feat(mod): Add data and visual extractor

.gitignore

@@ -17,3 +17,4 @@ itten.egg-info/
 tests/data
 .code-workspace.code-workspace
 scratchpad.py
+cittena.gsp

itten/corpus.py

+import sys
 from pathlib import Path
-from subprocess import Popen, PIPE, CalledProcessError, run
+from subprocess import PIPE, run
 from cv2 import imread, imwrite
+import pandas as pd
 import numpy as np
 from .movie import Movie
+from .views import UnivariateSequence, MultivariateSequence
+from .visuals import UnivariatePlot, MultivariatePlot
 
 _summary = {'seqmean': {'monochromatic': ['luminance'],
                         'saturation': ['saturation', 'colourfulness'],
@@ -43,6 +47,7 @@ class Corpus(object):
                       not('_' in str(d.name)[0])]
 
         for m in movie_dirs:
+            self.status = {}
             movie = m.name
             status[movie] = {}
 
@@ -101,6 +106,30 @@ class Corpus(object):
         """
         tasks = self._extraction_tasks()
 
+        for movie, missing in tasks.items():
+            # Extract frames
+            if 'frames' in missing:
+                self._extract_frames(movie)
+            # Extract data
+            try:
+                for strategy, contrasts in missing['data'].items():
+                    for ctrst, methods in contrasts.items():
+                        for meth in methods:
+                            print('{:15s} (data, {:18s}): {} missing with methods: {}'.format(movie, strategy, ctrst, meth))
+                            self._extract_data(movie, strategy, ctrst, meth)
+            except KeyError:
+                print('{0} (data): Nothing to do'.format(movie))
+            
+            # Extract visuals
+            try:
+                for strategy, contrasts in missing['visuals'].items():
+                    for ctrst, methods in contrasts.items():
+                        for meth in methods:
+                            print('{:15s} (visuals, {:15s}): {} missing with methods: {}'.format(movie, strategy, ctrst, meth))
+                            self._extract_visuals(movie, strategy, ctrst, meth)
+            except KeyError:
+                print('{0} (visuals): Nothing to do'.format(movie))
+
         # TODO Extractors implementieren
 
         return tasks
@@ -138,7 +167,8 @@ class Corpus(object):
                             if contrast not in ctrst_keys:
                                 missing.setdefault(m, {component: {}})
                                 missing[m][component].setdefault(view, {})
-                                missing[m][component][view][contrast] = (_summary[view][contrast])
+                                missing[m][component][view][contrast] = \
+                                    (_summary[view][contrast])
                             else:
                                 # For each method in the crrent view and
                                 # contrast look out if a key for that method
@@ -210,11 +240,53 @@ class Corpus(object):
         else:
             print('Finished!')
 
-    def _extract_data(self):
-        pass
-
-    def _extract_visuals(self):
-        pass
+    def _extract_data(self, movie, strategy, contrast, method):
+        try:
+            m = Movie(movie + '_', str(self.basedir / movie / 'frames' / '240p30') + '/')
+            if strategy == 'seqmean':
+                view = UnivariateSequence(m._frames)
+                view.seqmean(ctrst=contrast, method=method, frm_stp=4,)
+            elif strategy == 'distribution':
+                view = MultivariateSequence(m._frames)
+                view.populate(ctrst=contrast, method=method, frm_stp=4,)
+            title = movie + '_' + strategy + '_' + contrast + '_' + method + '_4fps'
+            Path(self.basedir / movie / 'data').mkdir(parents=True, exist_ok=True)
+            data = pd.DataFrame(view[:])
+            data.to_pickle(str(self.basedir / 'data' / (title + '.pkl')))
+        except:
+            e = sys.exc_info()[0]
+            print('{} (data, {}): {} ({}) raised an error:'.format(movie, strategy, contrast, method))
+            print('Error: {}'.format(e))
+
+    def _extract_visuals(self, movie, strategy, contrast, method):
+        m = Movie(movie + '_', str(self.basedir / movie / 'frames' / '240p30') + '/')
+        data = pd.read_pickle(str(self.basedir / movie / 'data' / (movie + '_' + strategy + '_' + contrast + '_' + method + '_4fps.pkl')))
+        try:
+            if strategy == 'seqmean':
+                view = UnivariateSequence(m._frames, input_array=data.to_numpy().flatten())
+                view._contrast = contrast
+                view._method = method
+                view.feature = 'mean'
+                viz = UnivariatePlot(view,)
+            elif strategy == 'distribution':
+                view = MultivariateSequence(m._frames, input_array=data.to_numpy())
+                view._bins = 16
+                view._threshold = 6000
+                view._contrast = contrast
+                view._method = method
+                viz = MultivariatePlot(view)
+            viz.plot(view)
+            title = movie + '_' +\
+                strategy + '_' +\
+                contrast + '_' +\
+                method +\
+                '_4fps'
+            file_name = title + '.png'
+            viz.saveplt(title=title, fname=self.basedir / movie / file_name)
+        except:
+            e = sys.exc_info()[0]
+            print('{} (visual, {}): {} ({}) raised an error:'.format(movie, strategy, contrast, method))
+            print('Error: {}'.format(e))
 
     def tableau(self, mode='contrast',
                 select=['distribution', 'monochromatic', 'luminance'],

itten/views.py

@@ -62,7 +62,7 @@ class View(np.ndarray):
             obj = np.asarray(input_array).view(cls).copy()
         obj._frames = frames
         obj._contrast = 2
-        obj._frame_step = 10
+        obj._frame_step = 4
         obj._bins = 256
         return obj
 
@@ -150,7 +150,7 @@ class MultivariateSequence(View):
         View.__array_finalize__(self, obj)
         self._threshold = getattr(obj, '_threshold', None)
 
-    def populate(self, ctrst='monochromatic', method='luminance', frm_stp=10,
+    def populate(self, ctrst='monochromatic', method='luminance', frm_stp=4,
                  bins=16, thrsh=6000, start=1, end=0):
         """doc (aus __new__ zusammentragen)
         """
@@ -254,7 +254,7 @@ class UnivariateSequence(View):
     # TODO: Kurve muß unbeding von cuts bereinigt werde und interpolation
     # funktioniert da nicht
     def seqmean(self, ctrst='monochromatic',
-                method='luminance', frm_stp=10, bins=256):
+                method='luminance', frm_stp=4, bins=256):
         """Creates a scatterplot for the dynamic of contrast across movie frames
 
            frm_fld: path to folder with movie images

itten/visuals.py

@@ -141,7 +141,7 @@ class MultivariatePlot(SequencePlot):
         self._ax, axt = self.ittenstyle(self._ax, view)
 
         # Plotten
-        axt.scatter(x, y, c=(0, 0, 0), s=thickness, linewidths=0)
+        axt.scatter(x, y, c='black', s=thickness, linewidths=0)
         # axt.scatter(x, y, c=y, cmap='Greys_r', s=thickness, linewidths=0)
         # # DOC: vmin/vmax sorgt für die Verteilung der Fraben der Colorm
         # ax.scatter(x, y, c=y, cmap='hsv', s=thickness, linewidths=0)