fix classify and detect for EPG audio read plugin

Author: bjarthur

src/classify

@@ -60,7 +60,7 @@ import importlib
 from lib import log_nvidia_smi_output, select_GPUs
 
 import tifffile
-from lib import compute_background
+from lib import compute_background, load_audio_read_plugin
 
 FLAGS = None
 
@@ -79,15 +79,8 @@ def main():
   sys.path.append(os.path.dirname(FLAGS.video_findfile))
   video_findfile = importlib.import_module(os.path.basename(FLAGS.video_findfile)).video_findfile
 
-  sys.path.append(os.path.dirname(FLAGS.audio_read_plugin))
-  audio_read_module = importlib.import_module(os.path.basename(FLAGS.audio_read_plugin))
-  def audio_read(wav_path, start_tic=None, stop_tic=None):
-      return audio_read_module.audio_read(wav_path, start_tic, stop_tic,
-                                          **FLAGS.audio_read_plugin_kwargs)
-  def audio_read_rec2ch(wavfile):
-      return audio_read_module.audio_read_rec2ch(wavfile, **FLAGS.audio_read_plugin_kwargs)
-  def audio_read_strip_rec(wavfile):
-      return audio_read_module.audio_read_strip_rec(wavfile, **FLAGS.audio_read_plugin_kwargs)
+  load_audio_read_plugin(FLAGS.audio_read_plugin, FLAGS.audio_read_plugin_kwargs)
+  from lib import  audio_read, trim_ext
 
   sys.path.append(os.path.dirname(FLAGS.video_read_plugin))
   video_read_module = importlib.import_module(os.path.basename(FLAGS.video_read_plugin))
@@ -249,12 +242,7 @@ def main():
     else:
       adjusted_probability = probability_matrix[:,ch]
     waveform = adjusted_probability*np.iinfo(np.int16).max
-    if len(audio_read_rec2ch(audio_read_strip_rec(FLAGS.wav))) > 1:
-      tmp = FLAGS.wav.split('-')
-      withext, rec = '-'.join(tmp[:-1]), tmp[-1]
-      withoutext = os.path.splitext(withext)[0]+'-'+rec
-    else:
-      withoutext = os.path.splitext(FLAGS.wav)[0]
+    withoutext = trim_ext(FLAGS.wav)
     filename = withoutext+'-'+labels[ch]+'.wav'
     wavfile.write(filename, int(sample_rate), waveform.astype('int16'))
 

src/detect-plugin.py

@@ -10,7 +10,6 @@
 
 import argparse
 import os
-import importlib
 import sys
 import csv
 import socket
@@ -64,11 +63,8 @@ def main():
   for key in sorted(flags.keys()):
     print('%s = %s' % (key, flags[key]))
 
-  sys.path.append(os.path.dirname(FLAGS.audio_read_plugin))
-  audio_read_module = importlib.import_module(os.path.basename(FLAGS.audio_read_plugin))
-  def audio_read(wav_path, start_tic=None, stop_tic=None):
-      return audio_read_module.audio_read(wav_path, start_tic, stop_tic,
-                                          **FLAGS.audio_read_plugin_kwargs)
+  load_audio_read_plugin(FLAGS.audio_read_plugin, FLAGS.audio_read_plugin_kwargs)
+  from lib import audio_read, trim_ext
 
   hyperparameter1 = int(FLAGS.parameters["my-simple-textbox"])
 
@@ -79,7 +75,7 @@ def audio_read(wav_path, start_tic=None, stop_tic=None):
   amplitude = scipy.signal.medfilt(song)
 
   basename = os.path.basename(FLAGS.filename)
-  with open(os.path.splitext(FLAGS.filename)[0]+'-detected.csv', 'w') as fid:
+  with open(trim_ext(FLAGS.filename)+'-detected.csv', 'w') as fid:
       csvwriter = csv.writer(fid, lineterminator='\n')
       for i in range(1,len(amplitude)-1):
           if amplitude[i] > hyperparameter1:

src/gui/controller.py

@@ -937,10 +937,10 @@ def pbfile_succeeded(pbfile, reftime):
     return recent_file_exists(pbfile, reftime, True)
 
 def detect_succeeded(wavfile, reftime):
-    logfile = wavfile[:-4]+'-detect.log'
+    logfile = M.trim_ext(wavfile)+'-detect.log'
     if not logfile_succeeded(logfile, reftime):
         return False
-    csvfile = wavfile[:-4]+'-detected.csv'
+    csvfile = M.trim_ext(wavfile)+'-detected.csv'
     if not csvfile_succeeded(csvfile, reftime):
         return False
     return True
@@ -955,7 +955,7 @@ async def detect_actuate():
 async def _detect_actuate(i, wavfiles, threads, results):
     wavfile = wavfiles.pop(0)
     currtime = time.time()
-    logfile = os.path.splitext(wavfile)[0]+'-detect.log'
+    logfile = M.trim_ext(wavfile)+'-detect.log'
     jobid = generic_actuate(M.detect_plugin+".py", logfile, \
                             M.detect_where,
                             M.detect_ncpu_cores,

src/gui/model.py

@@ -15,7 +15,7 @@
 import importlib
 
 sys.path.append(os.path.dirname(os.path.dirname(os.path.realpath(__file__))))
-from lib import get_srcrepobindirs
+from lib import get_srcrepobindirs, load_audio_read_plugin
 
 def parse_model_file(modelstr):
     filepath, filename = os.path.split(modelstr)
@@ -83,15 +83,6 @@ def isused(sound):
     return np.where([x['file']==sound['file'] and x['ticks']==sound['ticks'] \
                      for x in used_sounds])[0]
 
-def trim_ext(wavfile):
-    if len(audio_read_rec2ch(audio_read_strip_rec(wavfile))) > 1:
-        tmp = wavfile.split('-')
-        withext, rec = '-'.join(tmp[:-1]), tmp[-1]
-        withoutext = os.path.splitext(withext)[0]+'-'+rec
-    else:
-        withoutext = os.path.splitext(wavfile)[0]
-    return withoutext
-
 def save_annotations():
     global nrecent_annotations
     if nrecent_annotations>0:
@@ -259,7 +250,7 @@ def init(_bokeh_document, _configuration_file, _use_aitch):
     global context_width_sec0, context_offset_sec0
     global xcluster, ycluster, zcluster, ndcluster, tic2pix_max, snippet_width_pix, ilayer, ispecies, iword, inohyphen, ikind, nlayers, layers, species, words, nohyphens, kinds, used_labels, snippets_gap_sec, snippets_tic, snippets_gap_tic, snippets_decimate_by, snippets_pix, snippets_gap_pix, context_decimate_by, context_width_tic, context_offset_tic, context_sound, isnippet, xsnippet, ysnippet, file_nframes, context_midpoint_tic, ilabel, used_sounds, used_starts_sorted, used_stops, iused_stops_sorted, annotated_sounds, annotated_starts_sorted, annotated_stops, iannotated_stops_sorted, annotated_csvfiles_all, nrecent_annotations, clustered_sounds, clustered_activations, used_recording2firstsound, clustered_starts_sorted, clustered_stops, iclustered_stops_sorted, songexplorer_starttime, history_stack, history_idx, wizard, action, function, statepath, state, file_dialog_root, file_dialog_filter, nearest_sounds, status_ticker_queue, waitfor_job, dfs, remaining_isounds
     global user_changed_recording, user_copied_parameters
-    global audio_read, audio_read_exts, audio_read_rec2ch, audio_read_strip_rec
+    global audio_read, audio_read_exts, audio_read_rec2ch, audio_read_strip_rec, trim_ext
     global video_read, detect_labels, doubleclick_annotation, context_data, context_data_istart, model, video_findfile
     global detect_parameters, doubleclick_parameters, model_parameters, cluster_parameters
 
@@ -276,14 +267,8 @@ def init(_bokeh_document, _configuration_file, _use_aitch):
 
     sys.path.insert(0,srcdir)
 
-    sys.path.insert(0,os.path.dirname(audio_read_plugin))
-    audio_read_module = importlib.import_module(os.path.basename(audio_read_plugin))
-    audio_read_module.audio_read_init(**audio_read_plugin_kwargs)
-    def audio_read(wav_path, start_tic=None, stop_tic=None):
-        return audio_read_module.audio_read(wav_path, start_tic, stop_tic, **audio_read_plugin_kwargs)
-    def audio_read_exts(): return audio_read_module.audio_read_exts(**audio_read_plugin_kwargs)
-    def audio_read_rec2ch(wavfile): return audio_read_module.audio_read_rec2ch(wavfile, **audio_read_plugin_kwargs)
-    def audio_read_strip_rec(recfile): return audio_read_module.audio_read_strip_rec(recfile, **audio_read_plugin_kwargs)
+    load_audio_read_plugin(audio_read_plugin, audio_read_plugin_kwargs)
+    from lib import audio_read, audio_read_exts, audio_read_rec2ch, audio_read_strip_rec, trim_ext
 
     sys.path.insert(0,os.path.dirname(video_read_plugin))
     video_read_module = importlib.import_module(os.path.basename(video_read_plugin))

src/lib.py

@@ -20,13 +20,37 @@
 from scipy.io import wavfile
 import csv
 from datetime import datetime
+import importlib
 
 import tifffile
 
 import tensorflow as tf
 import platform
 from subprocess import run, PIPE, STDOUT
 
+def load_audio_read_plugin(audio_read_plugin, audio_read_plugin_kwargs):
+    sys.path.append(os.path.dirname(audio_read_plugin))
+    audio_read_module = importlib.import_module(os.path.basename(audio_read_plugin))
+    global audio_read, audio_read_exts, audio_read_rec2ch, audio_read_strip_rec
+    def audio_read(wav_path, start_tic=None, stop_tic=None):
+        return audio_read_module.audio_read(wav_path, start_tic, stop_tic,
+                                            **audio_read_plugin_kwargs)
+    def audio_read_exts():
+         return audio_read_module.audio_read_exts(**audio_read_plugin_kwargs)
+    def audio_read_rec2ch(wavfile):
+         return audio_read_module.audio_read_rec2ch(wavfile, **audio_read_plugin_kwargs)
+    def audio_read_strip_rec(recfile):
+         return audio_read_module.audio_read_strip_rec(recfile, **audio_read_plugin_kwargs)
+
+def trim_ext(wavfile):
+    if len(audio_read_rec2ch(audio_read_strip_rec(wavfile))) > 1:
+        tmp = wavfile.split('-')
+        withext, rec = '-'.join(tmp[:-1]), tmp[-1]
+        withoutext = os.path.splitext(withext)[0]+'-'+rec
+    else:
+        withoutext = os.path.splitext(wavfile)[0]
+    return withoutext
+
 def check_config(configuration_file):
     exec(open(configuration_file).read())
 

src/time-freq-threshold.py

@@ -30,7 +30,6 @@
 import argparse
 import os
 import numpy as np
-import importlib
 import skimage
 from skimage.morphology import closing, opening
 import nitime.utils as utils
@@ -43,7 +42,7 @@
 import socket
 import json
 
-from lib import combine_events
+from lib import combine_events, load_audio_read_plugin
 
 def _frequency_n_callback(M,V,C):
     C.time.sleep(0.5)
@@ -96,11 +95,8 @@ def main():
     time_scale = FLAGS.time_scale
     time_units = FLAGS.time_units
 
-    sys.path.append(os.path.dirname(FLAGS.audio_read_plugin))
-    audio_read_module = importlib.import_module(os.path.basename(FLAGS.audio_read_plugin))
-    def audio_read(wav_path, start_tic=None, stop_tic=None):
-        return audio_read_module.audio_read(wav_path, start_tic, stop_tic,
-                                            **FLAGS.audio_read_plugin_kwargs)
+    load_audio_read_plugin(FLAGS.audio_read_plugin, FLAGS.audio_read_plugin_kwargs)
+    from lib import audio_read, trim_ext
 
     time_sigma_signal, time_sigma_noise = [int(x) for x in FLAGS.parameters['time_sigma'].split(',')]
 
@@ -229,7 +225,7 @@ def bool2stamp(song_morphed, scale):
 
 
     basename = os.path.basename(FLAGS.filename)
-    with open(os.path.splitext(FLAGS.filename)[0]+'-detected.csv', 'w') as fid:
+    with open(trim_ext(FLAGS.filename)+'-detected.csv', 'w') as fid:
       csvwriter = csv.writer(fid, lineterminator='\n')
       for i in intervals_time_signal:
         csvwriter.writerow([basename,i[1],i[2],'detected','time'+i[3]])