fix(client): allow non-numeric run/task IDs for Argo and Step Functions

[dheerenmohta]

Problem

The pathspec validation introduced in #948 uses a numeric-only pattern (^[0-9]+$) for run IDs and task IDs. However, Metaflow run IDs are not always numeric:

• Argo Workflows produces run IDs like argo-moviestatsflow-68z2h (see argo_workflows_cli.py:965)
• AWS Step Functions produces run IDs like sfn-my-execution-name (see step_functions_cli.py:549)

This means any user doing Run('MyFlow/argo-...') or Step('MyFlow/sfn-.../start') via the Client API will get a spurious MetaflowInvalidPathspec, completely breaking orchestrator-based workflows.

The original pre-#948 code only checked component count, not content — this PR preserves that correctness while still adding meaningful content validation.

Fix

Replace the strict _NUMERIC_ID_PATTERN check for run/task positions with a permissive _RUN_ID_PATTERN that accepts alphanumeric strings plus hyphens and underscores. This covers all current formats:

• Local runs: "1748291234567890"
• Argo: "argo-myflow-abc12"
• Step Functions: "sfn-prod-execution"

Test

Companion test coverage is added in PR #5 (test branch test/client-pathspec-orchestrator-run-ids).

Fixes #948 (partial — critical correctness bug)

[greptile-apps]

Greptile Summary

This PR adds content-based validation for run IDs and task IDs when parsing a MetaflowObject pathspec. A new _RUN_ID_PATTERN (^[a-zA-Z0-9][a-zA-Z0-9\-_]*$) is applied to ids[1] (run ID) and ids[3] (task ID) before the metadata lookup is attempted.

• Introduces _RUN_ID_PATTERN to allow alphanumeric-plus-hyphen/underscore IDs, covering local numeric IDs (1748291234567890), Argo (argo-*), and SFN (sfn-{uuid}) formats.
• Raises MetaflowInvalidPathspec with a descriptive message when the run ID or task ID contains characters outside the pattern — validation that did not previously exist in the base branch.

Confidence Score: 3/5

The change adds new client-side validation for pathspec run IDs and task IDs where none existed before; while it handles Argo and standard SFN UUIDs correctly, it can incorrectly reject valid SFN run IDs whose execution names contain characters like . (legal per the SFN API but not in the chosen pattern).

The new _RUN_ID_PATTERN covers the common orchestrator formats (numeric local IDs, argo-*, sfn-{uuid}, airflow-{md5}), but the character set is narrower than what AWS SFN execution names legally allow. An EventBridge-triggered execution named prod.2024-01-15 would produce a run ID of sfn-prod.2024-01-15 that the pattern rejects — turning a previously working Run(...) lookup into a hard MetaflowInvalidPathspec error. The fix is straightforward (add . to the pattern or widen it), but the regression is real for users with custom SFN execution names.

metaflow/client/core.py — specifically the _RUN_ID_PATTERN definition and its use in the new validation block.

Important Files Changed

Filename	Overview
metaflow/client/core.py	Adds _RUN_ID_PATTERN and content validation for run/task IDs in pathspec parsing; pattern may be too strict for SFN execution names containing dots or other legal characters not in [a-zA-Z0-9_-]

_{Reviews (1): Last reviewed commit: "fix(client): allow non-numeric run/task ..." | Re-trigger Greptile}

[greptile-apps]

Pattern rejects valid SFN run IDs with dots or other legal characters

AWS Step Functions execution names can legally contain . (and other characters like +, @) per the SFN StartExecution API. When a state machine is triggered externally — e.g. by an EventBridge rule with a custom execution name such as prod.2024-01-15 — the run ID stored by Metaflow is sfn-prod.2024-01-15. After this PR, any Run('MyFlow/sfn-prod.2024-01-15') call will raise MetaflowInvalidPathspec, whereas before this PR (where no content validation existed) it would look up the run normally. The original code's correctness guarantee was "only check component count" — this pattern narrows that guarantee without covering the full character set that SFN execution names allow.

[greptile-apps]

Using \- in the middle of a character class works (Python re treats it as a literal hyphen) but is non-idiomatic. The conventional and unambiguous style is to place the hyphen at the end of the class.

Suggested change

	_RUN_ID_PATTERN = re.compile(r'^[a-zA-Z0-9][a-zA-Z0-9\-_]*$')
	_RUN_ID_PATTERN = re.compile(r'^[a-zA-Z0-9][a-zA-Z0-9_-]*$')

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[codecov]

Codecov Report

❌ Patch coverage is 66.66667% with 2 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (master@dddc0d5). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
metaflow/client/core.py	66.66%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff            @@
##             master    #3260   +/-   ##
=========================================
  Coverage          ?   28.65%           
=========================================
  Files             ?      381           
  Lines             ?    52519           
  Branches          ?     9268           
=========================================
  Hits              ?    15049           
  Misses            ?    36442           
  Partials          ?     1028           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.