From 520d0be5954ccdc45e4c43001d7becfe9370ede0 Mon Sep 17 00:00:00 2001
From: "Maxence G. de Montauzan" <maxence@gdemontauzan.fr>
Date: Sun, 1 Aug 2021 03:42:46 +0200
Subject: [PATCH] (back) Suggester V3: Process album in a separate way

(cherry picked from commit ebbeeccfb8535dbb67240d2c68c7dc9a4da7e7f8)
---
 suggester.es | 20 +++++++++++---------
 suggester.py | 47 ++++++++++++++++++++++++++++++++---------------
 2 files changed, 43 insertions(+), 24 deletions(-)

diff --git a/suggester.es b/suggester.es
index 53ac473..bcdeca3 100644
--- a/suggester.es
+++ b/suggester.es
@@ -55,22 +55,24 @@ GET itunes-suggest/_analyze
 
 POST itunes-suggest/_search
 {
-  "suggest": {
-    "name-suggest": {
-      "prefix": "sou",
-      "completion": {
-          "field": "artist_suggest"
-      }
+    "_source" : "artist",
+    "suggest": {
+        "name-suggest": {
+            "prefix": "sou",
+            "completion": {
+                "field": "artist_suggest"
+            }
+        }
     }
-  }
 }
 
 POST itunes-suggest/_search
 {
+    "_source" : "album",
     "suggest": {
         "name-suggest": {
-      "prefix": "trip",
-      "completion": {
+            "prefix": "new",
+            "completion": {
                 "field": "album_suggest",
                 "size": 20
             }
diff --git a/suggester.py b/suggester.py
index feff4c2..3cd76e4 100644
--- a/suggester.py
+++ b/suggester.py
@@ -26,7 +26,7 @@ def get_tokens(data: str) -> list:
         raise NoGoodDataException('Data is not correct to get tokens')
     return [t['token'] for t in r.json()['tokens']]
 
-def post_document(artist: str, artist_sugget: list, album: str, album_suggest: list) -> bool:
+def post_document(artist: str = None, artist_sugget: list = None, album: str = None, album_suggest: list = None) -> bool:
     element = {
         "artist_suggest" : artist_sugget,
         "artist": artist,
@@ -48,19 +48,36 @@ def post_document(artist: str, artist_sugget: list, album: str, album_suggest: l
     # print('Post_element - Element created: ' + el_id)
     return el_id
 
+if __name__ == '__main__':
+    # Using readlines()
+    with open('/home/budd/workspace/iTunes/es-artists.json', 'r') as artist_file:
+        artists_lines = artist_file.readlines()
 
-# Using readlines()
-itunes_file = open('/home/budd/workspace/iTunes/es-artists.json', 'r')
-lines = itunes_file.readlines()
+    with open('/home/budd/workspace/iTunes/es-albums.json', 'r') as artist_file:
+        albums_lines = artist_file.readlines()
 
-# Strips the newline character
-for line in lines:
-    data = json.loads(line)
-    if "Artist" in data:
-        try :
-            artist_input = get_tokens(data['Artist'])
-            album_input = get_tokens(data['Album'])
-            post_document(data['Artist'], artist_input, data['Album'], album_input)
-        except NoGoodDataException:
-            print('ERROR WITH DATA')
-            print(str(data))
+    # Strips the newline character
+    count = 0
+    for line in artists_lines:
+        data = json.loads(line)
+        if "Artist" in data:
+            try :
+                artist_input = get_tokens(data['Artist'])
+                post_document(artist=data['Artist'], artist_sugget=artist_input)
+                count += 1
+            except NoGoodDataException:
+                print('ERROR WITH DATA')
+                print(str(data))
+
+    for line in albums_lines:
+        data = json.loads(line)
+        if "Artist" in data:
+            try :
+                album_input = get_tokens(data['Album'])
+                post_document(album=data['Album'], album_suggest=album_input)
+                count += 1
+            except NoGoodDataException:
+                print('ERROR WITH DATA')
+                print(str(data))
+
+    print('Created documents: ' + str(count))