fix: prevent IndexError with integration URIs without .Arn

[hoffa]

Issue #, if available

N/A

Description of changes

Mitigates issue where transforming a template such as the included function_alias_with_http_api_events.yaml causes an uncaught IndexError.

This only happens when a function alias has at least two events, one of which uses the default method/path. get_integration_function_logical_id is only used here for an optimization for events with default method/path (see integration test):

serverless-application-model/samtranslator/model/eventsources/push.py

Lines 1074 to 1079 in 3bf8f61

	elif editor and resources_to_link.get("function").logical_id == editor.get_integration_function_logical_id(
	OpenApiEditor._DEFAULT_PATH, OpenApiEditor._X_ANY_METHOD
	):
	# Case where default exists for this function, and so the permissions for that will apply here as well
	# This can save us several CFN resources (not duplicating permissions)
	return

If a function has an HttpApi event with default method and path, it corresponds to */* in the AWS::Lambda::Permissions SourceArn for executing APIs, which encompasses every method and path. The optimization is skipping all other Permissions since they're already encompassed by the */*, and the permissions are always the same (lambda:InvokeFunction).

Ideally the same logic could be applied to function aliases. I didn't want to extend the regex further as it's already a hack that relies on a specific Fn::Sub format. The logical ID could maybe be tracked elsewhere, although this might be difficult if the input is the OpenAPI document.

Note: fixed code.zip as the function didn't run (was c.done() when it should've been c.succeed()).

Side effects

• SAM Translator won't crash with function_alias_with_http_api_events.yaml.
• The optimization used for Functions won't be used if it includes AutoPublishAlias (i.e. it'll create a separate AWS::Lambda::Permission for each HttpApi event).

Description of how you validated changes

pytest --no-cov integration/single/test_basic_function.py -vv -k test_function_with_http_api_events

Checklist

• Write/update tests
• make pr passes
• Update documentation
• Verify transformed template deploys and application functions as expected

Examples?

Please reach out in the comments, if you want to add an example. Examples will be
added to sam init through https://github.com/awslabs/aws-sam-cli-app-templates/

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[codecov-io]

Codecov Report

Merging #1901 (568666e) into develop (994fa65) will decrease coverage by 0.17%.
The diff coverage is 50.00%.

@@             Coverage Diff             @@
##           develop    #1901      +/-   ##
===========================================
- Coverage    94.04%   93.86%   -0.18%     
===========================================
  Files           87       89       +2     
  Lines         5695     5888     +193     
  Branches      1153     1206      +53     
===========================================
+ Hits          5356     5527     +171     
- Misses         156      165       +9     
- Partials       183      196      +13     

Impacted Files	Coverage Δ
samtranslator/open_api/open_api.py	91.91% <50.00%> (-0.46%)	⬇️
samtranslator/model/eventsources/pull.py	79.03% <0.00%> (-5.42%)	⬇️
samtranslator/model/s3_utils/uri_parser.py	69.23% <0.00%> (-2.20%)	⬇️
samtranslator/model/sam_resources.py	92.11% <0.00%> (-1.32%)	⬇️
samtranslator/model/s3.py	85.71% <0.00%> (ø)
samtranslator/model/lambda_.py	93.10% <0.00%> (ø)
samtranslator/region_configuration.py	100.00% <0.00%> (ø)
samtranslator/plugins/globals/globals.py	99.05% <0.00%> (ø)
samtranslator/model/eventbridge_utils.py	100.00% <0.00%> (ø)
samtranslator/model/packagetype.py	100.00% <0.00%> (ø)
... and 8 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 994fa65...568666e. Read the comment docs.

[jfuss]

It's hard for me to tell if this actually addresses the failure without adding additional tests.