Fix ID-targeted state load dropping resources with regex-escaped filter values

[michael-richey]

Summary

Resources with regex-metacharacters in their ids (e.g. dotted log-metric names) could be silently dropped during a sync run when both --minimize-reads and Operator=ExactMatch Name=id filters are in use.

ExactMatch filter values are regex-escaped by callers (Go regexp.QuoteMeta, Python re.escape, etc.) so they match literally once wrapped in ^...$ and compiled — this is required, otherwise svc.request.count would also match svcXrequestYcount. The --minimize-reads ID-targeted fast path (extract_exact_id_filters → _unwrap_exact_match_pattern) then reused the escaped pattern body as a literal storage key:

• stored blob key: logs_metrics.svc.request.count.json, keyed inside by svc.request.count
• looked-up key: svc\.request\.count ← never matches

State.get_by_ids / get_single returns None for the mismatch and skips it, so the resource never enters state.source, never reaches the dependency graph, and produces no create, no error, and no filtered outcome — it just disappears. Ids with no metacharacters were unaffected, which is why the failure was partial and easy to miss.

Fix

• _unwrap_exact_match_pattern now un-escapes the ^...$ body back to the literal id via a new _regex_literal_from_exact_match_body helper.
• If the body is not a provable literal — an unescaped metacharacter, a semantic or ambiguous alphanumeric escape such as \d, or a dangling backslash — it raises ValueError. extract_exact_id_filters already treats that as "not ID-targetable" and falls back to type-scoped loading, which matches via the compiled regex and stays correct.
• Escaped punctuation and doubled backslashes are recovered as literal ID characters.
• No behavior change for metacharacter-free ids (e.g. UUIDs): escaping is a no-op there.

Testing (red/green)

Added TestExactMatchEscapedValues in tests/unit/test_minimize_reads_id_targeted.py:

• escaped dots are recovered as literal ID characters
• plain literal IDs remain unchanged
• doubled backslashes recover a literal backslash
• genuine regex metacharacters trigger type-scoped fallback
• semantic escapes such as \d trigger type-scoped fallback instead of targeting the wrong ID
• escaped filters extract the original literal metric IDs
• State.get_by_ids resolves an escaped-value ID against the real blob key

The escaped-ID regression tests fail on main and pass with the fix. The semantic-escape regression fails on the initial PR head by extracting literal ID d from a pattern that matches digits, then passes after the conservative fallback change.

Focused suite: 28 passed.

Unit suite excluding the pre-existing @pytest.mark.experiment_subprocess tests: 1100 passed, 8 skipped, 10 deselected.

[Copilot]

Pull request overview

Fixes a --minimize-reads ID-targeted state loading bug where ExactMatch filters containing regex-escaped IDs (e.g., dotted metric IDs) could be treated as literal storage keys and silently fail to load matching resources.

Changes:

• Add _regex_literal_from_exact_match_body and update _unwrap_exact_match_pattern to unescape regex-escaped ExactMatch bodies back to literal IDs, with a conservative ValueError fallback for non-literal patterns.
• Add unit tests covering escaped dots, escaped backslashes, semantic escapes (e.g. \d), and an end-to-end State.get_by_ids resolution case.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
tests/unit/test_minimize_reads_id_targeted.py	Adds regression/unit tests for escaped ExactMatch ID handling and ID-targeted state load behavior.
datadog_sync/utils/configuration.py	Implements safe unescaping of ExactMatch regex bodies to recover literal IDs for ID-targeted loading.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.