Merge pull request #153 from ICAMS/process_fixes

Process fixes

calphy/postprocessing.py

@@ -25,6 +25,19 @@ def read_report(folder):
         data = yaml.safe_load(fin)
     return data
 
+def _extract_error(errfile):
+    error_code = None
+    if os.path.exists(errfile):
+        with open(errfile, 'r') as fin:
+            for line in fin:
+                if 'calphy.errors' in line:
+                    break
+            try:
+                error_code = line.split(':')[0].split('.')[-1]
+            except:
+                pass
+    return error_code
+
 def gather_results(mainfolder):
     """
     Gather results from all subfolders in a given folder into a Pandas DataFrame
@@ -52,6 +65,8 @@ def gather_results(mainfolder):
     datadict['free_energy'] = []
     datadict['reference_phase'] = []
     datadict['error_code'] = []
+    datadict['composition'] = []
+    datadict['calculation'] = []
 
     folders = next(os.walk(mainfolder))[1]
     for folder in folders:
@@ -77,6 +92,8 @@ def gather_results(mainfolder):
         datadict['temperature'].append(inp['temperature'])
         datadict['pressure'].append(inp['pressure'])
         datadict['reference_phase'].append(inp['reference_phase'])
+        datadict['composition'].append(None)
+        datadict['calculation'].append(folder)
 
         #check output file
         outfile = os.path.join(mainfolder, folder, 'report.yaml')
@@ -88,16 +105,7 @@ def gather_results(mainfolder):
             datadict['free_energy'].append(np.NaN)
             #check if error file is found
             errfile = os.path.join(os.getcwd(), mainfolder, folder+'.sub.err')
-            if os.path.exists(errfile):
-                with open(errfile, 'r') as fin:
-                    for line in fin:
-                        if 'calphy.errors' in line:
-                            break
-                    try:
-                        error_code = line.split(':')[0].split('.')[-1]
-                        datadict['error_code'][-1] = error_code
-                    except:
-                        pass
+            datadict['error_code'][-1] = _extract_error(errfile)
             continue;
 
         if mode in ['fe', 'alchemy', 'composition_scaling']:
@@ -108,13 +116,33 @@ def gather_results(mainfolder):
             out = yaml.safe_load(fin)
 
         datadict['free_energy'].append(out['results']['free_energy'])
-
+
+        #add normal composition
+        el_arr = np.array(out['input']['element'].split(' ')).astype(str)
+        comp_arr = np.array(out['input']['concentration'].split(' ')).astype(float)
+        composition = {x:y for x,y in zip(el_arr, comp_arr)}
+        datadict['composition'][-1] = composition
+
+        if mode == 'composition_scaling':
+            #we need to update composition
+            compdict = inp['composition_scaling']['output_chemical_composition']
+            maxatoms = np.sum([val for key, val in compdict.items()])
+            for key, val in compdict.items():
+                compdict[key] = val/maxatoms
+            datadict['composition'][-1] = compdict
+
         #parse extra info
         if mode in ['ts', 'tscale']:
-            datafile = os.path.join(os.getcwd(), folder, 'temperature_sweep.dat')
-            t, f = np.loadtxt(datafile, unpack=True, usecols=(0,1))
-            datadict['temperature'][-1] = t
-            datadict['free_energy'][-1] = f
-
+            datafile = os.path.join(os.getcwd(), mainfolder, folder, 'temperature_sweep.dat')
+            if os.path.exists(datafile):
+                datadict['status'].append('True')
+                t, f = np.loadtxt(datafile, unpack=True, usecols=(0,1))
+                datadict['temperature'][-1] = t
+                datadict['free_energy'][-1] = f
+            else:
+                datadict['status'].append('False')
+                errfile = os.path.join(os.getcwd(), mainfolder, folder+'.sub.err')
+                datadict['error_code'][-1] = _extract_error(errfile)
+
     df = pd.DataFrame(data=datadict)
     return df