Skip to content

Instantly share code, notes, and snippets.

@Kadajett

Kadajett/SemforaAnalysis.md

Created December 13, 2025 16:50
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save Kadajett/dbfd7ef3e626d9d1d63e97d8c5ed7e40 to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save Kadajett/dbfd7ef3e626d9d1d63e97d8c5ed7e40 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
SemforaAnalysis.md

Semfora Engine Codebase Analysis Report

Generated: December 13, 2025 Codebase: semfora-engine Analysis Tool: semfora-engine v0.1.0 (self-analysis)

Executive Summary

Metric Value
Total Files Analyzed 196
Total Modules 32
Total Symbols 1,571 (index) / 1,480 (analysis)
Total Lines of Code 41,429
Compression Ratio 62.9%
Indexing Time 1.267 seconds
Duplicate Clusters 66 clusters (140 duplicate functions)
Cognitive Complexity Avg 3.3

1. Indexing Performance

Metric Value
Files Processed 190
Source Size 11.5 MB
Index Time 1.267 seconds
Throughput ~150 files/second
Compression 62.9% token reduction

The sharded index system provides efficient caching with module-level granularity for incremental updates.

2. Cognitive Complexity Analysis

Top 15 Most Complex Functions

Rank Function Cognitive Nesting LOC Fan-Out File
1 build_call_graph_from_summaries 100 5 155 24 cache.rs
2 build_call_graph 87 5 97 20 shard.rs
3 load_module_summaries 66 6 105 27 cache.rs
4 collect_symbol_candidates 66 6 113 16 detectors
5 handle_query 61 2 356 0 socket_server
6 from_summaries 55 2 172 13 cache.rs
7 collect_files_recursive 54 7 64 18 mcp_server
8 start_with_cache 51 7 145 0 server
9 extract_candidate_from_decl 50 5 130 0 detectors
10 collect_files_recursive 48 7 60 16 socket_server
11 extract 44 2 80 41 extract.rs
12 run_shard 44 3 203 0 main.rs
13 load_layer 43 2 90 21 cache.rs
14 search_symbols 42 3 85 19 overlay.rs
15 run 40 3 126 0 main.rs

Complexity Scale:

  • 0-5: Simple
  • 6-10: Moderate
  • 11-20: Complex
  • 21+: Very Complex

Refactoring Candidates

Functions with cognitive complexity > 50 should be considered for refactoring:

  1. build_call_graph_from_summaries (100) - Extract helper functions for symbol lookup and edge building
  2. build_call_graph (87) - Similar to above, split graph construction logic
  3. load_module_summaries (66) - Separate file I/O from parsing logic
  4. collect_symbol_candidates (66) - Extract visitor pattern or iterative approach

3. Lines of Code Analysis

Highest LOC Functions

Rank Function LOC File
1 get_templates 6,642 benchmark_builder/templates.rs
2 handle_query 356 socket_server
3 run_shard 203 main.rs
4 from_summaries 172 cache.rs
5 build_call_graph_from_summaries 155 cache.rs
6 start_with_cache 145 server
7 extract_candidate_from_decl 130 detectors
8 run 126 main.rs
9 collect_symbol_candidates 113 detectors
10 load_module_summaries 105 cache.rs

Note: get_templates at 6,642 LOC contains embedded template strings for the benchmark builder - this is by design, not a refactoring candidate.

Module Metrics by LOC

Module Symbols Total LOC Avg CC Max CC
cache 144 3,755 3.7 100
detectors 113 2,500 3.8 31
mcp_server 82 2,192 4.8 34
detectors.javascript 100 2,007 4.0 32
socket_server 68 1,731 4.2 61
scripts 46 1,641 5.3 22
root 33 1,510 10.0 48
benches 48 1,477 5.5 24
detectors.javascript.core 37 1,231 8.3 66
toon 26 1,220 8.5 34
shard 31 1,023 10.6 87

4. Duplicate Function Detection

Configuration:

  • Threshold: 90%
  • Boilerplate Excluded: Yes
  • Total Signatures Analyzed: 1,295
  • Duplicate Clusters Found: 66
  • Total Duplicate Functions: 140

Exact Duplicates (100% Match) - HIGH PRIORITY

These are identical implementations that should be consolidated:

Function Locations Action
truncate_to_char_boundary common.rs, toon.rs, extract.rs Consolidate to single utility
extract_filename_stem python.rs, javascript/core.rs Move to shared module
parse_source javascript/core.rs, generic.rs Extract to common parser
collect_files main.rs, mcp_server/helpers.rs Deduplicate
default (Default impl) 5 locations Consider derive macro

Near Duplicates (90-97% Match)

Cluster Primary Duplicates Similarity
1 parse_and_extract_string parse_and_extract 95%
2 get_changed_files get_commit_changed_files 95%
3 test_load_layer_corrupted_symbols 5 similar test functions 90-94%
4 call_graph_path import_graph_path, module_graph_path 90%
5 layer_symbols_path layer_deleted_path, layer_moves_path 90%
6 test_compute_edit_insert 3 similar edit tests 92%

Divergent Duplicates (80-89% Match)

These share common patterns but have evolved differently:

Pattern Count Examples
Framework detection (is_*) 10 is_entry_point, is_component, is_service
Git operations (get_*) 6 get_current_branch, get_merge_base, etc.
Test assertions 15+ Similar test structure with different data
Layer operations 6 with_boilerplate_config, with_limit, etc.

Full Duplicate Cluster List (66 Clusters)

Click to expand all 66 clusters

Cluster 1: AnimatedEdge Components

  • Primary: AnimatedEdge (benchmark-visualizer)
  • Duplicates: StaticEdge (87%), NewEdge (87%)

Cluster 2: Script Utilities

  • Primary: run_cmd (realworld-test.py)
  • Duplicates: clear_cache (80%)

Cluster 3: Print Functions

  • Primary: print_status (realworld-test.py)
  • Duplicates: print_progress (87%)

Cluster 4: Benchmark Functions

  • Primary: benchmark_get_overview
  • Duplicates: benchmark_get_call_graph (88%)

Cluster 5: Filename Extraction (EXACT)

  • Primary: extract_filename_stem (python.rs)
  • Duplicates: extract_filename_stem (javascript/core.rs) - 100%

Cluster 6: Score Calculation

  • Primary: calculate_basic_score (python.rs)
  • Duplicates: calculate_symbol_score (go.rs) - 90%

Cluster 7-8: HCL Tests

  • 4 similar test functions for HCL parsing

Cluster 9: String Truncation (EXACT)

  • Primary: truncate_to_char_boundary (common.rs)
  • Duplicates: Same function in toon.rs, extract.rs - 100%

Cluster 10: Source Parsing (EXACT)

  • Primary: parse_source (javascript/core.rs)
  • Duplicates: parse_source (generic.rs) - 100%

Cluster 11-12: JavaScript Tests

  • Call attribution tests with minor variations

Cluster 13: Framework Detection Pattern (10 duplicates)

  • is_entry_point, is_middleware_file, is_vue_sfc, is_component (Vue/Angular), is_service, is_module, is_directive, is_pipe, is_composable, is_pinia_store
  • All follow same pattern: source.contains("pattern")

Cluster 14: Feature Detection

  • detect_from_source with 4 similar functions

Cluster 15: Vue SFC Tests

  • 4 test functions for Vue SFC extraction (93% similar)

Cluster 16: Export Detection

  • c_is_exported / kotlin_is_exported (90%)

Cluster 17-19: Default/New Implementations

  • Multiple simple constructors with similar patterns

Cluster 20-22: Search Tests

  • Test functions for file patterns and language filters

Cluster 23: File Collection (EXACT + NEAR)

  • collect_files (main.rs) = collect_files (helpers.rs) - 100%
  • collect_source_files (indexer.rs) - 93%

Cluster 24: Parse Functions

  • parse_and_extract_string / parse_and_extract (95%)

Cluster 25-30: Cache Path Functions

  • Multiple *_path functions returning cache paths

Cluster 31-34: Git Operations

  • get_current_branch, get_merge_base, get_parent_commit, etc.

Cluster 35-38: Test Functions

  • Multiple test clusters with similar assertion patterns

Cluster 39: Default Trait (EXACT)

  • default() implementation in 5 files - 100%

Cluster 40-66: Remaining Clusters

  • Mix of test functions, overlay operations, and utility functions

5. Call Graph Analysis

Entry Points (32 symbols)

Functions that are called from external sources or serve as module entry points.

Leaf Functions (121 symbols)

Functions that don't call any other internal functions - potential candidates for utility extraction.

Most Called Functions (Hotspots)

Function Callers Notes
Ok 106 Rust Result wrapper
Some 105 Rust Option wrapper
Vec::new 77 Collection initialization
PathBuf::from 65 Path construction
LayeredIndex::new 42 Index creation
make_test_symbol 42 Test helper
Default::default 41 Default trait
TempDir::new 38 Test directories

High Coupling (Most Outgoing Calls)

Symbol Outgoing Calls Risk
CacheMeta 83 High - consider splitting
McpDiffServer 80 High - facade pattern
main 68 Expected for entry point
DetectionResult 48 Moderate
BenchmarkResult 48 Moderate

Circular Dependencies Detected

5 circular dependency chains found in the call graph:

6b99d11b7375e677 → 82354bc187b12b66 → 6b99d11b7375e677
b880e2883646d4ea → 6b99d11b7375e677 → b880e2883646d4ea
6b99d11b7375e677 → 29205f009c7a6e4a → 6b99d11b7375e677
6b99d11b7375e677 → 5c084015448bd950 → 6b99d11b7375e677
6b99d11b7375e677 → 2381062eef1335bb → 6b99d11b7375e677

Note: These are in Python scripts (scripts/) and represent intentional recursive patterns, not problematic Rust dependencies.

6. Potentially Unused Code

Based on call graph analysis, the following patterns suggest potential dead code:

Leaf Functions Not Called from Entry Points

The analysis shows 121 leaf functions. Functions that are:

  1. Not test functions (test_*)
  2. Not trait implementations
  3. Not called by any other function

Should be reviewed for removal. Run --get-call-graph and cross-reference with entry points to identify specific candidates.

Unused Exports

The detectors/ module contains several language-specific functions that may not be invoked for all file types. This is expected behavior for a multi-language analyzer.

7. Recommendations

Immediate Actions (High Priority)

  1. Consolidate Exact Duplicates

    • Move truncate_to_char_boundary to a shared utils module
    • Unify extract_filename_stem implementations
    • Deduplicate collect_files between main.rs and helpers.rs

  2. Refactor High-Complexity Functions

    • build_call_graph_from_summaries (CC: 100) - Split into smaller functions
    • build_call_graph (CC: 87) - Extract graph building logic
    • load_module_summaries (CC: 66) - Separate I/O from parsing

Medium Priority

  1. Extract Common Patterns

    • Create trait for is_* framework detection functions
    • Consolidate git operation functions into trait with shared implementation
    • Unify cache path methods

  2. Test Deduplication

    • Consider parameterized tests for similar test clusters
    • Use test fixtures for common setup patterns

Future Improvements

  1. Reduce High Coupling

    • Consider splitting CacheMeta responsibilities
    • Apply facade pattern more consistently

  2. Documentation

    • Add module-level documentation for complex modules
    • Document circular dependency reasons in scripts

8. Appendix: Raw Metrics

Index Statistics

Files analyzed: 196
Modules: 32
Symbols: 1,571
Compression: 62.9%
Index time: 1.267s

Language Distribution

Language Files Symbols
Rust 150+ 1,200+
TypeScript/TSX 15+ 150+
Python 10+ 100+
Shell 5+ 50+

Risk Distribution

  • High Risk: ~50 symbols
  • Medium Risk: ~200 symbols
  • Low Risk: ~1,300 symbols

Report generated by semfora-engine's self-analysis capabilities.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment