[eks] Refactor script (#224)

Make the script more readable, mostly by:

- changing slightly the logic,
- using the Product and AutoConfig classes,
- removing the use of functions when unnecessary,
- a little bit of renaming and documentation.

Disabled web.archive.org has also been re-enabled because HTTP retry mechanism has been improved and should handle timeouts a lot better.

releases/eks.json

@@ -43,19 +43,73 @@
   "1.25-eks-3": "2023-05-05",
   "1.24-eks-6": "2023-05-05",
   "1.23-eks-8": "2023-05-05",
+  "1.22-eks-12": "2023-05-05",
+  "1.21-eks-17": "2023-05-05",
+  "1.20-eks-14": "2023-05-05",
   "1.26-eks-1": "2023-04-11",
   "1.25-eks-2": "2023-03-24",
   "1.24-eks-5": "2023-03-24",
   "1.23-eks-7": "2023-03-24",
+  "1.22-eks-11": "2023-03-24",
+  "1.21-eks-16": "2023-03-24",
   "1.25-eks-1": "2023-02-21",
   "1.24-eks-4": "2023-01-27",
   "1.23-eks-6": "2023-01-27",
+  "1.22-eks-10": "2023-01-27",
+  "1.21-eks-15": "2023-01-27",
   "1.24-eks-3": "2022-12-05",
   "1.23-eks-5": "2022-12-05",
+  "1.22-eks-9": "2022-12-05",
+  "1.21-eks-14": "2022-12-05",
+  "1.20-eks-12": "2022-12-05",
   "1.24-eks-2": "2022-11-18",
   "1.23-eks-4": "2022-11-18",
+  "1.22-eks-8": "2022-11-18",
+  "1.21-eks-13": "2022-11-18",
+  "1.20-eks-11": "2022-11-18",
   "1.24-eks-1": "2022-11-15",
   "1.23-eks-3": "2022-11-07",
+  "1.22-eks-7": "2022-11-07",
+  "1.21-eks-12": "2022-11-07",
+  "1.20-eks-10": "2022-11-07",
+  "1.21-eks-11": "2022-10-10",
   "1.23-eks-2": "2022-09-21",
-  "1.23-eks-1": "2022-08-11"
+  "1.22-eks-6": "2022-09-21",
+  "1.20-eks-9": "2022-09-21",
+  "1.22-eks-5": "2022-08-15",
+  "1.21-eks-10": "2022-08-15",
+  "1.20-eks-8": "2022-08-15",
+  "1.19-eks-11": "2022-08-15",
+  "1.23-eks-1": "2022-08-11",
+  "1.22-eks-4": "2022-07-21",
+  "1.21-eks-9": "2022-07-21",
+  "1.20-eks-7": "2022-07-21",
+  "1.22-eks-3": "2022-07-07",
+  "1.21-eks-8": "2022-07-07",
+  "1.22-eks-2": "2022-05-31",
+  "1.21-eks-7": "2022-05-31",
+  "1.20-eks-6": "2022-05-31",
+  "1.19-eks-10": "2022-05-31",
+  "1.21-eks-6": "2022-04-08",
+  "1.20-eks-5": "2022-04-08",
+  "1.19-eks-9": "2022-04-08",
+  "1.22-eks-1": "2022-04-04",
+  "1.20-eks-13": "2022-03-24",
+  "1.21-eks-5": "2022-03-10",
+  "1.20-eks-4": "2022-03-10",
+  "1.19-eks-8": "2022-03-10",
+  "1.21-eks-4": "2021-12-13",
+  "1.21-eks-3": "2021-11-08",
+  "1.20-eks-3": "2021-11-08",
+  "1.19-eks-7": "2021-11-08",
+  "1.21-eks-2": "2021-09-17",
+  "1.19-eks-6": "2021-09-17",
+  "1.20-eks-2": "2021-07-30",
+  "1.21-eks-1": "2021-07-19",
+  "1.19-eks-5": "2021-06-01",
+  "1.20-eks-1": "2021-05-18",
+  "1.19-eks-4": "2021-05-04",
+  "1.19-eks-3": "2021-04-14",
+  "1.19-eks-2": "2021-03-23",
+  "1.19-eks-1": "2021-02-16"
 }

src/common/endoflife.py

@@ -11,6 +11,7 @@ logging.basicConfig(format=logging.BASIC_FORMAT, level=logging.INFO)
 # 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_VERSION_PATTERN = re.compile(DEFAULT_VERSION_REGEX)
 DEFAULT_TAG_TEMPLATE = "{{major}}{% if minor %}.{{minor}}{% if patch %}.{{patch}}{% if tiny %}.{{tiny}}{%endif%}{%endif%}{%endif%}"
 
 PRODUCTS_PATH = os.environ.get("PRODUCTS_PATH", "website/products")

src/eks.py

@@ -1,45 +1,40 @@
-import re
 from bs4 import BeautifulSoup
 from common import http
 from common import dates
 from common import endoflife
 
-# Now that AWS no longer publishes docs on GitHub,
-# we use the Web Archive to still get the older versions
-# Keep older pages at top of the list
+"""Fetches EKS versions from AWS docs.
+Now that AWS no longer publishes docs on GitHub, we use the Web Archive to still get the older versions."""
+
 URLS = [
     # 1.19.eks.1
-    # Disabled, too much timed out.
-    # "https://web.archive.org/web/20221007150452/https://docs.aws.amazon.com/eks/latest/userguide/platform-versions.html",
+    "https://web.archive.org/web/20221007150452/https://docs.aws.amazon.com/eks/latest/userguide/platform-versions.html",
     # + 1.20
-    # Disabled, too much timed out.
-    # "https://web.archive.org/web/20230521061347/https://docs.aws.amazon.com/eks/latest/userguide/platform-versions.html",
+    "https://web.archive.org/web/20230521061347/https://docs.aws.amazon.com/eks/latest/userguide/platform-versions.html",
     # + latest
     "https://docs.aws.amazon.com/eks/latest/userguide/platform-versions.html",
 ]
 
+product = endoflife.Product("eks")
+print(f"::group::{product.name}")
 
-def parse_platforms_pages():
-    all_versions = {}
-    print("::group::eks")
-    for url in URLS:
-        response = http.fetch_url(url)
-        soup = BeautifulSoup(response.text, features="html5lib")
-        for tr in soup.select("#main-col-body")[0].findAll("tr"):
-            td = tr.find("td")
-            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 = 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}"
-                    all_versions[version] = d
-                    print(f"{version}: {d}")
-    print("::endgroup::")
-    return all_versions
+for version_list in http.fetch_urls(URLS):
+    version_list_soup = BeautifulSoup(version_list.text, features="html5lib")
+    for tr in version_list_soup.select("#main-col-body")[0].findAll("tr"):
+        cells = tr.findAll("td")
+        if not cells:
+            continue
 
+        k8s_version = cells[0].text.strip()
+        eks_version = cells[1].text.strip()
+        date_str = cells[-1].text.strip()
 
-versions = parse_platforms_pages()
-endoflife.write_releases('eks', versions)
+        k8s_version_match = endoflife.DEFAULT_VERSION_PATTERN.match(k8s_version)
+        if k8s_version_match:
+            date = dates.parse_date(date_str)
+            # K8S patch version is not kept to match versions on https://github.com/aws/eks-distro/tags.
+            version = f"{k8s_version_match.group('major')}.{k8s_version_match.group('minor')}-{eks_version.replace('.', '-')}"
+            product.declare_version(version, date)
+
+product.write()
+print("::endgroup::")