Graph-based subsystem extraction with pruning

CLAUDE.md

@@ -226,6 +226,7 @@ When tests are included, they follow the same user story organization.
 - JSON diagram files (existing persistence via Pydantic schemas) (014-iomarker-latex-rendering)
 - TypeScript 5.9 (frontend), Python 3.11+ (backend) + React 19.2.3, React Flow 11.11.4, anywidget (Jupyter widget framework), Pydantic (schema validation) (015-block-drag-detection)
 - Python 3.11+ + Pydantic 2.12+ (existing schema validation), python-control 0.10+ (existing) (017-diagram-label-indexing)
+- Python 3.11+ (existing Lynx requirement) + python-control 0.10+ (existing), no new dependencies required (018-graph-pruning-extraction)
 
 ## Key Components
 
@@ -606,6 +607,7 @@ blocks/
 - `js/src/test/` - Test configuration and setup files
 
 ## Recent Changes
+- 018-graph-pruning-extraction: Added Python 3.11+ (existing Lynx requirement) + python-control 0.10+ (existing), no new dependencies required
 - 017-diagram-label-indexing: Added Python 3.11+ + Pydantic 2.12+ (existing schema validation), python-control 0.10+ (existing)
 - **015-block-drag-detection**: Intelligent drag detection with 5-pixel movement threshold
   - Click-to-select (< 5px movement) vs drag-to-move (≥ 5px movement) behavior
@@ -618,7 +620,6 @@ blocks/
   - Performance: < 5ms overhead per drag operation, maintains 60 FPS
   - 717 total tests passing (407 Python + 310 frontend)
   - Frontend-only feature (no Python backend changes required)
-- **014-iomarker-latex-rendering**: Complete IOMarker LaTeX rendering with automatic indexing and Simulink-style renumbering
   - Automatic index display (0, 1, 2...) via LaTeX for InputMarker and OutputMarker blocks
   - Custom LaTeX override using existing useCustomLatex hook (checkbox + textarea UI)
   - Removed "Input/Output" text and "Type" dropdown from parameter panel for cleaner UX

pyproject.toml

@@ -4,7 +4,7 @@
 
 [project]
 name = "lynx-nb"
-version = "0.1.4"
+version = "0.1.5"
 description = "Block diagram editor for control systems"
 readme = "README.md"
 authors = [
@@ -13,7 +13,7 @@ authors = [
 requires-python = ">=3.11"
 dependencies = [
     "anywidget>=0.9.21",
-    "control>=0.10.1",
+    "control>=0.10.2",
     "numpy>=2.4.1",
     "pydantic>=2.12.5",
     "traitlets>=5.14.3",

specs/018-graph-pruning-extraction/checklists/requirements.md

@@ -0,0 +1,44 @@
+# Specification Quality Checklist: Graph-Based Subsystem Extraction
+
+**Purpose**: Validate specification completeness and quality before proceeding to planning
+**Created**: 2026-02-01
+**Feature**: [spec.md](../spec.md)
+
+## Content Quality
+
+- [x] No implementation details (languages, frameworks, APIs)
+- [x] Focused on user value and business needs
+- [x] Written for non-technical stakeholders
+- [x] All mandatory sections completed
+
+## Requirement Completeness
+
+- [x] No [NEEDS CLARIFICATION] markers remain
+- [x] Requirements are testable and unambiguous
+- [x] Success criteria are measurable
+- [x] Success criteria are technology-agnostic (no implementation details)
+- [x] All acceptance scenarios are defined
+- [x] Edge cases are identified
+- [x] Scope is clearly bounded
+- [x] Dependencies and assumptions identified
+
+## Feature Readiness
+
+- [x] All functional requirements have clear acceptance criteria
+- [x] User scenarios cover primary flows
+- [x] Feature meets measurable outcomes defined in Success Criteria
+- [x] No implementation details leak into specification
+
+## Validation Notes
+
+**Pass**: All checklist items validated successfully.
+
+- Specification is written in user-centric language without implementation details
+- All 12 functional requirements are testable and measurable
+- Success criteria focus on observable outcomes (TF order, execution time, state count)
+- Edge cases cover boundary conditions (same block, algebraic loops, multiple paths)
+- Scope clearly defines what is and isn't included
+- Dependencies and assumptions documented
+- No clarification markers present - all requirements are unambiguous
+
+**Ready for**: `/speckit.plan` - proceed to implementation planning phase

specs/018-graph-pruning-extraction/data-model.md

@@ -0,0 +1,123 @@
+<!--
+SPDX-FileCopyrightText: 2026 Jared Callaham <jared.callaham@gmail.com>
+
+SPDX-License-Identifier: GPL-3.0-or-later
+-->
+
+# Data Model: Graph-Based Subsystem Extraction
+
+**Feature**: 018-graph-pruning-extraction
+**Created**: 2026-02-01
+
+## Overview
+
+This feature operates on in-memory Diagram objects and does not introduce persistent data structures. The entities below represent transient computational structures used during extraction.
+
+## Entities
+
+### ConnectionGraph
+
+**Purpose**: Directed graph representation of diagram topology for analysis
+
+**Attributes**:
+- `nodes`: Set of block IDs (str) representing graph vertices
+- `forward_edges`: Dict[str, List[str]] - adjacency list for forward traversal (block_id → [connected_block_ids])
+- `backward_edges`: Dict[str, List[str]] - reverse adjacency list for backward traversal
+
+**Relationships**:
+- Built from Diagram.blocks and Diagram.connections
+- Used by path finding algorithms
+- Discarded after pruning completes
+
+**Validation Rules**:
+- Every edge (connection) must reference valid nodes (blocks)
+- Graph may contain cycles (feedback loops are valid)
+- Nodes and edges derived from validated Diagram (no independent validation needed)
+
+**Lifecycle**: Created → used for reachability analysis → discarded
+
+---
+
+### ReachabilityResult
+
+**Purpose**: Container for bidirectional reachability analysis output
+
+**Attributes**:
+- `forward_reachable`: Set[str] - block IDs reachable forward from source
+- `backward_reachable`: Set[str] - block IDs reachable backward from destination
+- `path_blocks`: Set[str] - intersection of forward and backward reachable sets
+- `source_block_id`: str - ID of block outputting source signal
+- `dest_block_id`: str - ID of block outputting destination signal
+
+**Relationships**:
+- Produced by `_find_reachable_blocks()` function
+- Consumed by `_prune_diagram()` function
+- References block IDs from original Diagram
+
+**Validation Rules**:
+- `source_block_id` MUST be in `forward_reachable`
+- `dest_block_id` MUST be in `backward_reachable`
+- `path_blocks` MUST be non-empty (otherwise no valid path exists)
+- `path_blocks` MUST contain both source and destination block IDs
+
+**State Transitions**:
+1. Initial: Empty sets
+2. Forward DFS: Populates `forward_reachable`
+3. Backward DFS: Populates `backward_reachable`
+4. Intersection: Computes `path_blocks`
+
+---
+
+### PrunedDiagram
+
+**Purpose**: Modified clone of original Diagram with only relevant blocks for extraction
+
+**Attributes**:
+- Inherits all attributes from `Diagram` class
+- `blocks`: Subset of original blocks (only those in `path_blocks`)
+- `connections`: Subset of original connections (only those between kept blocks)
+
+**Relationships**:
+- Cloned from original Diagram via `diagram._clone()`
+- Blocks removed using `diagram.remove_block(block_id)`
+- Passed to existing `_prepare_for_extraction()` for interconnect building
+
+**Validation Rules**:
+- Must contain at least source and destination blocks
+- All connections must reference existing blocks (auto-enforced by remove_block)
+- Must pass existing diagram validation (InputMarker, OutputMarker, connectivity)
+
+**Lifecycle**:
+1. Clone original Diagram
+2. Remove blocks not in `path_blocks`
+3. Pass to interconnect builder
+4. Discarded after extraction completes
+
+---
+
+## Data Flow
+
+```
+Original Diagram
+    ↓
+Signal Resolution (existing)
+    ↓
+Build ConnectionGraph
+    ↓
+Bidirectional DFS
+    ↓
+ReachabilityResult
+    ↓
+Clone & Prune → PrunedDiagram
+    ↓
+Interconnect Building (existing)
+    ↓
+Extracted Transfer Function
+```
+
+## Notes
+
+- No persistent storage required
+- All entities are transient (lifetime ~10-500ms during extraction)
+- Graph structures use native Python collections (set, dict, list)
+- No external graph library dependencies (NetworkX not needed)