Improve search results

add new dependency: unidecode

partitioncloud/modules/classes/user.py

@@ -182,7 +182,7 @@ class User():
             else:
                 self.accessible_partitions = db.execute(
                     """
-                    SELECT partition.uuid, partition.name,
+                    SELECT DISTINCT partition.uuid, partition.name,
                         partition.author, partition.body,
                         partition.user_id, partition.source
                     FROM partition
@@ -196,7 +196,7 @@ class User():
                         ON contient_user.user_id=?
                         AND album_id=album.id
                     UNION
-                        SELECT album.id FROM album
+                        SELECT DISTINCT album.id FROM album
                         JOIN groupe_contient_user
                         JOIN groupe_contient_album
                         ON groupe_contient_user.user_id=?

partitioncloud/modules/search.py

@@ -10,6 +10,7 @@ import os
 
 import pypdf
 import googlesearch
+from unidecode import unidecode
 
 from .db import get_db
 
@@ -20,20 +21,20 @@ def local_search(query, partitions):
     """
     Renvoie les 5 résultats les plus pertinents parmi une liste donnée
     """
-    query_words = [word.lower() for word in query.split(" ")]
+    query_words = [word.lower() for word in unidecode(query).split()]
     def score_attribution(partition):
         score = 0
         for word in query_words:
             if word != "":
-                if word in partition["name"].lower():
+                if word in unidecode(partition["name"]).lower():
                     score += 6
-                elif word in partition["author"].lower():
+                elif word in unidecode(partition["author"]).lower():
                     score += 4
-                elif word in partition["body"].lower():
+                elif word in unidecode(partition["body"]).lower():
                     score += 2
                 else:
-                    score -= 1
+                    score -= 6
-        for word in partition["name"].split(" "):
+        for word in unidecode(partition["name"]).split():
             if word != "" and word.lower() not in query_words:
                 score -= 1
         return score

requirements.txt

@@ -3,4 +3,5 @@ flask-babel
 google
 colorama
 pypdf
-qrcode
+qrcode
+unidecode

scripts/hooks/v1.py

@@ -187,4 +187,11 @@ def install_babel():
 """
 
 def install_pypdf():
-    utils.install_package("pypdf")
+    utils.install_package("pypdf")
+
+"""
+ v1.10.*
+"""
+
+def install_unidecode():
+    utils.install_package("unidecode")

scripts/migration.py

@@ -37,6 +37,7 @@ hooks = [
     ("v1.5.1", [("Move thumbnails", v1_hooks.move_thumbnails)]),
     ("v1.7.0", [("Install babel", v1_hooks.install_babel)]),
     ("v1.8.2", [("Install pypdf", v1_hooks.install_pypdf)])
+    ("v1.10.3", [("Install unidecode", v1_hooks.install_unidecode)])
 ]