mergeTermListJson

Merge/MergeTermListJson/README.md

+# MergeTwoJsonLists
+
+## About The project
+
+MergeTwoJsonLists get two Json term list from Gargantext and merge them.
+
+## Usage
+```shell
+python3 mergeTwoJsonLists.py list1.json list2.json
+```
+
+list1.json / list2.json -> GarganText Term list
+
+Output a new list of term : list1-merged.json
+
+
+## Date
+
+This script have been last updated the 2023/07/21.
+It can be outdated if the futur.
\ No newline at end of file

Merge/MergeTermListJson/mergeTwoJsonLists.py

+#######
+# mergeListV4.py
+# description : merge two list v4
+# licence : AGPL + CECILL v3
+# author : quentin lobbé - qlobbe@iscpif.fr
+#######
+
+# python3 mergeListV4.py list1.json list2.json
+
+import sys
+import json
+
+try :
+    pathList1 = sys.argv[1]
+    pathList2 = sys.argv[2]
+except :
+    print ("! args error\n Try : python3 mergeListV4.py list1.json list2.json\n")
+    sys.exit(0)
+
+def readJson(path) :
+	file = open(path)
+	return json.load(file)
+
+listJson1 = readJson(pathList1)
+listJson2 = readJson(pathList2)
+
+ngrams1 = listJson1['NgramsTerms']['data']
+ngrams2 = listJson2['NgramsTerms']['data']
+merged  = {}
+
+roots = []
+leafs = []
+
+
+# find the roots of list 1
+for root in ngrams1.keys() :
+
+	if ngrams1[root]['list'] == "MapTerm" :
+
+		roots.append(root)
+		leafs = leafs + ngrams1[root]['children']
+
+# merge list 2 in list 1
+for root in ngrams2.keys() :
+
+	if root in roots :
+
+		ngrams1[root]['children'] = list(set(ngrams1[root]['children'] + ngrams2[root]['children']))
+
+	else :
+
+		if root not in leafs :
+
+			ngrams1[root] = ngrams2[root]
+			children = ngrams1[root]['children']
+			ngrams1[root]['children'] = []
+			for child in children :
+				if child not in root or child not in leafs :
+					ngrams1[root]['children'].append(child)
+
+# clean the merged list
+for root in ngrams1.keys() :
+
+	if ngrams1[root]['list'] == "MapTerm" :
+
+		if len(ngrams1[root]['children']) == 0 :
+
+			ngrams1[root]['children'] = []
+
+		merged[root] = ngrams1[root]
+
+
+listJson1['NgramsTerms']['data'] = merged
+listJson1['Authors']['data']   = {}
+listJson1['Institutes']['data'] = {}
+listJson1['Sources']['data']   = {}
+
+name = ((pathList1.split('/')).pop()).split('.')
+
+newName = name[0] + '-merged.' + name[1]
+
+if len(pathList1.split('/')) == 1:
+	newPath = newName
+else:
+	newPath = '/'.join((pathList1.split('/'))[:-1]) + '/' + newName
+
+with open(newPath, 'w') as file :
+   
+   json.dump(listJson1, file, sort_keys=False, indent=4)	
+