feat: add transcription billing via SQS events#141
Open
AnthonyRonning wants to merge 2 commits intomasterfrom
Open
feat: add transcription billing via SQS events#141AnthonyRonning wants to merge 2 commits intomasterfrom
AnthonyRonning wants to merge 2 commits intomasterfrom
Conversation
WalkthroughAppends PCR snapshot entries to dev and prod history JSON files, adds optional Changes
Sequence DiagramsequenceDiagram
participant Client
participant TranscriptionHandler as Transcription<br/>Handler
participant Estimator as Duration<br/>Estimator
participant Provider as Transcription<br/>Provider
participant SQS as SQS<br/>Publisher
Client->>TranscriptionHandler: Upload/submit audio
TranscriptionHandler->>Provider: Send audio for transcription (stream/chunk)
Provider-->>TranscriptionHandler: Transcription result + provider_name
TranscriptionHandler->>Estimator: estimate_audio_duration_seconds(file_size, content_type)
Estimator-->>TranscriptionHandler: duration_seconds
TranscriptionHandler->>SQS: publish_transcription_usage_event(UsageEvent{audio_seconds,event_type,provider})
SQS-->>TranscriptionHandler: publish acknowledgment
TranscriptionHandler-->>Client: return transcription result (includes provider)
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing touches
🧪 Generate unit tests (beta)
Comment |
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
Greptile OverviewGreptile SummaryThis PR implements transcription billing by publishing SQS events with audio duration estimates. The implementation:
The duration estimation uses reasonable bitrate averages for common audio formats (MP3: 20KB/s, WAV: 176KB/s, OGG: 14KB/s, FLAC: 88KB/s, AAC: 20KB/s). The billing server will handle cost calculation based on the estimated Key observation: The implementation tracks only the first provider used when multiple chunks are processed via different providers (lines 1292-1300). This limitation was already noted in previous review threads. Confidence Score: 4/5
Important Files Changed
Sequence DiagramsequenceDiagram
participant Client
participant OpenAI Handler
participant AudioSplitter
participant Retry Logic
participant Provider (Tinfoil/Continuum)
participant SQS Publisher
participant Billing Server
Client->>OpenAI Handler: POST /v1/audio/transcriptions
OpenAI Handler->>OpenAI Handler: decode base64 audio
OpenAI Handler->>OpenAI Handler: validate file size
OpenAI Handler->>AudioSplitter: split audio into chunks
AudioSplitter-->>OpenAI Handler: return chunks
par Process chunks in parallel
OpenAI Handler->>Retry Logic: send_transcription_with_retries(chunk 0)
Retry Logic->>Provider (Tinfoil/Continuum): try primary provider
Provider (Tinfoil/Continuum)-->>Retry Logic: response
Retry Logic-->>OpenAI Handler: (response, provider_name)
and
OpenAI Handler->>Retry Logic: send_transcription_with_retries(chunk 1)
Retry Logic->>Provider (Tinfoil/Continuum): try primary/fallback
Provider (Tinfoil/Continuum)-->>Retry Logic: response
Retry Logic-->>OpenAI Handler: (response, provider_name)
end
OpenAI Handler->>OpenAI Handler: merge transcription results
OpenAI Handler->>OpenAI Handler: estimate_audio_duration_seconds()
OpenAI Handler->>SQS Publisher: publish_transcription_usage_event()
Note over SQS Publisher: event_type: "transcription"<br/>audio_seconds: estimated duration<br/>provider_name: first successful provider
SQS Publisher->>Billing Server: send UsageEvent to SQS queue
OpenAI Handler->>Client: return encrypted transcription
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
AnthonyRonning
force-pushed
the
feat/transcription-billing
branch
from
97d55df to
eaff978
Compare
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
src/web/openai.rs
Outdated
| estimated_cost: BigDecimal::from(0), | ||
| chat_time: Utc::now(), | ||
| is_api_request, | ||
| provider_name: "continuum".to_string(), |
There was a problem hiding this comment.
provider_name is hardcoded to "continuum" but transcriptions can be fulfilled by either Tinfoil or Continuum providers (see send_transcription_with_retries). Consider tracking which provider actually fulfilled the request, similar to how chat completions track successful_provider (line 633)
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!
Prompt To Fix With AI
This is a comment left during a code review.
Path: src/web/openai.rs
Line: 968:968
Comment:
`provider_name` is hardcoded to "continuum" but transcriptions can be fulfilled by either Tinfoil or Continuum providers (see `send_transcription_with_retries`). Consider tracking which provider actually fulfilled the request, similar to how chat completions track `successful_provider` (line 633)
<sub>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!</sub>
How can I resolve this? If you propose a fix, please make it concise.
Adds audio duration estimation based on file size and content type, then publishes SQS events with event_type='transcription' and audio_seconds for the billing server to process. - Tracks actual provider (tinfoil/continuum) that fulfilled request - Publishes asynchronously via tokio::spawn (non-blocking) - Rounds up to nearest second Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
AnthonyRonning
force-pushed
the
feat/transcription-billing
branch
from
eaff978 to
ab17c23
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
src/web/openai.rs (1)
1271-1285:⚠️ Potential issue | 🟡 MinorFix formatting to pass CI.
The pipeline indicates
cargo fmtcheck failed in this region. Runcargo fmt --allto fix the formatting issues.
Comment on lines
+1292
to
+1300
| let mut successful_provider = String::new(); | ||
| for result in results { | ||
| match result { | ||
| Ok(r) => successful_results.push(r), | ||
| Ok((index, value, provider)) => { | ||
| successful_results.push((index, value)); | ||
| // Track the provider (use the first one, or could use the last) | ||
| if successful_provider.is_empty() { | ||
| successful_provider = provider; | ||
| } |
There was a problem hiding this comment.
when multiple chunks succeed via different providers (e.g., chunk 0 via Tinfoil, chunk 1 via Continuum), only the first provider is tracked for billing. consider tracking all providers or using the most common one
Prompt To Fix With AI
This is a comment left during a code review.
Path: src/web/openai.rs
Line: 1292:1300
Comment:
when multiple chunks succeed via different providers (e.g., chunk 0 via Tinfoil, chunk 1 via Continuum), only the first provider is tracked for billing. consider tracking all providers or using the most common one
How can I resolve this? If you propose a fix, please make it concise.
Comment on lines
+43
to
+61
| fn estimate_audio_duration_seconds(file_size: usize, content_type: &str) -> f64 { | ||
| // Bytes per second for common audio formats (approximate averages) | ||
| let bytes_per_second: f64 = match content_type.to_lowercase().as_str() { | ||
| // MP3: typically 128-192kbps, use ~160kbps = 20KB/s | ||
| "audio/mpeg" | "audio/mp3" => 20_000.0, | ||
| // WAV: 44.1kHz, 16-bit stereo = 176.4KB/s | ||
| "audio/wav" | "audio/x-wav" | "audio/wave" => 176_400.0, | ||
| // OGG/Opus/WebM: typically ~96-128kbps = ~12-16KB/s, use 14KB/s | ||
| "audio/ogg" | "audio/opus" | "audio/webm" => 14_000.0, | ||
| // FLAC: lossless, roughly half of WAV = ~88KB/s | ||
| "audio/flac" | "audio/x-flac" => 88_000.0, | ||
| // AAC/M4A: similar to MP3, ~160kbps = 20KB/s | ||
| "audio/aac" | "audio/m4a" | "audio/mp4" | "audio/x-m4a" => 20_000.0, | ||
| // Default: assume MP3-like compression | ||
| _ => 20_000.0, | ||
| }; | ||
|
|
||
| file_size as f64 / bytes_per_second | ||
| } |
There was a problem hiding this comment.
consider adding test case for zero or very small file sizes to ensure duration estimation handles edge cases gracefully
Prompt To Fix With AI
This is a comment left during a code review.
Path: src/web/openai.rs
Line: 43:61
Comment:
consider adding test case for zero or very small file sizes to ensure duration estimation handles edge cases gracefully
How can I resolve this? If you propose a fix, please make it concise.
AnthonyRonning
force-pushed
the
feat/transcription-billing
branch
from
ab17c23 to
61bd304
Compare
Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds audio transcription billing by publishing SQS events with:
event_type: "transcription"audio_seconds: estimated duration based on file size and content typeDuration estimation lookup table:
The billing server will handle the actual cost calculation based on
audio_seconds.Includes unit tests for the duration estimation function.
Summary by CodeRabbit