Add a CLI for optimisation

solarcalc/cli.py

@@ -0,0 +1,36 @@
+import argparse as ap
+
+from solarcalc.spectopt import Light, Spectrum, BandCost, SimpleCost
+
+def parse_theory(tfile):
+    with open(tfile) as fh:
+        header = list(map(lambda x: x.strip(), fh.readline().split(",")))
+        domain = [int(x.rstrip("nm")) for x in header[5:]]
+        for line in fh:
+            cells = list(map(lambda x: x.strip(), line.split(",")))
+            date, mdate, temp, hum, tot, *wls = cells
+            wls = list(map(float, wls))
+            yield (date, mdate, temp, hum, tot, Spectrum(domain, wls))
+
+
+def optimise_for_light():
+    p = ap.ArgumentParser()
+    p.add_argument("-l", "--light", type=str, required=True,
+                   help="Light definition file")
+    p.add_argument("theoretical", type=str,
+                   help="Theoretical model output as a CSV")
+    args = p.parse_args()
+
+    light = Light(args.light, interpolation='linear')
+
+    cf = BandCost(bands=[(400, 500), (500, 600), (600, 700)], weights=[1, 0.1, 1])
+    light = Light("./data/spectra/growtainer-actual.csv", interpolation='linear')
+    print("datetime", "model_datetime", "temp", "humidity", "total_solar", *light.channels, sep="\t")
+    for line in parse_theory("./data/output.csv"):
+        want = line[5][min(light.wavelengths):max(light.wavelengths)]
+        opt = light.optimise_settings(want, cost_function=cf)
+        print(*line[0:5], *opt, sep="\t")
+
+
+if __name__ == "__main__":
+    optimise_for_light()

solarcalc/spectopt.py

@@ -15,8 +15,8 @@ class Spectrum(object):
     """Holds a spectral density curve, also integrate and interpolate"""
 
     def __init__(self, wavelengths, values):
-        self.wavelengths = wavelengths
-        self.values = values
+        self.wavelengths = np.array(wavelengths).astype(float)
+        self.values = np.array(values).astype(float)
 
     def interpolated(self, method='linear'):
         '''Create a function to interpolate spectrum to new wavelengths'''
@@ -39,13 +39,18 @@ def banded_integral(self, bands):
             integr.append(s.integral(*band))
         return integr
 
+    def __getitem__(self, item):
+        if isinstance(item, slice):
+            keep = np.logical_and(self.wavelengths >= item.start, self.wavelengths <= item.stop)
+            return Spectrum(self.wavelengths[keep], self.values[keep])
+
+
 class CostFunc(object):
     '''Base class of cost functions, designed to be used with Light.optimise_settings'''
 
     def __call__(self, weights, light, desired):
         got = light.light_output(weights, wavelengths=desired.wavelengths)
         cost = self._cost(got, desired)
-        print(type(self).__name__, weights, cost)
         return cost
 
 class SimpleCost(CostFunc):