Add text2sample generation task type

[Copilot]

Adds text2sample as a first-class task type throughout the stack, enabling short production-sample generation from plain-text descriptions (the workflow used by models like RoyalCities/Foundation-1). Differs from text2music in its DiT instruction and the fact that it always triggers LLM-driven metadata generation without requiring sample_mode=True from the caller.

API / generation pipeline

• constants.py: registers "text2sample" in TASK_INSTRUCTIONS ("Generate a music sample based on the given conditions:"), GENERATION_MODES_TURBO/BASE, and MODE_TO_TASK_TYPE ("Sample"→"text2sample").
• llm_generation_inputs.py: when task_type="text2sample", implicitly sets sample_mode=True and promotes req.prompt → sample_query when sample_query is empty, so callers can use the standard prompt field without extra flags.

# API usage — no sample_mode flag required
POST /release_task
{
  "task_type": "text2sample",
  "prompt": "punchy trap hi-hat loop at 140 BPM"
}

CLI

• text2sample is now option 7 in the interactive task menu.
• Automatically sets sample_mode=True; validation updated to allow --sample_mode/--sample_query for this task type.

python cli.py generate --task_type text2sample --sample_query "lofi piano chord loop, Bb minor, 85 BPM"

Gradio UI

• New Sample mode added to both turbo and base model mode lists, mapping to task_type="text2sample".
• Shows the same simple-mode query panel as Simple, plus a generate-button row (initially disabled). Button activates after Create Sample succeeds.
• After Create Sample in Sample mode the UI stays in Sample mode (preserving task_type="text2sample"); in Simple mode it continues to switch to Custom as before.
• i18n description strings added for all four supported languages.

OpenRouter server

• Message text is routed as sample_query when task_type="text2sample", consistent with sample_mode behavior.

Tests

• acestep/api/text2sample_test.py: 7 unit tests covering auto-enable of sample_mode, prompt promotion, non-regression for text2music, and constants correctness.

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• huggingface.co
  • Triggering command: /home/REDACTED/work/_temp/ghcca-node/node/bin/node /home/REDACTED/work/_temp/ghcca-node/node/bin/node --enable-source-maps /home/REDACTED/work/_temp/copilot-developer-action-main/dist/index.js (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

Original prompt

---

This section details on the original issue you should resolve

<issue_title>support text2sample generation</issue_title>
<issue_description>please support this in ace-step 1.5: https://huggingface.co/RoyalCities/Foundation-1</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes support text2sample generation #862

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[ChuxiJ]

Code Review

Overall: Well-structured PR that follows existing conventions. A few items to address before merge.

Key Issues

1. Medium - Restore trailing newlines in i18n JSON files: The diff removes trailing newlines from en.json, he.json, ja.json, and zh.json. This is unnecessary and may conflict with editor configs or linters that expect POSIX-compliant final newlines.

2. Medium - Missing edge case tests: Add tests for:

   • text2sample with empty prompt AND empty sample_query
   • text2sample with both prompt and sample_query provided

3. Low - Inconsistent getattr usage: getattr(req, "task_type", "text2music") is used on line ~78, but the next line uses req.task_type directly. Since GenerationRequest always has task_type, just use req.task_type for consistency.

4. Low - _did_promote_prompt_as_query flag is fragile: Consider replacing with an inline condition check at the consumption site to reduce coupling between distant code sections.

5. Low - Brittle test assertion: test_text2sample_promotes_prompt_as_query_when_sample_query_empty checks call_args without asserting call_args is not None first, so it could silently pass with None == None.

6. Low - Duplicate emoji in Streamlit config: text2sample reuses the same 🎼 emoji as "Complete Section". Consider using a distinct emoji.

Minor Notes

• The print() for disabling use_cot_lyrics should use the logging infrastructure instead
• OpenRouter callers may get different behavior than direct API callers for empty prompt+sample_query (no "NO USER INPUT" fallback)

No blocking issues. The main items are trailing newlines and edge case test coverage.