fix: make GitHub PR-description ticket selection deterministic

[dellch]

Fixes #2421

Summary

extract_ticket_links_from_pr_description() in pr_agent/tools/ticket_pr_compliance_check.py accumulates GitHub issue URLs in a set, then enforces the 3-ticket cap by slicing a list built from that set:

github_tickets = set()
...
if len(github_tickets) > 3:
    github_tickets = set(list(github_tickets)[:3])
...
return list(github_tickets)

Because a set has no defined iteration order (and Python randomizes string hashing per process via PYTHONHASHSEED), list(github_tickets)[:3] selects an arbitrary 3 of the referenced issues, and the selection can vary between runs. When a PR description references more than 3 issues, which issues get fetched for review context is therefore nondeterministic.

Fix

Track URLs in first-seen order while de-duplicating (a seen set plus an ordered list), then apply the cap by slicing the ordered list — mirroring what find_jira_tickets() in the same file already does. Selection is now stable and predictable. Behaviour is unchanged for 3 or fewer issues.

Tests

Structured as two commits — failing tests first, then the fix:

1. test: adds TestExtractTicketLinksFromPrDescription with two tests.
2. fix: makes them pass.

test_cap_selects_deterministic_first_seen_subset is the reliable regression guard: with more than 3 issues, the old set-based code returns an arbitrary subset that does not equal the first-seen subset. I verified it fails against the previous implementation on every hash seed PYTHONHASHSEED 0–11, and passes with the fix on all of them. (test_preserves_first_seen_order documents the intended ordering; its docstring notes it is not a reliable seed-independent guard on its own, which is why the cap test exists.)

Notes

This was originally surfaced by the Qodo review bot on a separate Jira-support PR, but the function predates that work and is unrelated to Jira, so it is raised here on its own.

Test plan

• pytest tests/unittest/test_extract_issue_from_branch.py
• New tests fail on the pre-fix code (verified across PYTHONHASHSEED 0–11) and pass with the fix

Co-Authored-By: Claude

[qodo-free-for-open-source-projects]

Review Summary by Qodo

Make GitHub PR-description ticket selection deterministic

🐞 Bug fix 🧪 Tests

Walkthroughs

Description

• Fix nondeterministic ticket selection in PR description parsing
  - Replace set-based accumulation with ordered list tracking
  - Preserve first-seen order while de-duplicating URLs
  - Cap selection now returns deterministic first 3 tickets
• Add comprehensive test suite for ticket extraction
  - Test first-seen ordering preservation
  - Test deterministic capped subset selection (reliable regression guard)

Diagram

flowchart LR
  A["PR Description"] -->|extract URLs| B["Ordered List + Seen Set"]
  B -->|deduplicate| C["Preserve First-Seen Order"]
  C -->|cap at 3| D["Deterministic Subset"]
  E["Tests"] -->|verify| D

Loading

File Changes

1. pr_agent/tools/ticket_pr_compliance_check.py 🐞 Bug fix +15/-6

Replace set-based ticket tracking with ordered list

• Replace set() accumulation with ordered list and seen set for deterministic ordering
• Add _add() helper function to track URLs in first-seen order while de-duplicating
• Change cap logic from set(list(github_tickets)[:3]) to github_tickets[:3]
• Return ordered list directly instead of converting from set

pr_agent/tools/ticket_pr_compliance_check.py

---

2. tests/unittest/test_extract_issue_from_branch.py 🧪 Tests +40/-1

Add tests for deterministic PR description ticket selection

• Import extract_ticket_links_from_pr_description function for testing
• Add MAX_TICKETS constant (3) matching function's hardcoded cap
• Add TestExtractTicketLinksFromPrDescription test class with two tests
• Test test_preserves_first_seen_order validates ordering and deduplication
• Test test_cap_selects_deterministic_first_seen_subset is reliable regression guard

tests/unittest/test_extract_issue_from_branch.py

---

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0) 🔗 Cross-repo conflicts (0)

Context used

✅ Tickets: 🎫 Nondeterministic ticket selection when a PR description references more than 3 issues

1. Single-quoted f-strings added ✓ Resolved 📘 Rule violation ⚙ Maintainability

Description

The updated GitHub ticket URL construction uses single-quoted f-strings, which violates the
repository’s Python string-quoting convention. This can cause avoidable lint/format churn and
inconsistent style across the codebase.

Code

pr_agent/tools/ticket_pr_compliance_check.py[R61-65]

Evidence

PR Compliance ID 11 requires following repository Python formatting conventions, including
preferring double quotes. The modified lines construct URLs using single-quoted f-strings (e.g.,
f'{base_url_html.strip("/")}/...').

AGENTS.md: Follow Repository Python Formatting and Import Ordering (Ruff/isort, 120-char lines, double quotes): AGENTS.md: Follow Repository Python Formatting and Import Ordering (Ruff/isort, 120-char lines, double quotes): AGENTS.md: Follow Repository Python Formatting and Import Ordering (Ruff/isort, 120-char lines, double quotes): AGENTS.md: Follow Repository Python Formatting and Import Ordering (Ruff/isort, 120-char lines, double quotes)
pr_agent/tools/ticket_pr_compliance_check.py[61-65]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
New/modified URL f-strings use single quotes, but the repository style requires preferring double quotes for Python string literals.
## Issue Context
These lines were modified in this PR, so they should be brought into compliance with the repo’s Ruff/style expectations to avoid lint failures and style drift.
## Fix Focus Areas
- pr_agent/tools/ticket_pr_compliance_check.py[61-65]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

Previous review results

Review updated until commit 03979aa

Results up to commit 8e08857

🐞 Bugs (0) 📘 Rule violations (1) 📎 Requirement gaps (0) 🔗 Cross-repo conflicts (0)

Context used

✅ Tickets: 🎫 Nondeterministic ticket selection when a PR description references more than 3 issues

1. Single-quoted f-strings added 📘 Rule violation ⚙ Maintainability

Description

The updated GitHub ticket URL construction uses single-quoted f-strings, which violates the
repository’s Python string-quoting convention. This can cause avoidable lint/format churn and
inconsistent style across the codebase.

Code

pr_agent/tools/ticket_pr_compliance_check.py[R61-65]

Evidence

PR Compliance ID 11 requires following repository Python formatting conventions, including
preferring double quotes. The modified lines construct URLs using single-quoted f-strings (e.g.,
f'{base_url_html.strip("/")}/...').

AGENTS.md: Follow Repository Python Formatting and Import Ordering (Ruff/isort, 120-char lines, double quotes)
pr_agent/tools/ticket_pr_compliance_check.py[61-65]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
New/modified URL f-strings use single quotes, but the repository style requires preferring double quotes for Python string literals.

## Issue Context
These lines were modified in this PR, so they should be brought into compliance with the repo’s Ruff/style expectations to avoid lint failures and style drift.

## Fix Focus Areas
- pr_agent/tools/ticket_pr_compliance_check.py[61-65]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[qodo-free-for-open-source-projects]

Code review by qodo was updated up to the latest commit 03979aa

[dellch]

Good catch — fixed in 03979aa.

To clarify the origin: those two single-quoted f-strings are verbatim from the existing code on main (the diff only wraps them in the new _add(...) helper, it did not introduce the quoting). But since this PR is already modifying those exact lines, and the rest of this file uses double-quoted f-strings (11 double vs the 2 single ones here — including the identical URL construction a few lines down), I normalized them to double quotes to match the local convention and remove the inconsistency:

_add(f"{base_url_html.strip(chr(47))}/{owner}/{repo}/issues/{issue_number}")

(inner quotes flipped to single so they do not clash with the outer double quotes). Tests still pass.