Add services, prompts, and utils modules

local_code_assistant/utils/context.py

@@ -0,0 +1,186 @@
+"""Context building utilities for Local Code Assistant."""
+
+import os
+from pathlib import Path
+from typing import Optional
+
+IGNORED_PATTERNS: set[str] = {
+    "*.pyc",
+    "__pycache__/*",
+    ".git/*",
+    "node_modules/*",
+    "*.min.js",
+    "*.min.css",
+    "*.swp",
+    "*.swo",
+    ".DS_Store",
+    "*.egg-info/*",
+    ".tox/*",
+    ".venv/*",
+    "venv/*",
+    ".env",
+    "*.log",
+}
+
+
+def should_ignore(path: Path) -> bool:
+    """Check if a path should be ignored.
+
+    Args:
+        path: Path to check.
+
+    Returns:
+        True if path should be ignored, False otherwise.
+    """
+    str_path = str(path)
+    str_path_lower = str_path.lower()
+
+    for pattern in IGNORED_PATTERNS:
+        if pattern.endswith("/*"):
+            prefix = pattern[:-2]
+            if prefix in str_path_lower:
+                return True
+        elif pattern.startswith("*"):
+            ext = pattern[1:]
+            if str_path_lower.endswith(ext):
+                return True
+        else:
+            if pattern in str_path_lower:
+                return True
+
+    return False
+
+
+def _is_valid_file(file_path: Path, extensions: Optional[list[str]]) -> bool:
+    """Check if file should be included.
+
+    Args:
+        file_path: Path to check.
+        extensions: Optional list of allowed extensions.
+
+    Returns:
+        True if file should be included.
+    """
+    ext = file_path.suffix.lower()
+
+    if extensions is not None and ext not in extensions:
+        return False
+
+    if file_path.stat().st_size > 100000:
+        return False
+
+    return True
+
+
+def scan_directory(
+    directory: Path,
+    max_files: int = 10,
+    max_chars: int = 50000,
+    extensions: Optional[list[str]] = None,
+) -> dict[str, str]:
+    """Scan a directory for source files.
+
+    Args:
+        directory: Directory to scan.
+        max_files: Maximum number of files to include.
+        max_chars: Maximum total characters to include.
+        extensions: Optional list of file extensions to include.
+
+    Returns:
+        Dictionary mapping file paths to content.
+    """
+    files: dict[str, str] = {}
+    total_chars = 0
+
+    if not directory.exists() or not directory.is_dir():
+        return files
+
+    for root, dirs, filenames in os.walk(directory):
+        root_path = Path(root)
+
+        dirs[:] = [d for d in dirs if not should_ignore(root_path / d)]
+
+        for filename in filenames:
+            file_path = root_path / filename
+
+            if should_ignore(file_path):
+                continue
+
+            if not _is_valid_file(file_path, extensions):
+                continue
+
+            try:
+                content = file_path.read_text(encoding="utf-8", errors="ignore")
+
+                if total_chars + len(content) > max_chars:
+                    continue
+
+                files[str(file_path)] = content
+                total_chars += len(content)
+
+                if len(files) >= max_files:
+                    return files
+
+            except Exception:
+                continue
+
+    return files
+
+
+def build_context_string(files: dict[str, str], max_files: int = 5) -> str:
+    """Build a context string from files.
+
+    Args:
+        files: Dictionary mapping file paths to content.
+        max_files: Maximum number of files to include.
+
+    Returns:
+        Context string for prompts.
+    """
+    sorted_files = sorted(files.items(), key=lambda x: x[0])
+
+    context_parts = []
+    count = 0
+
+    for file_path, content in sorted_files:
+        if count >= max_files:
+            break
+
+        context_parts.append(f"\n=== {file_path} ===\n{content}")
+        count += 1
+
+    return "\n".join(context_parts)
+
+
+def get_project_info(directory: Path) -> dict[str, str]:
+    """Get information about a project directory.
+
+    Args:
+        directory: Project directory path.
+
+    Returns:
+        Dictionary with project information.
+    """
+    info: dict[str, str] = {}
+
+    config_files = ["pyproject.toml", "package.json", "go.mod", "Cargo.toml", "Makefile"]
+
+    for config_file in config_files:
+        config_path = directory / config_file
+        if config_path.exists():
+            try:
+                if config_file == "pyproject.toml":
+                    info["build_system"] = "python"
+                elif config_file == "package.json":
+                    info["build_system"] = "javascript"
+                elif config_file == "go.mod":
+                    info["build_system"] = "go"
+                elif config_file == "Cargo.toml":
+                    info["build_system"] = "rust"
+                elif config_file == "Makefile":
+                    info["build_system"] = "make"
+                break
+            except Exception:
+                continue
+
+    return info