fix order of source issue in find_duplicate

Author: swairshah

application.py

@@ -83,11 +83,10 @@ def find_duplicate():
             doc_ids, dists = imagesearch.search(vec, thresh)
         else:
             doc_ids, dists = imagesearch.search(vec)
-
-        sources = [d.get('source') for d in db.docs.find({"doc_id" : {"$in" : doc_ids}})]
+        sources = {d.get('doc_id') : d.get('source') for d in db.docs.find({"doc_id" : {"$in" : doc_ids}})}
 
         if doc_ids is not None:
-            result = [{'doc_id' : doc_ids[i], 'dist' : dists[i], 'source' : sources[i]} for i in range(len(doc_ids))]
+            result = [{'doc_id' : doc_ids[i], 'dist' : dists[i], 'source' : sources[doc_ids[i]]} for i in range(len(doc_ids))]
             ret = {'failed' : 0, 'result' : result}
         else:
             ret = {'failed' : 0, 'result' : []}
@@ -96,10 +95,10 @@ def find_duplicate():
         duplicate_doc = db.docs.find_one({"text" : text})
         vec = doc2vec(text)
         doc_ids, dists = textsearch.search(vec)
-        sources = [d.get('source') for d in db.docs.find({"doc_id" : {"$in" : doc_ids}})]
+        sources = {d.get('doc_id') : d.get('source') for d in db.docs.find({"doc_id" : {"$in" : doc_ids}})}
 
         if doc_ids is not None:
-            result = [{'doc_id' : doc_ids[i], 'dist' : dists[i], 'source' : sources[i]} for i in range(len(doc_ids))]
+            result = [{'doc_id' : doc_ids[i], 'dist' : dists[i], 'source' : sources[doc_ids[i]]} for i in range(len(doc_ids))]
         else:
             result = []