From 64c56514b70aa510df485d7373ad47ba36b012d4 Mon Sep 17 00:00:00 2001
From: Sam <samuel.molina@enreda.coop>
Date: Wed, 17 Feb 2021 13:07:07 +0000
Subject: [PATCH] added command to extract top level tags

---
 core/management/commands/extractparenttags.py | 31 +++++++++++++++++++
 datasets/top_tags.txt                         | 22 +++++++++++++
 2 files changed, 53 insertions(+)
 create mode 100644 core/management/commands/extractparenttags.py
 create mode 100644 datasets/top_tags.txt

diff --git a/core/management/commands/extractparenttags.py b/core/management/commands/extractparenttags.py
new file mode 100644
index 0000000..7e14657
--- /dev/null
+++ b/core/management/commands/extractparenttags.py
@@ -0,0 +1,31 @@
+import logging
+
+from django.core.management.base import BaseCommand
+from django.conf import settings
+
+from core.models import TreeTag
+
+
+class Command(BaseCommand):
+
+    help = 'Extract top level tags'
+
+    def handle(self, *args, **kwargs):
+        # get all instances
+        tags = TreeTag.objects.all()
+        top_tags = []
+
+        print("Extracting top-level tags from TreeTag instances")
+        # extract tags with no ancestor
+        for tag in tags:
+            if not tag.get_ancestors():
+                top_tags.append(tag.name)
+
+        print("Saving top-level tags to file")
+        # save results to dataset/top_tags.txt
+        path = f"{settings.BASE_DIR}/../datasets/top_tags.txt"
+        with open(path, 'wt') as f:
+            f.writelines([tag + '\n' for tag in top_tags])
+
+        # print out results
+        logging.info(f"Extracted {len(top_tags)} to {path}")
diff --git a/datasets/top_tags.txt b/datasets/top_tags.txt
new file mode 100644
index 0000000..294d7cb
--- /dev/null
+++ b/datasets/top_tags.txt
@@ -0,0 +1,22 @@
+Alimentación, bebida y tabaco
+Arte y ocio
+Bebés y niños pequeños
+Bricolaje
+Cámaras y ópticas
+Casa y jardín
+Economía e industria
+Electrónica
+Elementos religiosos y ceremoniales
+Equipamiento deportivo
+# Google_Product_Taxonomy_Version: 2015-02-19
+Juegos y juguetes
+Maletas y bolsos de viaje
+Material de oficina
+Mobiliario
+Multimedia
+Productos para adultos
+Productos para mascotas y animales
+Ropa y accesorios
+Salud y belleza
+Software
+Vehículos y recambios