fix(inference-billing): fix inference billing when stream is true via API, add drag-and-drop functionality to deployed chat

[waleedlatif1]

Summary

fix inference billing when stream is true via API, add drag-and-drop functionality to deployed chat

Type of Change

• Bug fix
• New feature

Testing

Tested manually.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Preview	Comments	Updated (UTC)
docs	Skipped			Oct 12, 2025 2:55am

[greptile-apps]

Greptile Overview

Summary

Fixed inference billing when stream=true by deferring logging completion until after tokenization, and added drag-and-drop file upload to deployed chat interface.

Key Changes:

• Billing Fix: Deferred loggingSession.safeComplete() call in streaming execution flow to ensure cost/token data from processStreamingBlockLogs() is included in trace spans before logging
• Drag-and-Drop: Added drag event handlers (onDragEnter, onDragOver, onDragLeave, onDrop) to chat input with visual feedback via dragCounter state and conditional styling
• Added skipLoggingComplete parameter to executeWorkflow() options and _streamingMetadata field to ExecutionResult for passing logging session through streaming pipeline

The billing fix addresses a timing issue where streaming executions logged incomplete cost data because tokenization happened after logging completion.

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• Both features are well-implemented with clear separation of concerns. The billing fix properly defers logging until after tokenization without affecting non-streaming flows. The drag-and-drop implementation uses standard browser APIs with proper event handling and guards against streaming state. No breaking changes or security issues detected.
• No files require special attention

Important Files Changed

File Analysis

Filename	Score	Overview
apps/sim/lib/workflows/streaming.ts	5/5	Added deferred logging completion after tokenization to ensure cost data is calculated before logging
apps/sim/app/api/workflows/[id]/execute/route.ts	5/5	Added skipLoggingComplete parameter and streaming metadata to defer logging completion until after tokenization
apps/sim/executor/types.ts	5/5	Added _streamingMetadata to ExecutionResult for passing logging session data to streaming handler
apps/sim/app/chat/components/input/input.tsx	5/5	Implemented drag-and-drop file upload with visual feedback using drag counter and event handlers

Sequence Diagram

sequenceDiagram
participant Client
participant API as /api/workflows/[id]/execute
participant Execute as executeWorkflow()
participant Streaming as createStreamingResponse()
participant Tokenize as processStreamingBlockLogs()
participant TraceSpans as buildTraceSpans()
participant LogSession as loggingSession
Client->>API: POST with stream=true
API->>Streaming: createStreamingResponse()
Streaming->>Execute: executeWorkflow(skipLoggingComplete: true)
Execute->>Execute: Run workflow execution
Execute->>Execute: Build initial trace spans
Note over Execute,LogSession: Skip logging completion (skipLoggingComplete=true)
Execute->>Execute: Attach _streamingMetadata (loggingSession, processedInput)
Execute-->>Streaming: Return result with _streamingMetadata
Streaming->>Streaming: Update result.logs with streamed content
Streaming->>Tokenize: processStreamingBlockLogs(logs, streamedContent)
Tokenize->>Tokenize: Calculate tokens & cost for streamed content
Tokenize-->>Streaming: Updated logs with cost data
Streaming->>TraceSpans: buildTraceSpans(result)
TraceSpans-->>Streaming: traceSpans with cost data
Streaming->>LogSession: safeComplete(traceSpans, finalOutput)
Note over LogSession: Now includes accurate billing/cost data
LogSession-->>Streaming: Logging complete
Streaming->>Streaming: Clear _streamingMetadata
Streaming-->>Client: Stream response with accurate billing

Loading

_{4 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}