images/

[sergei-bronnikov]

https://bugtracker.codiodev.com/issue/codio-17369/Upgrade-BricksLLM-to-Support-OpenAI-Image-Generation

Summary by CodeRabbit

• New Features
  • Added image response metadata tracking to capture token usage, quality, and image dimensions.
  • Introduced support for new GPT image generation models (gpt-image-1, gpt-image-1.5, and variants).
  • Enhanced image cost estimation to use metadata-based calculations when available, with fallback to resolution-based pricing.

[destitutus]

@coderabbitai full review

[coderabbitai]

✅ Actions performed

Full review triggered.

[coderabbitai]

Walkthrough

Added support for capturing and leveraging OpenAI image response metadata to enable more accurate cost estimation. The changes introduce new types to represent image usage data, extend cost estimation interfaces to accept metadata, and implement logic to extract and propagate metadata through HTTP proxying and event logging.

Changes

Cohort / File(s)	Summary
Type Definitions internal/provider/openai/types.go	Added three new types: ImageResponseMetadata, ImageResponseUsage, and ImageResponseInputTokensDetails to represent image API response metadata including token usage and image dimensions.
Cost Estimation internal/provider/openai/cost.go, internal/message/handler.go	Extended image cost estimation to accept metadata parameter; added token-based cost maps for image models (gpt-image-1, gpt-image-1.5, etc.); implemented estimateImageByMetadata function for metadata-driven cost calculation with fallback to resolution-based pricing.
Metadata Extraction & Propagation internal/server/web/proxy/image.go, internal/server/web/proxy/proxy.go, internal/server/web/proxy/middleware.go	Added helpers to parse image response metadata from JSON, store in request context, and propagate to event logging; integrated metadata extraction into three image endpoints (generations, edits, variations).
Event Tracking internal/event/event_with_request_and_response.go	Added ImageResponseMetadata field to EventWithRequestAndContent struct to carry metadata through the event system.

Sequence Diagram(s)

sequenceDiagram
    participant Client
    participant Proxy
    participant OpenAI as OpenAI API
    participant Handler
    participant CostEstimator
    participant EventLog

    Client->>Proxy: POST /images/generations
    Proxy->>OpenAI: Forward request
    OpenAI->>Proxy: JSON response with metadata
    Proxy->>Proxy: Extract metadata (imageResponseMetadataFromBytes)
    Proxy->>Proxy: Store in context (setCtxImageResponseMetadata)
    Proxy->>Client: Return response
    
    Proxy->>Handler: Process event with metadata
    Handler->>CostEstimator: EstimateImagesCost(model, quality, resolution, metadata)
    
    alt Metadata available
        CostEstimator->>CostEstimator: estimateImageByMetadata (text + image tokens)
    else Fallback
        CostEstimator->>CostEstimator: Resolution-based pricing
    end
    
    CostEstimator->>Handler: Cost calculated
    Handler->>EventLog: Store event with ImageResponseMetadata
    EventLog->>EventLog: Log complete

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.
Title check	❓ Inconclusive	The pull request title 'images/' is too vague and does not clearly convey the main change, using only a generic prefix without descriptive content.	Use a more descriptive title that captures the main change, such as 'Add OpenAI image generation cost estimation and metadata tracking' or 'Extend support for OpenAI image operations with token-based cost calculation'.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch 17369_extend_support_image

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@internal/server/web/proxy/image.go`:
- Around line 126-143: The setter and getter use mismatched types:
setCtxImageResponseMetadata stores imageResponse.Usage (ImageResponseUsage) but
getCtxImageResponseMetadata type-asserts to openai.ImageResponseMetadata,
causing the assertion to fail; fix by storing the full metadata object in
setCtxImageResponseMetadata (store imageResponse itself) and update the getter's
type assertion to expect a *openai.ImageResponseMetadata (or the concrete value
you store) so getCtxImageResponseMetadata returns the stored metadata pointer
correctly.

🧹 Nitpick comments (1)

internal/provider/openai/cost.go (1)

472-506: Metadata-based cost estimation looks good, but silently ignores missing cost entries.

The function correctly calculates costs from text tokens, image tokens, and output tokens. However, the ok return values from cost map lookups (lines 486, 494, 502) are ignored. If a model is in imageModelsWithTokensCost but missing from one of the token cost maps, the cost for that component will silently be 0.

Consider adding validation or logging when a model passes the initial check but is missing from a specific cost map.

♻️ Optional: Add logging for missing cost entries

 	textInputCost, _ := textInputCostMap[model]
+	// Note: If model is missing, textInputCost defaults to 0
 	totalCost += (float64(textInputTokens) / 1000) * textInputCost

Or return an error if the lookup fails:

-	textInputCost, _ := textInputCostMap[model]
+	textInputCost, ok := textInputCostMap[model]
+	if !ok {
+		return 0, fmt.Errorf("model %s not found in prompt cost map", model)
+	}