From e09ccbaac956b82b5167a4a6c63ddd56624b8a02 Mon Sep 17 00:00:00 2001
From: nesquena-hermes <nesquena+hermes@gmail.com>
Date: Fri, 7 Aug 2026 19:53:02 -0700
Subject: [PATCH] Release exp-v0.52.174: regression coverage for empty
 tool_calls drop on mirror path (#6803) (#6844)

* test(streaming): cover empty tool_calls drop in _api_safe_message_positions (#6796)

* Release exp-v0.52.174: regression coverage for empty tool_calls drop on mirror path (#6803)

---------

Co-authored-by: Webtecnica Bot <bot@webtecnica.com.br>
Co-authored-by: nesquena-hermes <nesquena-hermes@users.noreply.github.com>
---
 CHANGELOG.md                               |  2 ++
 tests/test_context_history_sanitization.py | 41 ++++++++++++++++++++++
 2 files changed, 43 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index a2951168..56f424d8 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,8 @@
 
 - **Internal: clarified the transcript-virtualization boot comment (opt-in default retained).** The comment above the boot-time `_virtualizeTranscript` apply now records that the #4346 Phase B footer-jitter fix resolved the original scroll-up flicker root cause, while virtualization stays opt-in (default OFF) until battle-tested further. Zero behavior change — the apply predicate, settings-load fallback, checkbox, and server default all remain opt-in/OFF and in agreement. Thanks @webtecnica. (#6318, #6155)
 
+- **Internal: added regression coverage for the empty-`tool_calls` drop on the metadata-restoration mirror path.** `_api_safe_message_positions()` must apply the same `tool_calls: []` drop as the main API sanitizer (#5737) so metadata restoration never treats a `tool_calls: []` row as a distinct message; a new test pins that empty arrays lose the key while populated tool-call chains (and their tool results) survive untouched. Test-only, no production change. Thanks @webtecnica. (#6803, #6796)
+
 - **The busy-time send behavior is now called "Default message mode," and new installs default to Steer.** The Settings → Preferences control formerly labeled "Busy input mode" is renamed to "Default message mode," and a fresh install now defaults to **Steer** (inject a mid-turn correction without interrupting) instead of Queue. Your existing choice is preserved — if you ever saved settings, your current mode (Queue/Interrupt/Steer) is migrated as-is and unchanged; only never-configured installs pick up the new Steer default. The saved preference still survives a reload or a brief server outage (the localStorage mirror from the previous release is intact). Thanks @rodboev. (#5162, #5145)
 
 ### Added
diff --git a/tests/test_context_history_sanitization.py b/tests/test_context_history_sanitization.py
index f412e425..a65b20fa 100644
--- a/tests/test_context_history_sanitization.py
+++ b/tests/test_context_history_sanitization.py
@@ -7,6 +7,7 @@ from types import SimpleNamespace
 from unittest.mock import patch
 
 from api.streaming import (
+    _api_safe_message_positions,
     _compact_image_parts_for_persistence,
     _compact_session_image_parts_for_persistence,
     _restore_reasoning_metadata,
@@ -309,6 +310,46 @@ def test_sanitize_messages_drops_empty_tool_calls_array():
             "no message may ship an empty tool_calls array"
 
 
+def test_api_safe_message_positions_drops_empty_tool_calls_array():
+    """#5737 mirror path: _api_safe_message_positions must apply the same
+    empty-tool_calls drop as _sanitize_messages_for_api, so metadata
+    restoration (which aligns rows against these positions) never treats a
+    tool_calls: [] row as a different message. Empty array -> key removed;
+    populated tool_calls chain -> preserved."""
+    messages = [
+        {"role": "user", "content": "hi"},
+        # Empty tool_calls: [] must be dropped (would 400 on strict providers).
+        {"role": "assistant", "content": "thinking out loud", "tool_calls": []},
+        {"role": "user", "content": "go on"},
+        # Populated tool_calls (+ its tool result) must be preserved untouched.
+        {
+            "role": "assistant",
+            "content": "",
+            "tool_calls": [
+                {"type": "function", "id": "call-9",
+                 "function": {"name": "terminal", "arguments": "{}"}},
+            ],
+        },
+        {"role": "tool", "tool_call_id": "call-9", "content": "ok"},
+    ]
+
+    positions = _api_safe_message_positions(messages)
+
+    sanitized = [msg for _idx, msg in positions]
+    # The empty-tool_calls assistant message survives, but WITHOUT the key.
+    empty_asst = next(m for m in sanitized if m.get("content") == "thinking out loud")
+    assert empty_asst["role"] == "assistant"
+    assert "tool_calls" not in empty_asst, "empty tool_calls: [] must be dropped"
+    # The populated chain is preserved.
+    populated = next(m for m in sanitized if m.get("role") == "assistant" and m.get("tool_calls"))
+    assert populated["tool_calls"][0]["id"] == "call-9"
+    assert any(m.get("tool_call_id") == "call-9" for m in sanitized)
+    # No row that reaches the API carries an empty tool_calls array.
+    for _idx, msg in positions:
+        assert not ("tool_calls" in msg and not msg["tool_calls"]), \
+            "no API-safe position may ship an empty tool_calls array"
+
+
 def test_gateway_conversation_history_no_oob():
     from api.config import STREAM_PARTIAL_TEXT, STREAM_REASONING_TEXT
     from api.gateway_chat import _STREAM_RUN_IDS, _run_gateway_runs_api_streaming