[red-hat-jboss-eap] Automation version retrieval (#414)

red-hat-jboss-eap-7.py retrieves all the 7.x versions based on https://access.redhat.com/articles/2332721.

red-hat-jboss-eap-8.py retrieves only the last 8.x version based on https://maven.repository.redhat.com/ga/org/jboss/eap/channels/eap-8.0/maven-metadata.xml (the version date is approximated using the lastUpdated tag).

---------

Co-authored-by: Denis Fuenzalida 🐙 <defuenza@microsoft.com>
Co-authored-by: Marc Wrobel <marc.wrobel@gmail.com>

src/common/dates.py

@@ -54,6 +54,7 @@ def parse_datetime(text: str, formats: list[str] = frozenset([
     "%Y-%m-%dT%H:%M:%S.%f%z",    # 2023-05-01T08:32:34.123456Z
     "%Y/%m/%d %H:%M:%S",         # 2023/05/01 08:32:34
     "%a %d %b %Y %H:%M:%S %Z",   # Wed, 01 Jan 2020 00:00:00 GMT
+    "%Y%m%d%H%M%S",              # 20230501083234
 ]), to_utc: bool = True) -> datetime:
     """Parse a given text representing a datetime using a list of formats,
     optionally converting it to UTC.

src/red-hat-jboss-eap-7.py

@@ -0,0 +1,37 @@
+import logging
+
+from bs4 import BeautifulSoup
+from common import dates, http, releasedata
+
+"""Fetches RedHat JBoss EAP version data for JBoss 7"""
+
+with releasedata.ProductData("red-hat-jboss-eap") as product_data:
+    response = http.fetch_url("https://access.redhat.com/articles/2332721")
+    soup = BeautifulSoup(response.text, features="html5lib")
+
+    for h4 in soup.find_all("h4"):
+        title = h4.get_text(strip=True)
+        if not title.startswith("7."):
+            continue
+
+        release = title[:3]
+        version_table = h4.find_next("table")
+        if not version_table:
+            logging.warning(f"Version table not found for {title}")
+            continue
+
+        for (i, row) in enumerate(version_table.find_all("tr")):
+            if i == 0:  # Skip the first row (header)
+                continue
+
+            columns = row.find_all("td")
+            # Get the version name without the content of the <sup> tag, if present
+            name_str = ''.join([content for content in columns[0].contents if isinstance(content, str)]).strip()
+            date_str = columns[1].text.strip()
+
+            if date_str == "TBD": # Placeholder for a future release
+                continue
+
+            name = name_str.replace("GA", "Update 0").replace("Update ", release + ".")
+            date = dates.parse_date(date_str)
+            product_data.declare_version(name, date)

src/red-hat-jboss-eap-8.py

@@ -0,0 +1,20 @@
+import re
+from xml.dom.minidom import parseString
+
+from common import dates, http, releasedata
+
+"""Fetches the latest RedHat JBoss EAP version data for JBoss 8.0"""
+
+with releasedata.ProductData("red-hat-jboss-eap") as product_data:
+    response = http.fetch_url("https://maven.repository.redhat.com/ga/org/jboss/eap/channels/eap-8.0/maven-metadata.xml")
+
+    xml = parseString(response.text)
+    versioning = xml.getElementsByTagName("metadata")[0].getElementsByTagName("versioning")[0]
+
+    latest_str = versioning.getElementsByTagName("latest")[0].firstChild.nodeValue
+    latest_name = "8.0." + re.match(r"^..(.*)\.GA", latest_str).group(1)
+
+    latest_date_str = versioning.getElementsByTagName("lastUpdated")[0].firstChild.nodeValue
+    latest_date = dates.parse_datetime(latest_date_str)
+
+    product_data.declare_version(latest_name, latest_date)