From 171fa645d2f8dc00eec452859357e514eb9fb018 Mon Sep 17 00:00:00 2001
From: Daoud Clarke <daoud.clarke@gmail.com>
Date: Sun, 23 Jan 2022 22:04:30 +0000
Subject: [PATCH] Add script to export top domains

---
 analyse/export_top_domains.py | 13 +++++++++++++
 mwmbl/indexer/paths.py        |  2 ++
 2 files changed, 15 insertions(+)
 create mode 100644 analyse/export_top_domains.py

diff --git a/analyse/export_top_domains.py b/analyse/export_top_domains.py
new file mode 100644
index 0000000..2794804
--- /dev/null
+++ b/analyse/export_top_domains.py
@@ -0,0 +1,13 @@
+import json
+
+from mwmbl.indexer.paths import TOP_DOMAINS_JSON_PATH
+from mwmbl.tinysearchengine.hn_top_domains_filtered import DOMAINS
+
+
+def export_top_domains_to_json():
+    with open(TOP_DOMAINS_JSON_PATH, 'w') as output_file:
+        json.dump(DOMAINS, output_file, indent=2)
+
+
+if __name__ == '__main__':
+    export_top_domains_to_json()
diff --git a/mwmbl/indexer/paths.py b/mwmbl/indexer/paths.py
index f9cf1e0..41e4672 100644
--- a/mwmbl/indexer/paths.py
+++ b/mwmbl/indexer/paths.py
@@ -20,3 +20,5 @@ DOMAINS_TITLES_QUEUE_NAME = 'domains-title-queue-fs'
 DOMAINS_PATH = os.path.join(DATA_DIR, 'top10milliondomains.csv.gz')
 
 INDEX_PATH = Path(__file__).parent / 'data' / 'index.tinysearch'
+
+TOP_DOMAINS_JSON_PATH = DATA_DIR / 'hn-top-domains.json'