code-analyser 1.2.0__tar.gz → 1.4.0__tar.gz

{code_analyser-1.2.0 → code_analyser-1.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: code-analyser
-Version: 1.2.0
+Version: 1.4.0
 Summary: Source code analyser — part of the analyser family
 License-File: LICENSE
 Requires-Python: >=3.10
@@ -22,6 +22,8 @@ Requires-Dist: httpx>=0.27.0; extra == 'dev'
 Requires-Dist: pytest-cov>=4.1.0; extra == 'dev'
 Requires-Dist: pytest>=7.4.0; extra == 'dev'
 Requires-Dist: ruff>=0.4.0; extra == 'dev'
+Provides-Extra: embeddings
+Requires-Dist: lens-embed[text]>=0.1.1; extra == 'embeddings'
 Provides-Extra: llm
 Requires-Dist: anthropic>=0.7.0; extra == 'llm'
 Description-Content-Type: text/markdown

code_analyser-1.4.0/examples/basic_usage.md

@@ -0,0 +1,43 @@
+# code-analyser — basic usage
+
+Analyse a source file (`.py`, `.js`, `.ts`, `.html`, `.css`, `.sql`, `.ipynb`, or a `.zip` of them) for structural code signals.
+
+## Install
+
+```bash
+pip install code-analyser
+```
+
+## CLI
+
+```bash
+# Human-readable table
+code-analyser app.py
+
+# JSON output
+code-analyser app.py --json
+
+# Include LLM quality signals (requires the [llm] extra)
+code-analyser app.py --llm
+```
+
+## Python
+
+```python
+from code_analyser import CodeAnalyser
+
+result = CodeAnalyser().analyse("app.py")
+print(result.languages_detected, result.file_count)
+for f in result.files:
+    print(f.filename, f.language, f.metrics)
+```
+
+## HTTP
+
+```bash
+# Start the server (default port 8004)
+code-analyser serve
+
+# Analyse a source file via multipart upload
+curl -F file=@app.py http://localhost:8004/analyse
+```

{code_analyser-1.2.0 → code_analyser-1.4.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "code-analyser"
-version = "1.2.0"
+version = "1.4.0"
 description = "Source code analyser — part of the analyser family"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -26,6 +26,9 @@ dependencies = [
 
 [project.optional-dependencies]
 llm = ["anthropic>=0.7.0"]
+embeddings = [
+    "lens-embed[text]>=0.1.1",
+]
 dev = [
     "pytest>=7.4.0",
     "pytest-cov>=4.1.0",
@@ -37,6 +40,7 @@ dev = [
 # the published wheel, which keeps the plain `lens-contract>=0.2.0` PyPI pin.
 [tool.uv.sources]
 lens-contract = { path = "../lens-contract", editable = true }
+lens-embed = { path = "../lens-embed", editable = true }
 
 [project.scripts]
 code-analyser = "code_analyser.cli:main"

code_analyser-1.4.0/src/code_analyser/__init__.py

@@ -0,0 +1,21 @@
+from importlib.metadata import version as _v
+from pathlib import Path
+
+from .manifest import MANIFEST
+from .models import CodeAnalysis
+from .pipeline import CodeAnalyser
+
+__version__ = _v("code-analyser")
+del _v
+
+
+def analyse(path: str | Path, *, llm: bool = False) -> CodeAnalysis:
+    """Analyse ``path`` and return a :class:`CodeAnalysis`.
+
+    Module-level convenience for the family's canonical call shape — equivalent
+    to ``CodeAnalyser().analyse(path)``.
+    """
+    return CodeAnalyser().analyse(Path(path), llm=llm)
+
+
+__all__ = ["CodeAnalyser", "CodeAnalysis", "analyse", "MANIFEST", "__version__"]

code_analyser-1.4.0/src/code_analyser/embedding.py

@@ -0,0 +1,25 @@
+"""Source-code embedding via the family's shared helper (lens-embed).
+
+A single pinned model across the family means this vector is comparable to
+other members' vectors — the basis for cross-artefact and cohort-distinctiveness
+signals computed downstream. Opt-in and degradable: install the [embeddings]
+extra to populate it; without it (or on any failure) this returns None.
+"""
+
+from __future__ import annotations
+
+
+def embed_document(text: str) -> list[float] | None:
+    """Pooled, L2-normalised vector, or None if embeddings are off."""
+    if not text or not text.strip():
+        return None
+    try:
+        from lens_embed import backend_available, embed_long_text
+    except ImportError:
+        return None
+    if not backend_available("text"):
+        return None
+    try:
+        return embed_long_text(text)
+    except Exception:
+        return None

{code_analyser-1.2.0 → code_analyser-1.4.0}/src/code_analyser/models.py

@@ -191,3 +191,6 @@ class CodeAnalysis(BaseModel):
     files: list[FileAnalysis]
     cross_file: CrossFileSignals
     llm_signals: TopLevelLLMSignals | None = None
+    # Pooled, L2-normalised source vector from lens-embed (pinned all-MiniLM-L6-v2).
+    # Comparable across members; None unless [embeddings] installed.
+    embedding: list[float] | None = None

{code_analyser-1.2.0 → code_analyser-1.4.0}/src/code_analyser/pipeline.py

@@ -3,6 +3,7 @@ import zipfile
 from pathlib import Path
 
 from .detect import detect_language
+from .embedding import embed_document
 from .models import (
     CodeAnalysis, CrossFileSignals, FileAnalysis,
     FileLLMSignals, TopLevelLLMSignals,
@@ -79,6 +80,9 @@ class CodeAnalyser:
             frameworks_detected=sorted(all_frameworks),
         )
 
+        combined_source = "\n\n".join(_decode(content) for _, content in pairs)
+        embedding = embed_document(combined_source)
+
         return CodeAnalysis(
             input=path.name,
             file_count=len(files),
@@ -86,6 +90,7 @@ class CodeAnalyser:
             files=files,
             cross_file=cross,
             llm_signals=llm_top,
+            embedding=embedding,
         )
 
 

code_analyser-1.4.0/tests/test_embedding.py

@@ -0,0 +1,25 @@
+"""Code embedding wiring — field presence + graceful degradation."""
+from __future__ import annotations
+import importlib.util
+import pytest
+from code_analyser.embedding import embed_document
+from code_analyser.models import CodeAnalysis
+
+_TEXT = importlib.util.find_spec("lens_embed") is not None and importlib.util.find_spec("sentence_transformers") is not None
+
+def test_field_default_none():
+    assert "embedding" in CodeAnalysis.model_fields
+    assert CodeAnalysis.model_fields["embedding"].default is None
+
+def test_empty_is_none():
+    assert embed_document("") is None
+    assert embed_document("  \n ") is None
+
+@pytest.mark.skipif(_TEXT, reason="embeddings extra installed")
+def test_none_without_backend():
+    assert embed_document("def add(a, b):\n    return a + b\n") is None
+
+@pytest.mark.skipif(not _TEXT, reason="needs [embeddings]")
+def test_vector_with_backend():
+    v = embed_document("def add(a, b):\n    return a + b\n\n" * 5)
+    assert isinstance(v, list) and len(v) == 384

code_analyser-1.4.0/tests/test_public_api.py

@@ -0,0 +1,29 @@
+"""The canonical public surface every family analyser exposes.
+
+See lens-analysers/CONVENTIONS.md: each `-analyser` engine exports its
+`<Name>Analyser` class, the `<Name>Analysis` result model, a module-level
+`analyse()` convenience function, `MANIFEST`, and `__version__`.
+"""
+
+from __future__ import annotations
+
+import code_analyser
+
+
+def test_canonical_surface_importable():
+    from code_analyser import (  # noqa: F401
+        MANIFEST,
+        CodeAnalyser,
+        CodeAnalysis,
+        analyse,
+    )
+
+    assert callable(analyse)
+    assert callable(CodeAnalyser)
+    assert MANIFEST["name"] == "code-analyser"
+    assert isinstance(code_analyser.__version__, str)
+
+
+def test_surface_in_dunder_all():
+    for name in ("CodeAnalyser", "CodeAnalysis", "analyse", "MANIFEST", "__version__"):
+        assert name in code_analyser.__all__