file_path, squeeze_me=True)
y = x['eod'].reshape(x['eod'].size)
for n in y.dtype.names:
c = keys.get(n, 0)
keys[n] = c + 1
t = type(y[n][0])
c = types.get(t, 0)
types[t] = c + 1
t = y['wave'][0].dtype
c = data_types.get(t, 0)
data_types[t] = c + 1
# each file contains several "wave" and "time" arrays for plotting the EODs.
# within a file they might differ in size!
# print all keys found in the data with their frequency:
print('keys:')
for k in keys:
print(f' {100*keys[k]/len(pathes):3.0f}%', k)
print()
# print all wave data types with their frequency:
print('data types:')
for t in data_types:
print(f' {data_types[t]:5d}', t)
print()
# 226 float64
# 57 int16
# print types of all fields with their frequency:
print('field types:')
for t in types:
print(f' {types[t]:5d}', t)
print()
# 2845 <class 'int'>
# 4002 <class 'numpy.ndarray'>
#14681 <class 'str'>
# 1478 <class 'float'>
if __name__ == '__main__':
analyse_hopkins(sys.argv[1:])
for file_path in sys.argv[1:]:
print(file_path)
data, md = load_hopkins(file_path)
print_metadata(md, ' ')
fig, ax = plt.subplots()
ax.set_title(md.get('speciesIDweb', ''))
for k in range(len(data)):
ax.plot(1000*data[k][:, 0], data[k][:, 1])
ax.set_xlabel('Time [ms]')
plt.show()
print()Functions
def load_hopkins(file_path)-
Load a Hopkins file containing a few EOD pulses.
Parameters
file_path:str- The mat file with the data.
Returns
data:listof2-D ndarrays- A list of single EOD pulses. First column is time in seconds, second column EOD waveform.
md:nested dict- Metadata.
Raises
FileNotFoundErrorfile_pathdoes not exist.ValueErrorfile_pathis not a valid mat file.
Expand source code
def load_hopkins(file_path): """ Load a Hopkins file containing a few EOD pulses. Parameters ---------- file_path: str The mat file with the data. Returns ------- data: list of 2-D ndarrays A list of single EOD pulses. First column is time in seconds, second column EOD waveform. md: nested dict Metadata. Raises ------ FileNotFoundError `file_path` does not exist. ValueError `file_path` is not a valid mat file. """ x = loadmat(file_path, squeeze_me=True) if not 'eod' in x or not hasattr(x['eod'], 'size'): return [], {} y = x['eod'].reshape(x['eod'].size) if not 'wave' in y.dtype.names or not 'time' in y.dtype.names: return [], {} # assemble data: data = [] for k in range(len(y['wave'])): eod = np.zeros((len(y['wave'][k]), 2)) eod[:, 0] = y['time'][k] eod[:, 1] = (y['wave'][k]).astype(float) data.append(eod) # assemble metadata: md = {} eod_md = [] for n in y.dtype.names: t = type(y[n][0]) if not t is np.ndarray: # some metadata may or may not differ between EODs: for k in range(len(y[n])): if y[n][k] and n != 'eodnum' and y[n][k] != y[n][0]: while len(eod_md) < len(y[n]): eod_md.append({}) for k in range(len(y[n])): v = y[n][k] if isinstance(v, str): v = v.replace('Date:', '') v = v.replace('Time:', '') v = v.replace('Time', '') v = v.strip() eod_md[k][n] = v break else: v = y[n][0] if isinstance(v, str): v = v.replace('Date:', '') v = v.replace('Time:', '') v = v.replace('Time', '') v = v.replace(' T: ', 'T') v = v.strip() md[n] = v for k in range(len(eod_md)): md[f'EOD{k}'] = eod_md[k] return data, md def analyse_hopkins(pathes)-
Analyze the content of Hopkins files.
Prints out some statistics about the field names and types.
Parameters
pathes:listofstr- Files to be analyzed.
Expand source code
def analyse_hopkins(pathes): """ Analyze the content of Hopkins files. Prints out some statistics about the field names and types. Parameters ---------- pathes: list of str Files to be analyzed. """ keys = {} types = {} data_types = {} for file_path in pathes: x = loadmat(file_path, squeeze_me=True) y = x['eod'].reshape(x['eod'].size) for n in y.dtype.names: c = keys.get(n, 0) keys[n] = c + 1 t = type(y[n][0]) c = types.get(t, 0) types[t] = c + 1 t = y['wave'][0].dtype c = data_types.get(t, 0) data_types[t] = c + 1 # each file contains several "wave" and "time" arrays for plotting the EODs. # within a file they might differ in size! # print all keys found in the data with their frequency: print('keys:') for k in keys: print(f' {100*keys[k]/len(pathes):3.0f}%', k) print() # print all wave data types with their frequency: print('data types:') for t in data_types: print(f' {data_types[t]:5d}', t) print() # 226 float64 # 57 int16 # print types of all fields with their frequency: print('field types:') for t in types: print(f' {types[t]:5d}', t) print() # 2845 <class 'int'> # 4002 <class 'numpy.ndarray'> #14681 <class 'str'> # 1478 <class 'float'>