From 64e9d682150e2f12e52b2193759f2e4c54b98460 Mon Sep 17 00:00:00 2001
From: Marc Wrobel <marc.wrobel@gmail.com>
Date: Thu, 15 Aug 2024 14:22:32 +0200
Subject: [PATCH] [lua] Automate release and version retrieval

---
 src/lua.py | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)
 create mode 100644 src/lua.py

diff --git a/src/lua.py b/src/lua.py
new file mode 100644
index 000000000..3c87efdeb
--- /dev/null
+++ b/src/lua.py
@@ -0,0 +1,24 @@
+import re
+
+from bs4 import BeautifulSoup
+from common import dates, http, releasedata
+
+"""Fetches Lua releases from lua.org."""
+
+RELEASED_AT_PATTERN = re.compile(r"Lua\s*(?P<release>\d+\.\d+)\s*was\s*released\s*on\s*(?P<release_date>\d+\s*\w+\s*\d{4})")
+VERSION_PATTERN = re.compile(r"(?P<version>\d+\.\d+\.\d+),\s*released\s*on\s*(?P<version_date>\d+\s*\w+\s*\d{4})")
+
+with releasedata.ProductData("lua") as product_data:
+    page = http.fetch_url("https://www.lua.org/versions.html")
+    soup = BeautifulSoup(page.text, 'html.parser')
+    page_text = soup.text # HTML is broken, no way to parse it with beautifulsoup
+
+    for release_match in RELEASED_AT_PATTERN.finditer(page_text):
+        release = release_match.group('release')
+        release_date = dates.parse_date(release_match.group('release_date'))
+        product_data.get_release(release).set_release_date(release_date)
+
+    for version_match in VERSION_PATTERN.finditer(page_text):
+        version = version_match.group('version')
+        version_date = dates.parse_date(version_match.group('version_date'))
+        product_data.declare_version(version, version_date)