Add AWS image name parser

Parse additional information about AWS images using their name.

Fixes #34

Signed-off-by: Major Hayden <major@redhat.com>

src/rhelocator/update_images.py

@@ -1,6 +1,8 @@
 """Update images from public cloud APIs."""
 from __future__ import annotations
 
+import re
+
 import boto3
 import requests
 
@@ -71,6 +73,44 @@ def get_aws_all_images(image_type: str = "hourly") -> dict[str, list[dict[str, s
     return images_per_region
 
 
+def parse_aws_image_name(image_name: str) -> dict[str, str]:
+    """Parse an AWS image name and return extra data about the image.
+
+    Regex101: https://regex101.com/r/mXCl73/1
+
+    Args:
+        image_name: String containing the image name, such as:
+                    RHEL-9.0.0_HVM_BETA-20211026-x86_64-10-Hourly2-GP2
+
+    Returns:
+        Dictionary with additional information about the image.
+    """
+    # See Regex101 link above to tinker with this regex. Each group is named to make
+    # it easier to handle parsed data. Explanation of names:
+    #
+    #     intprod = internal product (such as HA)
+    #     extprod = external product (such as SAP)
+    #     version = RHEL version (such as 9.0.0)
+    #     virt = virtualization type (such as HVM)
+    #     beta = beta vs non-beta release
+    #     date = date image was produced
+    #     arch = architecture (such as x86_64 or arm64)
+    #     release = release number of the image
+    #     billing = Hourly2 or Access2  #noqa: E800
+    #     storage = storage type (almost always GP2)
+    #
+    aws_image_name_regex = (
+        r"RHEL_*(?P<intprod>\w*)?-*(?P<extprod>\w*)?-(?P<version>[\d\-\.]*)_"
+        r"(?P<virt>[A-Z]*)_*(?P<beta>\w*)?-(?P<date>\d+)-(?P<arch>\w*)-"
+        r"(?P<release>\d+)-(?P<billing>[\w\d]*)-(?P<storage>\w*)"
+    )
+    matches = re.match(aws_image_name_regex, image_name, re.IGNORECASE)
+    if matches:
+        return matches.groupdict()
+
+    return {}
+
+
 def get_azure_access_token() -> str:
     """Authenticate with Azure and return the access token to use with API requests.
 

tests/test_update_images.py

@@ -65,6 +65,82 @@ def test_get_aws_all_images(mock_aws_regions, mock_aws_images) -> None:
         assert billing_codes == {config.AWS_HOURLY_BILLING_CODE}
 
 
+def test_parse_aws_image_name_basic():
+    """Test parsing an AWS image name with a very basic image."""
+    image_name = "RHEL-7.9_HVM-20220512-x86_64-1-Hourly2-GP2"
+    data = update_images.parse_aws_image_name(image_name)
+
+    assert isinstance(data, dict)
+
+    assert not data["intprod"]
+    assert not data["extprod"]
+    assert data["version"] == "7.9"
+    assert data["virt"] == "HVM"
+    assert not data["beta"]
+    assert data["date"] == "20220512"
+    assert data["arch"] == "x86_64"
+    assert data["release"] == "1"
+    assert data["billing"] == "Hourly2"
+    assert data["storage"] == "GP2"
+
+
+def test_parse_aws_image_name_beta():
+    """Test parsing an AWS image name for a beta arm64 image."""
+    image_name = "RHEL-9.1.0_HVM_BETA-20220829-arm64-0-Hourly2-GP2"
+    data = update_images.parse_aws_image_name(image_name)
+
+    assert isinstance(data, dict)
+
+    assert not data["intprod"]
+    assert not data["extprod"]
+    assert data["version"] == "9.1.0"
+    assert data["virt"] == "HVM"
+    assert data["beta"] == "BETA"
+    assert data["date"] == "20220829"
+    assert data["arch"] == "arm64"
+    assert data["release"] == "0"
+    assert data["billing"] == "Hourly2"
+    assert data["storage"] == "GP2"
+
+
+def test_parse_aws_image_name_internal_product():
+    """Test parsing an AWS image name with an internal product."""
+    image_name = "RHEL_HA-8.5.0_HVM-20211103-x86_64-0-Hourly2-GP2"
+    data = update_images.parse_aws_image_name(image_name)
+
+    assert isinstance(data, dict)
+
+    assert data["intprod"] == "HA"
+    assert not data["extprod"]
+    assert data["version"] == "8.5.0"
+    assert data["virt"] == "HVM"
+    assert not data["beta"]
+    assert data["date"] == "20211103"
+    assert data["arch"] == "x86_64"
+    assert data["release"] == "0"
+    assert data["billing"] == "Hourly2"
+    assert data["storage"] == "GP2"
+
+
+def test_parse_aws_image_name_external_product():
+    """Test parsing an AWS image name for a beta arm64 image."""
+    image_name = "RHEL-SAP-8.2.0_HVM-20211007-x86_64-0-Hourly2-GP2"
+    data = update_images.parse_aws_image_name(image_name)
+
+    assert isinstance(data, dict)
+
+    assert not data["intprod"]
+    assert data["extprod"] == "SAP"
+    assert data["version"] == "8.2.0"
+    assert data["virt"] == "HVM"
+    assert not data["beta"]
+    assert data["date"] == "20211007"
+    assert data["arch"] == "x86_64"
+    assert data["release"] == "0"
+    assert data["billing"] == "Hourly2"
+    assert data["storage"] == "GP2"
+
+
 @patch("rhelocator.update_images.requests.post")
 def test_get_azure_access_token(mock_post) -> None:
     """Test retrieving Azure locations."""