Skip to content

refactor: extract codex_responses logic into dedicated adapter#12975

Closed
kshitijk4poor wants to merge 1 commit into
mainfrom
refactor/extract-codex-adapter
Closed

refactor: extract codex_responses logic into dedicated adapter#12975
kshitijk4poor wants to merge 1 commit into
mainfrom
refactor/extract-codex-adapter

Conversation

@kshitijk4poor

@kshitijk4poor kshitijk4poor commented Apr 20, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator

Summary

Extract 10 Responses API format-conversion and normalization functions from run_agent.py into agent/codex_responses_adapter.py.

run_agent.py: 12,410 → 11,845 lines (-565 lines)

What moved

Function Purpose Lines
_deterministic_call_id Deterministic tool call ID generation (cache-safe) 13
_split_responses_tool_id Composite ID splitting (call_id|response_item_id) 16
_derive_responses_function_call_id call_ → fc_ prefix conversion 29
_responses_tools Chat Completions tool schema → Responses format 22
_chat_messages_to_responses_input Message format conversion 109
_preflight_codex_input_items Input item normalization 93
_preflight_codex_api_kwargs API kwargs validation/cleaning 143
_extract_responses_message_text Text extraction from response items 16
_extract_responses_reasoning_text Reasoning extraction from response items 16
_normalize_codex_response Full response normalization 164

Design

  • All extracted functions are stateless module-level functions with zero self references
  • AIAgent methods remain as thin one-line wrappers that delegate to the adapter
  • All callers (tests, gateway, CLI) are completely unchanged
  • Brings codex_responses in line with anthropic_adapter.py and bedrock_adapter.py

Bugs fixed during testing

  • Missing uuid import — _derive_responses_function_call_id would crash on the uuid.uuid4() fallback path (both call_id and response_item_id empty/None)
  • Missing DEFAULT_AGENT_IDENTITY import — _preflight_codex_api_kwargs would crash when instructions field is empty
  • Stray @staticmethod decorators — left over from extraction, harmless on Python 3.10+ but wrong on module-level functions
  • Redundant import hashlib inside _deterministic_call_id body (already imported at module level)
  • Missing blank line spacing between method definitions in both files

Test plan

  • py_compile passes on both files
  • 231 codex/responses-related tests pass (-k 'codex or responses')
  • Full test suite: 12,959 passed (140 failed + 360 errors are pre-existing/unrelated)
  • 56 E2E tests verifying adapter functions produce identical output including edge cases (empty inputs, uuid fallback, DEFAULT_AGENT_IDENTITY fallback, reasoning dedup, etc.)

@kshitijk4poor
refactor: extract codex_responses logic into dedicated adapter
b277962 
Move 10 Responses API format-conversion and normalization functions from
run_agent.py into agent/codex_responses_adapter.py. All functions are now
stateless module-level functions with zero self references.

The AIAgent methods remain as thin one-line wrappers that delegate to the
adapter, so all callers (tests, gateway, CLI) are unchanged.

Functions extracted:
- _deterministic_call_id: deterministic tool call ID generation
- _split_responses_tool_id: composite ID splitting
- _derive_responses_function_call_id: call_ to fc_ prefix conversion
- _responses_tools: chat completions tool schema → Responses format
- _chat_messages_to_responses_input: message format conversion
- _preflight_codex_input_items: input item normalization
- _preflight_codex_api_kwargs: API kwargs validation/cleaning
- _extract_responses_message_text: text extraction from response items
- _extract_responses_reasoning_text: reasoning extraction
- _normalize_codex_response: full response normalization

This brings codex_responses in line with anthropic_adapter.py and
bedrock_adapter.py which already have their own adapter files.

run_agent.py: 12410 → 11845 lines (-565 net)
@kshitijk4poor kshitijk4poor force-pushed the refactor/extract-codex-adapter branch from be8ba8f to b277962 Compare April 20, 2026 11:07
teknium1 pushed a commit that referenced this pull request Apr 20, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
a26d116 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR #12969) that the original PR
was missing.

Based on PR #12975 by @kshitijk4poor.
@teknium1 teknium1 mentioned this pull request Apr 20, 2026
Merged
@kshitijk4poor kshitijk4poor mentioned this pull request Apr 20, 2026
Closed
teknium1 pushed a commit that referenced this pull request Apr 20, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
ff56beb 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR #12969) that the original PR
was missing.

Based on PR #12975 by @kshitijk4poor.
@teknium1

teknium1 commented Apr 20, 2026

Copy link
Copy Markdown
Contributor

Merged via PR #13018 — your extraction was cherry-picked onto current main with multimodal inline image support (PR #12969) incorporated. Your authorship is preserved in the git log. Thanks for the clean refactor!

#13018

@teknium1 teknium1 closed this Apr 20, 2026
This was referenced Apr 21, 2026
Merged
Merged
Closed
Closed
Closed
@kshitijk4poor kshitijk4poor mentioned this pull request Apr 23, 2026
Closed
25 tasks
ulasbilgen pushed a commit to ulasbilgen/hermes-adhd-agent that referenced this pull request May 1, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
8eff7b4 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR NousResearch#12969) that the original PR
was missing.

Based on PR NousResearch#12975 by @kshitijk4poor.
aj-nt pushed a commit to aj-nt/hermes-agent that referenced this pull request May 1, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
ca926a8 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR NousResearch#12969) that the original PR
was missing.

Based on PR NousResearch#12975 by @kshitijk4poor.
Luminet2023 pushed a commit to Luminet2023/hermes-agent that referenced this pull request May 1, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
a7618e1 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR NousResearch#12969) that the original PR
was missing.

Based on PR NousResearch#12975 by @kshitijk4poor.
02356abc pushed a commit to 02356abc/hermes-agent that referenced this pull request May 14, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
3fe7df4 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR NousResearch#12969) that the original PR
was missing.

Based on PR NousResearch#12975 by @kshitijk4poor.
This was referenced May 16, 2026
Closed
Closed
gweeteve pushed a commit to gweeteve/hermes-agent that referenced this pull request Jun 2, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
49d4621 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR NousResearch#12969) that the original PR
was missing.

Based on PR NousResearch#12975 by @kshitijk4poor.
Egavasyug pushed a commit to Egavasyug/hermes-agent that referenced this pull request Jun 10, 2026
@kshitijk4poor @teknium1
refactor: extract codex_responses logic into dedicated adapter
d1cca3e 
Extract 12 Codex Responses API format-conversion and normalization functions
from run_agent.py into agent/codex_responses_adapter.py, following the
existing pattern of anthropic_adapter.py and bedrock_adapter.py.

run_agent.py: 12,550 → 11,865 lines (-685 lines)

Functions moved:
- _chat_content_to_responses_parts (multimodal content conversion)
- _summarize_user_message_for_log (multimodal message logging)
- _deterministic_call_id (cache-safe fallback IDs)
- _split_responses_tool_id (composite ID splitting)
- _derive_responses_function_call_id (fc_ prefix conversion)
- _responses_tools (schema format conversion)
- _chat_messages_to_responses_input (message format conversion)
- _preflight_codex_input_items (input validation)
- _preflight_codex_api_kwargs (API kwargs validation)
- _extract_responses_message_text (response text extraction)
- _extract_responses_reasoning_text (reasoning extraction)
- _normalize_codex_response (full response normalization)

All functions are stateless module-level functions. AIAgent methods remain
as thin one-line wrappers. Both module-level helpers are re-exported from
run_agent.py for backward compatibility with existing test imports.

Includes multimodal inline image support (PR NousResearch#12969) that the original PR
was missing.

Based on PR NousResearch#12975 by @kshitijk4poor.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@kshitijk4poor @teknium1