code2logic 1.0.1__tar.gz → 1.0.3__tar.gz

{code2logic-1.0.1 → code2logic-1.0.3}/CHANGELOG.md

@@ -7,6 +7,17 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+- `code2logic llm ...` command group for managing LLM provider configuration.
+- `code2logic llm set-provider auto` for automatic provider fallback.
+- Provider and model priority management:
+  - provider priorities (stored in `~/.code2logic/llm_config.json`)
+  - model priorities by exact model string and prefix (family)
+  - priority modes: `provider-first`, `model-first`, `mixed`
+
+### Changed
+- Provider auto-selection order now follows configured priorities.
+
 ## [1.0.1] - 2026-01-03
 
 ### Changed

{code2logic-1.0.1 → code2logic-1.0.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: code2logic
-Version: 1.0.1
+Version: 1.0.3
 Summary: Convert source code to logical representation for LLM analysis
 Project-URL: Homepage, https://github.com/softreck/code2logic
 Project-URL: Documentation, https://code2logic.readthedocs.io
@@ -31,6 +31,7 @@ Requires-Python: >=3.9
 Provides-Extra: dev
 Requires-Dist: black>=23.0; extra == 'dev'
 Requires-Dist: build>=1.0.0; extra == 'dev'
+Requires-Dist: bumpver>=2023.1129; extra == 'dev'
 Requires-Dist: mypy>=1.0; extra == 'dev'
 Requires-Dist: pre-commit>=3.0; extra == 'dev'
 Requires-Dist: pytest-cov>=4.0; extra == 'dev'
@@ -74,7 +75,8 @@ Description-Content-Type: text/markdown
 
 **Convert source code to logical representation for LLM analysis.**
 
-Code2Logic analyzes codebases and generates compact, LLM-friendly representations with semantic understanding. Perfect for feeding project context to AI assistants, building code documentation, or analyzing code structure.
+Code2Logic analyzes codebases and generates compact, LLM-friendly representations with semantic understanding. 
+Perfect for feeding project context to AI assistants, building code documentation, or analyzing code structure.
 
 ## ✨ Features
 
@@ -146,6 +148,25 @@ hubs = [p for p, n in project.dependency_metrics.items() if n.is_hub]
 print(f"Key modules: {hubs}")
 ```
 
+### Organized Imports
+
+```python
+# Core analysis
+from code2logic.core import ProjectInfo, ProjectAnalyzer, analyze_project
+
+# Format generators
+from code2logic.formats import (
+    YAMLGenerator, JSONGenerator, TOONGenerator,
+    LogicMLGenerator, GherkinGenerator
+)
+
+# LLM clients
+from code2logic.llm import get_client, BaseLLMClient
+
+# Development tools
+from code2logic.tools import run_benchmark, CodeReviewer
+```
+
 ## 📋 Output Formats
 
 ### Markdown (default)
@@ -187,6 +208,48 @@ Library Status:
   rapidfuzz: ✓
   nltk: ✗
   spacy: ✗
+
+### LLM Configuration
+
+Manage LLM providers, models, API keys, and routing priorities:
+
+```bash
+code2logic llm status
+code2logic llm set-provider auto
+code2logic llm set-model openrouter nvidia/nemotron-3-nano-30b-a3b:free
+code2logic llm key set openrouter <OPENROUTER_API_KEY>
+code2logic llm priority set-provider openrouter 10
+code2logic llm priority set-mode provider-first
+code2logic llm priority set-llm-model nvidia/nemotron-3-nano-30b-a3b:free 5
+code2logic llm priority set-llm-family nvidia/ 5
+code2logic llm config list
+```
+
+Notes:
+
+- `code2logic llm set-provider auto` enables automatic fallback selection: providers are tried in priority order.
+- API keys should be stored in `.env` (or environment variables), not in `litellm_config.yaml`.
+
+#### Priority modes
+
+You can choose how automatic fallback ordering is computed:
+
+- `provider-first`
+  providers are ordered by provider priority (defaults + overrides)
+- `model-first`
+  providers are ordered by priority rules for the provider's configured model (exact/prefix)
+- `mixed`
+  providers are ordered by the best (lowest) priority from either provider priority or model rules
+
+Configure the mode:
+
+```bash
+code2logic llm priority set-mode provider-first
+code2logic llm priority set-mode model-first
+code2logic llm priority set-mode mixed
+```
+
+Model priority rules are stored in `~/.code2logic/llm_config.json`.
 ```
 
 ### Python API
@@ -339,6 +402,8 @@ python examples/09_async_benchmark.py --folder tests/samples/
 
 # Function-level reproduction
 python examples/10_function_reproduction.py --multi-lang
+
+python examples/15_unified_benchmark.py --folder tests/samples/
 ```
 
 ## 🤝 Contributing
@@ -351,8 +416,21 @@ MIT License - see [LICENSE](LICENSE) for details.
 
 ## 📚 Documentation
 
-- [API Documentation](DOCS.md) - Complete API reference
-- [Refactoring Plan](TODO.md) - Development roadmap
+- [00 - Docs Index](docs/00-index.md) - Documentation home (start here)
+- [01 - Getting Started](docs/01-getting-started.md) - Install and first steps
+- [02 - Configuration](docs/02-configuration.md) - API keys, environment setup
+- [03 - CLI Reference](docs/03-cli-reference.md) - Command-line usage
+- [04 - Python API](docs/04-python-api.md) - Programmatic usage
+- [05 - Output Formats](docs/05-output-formats.md) - Format comparison and usage
+- [06 - Format Specifications](docs/06-format-specifications.md) - Detailed format specs
+- [07 - TOON Format](docs/07-toon.md) - Token-Oriented Object Notation
+- [08 - LLM Integration](docs/08-llm-integration.md) - OpenRouter/Ollama/LiteLLM
+- [09 - LLM Comparison](docs/09-llm-comparison-report.md) - Provider/model comparison
+- [10 - Benchmarking](docs/10-benchmark.md) - Benchmark methodology and results
+- [11 - Repeatability](docs/11-repeatability.md) - Repeatability testing
+- [12 - Examples](docs/12-examples.md) - Usage workflows and examples
+- [13 - Architecture](docs/13-architecture.md) - System design and components
+- [14 - Format Analysis](docs/14-format-analysis.md) - Deeper format evaluation
 
 ## 🔗 Links
 

{code2logic-1.0.1 → code2logic-1.0.3}/README.md

@@ -6,7 +6,8 @@
 
 **Convert source code to logical representation for LLM analysis.**
 
-Code2Logic analyzes codebases and generates compact, LLM-friendly representations with semantic understanding. Perfect for feeding project context to AI assistants, building code documentation, or analyzing code structure.
+Code2Logic analyzes codebases and generates compact, LLM-friendly representations with semantic understanding. 
+Perfect for feeding project context to AI assistants, building code documentation, or analyzing code structure.
 
 ## ✨ Features
 
@@ -78,6 +79,25 @@ hubs = [p for p, n in project.dependency_metrics.items() if n.is_hub]
 print(f"Key modules: {hubs}")
 ```
 
+### Organized Imports
+
+```python
+# Core analysis
+from code2logic.core import ProjectInfo, ProjectAnalyzer, analyze_project
+
+# Format generators
+from code2logic.formats import (
+    YAMLGenerator, JSONGenerator, TOONGenerator,
+    LogicMLGenerator, GherkinGenerator
+)
+
+# LLM clients
+from code2logic.llm import get_client, BaseLLMClient
+
+# Development tools
+from code2logic.tools import run_benchmark, CodeReviewer
+```
+
 ## 📋 Output Formats
 
 ### Markdown (default)
@@ -119,6 +139,48 @@ Library Status:
   rapidfuzz: ✓
   nltk: ✗
   spacy: ✗
+
+### LLM Configuration
+
+Manage LLM providers, models, API keys, and routing priorities:
+
+```bash
+code2logic llm status
+code2logic llm set-provider auto
+code2logic llm set-model openrouter nvidia/nemotron-3-nano-30b-a3b:free
+code2logic llm key set openrouter <OPENROUTER_API_KEY>
+code2logic llm priority set-provider openrouter 10
+code2logic llm priority set-mode provider-first
+code2logic llm priority set-llm-model nvidia/nemotron-3-nano-30b-a3b:free 5
+code2logic llm priority set-llm-family nvidia/ 5
+code2logic llm config list
+```
+
+Notes:
+
+- `code2logic llm set-provider auto` enables automatic fallback selection: providers are tried in priority order.
+- API keys should be stored in `.env` (or environment variables), not in `litellm_config.yaml`.
+
+#### Priority modes
+
+You can choose how automatic fallback ordering is computed:
+
+- `provider-first`
+  providers are ordered by provider priority (defaults + overrides)
+- `model-first`
+  providers are ordered by priority rules for the provider's configured model (exact/prefix)
+- `mixed`
+  providers are ordered by the best (lowest) priority from either provider priority or model rules
+
+Configure the mode:
+
+```bash
+code2logic llm priority set-mode provider-first
+code2logic llm priority set-mode model-first
+code2logic llm priority set-mode mixed
+```
+
+Model priority rules are stored in `~/.code2logic/llm_config.json`.
 ```
 
 ### Python API
@@ -271,6 +333,8 @@ python examples/09_async_benchmark.py --folder tests/samples/
 
 # Function-level reproduction
 python examples/10_function_reproduction.py --multi-lang
+
+python examples/15_unified_benchmark.py --folder tests/samples/
 ```
 
 ## 🤝 Contributing
@@ -283,8 +347,21 @@ MIT License - see [LICENSE](LICENSE) for details.
 
 ## 📚 Documentation
 
-- [API Documentation](DOCS.md) - Complete API reference
-- [Refactoring Plan](TODO.md) - Development roadmap
+- [00 - Docs Index](docs/00-index.md) - Documentation home (start here)
+- [01 - Getting Started](docs/01-getting-started.md) - Install and first steps
+- [02 - Configuration](docs/02-configuration.md) - API keys, environment setup
+- [03 - CLI Reference](docs/03-cli-reference.md) - Command-line usage
+- [04 - Python API](docs/04-python-api.md) - Programmatic usage
+- [05 - Output Formats](docs/05-output-formats.md) - Format comparison and usage
+- [06 - Format Specifications](docs/06-format-specifications.md) - Detailed format specs
+- [07 - TOON Format](docs/07-toon.md) - Token-Oriented Object Notation
+- [08 - LLM Integration](docs/08-llm-integration.md) - OpenRouter/Ollama/LiteLLM
+- [09 - LLM Comparison](docs/09-llm-comparison-report.md) - Provider/model comparison
+- [10 - Benchmarking](docs/10-benchmark.md) - Benchmark methodology and results
+- [11 - Repeatability](docs/11-repeatability.md) - Repeatability testing
+- [12 - Examples](docs/12-examples.md) - Usage workflows and examples
+- [13 - Architecture](docs/13-architecture.md) - System design and components
+- [14 - Format Analysis](docs/14-format-analysis.md) - Deeper format evaluation
 
 ## 🔗 Links
 

{code2logic-1.0.1 → code2logic-1.0.3}/code2logic/__init__.py

@@ -18,7 +18,7 @@ Example:
     >>> print(output)
 """
 
-__version__ = "1.0.1"
+__version__ = "1.0.3"
 __author__ = "Softreck"
 __email__ = "info@softreck.dev"
 __license__ = "MIT"
@@ -146,6 +146,46 @@ from .prompts import (
     get_review_prompt,
     get_fix_prompt,
 )
+from .schemas import (
+    validate_yaml,
+    validate_logicml,
+    validate_markdown,
+    validate_json,
+    YAMLSchema,
+    LogicMLSchema,
+    MarkdownSchema,
+    JSONSchema,
+)
+from .quality import (
+    QualityAnalyzer,
+    QualityReport,
+    QualityIssue,
+    analyze_quality,
+    get_quality_summary,
+)
+from .similarity import get_refactoring_suggestions
+from .errors import (
+    ErrorHandler,
+    ErrorType,
+    ErrorSeverity,
+    AnalysisError,
+    AnalysisResult,
+    create_error_handler,
+)
+from .reproducer import (
+    SpecReproducer,
+    SpecValidator,
+    ReproductionResult,
+    FileValidation,
+    reproduce_project,
+    validate_files,
+)
+from .toon_format import (
+    TOONGenerator,
+    TOONParser,
+    generate_toon,
+    parse_toon,
+)
 from .chunked_reproduction import (
     ChunkedReproducer,
     ChunkedResult,
@@ -155,6 +195,23 @@ from .chunked_reproduction import (
     auto_chunk_reproduce,
     get_llm_limit,
 )
+from .llm_profiler import (
+    LLMProfiler,
+    LLMProfile,
+    AdaptiveChunker,
+    profile_llm,
+    get_profile,
+    get_or_create_profile,
+    get_adaptive_chunker,
+    load_profiles,
+    save_profile,
+)
+from .terminal import (
+    ShellRenderer,
+    render,
+    get_renderer,
+    set_renderer,
+)
 
 __all__ = [
     # Version
@@ -262,4 +319,19 @@ __all__ = [
     "MarkdownSpec",
     "generate_markdown_hybrid",
     "generate_file_markdown",
+    # LLM Profiler
+    "LLMProfiler",
+    "LLMProfile",
+    "AdaptiveChunker",
+    "profile_llm",
+    "get_profile",
+    "get_or_create_profile",
+    "get_adaptive_chunker",
+    "load_profiles",
+    "save_profile",
+    # Terminal Rendering
+    "ShellRenderer",
+    "render",
+    "get_renderer",
+    "set_renderer",
 ]

code2logic-1.0.3/code2logic/benchmarks/__init__.py

@@ -0,0 +1,19 @@
+from .common import (
+    create_single_project,
+    generate_spec,
+    generate_spec_token,
+    get_async_reproduction_prompt,
+    get_token_reproduction_prompt,
+    get_simple_reproduction_prompt,
+)
+from .results import (
+    BenchmarkResult,
+    BenchmarkConfig,
+    FileResult,
+    FunctionResult,
+    FormatResult,
+)
+from .runner import (
+    BenchmarkRunner,
+    run_benchmark,
+)

code2logic-1.0.3/code2logic/benchmarks/common.py

@@ -0,0 +1,258 @@
+from __future__ import annotations
+
+from datetime import datetime
+from pathlib import Path
+import json
+
+from ..gherkin import GherkinGenerator
+from ..generators import JSONGenerator, YAMLGenerator
+from ..logicml import LogicMLGenerator
+from ..markdown_format import MarkdownHybridGenerator
+from ..toon_format import TOONGenerator
+from ..models import ProjectInfo
+
+
+def create_single_project(module_info, file_path: Path) -> ProjectInfo:
+    return ProjectInfo(
+        name=file_path.name,
+        root_path=str(file_path.parent),
+        languages={"python": 1},
+        modules=[module_info],
+        dependency_graph={},
+        dependency_metrics={},
+        entrypoints=[],
+        similar_functions={},
+        total_files=1,
+        total_lines=module_info.lines_total,
+        generated_at=datetime.now().isoformat(),
+    )
+
+
+def generate_spec(project: ProjectInfo, fmt: str) -> str:
+    if fmt == "gherkin":
+        gen = GherkinGenerator()
+        return gen.generate(project)
+    if fmt == "yaml":
+        gen = YAMLGenerator()
+        return gen.generate(project, detail="full")
+    if fmt == "markdown":
+        gen = MarkdownHybridGenerator()
+        spec = gen.generate(project)
+        return spec.content
+    if fmt == "json":
+        gen = JSONGenerator()
+        return gen.generate(project, detail="full")
+    if fmt == "logicml":
+        gen = LogicMLGenerator()
+        spec = gen.generate(project)
+        return spec.content
+    if fmt == "toon":
+        gen = TOONGenerator()
+        return gen.generate(project, detail="full")
+    return ""
+
+
+def _generate_token_json(project: ProjectInfo) -> str:
+    """Generate compact, token-friendly JSON spec (used by examples/11_token_benchmark.py)."""
+    data = {
+        "project": project.name,
+        "files": project.total_files,
+        "lines": project.total_lines,
+        "modules": [],
+    }
+
+    for m in project.modules:
+        module: dict = {
+            "path": m.path,
+            "language": m.language,
+            "imports": m.imports[:10],
+            "exports": m.exports[:10],
+        }
+
+        if m.classes:
+            module["classes"] = []
+            for c in m.classes[:20]:
+                cls = {
+                    "name": c.name,
+                    "bases": c.bases,
+                    "doc": (c.docstring[:80] if c.docstring else ""),
+                    "properties": c.properties[:15],
+                    "methods": [
+                        {
+                            "name": method.name,
+                            "params": method.params[:5],
+                            "returns": method.return_type or "None",
+                            "doc": (method.intent[:50] if method.intent else ""),
+                            "async": method.is_async,
+                        }
+                        for method in c.methods[:15]
+                    ],
+                }
+                module["classes"].append(cls)
+
+        if m.functions:
+            module["functions"] = [
+                {
+                    "name": f.name,
+                    "params": f.params[:6],
+                    "returns": f.return_type or "None",
+                    "doc": (f.intent[:60] if f.intent else ""),
+                    "async": f.is_async,
+                    "lines": f.lines,
+                }
+                for f in m.functions[:20]
+            ]
+
+        data["modules"].append(module)
+
+    return json.dumps(data, indent=2)
+
+
+def _generate_token_json_compact(project: ProjectInfo) -> str:
+    data = json.loads(_generate_token_json(project))
+    return json.dumps(data, separators=(",", ":"))
+
+
+def generate_spec_token(project: ProjectInfo, fmt: str) -> str:
+    """Generate spec optimized for token benchmark (keeps historical behavior).
+
+    Notes:
+    - json/json_compact use the token-friendly JSON representation.
+    - other formats delegate to generate_spec.
+    """
+    if fmt == "json":
+        return _generate_token_json(project)
+    if fmt == "json_compact":
+        return _generate_token_json_compact(project)
+    return generate_spec(project, fmt)
+
+
+def get_async_reproduction_prompt(spec: str, fmt: str, file_name: str, with_tests: bool = False) -> str:
+    base_prompts = {
+        "gherkin": f"""Generate Python code from this Gherkin/BDD specification.
+Implement all scenarios as working, production-ready code.
+
+{spec[:6000]}
+
+Requirements:
+- Generate complete, working Python code for {file_name}
+- Include all imports
+- Use type hints
+- Add docstrings""",
+        "yaml": f"""Generate Python code from this YAML specification.
+Match the structure exactly with all classes and functions.
+
+{spec[:6000]}
+
+Requirements:
+- Generate complete, working Python code for {file_name}
+- Include all imports  
+- Use type hints
+- Implement all methods with actual logic""",
+        "markdown": f"""Generate Python code from this Markdown specification.
+It contains embedded Gherkin (behaviors) and YAML (structures).
+
+{spec[:6000]}
+
+Requirements:
+- Generate complete, working Python code for {file_name}
+- Include all imports
+- Implement all classes and functions
+- Use type hints throughout""",
+    }
+
+    prompt = base_prompts.get(fmt, base_prompts["yaml"])
+
+    if with_tests:
+        prompt += """
+
+IMPORTANT: Also generate a unittest test class at the end of the file.
+Include tests for each function/method with at least 2 test cases each.
+Use unittest.TestCase as base class.
+Name the test class Test<ClassName> or TestFunctions."""
+
+    return prompt
+
+
+def get_token_reproduction_prompt(spec: str, fmt: str, file_name: str) -> str:
+    format_hints = {
+        "json": "Parse the JSON structure and implement all classes and functions.",
+        "json_compact": "Parse the compact JSON and implement all elements.",
+        "yaml": "Parse the YAML structure and implement all classes and functions with exact signatures.",
+        "gherkin": "Implement scenarios as SIMPLE, MINIMAL Python code. NO extra error classes, NO over-engineering. Keep code short and direct.",
+        "markdown": "Parse embedded Gherkin (behaviors) and YAML (structures).",
+        "logicml": """Parse LogicML and generate VALID Python code:
+- 'sig: (params) -> Type' = def func(params) -> Type
+- 'sig: async (params)' = async def func(params)
+- 'sig: @property (self)' = @property decorator
+- 'bases: [BaseModel]' = class X(BaseModel) with Field()
+- 'type: re-export' = from .module import X
+CRITICAL: Ensure valid syntax - balanced brackets, proper indentation, no undefined variables.""",
+        "toon": """Parse TOON (Token-Oriented Object Notation) format carefully:
+
+STRUCTURE:
+- 'imports[N]: mod1,mod2' = import statements to include
+- 'classes[N]{name,bases,decorators,props,methods}:' = class definitions
+- 'functions[N]{name,sig,decorators,async,category,lines}:' = function definitions
+- 'function_docs:' section = docstrings/intent for each function
+
+SIGNATURE FORMAT '(params)->ReturnType':
+- 'sig: (self;x: int;y: str)->bool' = def func(self, x: int, y: str) -> bool
+- 'sig: (self)->None' = def func(self) -> None
+- Semicolons separate params, '->' indicates return type
+
+DECORATORS:
+- 'decorators: @property' = add @property decorator
+- 'decorators: @staticmethod|@cache' = multiple decorators
+
+CRITICAL: Use imports[], function_docs, and exact signatures to reproduce code accurately.""",
+    }
+
+    max_spec = 5000
+    spec_truncated = spec[:max_spec] if len(spec) > max_spec else spec
+
+    prompt = f"""Generate Python code from this {fmt.upper()} specification.
+{format_hints.get(fmt, '')}
+
+{spec_truncated}
+
+Requirements:
+- Complete, working Python code for {file_name}
+- Include imports and type hints
+- Implement all functions with actual logic
+
+```python
+"""
+    return prompt
+
+
+def get_simple_reproduction_prompt(spec: str, fmt: str, file_name: str) -> str:
+    prompts = {
+        "gherkin": f"""Generate Python code from this Gherkin/BDD specification.
+Implement all scenarios as working code.
+
+{spec[:5000]}
+
+Generate complete Python code for {file_name}:""",
+        "yaml": f"""Generate Python code from this YAML specification.
+Match the structure exactly.
+
+{spec[:5000]}
+
+Generate complete Python code for {file_name}:""",
+        "markdown": f"""Generate Python code from this Markdown specification.
+It contains embedded Gherkin and YAML sections.
+
+{spec[:5000]}
+
+Generate complete Python code for {file_name}:""",
+        "logicml": f"""Generate Python code from this LogicML specification.
+'sig:' = EXACT function signature, 'does:' = docstring, 'attrs:' = class attributes.
+Match signatures EXACTLY.
+
+{spec[:5000]}
+
+Generate complete Python code for {file_name}:""",
+    }
+
+    return prompts.get(fmt, prompts["yaml"])