Add CORPUS data and temporary python script

tmp/CORPUStidy.py

@@ -0,0 +1,27 @@
+#! /usr/bin/python3
+
+import json
+
+print("Opening CORPUSExtract.json")
+with open("./CORPUSExtract.json", "r") as input_file:
+    dict = json.load(input_file)
+    list = dict['TIPLOCDATA']
+
+cleanList = []
+
+print("Processing data")
+for item in list:
+    if item["3ALPHA"] != ' ' and item["STANOX"] != ' ':
+        item.pop("UIC")
+        item.pop("NLCDESC16")
+        item.pop("NLC")
+        cleanList.append(item)
+
+print("Saving data")
+
+cleanDict = {"data":cleanList}
+with open("CorpusClean.json", "w") as output_file:
+    output_file.write(json.dumps(cleanDict))
+
+print(cleanList)
+print("Processed.")