From ba9c57606728d59bbeee11625a3d39e793a59153 Mon Sep 17 00:00:00 2001
From: Chris Woodward <cw00dw0rd@gmail.com>
Date: Fri, 19 Jun 2020 20:10:44 -0400
Subject: [PATCH] Updated savers object creation to bulk save

---
 pyGeno/backends/arangodb/configuration.py |  2 +-
 pyGeno/backends/arangodb/savers.py        | 26 ++++++++++++++---------
 2 files changed, 17 insertions(+), 11 deletions(-)

diff --git a/pyGeno/backends/arangodb/configuration.py b/pyGeno/backends/arangodb/configuration.py
index 6702253..4a2b66e 100644
--- a/pyGeno/backends/arangodb/configuration.py
+++ b/pyGeno/backends/arangodb/configuration.py
@@ -55,7 +55,7 @@ def prompt_setup(self):
         args = {
             "username": "root",
             "password": "root",
-            "arangoURL": "http://localhost:8529/"
+            "arangoURL": "http://127.0.0.1:8529/"
         }
         print("## ArangoDB backend setup")
         for key, value in args.items():
diff --git a/pyGeno/backends/arangodb/savers.py b/pyGeno/backends/arangodb/savers.py
index f10bf72..0b022fc 100644
--- a/pyGeno/backends/arangodb/savers.py
+++ b/pyGeno/backends/arangodb/savers.py
@@ -1,7 +1,7 @@
 from ..backend_abs import GenomeSaver_ABS
 from pyGeno.configuration import system_message
 from .objects import schemas
-
+import pyArango
 class GenomeSaver(GenomeSaver_ABS):
     """
     Saves genome into database
@@ -14,7 +14,7 @@ def __init__(self, database_configuration):
     def init_db(self):
         for col in schemas.ALL_COLLECTIONS:
             colname = col.__name__
-            if colname not in self.db:
+            if not self.db.hasCollection(str(colname)):
                 self.db.createCollection(colname)
             else :
                 print("TRUNCATING (temporary for tests, should be removed)", colname)
@@ -22,16 +22,22 @@ def init_db(self):
     
     def create_objects(self):
         for colname, objs in self.data.items():
+            i = 0
+            c = 0
+            docs = []
+            doc = {}
             for key, obj in objs.items():
-                try :
-                    doc = self.db[colname][key]
-                except :
-                    doc = self.db[colname].createDocument()
-                    doc["_key"] = key
-                doc.set(obj)
+                doc = self.db[colname].createDocument_(obj)
+                doc['_key'] = key
                 if colname in self.accepted_contigs:
                     doc["contig"] = self.get_subsequence(doc["seqname"], doc["start"], doc["end"])
-                doc.save()
+                docs.append(doc)
+                i = i + 1
+                c = c + 1
+                if i == len(objs) or i == 1000 or c + i == len(objs):
+                    self.db[colname].bulkSave(docs, onDuplicate='ignore')
+                    docs = []
+                    i = 0
 
     def create_links(self):
         def _get_collection(from_type, to_type):
@@ -56,4 +62,4 @@ def _get_collection(from_type, to_type):
     def save(self) :
         self.init_db()
         self.create_objects()
-        self.create_links()
+        #self.create_links()