[eks] Use the new platform-versions page.

- See https://github.com/awsdocs/amazon-eks-user-guide/issues/603
- Tracking eks release numbers makes more sense here

releases/eks.json

@@ -1,44 +1,50 @@
 {
-  "1.23.7": "2022-08-12",
-  "1.22.10": "2022-07-19",
-  "1.21.13": "2022-07-19",
-  "1.20.15": "2022-05-11",
-  "1.19.16": "2022-05-11",
-  "1.22.9": "2022-06-13",
-  "1.21.12": "2022-06-13",
-  "1.22.6": "2022-04-05",
-  "1.21.9": "2022-05-11",
-  "1.21.5": "2022-02-22",
-  "1.20.7": "2021-08-03",
-  "1.19.15": "2022-02-22",
-  "1.18.16": "2021-04-28",
-  "1.21.2": "2021-07-20",
-  "1.19.8": "2021-04-26",
-  "1.17.17": "2021-04-28",
-  "1.16.15": "2020-11-27",
-  "1.20.4": "2021-05-19",
-  "1.15.12": "2020-11-27",
-  "1.18.9": "2020-11-27",
-  "1.17.12": "2020-11-27",
-  "1.19.6": "2021-02-20",
-  "1.14.9": "2020-01-02",
-  "1.18.8": "2020-10-27",
-  "1.17.9": "2020-08-17",
-  "1.16.13": "2020-08-17",
-  "1.15.11": "2020-04-14",
-  "1.17.6": "2020-07-31",
-  "1.16.8": "2020-05-05",
-  "1.16.12": "2020-07-10",
-  "1.13.12": "2020-01-07",
-  "1.12.10": "2019-09-05",
-  "1.15.10": "2020-03-13",
-  "1.14.7": "2019-11-01",
-  "1.13.10": "2019-09-16",
-  "1.11.10": "2019-09-05",
-  "1.14.6": "2019-09-16",
-  "1.13.8": "2019-09-05",
-  "1.13.7": "2019-06-19",
-  "1.12.6": "2019-06-19",
-  "1.11.8": "2019-06-19",
-  "1.10.13": "2019-06-19"
+  "1.23-eks-1": "2022-08-11",
+  "1.22-eks-5": "2022-08-15",
+  "1.22-eks-4": "2022-07-21",
+  "1.22-eks-3": "2022-07-07",
+  "1.22-eks-2": "2022-05-31",
+  "1.22-eks-1": "2022-04-04",
+  "1.21-eks-10": "2022-08-15",
+  "1.21-eks-9": "2022-07-21",
+  "1.21-eks-8": "2022-07-07",
+  "1.21-eks-7": "2022-05-31",
+  "1.21-eks-6": "2022-04-08",
+  "1.21-eks-5": "2022-03-10",
+  "1.21-eks-4": "2021-12-13",
+  "1.21-eks-3": "2021-11-08",
+  "1.21-eks-2": "2021-09-17",
+  "1.21-eks-1": "2021-07-19",
+  "1.20-eks-8": "2022-08-15",
+  "1.20-eks-7": "2022-07-21",
+  "1.20-eks-6": "2022-05-31",
+  "1.20-eks-5": "2022-04-08",
+  "1.20-eks-4": "2022-03-10",
+  "1.20-eks-3": "2021-11-08",
+  "1.20-eks-2": "2021-07-30",
+  "1.20-eks-1": "2021-05-18",
+  "1.19-eks-11": "2022-08-15",
+  "1.19-eks-10": "2022-05-31",
+  "1.19-eks-9": "2022-04-08",
+  "1.19-eks-8": "2022-03-10",
+  "1.19-eks-7": "2021-11-08",
+  "1.19-eks-6": "2021-09-17",
+  "1.19-eks-5": "2021-06-01",
+  "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",
+  "1.18-eks-13": "2022-08-15",
+  "1.18-eks-12": "2022-05-31",
+  "1.18-eks-11": "2022-04-08",
+  "1.18-eks-10": "2022-03-10",
+  "1.18-eks-9": "2021-11-08",
+  "1.18-eks-8": "2021-09-17",
+  "1.18-eks-7": "2021-06-01",
+  "1.18-eks-6": "2021-05-04",
+  "1.18-eks-5": "2021-04-26",
+  "1.18-eks-4": "2021-03-16",
+  "1.18-eks-3": "2020-12-01",
+  "1.18-eks-2": "2020-11-12",
+  "1.18-eks-1": "2020-10-13"
 }

src/eks.py

@@ -1,40 +1,36 @@
-import pygit2
+import urllib.request
+import datetime
+import markdown
 import re
-from datetime import datetime
 import json
+from datetime import datetime
+from bs4 import BeautifulSoup
 
-REPO_CLONE_URL = 'https://github.com/awsdocs/amazon-eks-user-guide.git'
-PATH = 'doc_source/kubernetes-versions.md'
-TEMP_REPO_PATH = '/tmp/eks-docs'
-REGEX = r"^\+ `?(?P<major>\d+)\\?\.(?P<minor>\d+)\\?\.(?P<patch>\d+)`?$"
+URL = "https://raw.githubusercontent.com/awsdocs/amazon-eks-user-guide/master/doc_source/platform-versions.md"
+REGEX = r"^(?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)$"
 
-versions = {}
 
-def add_versions(c_versions, commit):
-  for v in c_versions:
-    if v not in versions:
-      version_string = "%s.%s.%s" % v
-      date = datetime.fromtimestamp(commit.commit_time).strftime('%Y-%m-%d')
-      versions[version_string] = date
+def parse_platforms_page():
+    versions = {}
+    with urllib.request.urlopen(URL, data=None, timeout=5) as contents:
+        html = markdown.markdown(contents.read().decode("utf-8"), extensions=["tables"])
+        soup = BeautifulSoup(html, features="html5lib")
+        for tr in soup.findAll("tr"):
+            td = tr.find("td")
+            if td and re.match(REGEX, td.text):
+                version = td.text
+                data = tr.findAll("td")
+                date = data[-1].text
+                if len(date) > 0:
+                    d = datetime.strptime(date, "%B %d, %Y").strftime("%Y-%m-%d")
+                    k8s_version = ".".join(data[0].text.split(".")[:-1])
+                    eks_version = data[1].text.replace(".", "-")
+                    version = "%s-%s" % (k8s_version, eks_version)
+                    versions[version] = d
+    return versions
 
-def get_versions(markdown):
-  return re.findall(REGEX, markdown, re.MULTILINE)
-
-pygit2.clone_repository(REPO_CLONE_URL, TEMP_REPO_PATH)
-repo = pygit2.Repository(TEMP_REPO_PATH)
-prev = None
-tree_list = []
-for cur in repo.walk(repo.head.target):
-  if prev is not None:
-    for d in cur.tree.diff_to_tree(prev.tree).deltas:
-      if(d.new_file.path ==  PATH and PATH in cur.tree):
-        contents = cur.tree[PATH].data.decode('UTF-8')
-        add_versions(get_versions(contents), cur)
-
-  if cur.parents:
-    prev = cur
-    cur = cur.parents[0]
-
-with open('releases/eks.json', 'w') as f:
-  f.write(json.dumps(versions, indent=2))
 
+if __name__ == "__main__":
+    versions = parse_platforms_page()
+    with open("releases/eks.json", "w") as f:
+        f.write(json.dumps(versions, indent=2))