fix: Gracefully handle missing speech profiles bucket

[neooriginal]

Fully tested e2e

This pull request improves the robustness and error handling of all speech profile storage operations in backend/utils/other/storage.py. It introduces a helper function to safely access the speech profiles bucket, adds warnings and explicit error messages when the bucket is not configured, and ensures that all relevant functions gracefully handle missing configuration instead of failing unexpectedly.

Speech profile bucket access and error handling:

• Added the _get_speech_profiles_bucket helper function to centralize logic for accessing the speech_profiles_bucket, including warning logging and raising errors when required. All speech profile functions now use this helper to prevent failures when the bucket is not configured.
• Updated all speech profile-related functions (such as upload_profile_audio, get_user_has_speech_profile, get_profile_audio_if_exists, etc.) to use _get_speech_profiles_bucket and handle missing bucket configuration gracefully by returning early or raising informative exceptions. [1] [2] [3] [4] [5] [6] [7] [8]
• Added explicit error raising in download_speech_profile_bytes and safe return in delete_speech_profile_blob if the speech profiles bucket is not configured, improving clarity and preventing silent failures. [1] [2]

Environment variable handling:

• Improved the parsing of the BUCKET_SPEECH_PROFILES environment variable to correctly handle empty or whitespace-only values, ensuring the configuration is robust against misconfiguration.

[Copilot]

Pull request overview

This PR improves robustness of speech profile storage operations by centralizing access to the speech profiles GCS bucket and handling missing bucket configuration more gracefully across read/write/delete paths.

Changes:

• Parse BUCKET_SPEECH_PROFILES more robustly (treat empty/whitespace-only values as unset).
• Add _get_speech_profiles_bucket() helper to centralize bucket resolution + one-time warning + optional hard failure.
• Update speech-profile-related functions to early-return (or raise) when the bucket is not configured.

---

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

[neooriginal]

@copilot apply changes based on the comments in this thread

[greptile-apps]

Greptile Summary

This PR improves robustness for deployments where BUCKET_SPEECH_PROFILES is not configured by centralizing bucket access through a _get_speech_profiles_bucket() helper. Read/list/delete functions now return early with safe empty values, while upload functions raise RuntimeError to prevent silent failures.

Key changes:

• New _get_speech_profiles_bucket(required: bool) helper that emits a one-time warning and returns None (or raises RuntimeError) when the bucket env var is absent.
• All speech profile functions updated to use the helper instead of calling storage_client.bucket() directly.
• Env-var parsing hardened: (os.getenv('BUCKET_SPEECH_PROFILES') or '').strip() or None correctly normalises whitespace-only values to None.
• Issue: download_speech_profile_bytes raises BlobNotFound (google.cloud.exceptions.NotFound) when the bucket is unconfigured. Callers like speaker_sample_migration.py specifically catch NotFound to mean "blob is gone — delete Firestore reference", so a missing bucket would silently mark all speech sample paths for Firestore deletion. Using RuntimeError (as the rest of this PR does) would land in the broader except Exception handler instead, preventing unintended data loss.

Confidence Score: 4/5

• Safe to merge after fixing the BlobNotFound exception type in download_speech_profile_bytes; the rest of the error-handling refactor is well-structured
• One P1 finding: raising BlobNotFound instead of RuntimeError in download_speech_profile_bytes can trigger the "delete from Firestore" branch in the migration path when the bucket is simply unconfigured, risking metadata loss. All other changes are correct and an improvement over the prior crash-on-None behaviour.
• backend/utils/other/storage.py — specifically the download_speech_profile_bytes function around line 1138

Important Files Changed

Filename	Overview
backend/utils/other/storage.py	Adds _get_speech_profiles_bucket helper and gracefully handles missing BUCKET_SPEECH_PROFILES across all speech profile functions; raises BlobNotFound in download_speech_profile_bytes which conflicts with caller exception handling in the migration path

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Speech profile function called] --> B{_get_speech_profiles_bucket}
    B -->|bucket configured| C[Return storage_client.bucket obj]
    B -->|not configured + required=False| D[Log one-time warning\nReturn None]
    B -->|not configured + required=True| E[Log one-time warning\nRaise RuntimeError]
    C --> F[Proceed with GCS operation]
    D --> G{Function type}
    G -->|get_user_has_speech_profile| H[return False]
    G -->|get_profile_audio_if_exists| I[return None]
    G -->|get_additional_profile_recordings| J[return empty list]
    G -->|delete_* functions| K[return early / no-op]
    E --> L[Caller receives RuntimeError\n→ HTTP 500]

    M[download_speech_profile_bytes] --> N{speech_profiles_bucket set?}
    N -->|No| O[raise BlobNotFound ⚠️\nSame type as 'blob missing']
    N -->|Yes| P[download_blob_bytes]
    O --> Q[Migration catches NotFound\n→ marks for Firestore deletion ⚠️]

Loading

_{Reviews (1): Last reviewed commit: "Merge branch 'BasedHardware:main' into m..." | Re-trigger Greptile}