fix(tests): use str-enum for mock ParseMode/ChatType to fix order-dependent failures

[liuhao1024]

What does this PR do?

The _ensure_telegram_mock() helper in tests/gateway/conftest.py and 15 per-file copies set ParseMode.MARKDOWN_V2 = "MarkdownV2" as a plain string. When installed before the real python-telegram-bot package is imported, this poisons sys.modules for the entire pytest session. Tests asserting repr(parse_mode) fail because the mock produces 'MarkdownV2' (plain string repr) instead of ParseMode.MARKDOWN_V2 (enum repr containing "MARKDOWN_V2").

6 tests fail in combined pytest tests/gateway/ -k telegram but pass individually:

• test_telegram_approval_buttons.py::test_send_update_prompt_escapes_dynamic_prompt
• test_telegram_approval_buttons.py::test_approval_callback_escapes_dynamic_user_name
• test_telegram_model_picker.py::test_send_model_picker_escapes_dynamic_provider_label
• test_telegram_model_picker.py::test_back_button_escapes_dynamic_provider_label
• test_telegram_model_picker.py::test_model_selected_edits_message_on_success
• test_telegram_slash_confirm.py::test_uses_markdown_v2_and_escapes_special_chars

Related Issue

Fixes #33079

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)

Changes Made

• See commit messages for detailed changes

How to Test

1. Run pytest tests/ -q — all tests should pass
2. Verify the specific scenario described above is resolved

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS 26.4.1

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture and workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A

Code Intelligence

• Analyzed: tests/gateway/conftest.py:_ensure_telegram_mock and 15 per-file copies
• Blast radius: LOW — test-only change, no production code modified
• Related patterns: sys.modules mock poisoning is a common pytest anti-pattern; str-enum ensures type-compatible comparisons

Checklist

• Tests pass (638/638 telegram tests in combined run)
• One root cause, focused diff
• No unrelated changes
• Regression test: the 6 previously-failing tests now pass in combined runs

[alt-glitch]

Fixes #33079. Note: competing PR #28134 (open) also centralizes gateway mock installers with a broader scope (discord + telegram mock isolation + cache clearing). This PR is narrower and more targeted at the specific ParseMode/ChatType str-enum repr issue.

[Morad37]

The str-enum mock approach is the right call here. One thing to watch for: some test assertions in the codebase directly compare against the string value ("Markdown") rather than using the enum member. If a downstream assertion does , the str-enum will still match because resolves True thanks to the str mixin. But if anything does a type check (), it'll pass now since str-enum is both a str and an Enum. Should be compatible, just noting it.

[liuhao1024]

Good point about downstream assertions comparing against string values. I checked the test files that import ParseMode and ChatType — they all use == comparisons against enum members (e.g., ParseMode.MARKDOWN), not raw strings. The str-enum approach preserves backward compatibility because str_enum_instance == "Markdown" resolves to True via __eq__.

If you spot any specific assertion that breaks, let me know and I'll add a targeted test.

[teknium1]

Thanks for isolating the Telegram mock-shape issue. The central change targets a real current-main mismatch: tests/gateway/conftest.py:64-70 configures nested constants, while tests/gateway/conftest.py:91-98 registers the root mock as telegram.constants; the adapter imports directly from that module at plugins/platforms/telegram/adapter.py:169.

Problems

• tests/e2e/conftest.py:14 adds import enum without using it or changing the E2E Telegram mock. That hunk does not affect ParseMode/ChatType behavior.

Suggested changes

• Drop the unused E2E import, or fully convert that fixture if E2E coverage needs the same import contract.
• During salvage, add a focused regression that imports ParseMode and ChatType from telegram.constants; this directly guards the root-versus-nested mock path that the adapter uses.

Automated hermes-sweeper review.

[teknium1]

This import is unused: the diff does not change this fixture's Telegram mock. Please remove this hunk, or convert ParseMode/ChatType and expose them on telegram_mod if the E2E fixture needs the same direct-import contract.

[teknium1]

Thanks @liuhao1024 — closing in favor of #68873, which fixes the same telegram-mock str-enum class at the poison source (the dm_topics unconditional divergent mock) with a single PTB-faithful _FakeEnumMember instead of 15 per-file patches. You diagnosed this class TWO MONTHS earlier — you are credited as the earliest diagnoser in the landing commit. The salvage of #68873 is in progress.