Research: D-Grade Complexity Hotspots

Research – D-Grade Complexity Hotspots

Date: 2026-02-01 Phase: Research

Overview

4 functions exceed McCabe complexity threshold (25):

1. dispatch_tools (Score: 29)

File: src/tunacode/core/agents/agent_components/orchestrator/tool_dispatcher.py:219

Signature

async def dispatch_tools(
    parts: list[Any],
    node: Any,
    state_manager: StateManagerProtocol,
    tool_callback: ToolCallback | None,
    _tool_result_callback: ToolResultCallback | None,
    tool_start_callback: ToolStartCallback | None,
    response_state: ResponseState | None,
) -> ToolDispatchResult

Complexity Breakdown

Nested Complexity Sources

• Triple-nested: debug_mode and _is_suspicious_tool_name(...) inside loop (line 260)
• Triple-nested: is_processing_tools and response_state and can_transition_to(...) (lines 315-320)
• Double-nested with loop: if tool_start_callback inside if tool_tasks and tool_callback (line 301)

Call Graph

ui/app.py → core/agents/main.py:process_request() → RequestOrchestrator.run()
    → process_node() → dispatch_tools()

Extractable Subroutines

• _process_native_tool_calls(parts, tool_callback, debug_mode) → handle primary loop
• _handle_fallback_parsing(tool_callback, response_parts) → fallback extraction
• _execute_tool_batch(tool_tasks, tool_callback, start_callback) → tool execution
• _finalize_dispatch_state(response_state, is_processing) → state transition

2. block_anchor_replacer (Score: 27)

File: src/tunacode/tools/utils/text_match.py:162

Signature

def block_anchor_replacer(content: str, find: str) -> Generator[str, None, None]

Complexity Breakdown

Key Issue: Mutually Exclusive Paths

Two large branches with independent complexity:

• Single candidate path: 11 decision points (lines 214-241)
• Multiple candidates path: 11 decision points (lines 243-276)

Both paths share similar logic (similarity calculation) but are duplicated.

Nested Complexity Sources

• Triple-nested: if max_len == 0 inside loop inside if lines_to_check > 0 inside if len(candidates) == 1
• Double-nested loops: for i in first_indices → for j in last_indices

Call Graph

tools/update_file.py:update_file() → text_match.py:replace()
    → block_anchor_replacer() (as generator)

Extractable Subroutines

• _calculate_similarity(content_lines, search_lines, start, end) → used in BOTH paths
• _build_candidate_pairs(first_indices, last_indices) → nested loop extraction
• _find_best_match(candidates, content_lines, search_lines) → multiple candidate logic

3. process_node (Score: 24)

File: src/tunacode/core/agents/agent_components/orchestrator/orchestrator.py:180

Signature

async def process_node(
    node: Any,
    tool_callback: ToolCallback | None,
    state_manager: StateManagerProtocol,
    _tool_buffer: ToolBuffer | None = None,
    _streaming_callback: StreamingCallback | None = None,
    response_state: ResponseState | None = None,
    tool_result_callback: ToolResultCallback | None = None,
    tool_start_callback: ToolStartCallback | None = None,
) -> tuple[bool, str | None]

Complexity Breakdown

Key Issue: Long Conditional Chain

if model_response is not None block (lines 231-285) contains:

• Debug logging branch
• Response state processing
• Content extraction loop with 2 branches
• Content combination block
• Empty response detection
• Tool dispatch call

Nested Complexity Sources

• Triple-nested: if content.strip() inside loop inside if response_state inside if model_response
• Triple condition: if response_state and can_transition_to(...) and not is_completed()

Call Graph

core/agents/main.py:RequestOrchestrator.run() → process_node()
    → dispatch_tools() (complexity 29)

Extractable Subroutines

• _process_model_response(model_response, response_state, debug_mode) → entire lines 231-285
• _extract_content_parts(response_parts) → content extraction loop
• _detect_empty_response(parts_result, response_state) → empty detection logic

4. is_ignored (Score: 23)

File: src/tunacode/configuration/ignore_patterns.py:84

Signature

def is_ignored(rel_path: str, name: str, patterns: Iterable[str]) -> bool

Complexity Breakdown

Key Issue: Deep Nesting in Rooted Patterns

for pattern in patterns:           # level 1
    if starts_with("/"):           # level 2
        if fnmatch(...):           # level 3
            if is_dir_pattern:     # level 4
                if exact_match:    # level 5 (!)

Nested Complexity Sources

• Quadruple-nested: if rel_path == match_pattern or ... inside if is_dir_pattern inside if fnmatch(...) inside loop
• Triple-nested: if fnmatch.fnmatch(path_parts[i], ...) inside inner loop inside if is_dir_pattern or...
• Nested loop: for i in range(limit) inside for pattern in patterns

Call Graph

utils/system/gitignore.py:72,78 → is_ignored()
infrastructure/file_filter.py:37 → is_ignored() (different signature)
tools/glob.py, tools/grep.py, tools/list_dir.py → use FileFilter → is_ignored()

Note: Two Implementations

1. configuration/ignore_patterns.py:84 - Pure function (this one, score 23)
2. infrastructure/file_filter.py:37 - Method on FileFilter class (different signature)

Extractable Subroutines

• _match_rooted_pattern(rel_path, pattern, is_dir_pattern) → entire lines 100-110
• _match_path_components(path_parts, pattern, is_dir_pattern) → lines 118-124

Common Patterns Across All Functions

1. Long Sequential Branches

All functions have 15+ sequential decision points in a single function body.

2. Deep Nesting (3-4 levels)

3. Loops with Multiple Internal Branches

4. No Subroutine Extraction

All functions keep related logic inline rather than extracting helper functions.

Dependency Layer Map

UI Layer
  └── ui/app.py → process_request()
        │
Core Layer
  └── core/agents/main.py → RequestOrchestrator.run()
        │
        ├── orchestrator/orchestrator.py:process_node() [24]
        │     └── orchestrator/tool_dispatcher.py:dispatch_tools() [29]
        │
Tools Layer
  └── tools/update_file.py → update_file()
        └── tools/utils/text_match.py:block_anchor_replacer() [27]

Configuration Layer
  └── configuration/ignore_patterns.py:is_ignored() [23]
        ↑ used by
        ├── utils/system/gitignore.py
        ├── infrastructure/file_filter.py
        └── tools/{glob,grep,list_dir}.py (via FileFilter)

Test Coverage

Files Touched (Full Paths)

dispatch_tools

• /home/tuna/tunacode/src/tunacode/core/agents/agent_components/orchestrator/tool_dispatcher.py
• /home/tuna/tunacode/src/tunacode/core/agents/agent_components/orchestrator/orchestrator.py
• /home/tuna/tunacode/src/tunacode/core/agents/main.py

block_anchor_replacer

• /home/tuna/tunacode/src/tunacode/tools/utils/text_match.py
• /home/tuna/tunacode/src/tunacode/tools/update_file.py

process_node

• /home/tuna/tunacode/src/tunacode/core/agents/agent_components/orchestrator/orchestrator.py
• /home/tuna/tunacode/src/tunacode/core/agents/main.py
• /home/tuna/tunacode/src/tunacode/ui/app.py

is_ignored

• /home/tuna/tunacode/src/tunacode/configuration/ignore_patterns.py
• /home/tuna/tunacode/src/tunacode/infrastructure/file_filter.py
• /home/tuna/tunacode/src/tunacode/utils/system/gitignore.py