fix(responses): parse standard SSE framing

[0YHR0]

Description

Make all existing translator SSE parsers accept standard event framing while they inspect streamed events for Gateway observability and protocol translation.

Root Cause

Several parsers required a space in data: lines, although SSE permits data: without that optional space. The Responses and Speech parsers also recognized only LF event boundaries, although SSE permits CRLF boundaries.

Impact

Client response bytes remain unchanged for passthrough translators. The missing parsing affects Gateway's extracted token usage, resolved response model, tracing chunks, and translated streams.

Fix

Use shared helpers to parse data: fields and LF or CRLF event boundaries across the affected stream parsers.

Related Issue

Fixes #2463

Testing

• go test ./internal/translator -count=1

[dosubot]

📄 Knowledge review

Dosu skipped reviewing this PR because your organization has used its 200 included credits for the month. Your usage will reset on 2026-08-01. To have Dosu review this PR before then, ask your organization admin to upgrade to a pro account.

---

[codecov-commenter]

Codecov Report

❌ Patch coverage is 93.75000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 84.94%. Comparing base (e989693) to head (0ebee32).

Files with missing lines	Patch %	Lines
internal/translator/openai_responses.go	93.75%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2464   +/-   ##
=======================================
  Coverage   84.93%   84.94%           
=======================================
  Files         154      154           
  Lines       22423    22433   +10     
=======================================
+ Hits        19046    19056   +10     
  Misses       2222     2222           
  Partials     1155     1155           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[kanurag94]

LGTM
One thing I noticed is that this bug exists in other translators too.

- openai_completions.go:157
- openai_openai.go:195
- openai_speech.go:154
- openai_transcription.go:137
- anthropic_anthropic.go:146
- openai_gcpvertexai.go:318

They split on a single \n and not \n\n. A backend sending data:{} will break them the same way.
Is it possible to create a followup/ or fix them here (create a shared helper).

[0YHR0]

@kanurag94 Thanks, I took the shared-helper approach in 4c81fc2. It now accepts both data field forms across the existing stream parsers, and shares LF/CRLF event-boundary handling for the parsers that buffer full events. I also added helper coverage and ran go test ./internal/translator -count=1. Could you please take another look?