Use a default regex when possible (#205)

src/cgit.py

@@ -13,14 +13,6 @@ do not support partial clone so we cannot.
 """
 
 METHOD = 'cgit'
-# Same as used in Ruby (update.rb)
-DEFAULT_TAG_TEMPLATE = (
-    "{{major}}{% if minor %}.{{minor}}{% if patch %}.{{patch}}{%if tiny %}.{{tiny}}{%endif%}{%endif%}{%endif%}"
-)
-DEFAULT_VERSION_REGEX = (
-    r"^v?(?P<major>\d+)\.(?P<minor>\d+)\.?(?P<patch>\d+)?\.?(?P<tiny>\d+)?$"
-)
-
 
 def make_bs_request(url):
     response = endoflife.fetch_url(url + '/refs/tags')
@@ -56,8 +48,8 @@ def update_product(product_name, configs):
     versions = {}
 
     for config in configs:
-        t = config.get("template", DEFAULT_TAG_TEMPLATE)
+        t = config.get("template", endoflife.DEFAULT_TAG_TEMPLATE)
-        regex = config.get("regex", DEFAULT_VERSION_REGEX)
+        regex = config.get("regex", endoflife.DEFAULT_VERSION_REGEX)
         versions = versions | fetch_releases(config[METHOD], regex, t)
 
     endoflife.write_releases(product_name, versions)

src/common/endoflife.py

@@ -12,6 +12,11 @@ from urllib3.util import Retry
 # See https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/User-Agent.
 USER_AGENT = 'Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101 Firefox/102.0'
 
+# Handle versions having at least 2 digits (ex. 1.2) and at most 4 digits (ex. 1.2.3.4), with an optional leading "v".
+# Major version must be >= 1.
+DEFAULT_VERSION_REGEX = r"^v?(?P<major>[1-9]\d*)\.(?P<minor>\d+)(\.(?P<patch>\d+)(\.(?P<tiny>\d+))?)?$"
+DEFAULT_TAG_TEMPLATE = "{{major}}.{{minor}}{% if patch %}.{{patch}}{% if tiny %}.{{tiny}}{%endif%}{%endif%}"
+
 
 def load_product(product_name, pathname="website/products") -> frontmatter.Post:
     """Load the product's file frontmatter.

src/docker_hub.py

@@ -4,7 +4,6 @@ import sys
 from common import endoflife
 
 METHOD = "docker_hub"
-REGEX = r"^(?:(\d+\.(?:\d+\.)*\d+))$"
 
 
 def fetch_releases(url, regex, releases):
@@ -35,7 +34,7 @@ def update_product(product_name, configs):
 
     for config in configs:
         url = f"https://hub.docker.com/v2/repositories/{config[METHOD]}/tags?page_size=100&page=1"
-        config = {"regex": REGEX} | config
+        config = {"regex": endoflife.DEFAULT_VERSION_REGEX} | config
         fetch_releases(url, config["regex"], versions)
 
     endoflife.write_releases(product_name, versions)

src/eks.py

@@ -1,6 +1,6 @@
-import datetime as dt
 import re
 from bs4 import BeautifulSoup
+from common import dates
 from common import endoflife
 
 # Now that AWS no longer publishes docs on GitHub,
@@ -16,7 +16,6 @@ URLS = [
     # + latest
     "https://docs.aws.amazon.com/eks/latest/userguide/platform-versions.html",
 ]
-REGEX = r"^(?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)$"
 
 
 def parse_platforms_pages():
@@ -27,11 +26,11 @@ def parse_platforms_pages():
         soup = BeautifulSoup(response, features="html5lib")
         for tr in soup.select("#main-col-body")[0].findAll("tr"):
             td = tr.find("td")
-            if td and re.match(REGEX, td.text.strip()):
+            if td and re.match(endoflife.DEFAULT_VERSION_REGEX, td.text.strip()):
                 data = tr.findAll("td")
                 date = data[-1].text.strip()
                 if len(date) > 0:
-                    d = dt.datetime.strptime(date, "%B %d, %Y").strftime("%Y-%m-%d")
+                    d = dates.parse_date(date).strftime("%Y-%m-%d")
                     k8s_version = ".".join(data[0].text.strip().split(".")[:-1])
                     eks_version = data[1].text.strip().replace(".", "-")
                     version = f"{k8s_version}-{eks_version}"

src/git.py

@@ -16,11 +16,9 @@ identically named groups (as used in the mariadb product).
 # craft-cms, exim, gerrit, jquery, kdeplasma, kirby, logstash, nexus, silverstripe
 # and tarantool versions.
 METHOD = 'git'
-DEFAULT_VERSION_REGEX = r"^v?(?P<major>[1-9]\d*)\.(?P<minor>\d+)(\.(?P<patch>\d+)(\.(?P<tiny>\d+))?)?$"
-DEFAULT_TAG_TEMPLATE = "{{major}}.{{minor}}{% if patch %}.{{patch}}{% if tiny %}.{{tiny}}{%endif%}{%endif%}"
 
 
-def fetch_releases(product_name, url, regex, template):
+def fetch_releases(url, regex, template):
     releases = {}
 
     git = Git(url)
@@ -40,10 +38,10 @@ def update_product(product_name, configs):
     versions = {}
 
     for config in configs:
-        t = config.get("template", DEFAULT_TAG_TEMPLATE)
+        t = config.get("template", endoflife.DEFAULT_TAG_TEMPLATE)
-        regex = config.get("regex", DEFAULT_VERSION_REGEX)
+        regex = config.get("regex", endoflife.DEFAULT_VERSION_REGEX)
         regex = regex.replace("(?<", "(?P<")  # convert ruby regex to python regex
-        versions = versions | fetch_releases(product_name, config[METHOD], regex, t)
+        versions = versions | fetch_releases(config[METHOD], regex, t)
 
     endoflife.write_releases(product_name, versions)
 

src/maven.py

@@ -5,13 +5,12 @@ import sys
 from common import endoflife
 
 METHOD = "maven"
-VERSION_REGEX = r'^\d+\.\d+(\.\d+)?$'
 
 
 # TODO: Add support for custom regexes
 # Hasn't been needed yet, so only write if we need it
 def valid_version(version):
-    if re.match(VERSION_REGEX, version):
+    if re.match(endoflife.DEFAULT_VERSION_REGEX, version):
         return True
     return False
 

src/npm.py

@@ -4,7 +4,6 @@ import sys
 from common import endoflife
 
 METHOD = "npm"
-REGEX = r"^(?:(\d+\.(?:\d+\.)*\d+))$"
 
 
 def fetch_releases(npm_id, regex):
@@ -34,7 +33,7 @@ def update_product(product_name, configs):
     versions = {}
 
     for config in configs:
-        config = {"regex": REGEX} | config
+        config = {"regex": endoflife.DEFAULT_VERSION_REGEX} | config
         versions = versions | fetch_releases(config[METHOD], config["regex"])
 
     endoflife.write_releases(product_name, versions)

src/pypi.py

@@ -5,10 +5,6 @@ from common import dates
 from common import endoflife
 
 METHOD = "pypi"
-DEFAULT_TAG_TEMPLATE = (  # Same as used in Ruby (update.rb)
-    "{{major}}{% if minor %}.{{minor}}{% if patch %}.{{patch}}{%endif%}{%endif%}"
-)
-REGEX = r"^(?:(\d+\.(?:\d+\.)*\d+))$"
 
 
 def fetch_releases(pypi_id, regex):
@@ -38,7 +34,7 @@ def update_product(product_name, configs):
     versions = {}
 
     for config in configs:
-        config = {"regex": REGEX} | config
+        config = {"regex": endoflife.DEFAULT_VERSION_REGEX} | config
         versions = versions | fetch_releases(config[METHOD], config["regex"])
 
     endoflife.write_releases(product_name, versions)

src/rockylinux.py

@@ -3,7 +3,6 @@ from common import dates
 from common import endoflife
 
 URL = "https://raw.githubusercontent.com/rocky-linux/wiki.rockylinux.org/development/docs/include/releng/version_table.md"
-REGEX = r"^(\d+\.\d+)$"
 
 
 def parse_date(date_str):
@@ -17,7 +16,7 @@ def parse_markdown_table(table_text):
 
     for line in lines:
         items = line.split('|')
-        if len(items) >=5 and re.match(REGEX, items[1].strip()):
+        if len(items) >=5 and re.match(endoflife.DEFAULT_VERSION_REGEX, items[1].strip()):
             version = items[1].strip()
             date = parse_date(items[3])
             print(f"{version}: {date}")

src/unrealircd.py

@@ -3,7 +3,6 @@ import re
 from common import endoflife
 
 URL = "https://www.unrealircd.org/docwiki/index.php?title=History_of_UnrealIRCd_releases&action=raw"
-REGEX = r"^(?:(\d+\.(?:\d+\.)*\d+))$"
 
 print("::group::unrealircd")
 response = endoflife.fetch_url(URL)
@@ -14,7 +13,7 @@ for tr in wikicode.ifilter_tags(matches=lambda node: node.tag == "tr"):
     items = tr.contents.filter_tags(matches=lambda node: node.tag == "td")
     if len(items) >= 2:
         maybe_version = items[0].__strip__()
-        if re.match(REGEX, maybe_version):
+        if re.match(endoflife.DEFAULT_VERSION_REGEX, maybe_version):
             maybe_date = items[1].__strip__()
             if re.match(r"\d{4}-\d{2}-\d{2}", maybe_date):
                 versions[maybe_version] = maybe_date