cecli-dev 1.8.0__tar.gz → 1.8.1rc2__tar.gz

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cecli-dev
-Version: 1.8.0
+Version: 1.8.1rc2
 Summary: we also can't pronounce cecli
 Project-URL: Homepage, https://github.com/dwash96/cecli
 Project-URL: Documentation, https://github.com/dwash96/cecli

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '1.8.0'
-__version_tuple__ = version_tuple = (1, 8, 0)
+__version__ = version = '1.8.1rc2'
+__version_tuple__ = version_tuple = (1, 8, 1, 'rc2')
 
-__commit_id__ = commit_id = 'gbf30ff4c0'
+__commit_id__ = commit_id = 'g9c5114f1f'

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/coders/base_coder.py

@@ -2253,6 +2253,31 @@ class Coder(metaclass=UsageMeta):
                 )
                 return
 
+        # Compaction can only shrink the conversation buckets counted above.
+        # If the fixed context (files added to the chat, repo map, system
+        # prompt, rules, tool schemas) already meets the limit on its own,
+        # summarizing cannot bring the context under the limit. Bail out with
+        # guidance instead of churning every turn, which is what a large file
+        # added to the chat otherwise causes.
+        non_compactable_tokens = all_tokens - compactable_tokens
+
+        if (
+            self.context_compaction_max_tokens
+            and non_compactable_tokens >= self.context_compaction_max_tokens
+        ):
+            if getattr(self, "_compaction_noop_warning_time", 0) + 300 < time.time():
+                self._compaction_noop_warning_time = time.time()
+                self.io.tool_warning(
+                    "Compaction cannot reduce the context: the fixed context (files in the"
+                    f" chat, repo map, prompts) is already ~{non_compactable_tokens:,}"
+                    f" tokens, at or above the ~{self.context_compaction_max_tokens:,} token"
+                    " compaction limit."
+                )
+                self.io.tool_output("- Use /drop to remove unneeded files from the chat")
+                self.io.tool_output("- Use /clear to clear the chat history")
+
+            return
+
         message_tokens = done_tokens + cur_tokens
         file_tokens = diff_tokens + file_context_tokens
         combined_tokens = done_tokens + cur_tokens + diff_tokens + file_context_tokens
@@ -3996,7 +4021,7 @@ class Coder(metaclass=UsageMeta):
         if (
             not len(self.partial_response_content)
             and not len(self.partial_response_tool_calls)
-            and not len(self.partial_response_reasoning_content)
+            and not _is_meaningful_reasoning(self.partial_response_reasoning_content)
         ):
             self.empty_response = True
             return
@@ -4126,7 +4151,8 @@ class Coder(metaclass=UsageMeta):
 
                             text += reasoning_content
                             self.got_reasoning_content = True
-                            received_content = True
+                            if _is_meaningful_reasoning(reasoning_content):
+                                received_content = True
 
                         self.token_profiler.on_token()
                         self.io.update_spinner_suffix(reasoning_content)
@@ -4202,7 +4228,14 @@ class Coder(metaclass=UsageMeta):
             self.io.tool_warning("Execution stopped by on message hook")
             return
 
-        if not received_content and len(self.partial_response_tool_calls) == 0:
+        # Treat the response as empty when nothing was received, or when the
+        # only thing received was reasoning made entirely of non-alphanumeric
+        # characters (e.g. moonshotai/kimi-k3 returning "!!!!").
+        if (
+            not received_content
+            and len(self.partial_response_tool_calls) == 0
+            and not _is_meaningful_reasoning(self.partial_response_reasoning_content)
+        ):
             self.empty_response = True
             return
 
@@ -5391,3 +5424,15 @@ def _first_usage_tokens(usage: object, paths: list[str], default: int = 0) -> in
         if value is not None:
             return value
     return default
+
+
+def _is_meaningful_reasoning(text):
+    """Return True if reasoning text contains at least one alphanumeric character.
+
+    Some providers (e.g. moonshotai/kimi-k3) occasionally return completions
+    with empty ``content`` and a ``reasoning_content`` made entirely of
+    punctuation (e.g. ``"!!!!"``). Those responses are effectively empty, so
+    the empty-response detector only lets reasoning count as response
+    content when it holds at least one alphanumeric character.
+    """
+    return bool(text) and any(ch.isalnum() for ch in text)

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/helpers/llms/domains/bedrock.py

@@ -194,7 +194,7 @@ def bedrock_endpoint(resolved: Dict[str, Any]) -> str:
         )
 
     route = resolved["route"]
-    return f"https://bedrock-runtime.{region}.amazonaws.com/model/{route}/converse"
+    return f"{resolved['api_base']}/model/{route}/converse"
 
 
 def _signed_headers(

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/helpers/llms/litellm_compat.py

@@ -19,6 +19,7 @@ from __future__ import annotations
 
 import asyncio
 import json
+import math
 import os
 import re
 import warnings
@@ -1096,15 +1097,11 @@ class _LiteLLMFacade:
         """Estimate tokens for ``text`` without loading a BPE tokenizer.
 
         tiktoken's cl100k_base/o200k_base encodings cost ~30-95MB of RSS the
-        first time they are loaded, so we approximate instead: roughly one token
-        per four ASCII characters (the classic rule of thumb), plus one token per
-        non-ASCII character (CJK/emoji tokenize at about one token per char), and
-        never fewer tokens than there are whitespace-delimited word/punctuation
-        runs (dense code with short tokens is otherwise undercounted).
-
-        The estimate lands within ~±30% of cl100k_base on code and prose, which
-        is plenty for context-window checks, cost display, and file-size
-        warnings.
+        first time they are loaded, so we approximate instead with a regex
+        pre-tokenizer patterned on OpenAI's o200k_base (see
+        ``_estimate_token_count``). The estimate lands within ~±30% of
+        cl100k_base on code and prose, which is plenty for context-window
+        checks, cost display, and file-size warnings.
 
         Returns a ``range`` whose length is the estimated token count; the
         individual values are dummy ids (callers only use ``len()``/iteration).
@@ -1244,23 +1241,41 @@ __all__ = ["litellm"]
 # Token-count estimator (tiktoken-free)
 # ---------------------------------------------------------------------------
 
-_WORD_OR_PUNCT_RE = re.compile(r"\w+|[^\w\s]")
-_NON_ASCII_RE = re.compile(r"[^\x00-\x7f]")
+# Standard ``re`` approximation of OpenAI's o200k_base pre-tokenizer pattern.
+# The letter class is Unicode-aware (``[^\W\d_]``) so non-Latin scripts are
+# counted rather than dropped.
+_O200K_APPROX_PATTERN = re.compile(
+    r"""'(?:[sdmt]|ll|ve|re)| ?[^\W\d_]+| ?\d{1,3}| ?[^\s\w]+|\s+""", re.UNICODE
+)
+
+#: Bytes charged per token for chunks too long to be a single token, calibrated
+#: against o200k_base (~5 bytes/token on code, prose and JSON).
+_CHARS_PER_TOKEN = 5.0
 
 
 def _estimate_token_count(text: str) -> int:
     """Estimate the number of tokens in ``text`` without a BPE tokenizer.
 
-    Uses the classic ~4 chars/token rule of thumb, lifted for non-ASCII text
-    (CJK/emoji tokenize at roughly one token per character) and floored by the
-    number of word/punctuation runs so dense code with many short tokens isn't
-    undercounted.
+    Splits ``text`` with a regex approximation of OpenAI's o200k_base
+    pre-tokenizer, then accumulates a fractional token weight per chunk:
+    whitespace runs collapse to a single token, numbers and short chunks
+    (<= 4 bytes) count as one, and longer chunks are charged at roughly
+    ``_CHARS_PER_TOKEN`` bytes per token. The total is rounded up and floored
+    at one token for any non-empty input.
     """
 
     if not text:
         return 0
 
-    units = len(_WORD_OR_PUNCT_RE.findall(text))
-    non_ascii = len(_NON_ASCII_RE.findall(text))
+    tokens = 0.0
+    for chunk in _O200K_APPROX_PATTERN.findall(text):
+        byte_len = len(chunk.encode("utf-8"))
+
+        if chunk.isspace():
+            tokens += 1.0
+        elif chunk.strip().isdigit() or byte_len <= 4:
+            tokens += 1.0
+        else:
+            tokens += max(1.0, byte_len / _CHARS_PER_TOKEN)
 
-    return max(1, len(text) // 4, units) + non_ascii
+    return max(1, math.ceil(tokens - 1e-9))

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/helpers/workflows/config.py

@@ -10,8 +10,10 @@ Resolution order, highest precedence first:
    ``None``, ``''``, ``'{}'``, ``'[]'`` and ``'null'`` count as *absent* so that
    lower-precedence sources still load.
 2. ``<git_root or cwd>/.cecli.workflows.yml`` / ``.cecli.workflows.yaml``.
-3. ``merged_config["workflows"]`` — the deep-merged ``.cecli.conf.yml`` value.
-4. ``~/.cecli/workflows/*.yml`` / ``*.yaml`` — globbed in stable sorted order.
+3. ``<git_root or cwd>/.cecli/workflows/*.yml`` / ``*.yaml`` — globbed in stable sorted order.
+4. ``<git_root or cwd>/.agents/workflows/*.yml`` / ``*.yaml`` — globbed in stable sorted order.
+5. ``merged_config["workflows"]`` — the deep-merged ``.cecli.conf.yml`` value.
+6. ``~/.cecli/workflows/*.yml`` / ``*.yaml`` — globbed in stable sorted order.
 
 Each source may hold a single workflow mapping or a ``workflows: [...]`` list.
 On a duplicate ``name`` the higher-precedence definition wins and the loser is
@@ -41,6 +43,8 @@ WORKFLOW_FILENAMES = (".cecli.workflows.yml", ".cecli.workflows.yaml")
 
 WORKFLOW_DIR = Path(".cecli") / "workflows"
 
+AGENTS_WORKFLOW_DIR = Path(".agents") / "workflows"
+
 # Values argparse/deep-merge normalization leaves behind when nothing was set.
 _ABSENT_STRINGS = ("", "{}", "[]", "null", "~")
 
@@ -93,6 +97,8 @@ def resolve_workflow_configs(
         ordered += _entries_from_arg(str(config_arg))
 
     ordered += _entries_from_local(git_root)
+    ordered += _entries_from_local_dir(git_root)
+    ordered += _entries_from_local_agents_dir(git_root)
 
     if isinstance(merged_config, dict):
         ordered += _entries(merged_config.get("workflows"), None, _CONFIG_LABEL)
@@ -289,9 +295,28 @@ def _entries_from_local(git_root: Optional[Any]) -> List[_Discovered]:
     return []
 
 
+def _entries_from_local_dir(git_root: Optional[Any]) -> List[_Discovered]:
+    """Resolve a repo-local ``.cecli/workflows/*.yml|*.yaml`` directory."""
+    base = Path(git_root).expanduser() if git_root else Path.cwd()
+
+    return _entries_from_dir(base / WORKFLOW_DIR)
+
+
+def _entries_from_local_agents_dir(git_root: Optional[Any]) -> List[_Discovered]:
+    """Resolve a repo-local ``.agents/workflows/*.yml|*.yaml`` directory."""
+    base = Path(git_root).expanduser() if git_root else Path.cwd()
+
+    return _entries_from_dir(base / AGENTS_WORKFLOW_DIR)
+
+
 def _entries_from_home() -> List[_Discovered]:
     """Resolve ``~/.cecli/workflows/*.yml|*.yaml`` in stable sorted order."""
-    root = Path.home() / WORKFLOW_DIR
+
+    return _entries_from_dir(Path.home() / WORKFLOW_DIR)
+
+
+def _entries_from_dir(root: Path) -> List[_Discovered]:
+    """Resolve ``*.yml`` / ``*.yaml`` files under *root* in stable sorted order."""
     if not root.is_dir():
         return []
 

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/website/docs/config/retries.md

@@ -11,6 +11,7 @@ Cecli can be configured to retry failed API calls. This is useful for handling i
 - `retry-timeout`: The timeout in seconds for each retry.
 - `retry-backoff-factor`: The backoff factor to use between retries.
 - `retry-on-unavailable`: Whether to retry on 503 Service Unavailable errors.
+- `retry-on-empty`: Whether to retry when the model returns an empty response. A response is considered empty when it has no content, no tool calls, and no *meaningful* reasoning. Reasoning that contains no alphanumeric characters (for example a `reasoning_content` of `"!!!!"`) does not count as a response, so it is retried like any other empty response.
 - `retry-on-unauthorized`: Whether to retry on 401 Unauthorized (and 403 Forbidden) errors. Default: false.
 
 Example usage in `.cecli.conf.yml`:

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/website/docs/config/scheduler.md

@@ -1,7 +1,7 @@
 ---
 parent: Configuration
 nav_order: 48
-description: The scheduler turns workflow schedule: entries into runs, coordinating any number of cecli instances through a shared SQLite database.
+description: The scheduler turns workflow schedule entries into runs, coordinating any number of cecli instances through a shared SQLite database.
 ---
 # Scheduler
 

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli/website/docs/config/workflows.md

@@ -17,12 +17,14 @@ Preferred: one YAML file per workflow in `~/.cecli/workflows/`:
     morning-triage.yml
 ```
 
-Also supported: inline under a `workflows:` key in `.cecli.conf.yml`, or a repo-local `.cecli.workflows.yml`. Each source may hold a single workflow mapping, a `workflows: [...]` list, or a plain list of mappings. Resolution precedence (highest first):
+Also supported: a repo-local `.cecli/workflows/` or `.agents/workflows/` directory (one YAML file per workflow), inline under a `workflows:` key in `.cecli.conf.yml`, or a repo-local `.cecli.workflows.yml`. Each source may hold a single workflow mapping, a `workflows: [...]` list, or a plain list of mappings. Resolution precedence (highest first):
 
 1. **CLI argument** — an inline JSON/YAML string or a file path passed to `--workflows`.
 2. **Local file** — `.cecli.workflows.yml` / `.cecli.workflows.yaml` in the git root (or working directory).
-3. **Merged config** — the `workflows:` key in `.cecli.conf.yml`.
-4. **Global files** — `~/.cecli/workflows/*.yml` / `*.yaml`, loaded in sorted order.
+3. **Local directory** — `.cecli/workflows/*.yml` / `*.yaml` in the git root (or working directory), loaded in sorted order.
+4. **Standard local directory** — `.agents/workflows/*.yml` / `*.yaml` in the git root (or working directory), loaded in sorted order.
+5. **Merged config** — the `workflows:` key in `.cecli.conf.yml`.
+6. **Global files** — `~/.cecli/workflows/*.yml` / `*.yaml`, loaded in sorted order.
 
 On a duplicate `name`, the higher-precedence definition wins and the loser is skipped with a warning. A source that fails to parse is skipped with a warning as well — a malformed workflows file never crashes startup.
 

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli_dev.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cecli-dev
-Version: 1.8.0
+Version: 1.8.1rc2
 Summary: we also can't pronounce cecli
 Project-URL: Homepage, https://github.com/dwash96/cecli
 Project-URL: Documentation, https://github.com/dwash96/cecli

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli_dev.egg-info/SOURCES.txt

@@ -686,6 +686,7 @@ tests/basic/test_custom_commands.py
 tests/basic/test_deprecated.py
 tests/basic/test_editblock.py
 tests/basic/test_editor.py
+tests/basic/test_empty_response.py
 tests/basic/test_exceptions.py
 tests/basic/test_file_system_service.py
 tests/basic/test_find_or_blocks.py

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/cecli_dev.egg-info/scm_file_list.json

@@ -926,6 +926,7 @@
     "tests/basic/test_deprecated.py",
     "tests/basic/test_editblock.py",
     "tests/basic/test_editor.py",
+    "tests/basic/test_empty_response.py",
     "tests/basic/test_exceptions.py",
     "tests/basic/test_file_system_service.py",
     "tests/basic/test_find_or_blocks.py",

cecli_dev-1.8.1rc2/cecli_dev.egg-info/scm_version.json

@@ -0,0 +1,8 @@
+{
+  "tag": "1.8.1rc2",
+  "distance": 0,
+  "node": "g9c5114f1f4ccf75c0b68aca1830a9232d48f1ed0",
+  "dirty": false,
+  "branch": "HEAD",
+  "node_date": "2026-10-14"
+}

cecli_dev-1.8.1rc2/tests/basic/test_empty_response.py

@@ -0,0 +1,223 @@
+"""Tests for empty-response detection with non-meaningful reasoning content.
+
+Regression coverage for the ``retry-on-empty`` fix: a completion with empty
+``content``, no tool calls, and a ``reasoning_content`` made entirely of
+non-alphanumeric characters (e.g. moonshotai/kimi-k3 returning ``"!!!!"``) must
+be classified as an empty response so the existing retry loop engages.
+"""
+
+from types import SimpleNamespace
+
+import pytest
+
+from cecli.coders.base_coder import Coder, _is_meaningful_reasoning
+from cecli.helpers.threading import ThreadSafeEvent
+from cecli.llm import litellm
+
+
+# --------------------------------------------------------------------------- #
+# Test doubles
+# --------------------------------------------------------------------------- #
+class _AlwaysSetEvent:
+    def is_set(self):
+        return True
+
+
+class _FakeIO:
+    def __init__(self):
+        self.confirmation_in_progress_event = _AlwaysSetEvent()
+        self.assistant_outputs = []
+
+    def tool_error(self, *a, **k):
+        pass
+
+    def tool_warning(self, *a, **k):
+        pass
+
+    def update_spinner_suffix(self, *a, **k):
+        pass
+
+    def reset_streaming_response(self):
+        pass
+
+    def stream_output(self, *a, **k):
+        pass
+
+    def ai_output(self, *a, **k):
+        pass
+
+    def tool_output(self, *a, **k):
+        pass
+
+    def assistant_output(self, *a, **k):
+        self.assistant_outputs.append(a)
+
+
+class _FakeTokenProfiler:
+    def start(self):
+        pass
+
+    def on_token(self):
+        pass
+
+    def on_error(self):
+        pass
+
+    def add_to_usage_report(self, *a, **k):
+        return a[0] if a else ""
+
+
+def _make_coder(stream=False):
+    coder = Coder.__new__(Coder)
+    coder.stream = stream
+    coder.verbose = False
+    coder.args = SimpleNamespace(debug=False, show_thinking=False)
+    coder.io = _FakeIO()
+    coder.interrupt_event = ThreadSafeEvent()
+    coder.pretty = False
+    coder.reasoning_tag_name = "THINKING"
+    coder.got_reasoning_content = False
+    coder.ended_reasoning_content = False
+    coder.empty_response = False
+    coder.tool_reflection = False
+    coder.partial_response_content = ""
+    coder.partial_response_reasoning_content = ""
+    coder.partial_response_chunks = []
+    coder.partial_response_tool_calls = []
+    coder.partial_response_function_call = dict()
+    coder.partial_response_consolidated = None
+    coder.multi_response_content = ""
+    coder.chat_completion_response_hashes = []
+    coder._streaming_buffer_length = 0
+    coder.token_profiler = _FakeTokenProfiler()
+    coder._output_loop_detected = False
+    coder._output_loop_message = ""
+    coder._has_empty_reflected = False
+    coder.edit_format = "code"
+    return coder
+
+
+def _tc(index, call_id, name, arguments):
+    return litellm.ChatCompletionMessageToolCall(
+        id=call_id,
+        function=litellm.Function(arguments=arguments or "", name=name),
+        type="function",
+        index=index,
+    )
+
+
+def _non_streaming_completion(content="", reasoning=None, tool_calls=None):
+    message = {"role": "assistant", "content": content}
+    if reasoning is not None:
+        message["reasoning_content"] = reasoning
+    if tool_calls is not None:
+        message["tool_calls"] = tool_calls
+    return litellm.ModelResponse(
+        id="test-completion",
+        created=0,
+        model="gpt-test",
+        object="chat.completion",
+        choices=[{"finish_reason": "stop", "index": 0, "message": message}],
+        usage={"completion_tokens": 1, "prompt_tokens": 1, "total_tokens": 2},
+    )
+
+
+def _reasoning_chunk(text):
+    delta = litellm.Delta(role="assistant", content=None, reasoning_content=text)
+    choice = litellm.StreamChoice(finish_reason=None, index=0, delta=delta)
+    return litellm.StreamChunk(
+        id="cmpl-test", created=1000, model="gpt-test", choices=[choice], usage=None
+    )
+
+
+async def _agen(chunks):
+    for chunk in chunks:
+        yield chunk
+
+
+# --------------------------------------------------------------------------- #
+# _is_meaningful_reasoning unit tests
+# --------------------------------------------------------------------------- #
+def test_is_meaningful_reasoning_empty_and_none():
+    assert _is_meaningful_reasoning("") is False
+    assert _is_meaningful_reasoning(None) is False
+
+
+def test_is_meaningful_reasoning_punctuation_only():
+    assert _is_meaningful_reasoning("!!!!") is False
+    assert _is_meaningful_reasoning("...?!,;:") is False
+
+
+def test_is_meaningful_reasoning_whitespace_only():
+    assert _is_meaningful_reasoning("   \n\t  ") is False
+
+
+def test_is_meaningful_reasoning_normal_text():
+    assert _is_meaningful_reasoning("Let me think about this") is True
+
+
+def test_is_meaningful_reasoning_mixed_punctuation_and_alnum():
+    assert _is_meaningful_reasoning("42?") is True
+    assert _is_meaningful_reasoning("!!!a!!!") is True
+
+
+# --------------------------------------------------------------------------- #
+# Non-streaming path (show_send_output)
+# --------------------------------------------------------------------------- #
+@pytest.mark.asyncio
+async def test_junk_reasoning_is_empty_non_streaming():
+    coder = _make_coder(stream=False)
+    await coder.show_send_output(_non_streaming_completion(content="", reasoning="!" * 40))
+    assert coder.empty_response is True
+
+
+@pytest.mark.asyncio
+async def test_meaningful_reasoning_is_not_empty_non_streaming():
+    coder = _make_coder(stream=False)
+    await coder.show_send_output(
+        _non_streaming_completion(content="", reasoning="Let me think about this")
+    )
+    assert coder.empty_response is False
+
+
+@pytest.mark.asyncio
+async def test_tool_calls_are_not_empty_non_streaming():
+    coder = _make_coder(stream=False)
+    await coder.show_send_output(
+        _non_streaming_completion(
+            content="",
+            reasoning="",
+            tool_calls=[_tc(0, "call_1", "Local--ls", "{}")],
+        )
+    )
+    assert coder.empty_response is False
+
+
+@pytest.mark.asyncio
+async def test_mixed_reasoning_is_not_empty_non_streaming():
+    coder = _make_coder(stream=False)
+    await coder.show_send_output(_non_streaming_completion(content="", reasoning="42?"))
+    assert coder.empty_response is False
+
+
+# --------------------------------------------------------------------------- #
+# Streaming path (show_send_output_stream)
+# --------------------------------------------------------------------------- #
+@pytest.mark.asyncio
+async def test_junk_reasoning_is_empty_streaming():
+    coder = _make_coder(stream=True)
+    coder.args.show_thinking = True
+    async for _ in coder.show_send_output_stream(_agen([_reasoning_chunk("!" * 40)])):
+        pass
+    assert coder.empty_response is True
+
+
+@pytest.mark.asyncio
+async def test_meaningful_reasoning_is_not_empty_streaming():
+    coder = _make_coder(stream=True)
+    coder.args.show_thinking = True
+    async for _ in coder.show_send_output_stream(
+        _agen([_reasoning_chunk("Let me think about this")])
+    ):
+        pass
+    assert coder.empty_response is False

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/tests/helpers/observations/test_observation_service.py

@@ -247,3 +247,48 @@ async def test_run_observation_uses_formatted_chat_chunks():
     # The observation prompt is appended as the final user message
     assert sent_messages[-1]["role"] == "user"
     assert "Observation Prompt" in sent_messages[-1]["content"]
+
+
+@pytest.mark.asyncio
+async def test_compaction_bails_out_when_fixed_context_dominates():
+    """A large file in the chat cannot be compacted away, so bail instead of looping."""
+    from cecli.coders.base_coder import Coder
+
+    coder = MagicMock(spec=Coder)
+    coder.uuid = "test-coder-fixed-context"
+    coder.enable_context_compaction = True
+    coder.context_compaction_max_tokens = 1000
+    coder.context_compaction_summary_tokens = 100
+    coder.io = MagicMock()
+    coder.gpt_prompts = MagicMock()
+    coder._rate_limit_sleep = AsyncMock()
+
+    # done/cur/diff/file_context are empty, so the entire context is the fixed
+    # context (chat files, repo map, prompts). Summarizing cannot shrink it.
+    coder.summarizer = MagicMock()
+    coder.summarizer.count_tokens.side_effect = [0, 0, 0, 0, 6000]
+    coder.summarizer.summarize_all_as_text = AsyncMock(return_value="Summary Text")
+
+    mock_conv_manager = MagicMock()
+    mock_conv_manager.get_messages_dict.return_value = []
+
+    obs = MagicMock()
+    obs.check_and_trigger = AsyncMock()
+
+    with (
+        patch(
+            "cecli.coders.base_coder.ConversationService.get_manager",
+            return_value=mock_conv_manager,
+        ),
+        patch(
+            "cecli.coders.base_coder.ObservationService.get_instance",
+            return_value=obs,
+        ),
+    ):
+        await Coder.compact_context_if_needed(coder, force=True)
+
+    # Compaction must not summarize anything...
+    coder.summarizer.summarize_all_as_text.assert_not_called()
+    # ...and must tell the user why, with a remedy.
+    coder.io.tool_warning.assert_called_once()
+    coder.io.tool_output.assert_any_call("- Use /drop to remove unneeded files from the chat")

{cecli_dev-1.8.0 → cecli_dev-1.8.1rc2}/tests/helpers/test_llms_hyperscalers.py

@@ -153,6 +153,19 @@ def test_bedrock_endpoint_requires_region():
         raise AssertionError("expected ValueError without a region")
 
 
+def test_bedrock_endpoint_built_from_resolved_api_base(monkeypatch):
+    # The Converse URL is assembled from resolved["api_base"] verbatim rather
+    # than a hardcoded public host, so a custom endpoint (VPC endpoint, proxy,
+    # LocalStack, ...) resolved onto api_base is honored.
+    monkeypatch.setenv("AWS_REGION_NAME", "us-east-1")
+    resolved = {
+        "route": "anthropic.claude-sonnet",
+        "api_base": "https://bedrock.internal.example.com",
+    }
+    url = bedrock_endpoint(resolved)
+    assert url == ("https://bedrock.internal.example.com/model/anthropic.claude-sonnet/converse")
+
+
 # ---------------------------------------------------------------------------
 # Chat-family wiring: signer hook + extra_query
 # ---------------------------------------------------------------------------