Add ECS AMI mapper

2 files changed, 76 insertions, 0 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..34422ef
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1 @@
+/.py3
diff --git a/make_ecs_ami_yaml.py b/make_ecs_ami_yaml.py
new file mode 100755
index 0000000..dda23da
--- /dev/null
+++ b/make_ecs_ami_yaml.py
@@ -0,0 +1,75 @@
+#!/usr/bin/env python3
+#
+# Create a YAML mapping that is sutible for use in a CloudFormation template of
+# the current ECS AMIs based on those published in the documentation. The
+# script attempts to be as fault-tolerant as possible while still producing
+# sane data.
+#
+# Assumptions:
+# The script looks for a table where the first row (td) contains a set of
+# headers (th), one of which has the text 'AMI ID' and the other of which is
+# 'Region'. Given these it will extract the region/ami-id pairs.
+#
+
+import re
+import yaml
+import requests
+from bs4 import BeautifulSoup
+
+AMI_URL = ("https://docs.aws.amazon.com/AmazonECS/latest/"
+           "developerguide/ecs-optimized_AMI.html")
+
+AMI_RE = re.compile("^ami-[0-9a-f]+$")
+
+soup = BeautifulSoup(requests.get(AMI_URL).text, "html.parser")
+
+# Find the AMI table and column indicies
+ami_table, id_index, region_index = None, None, None
+for table in soup.find_all("table"):
+    # Only want header rows for identification
+    ths = table.tr.find_all('th')
+    if len(ths) < 2:
+        continue
+
+    # Pick out the column mappings and hold onto their indicies
+    for i, th in enumerate(ths):
+        if th.text.lower() == "region":
+            region_index = i
+
+        if th.text.lower() == "ami id":
+            ami_table = table
+            id_index = i
+
+    # Indicies may be zero
+    if ami_table and region_index != None and id_index != None:
+        break
+
+# Indicies may be zero
+if ami_table == None or id_index == None or region_index == None:
+    raise Exception("No AMI table found")
+
+# Build the AMI map
+ami_map = {}
+for row in ami_table.find_all("tr"):
+    # Must not be a header (th) row
+    tds = row.find_all("td")
+    if len(tds) < max(id_index, region_index):
+        continue
+
+    region = tds[region_index].text
+    ami_id = tds[id_index].text
+
+    # Sanity check the AMI ID in-case we mis-identified a row
+    if not AMI_RE.match(ami_id):
+        continue
+
+    ami_map[region] = {"AMI": ami_id}
+
+# Validate that there at least enough AMIs (min 4 because there are currently 4
+# big US regions)
+if len(ami_map) < 4:
+    raise Exception("Not enough AMIs found")
+
+# Output AMI map as YAML
+print(yaml.dump({"ECSRegionImages": ami_map},
+    default_flow_style=False, explicit_start=True))