remdb 0.3.103__py3-none-any.whl → 0.3.141__py3-none-any.whl

rem/utils/schema_loader.py

@@ -195,6 +195,8 @@ def load_agent_schema(
     """
     Load agent schema from YAML file with unified search logic and caching.
 
+    Schema names are case-invariant - "Rem", "rem", "REM" all resolve to the same schema.
+
     Filesystem schemas are cached indefinitely (immutable, versioned with code).
     Database schemas (future) will be cached with TTL for invalidation.
 
@@ -218,8 +220,8 @@ def load_agent_schema(
     9. Database LOOKUP: schemas table (if enable_db_fallback=True and user_id provided)
 
     Args:
-        schema_name_or_path: Schema name or file path
-            Examples: "rem-query-agent", "contract-analyzer", "./my-schema.yaml"
+        schema_name_or_path: Schema name or file path (case-invariant for names)
+            Examples: "rem-query-agent", "Contract-Analyzer", "./my-schema.yaml"
         use_cache: If True, uses in-memory cache for filesystem schemas
         user_id: User ID for database schema lookup (required for DB fallback)
         enable_db_fallback: If True, falls back to database LOOKUP when file not found
@@ -232,8 +234,8 @@ def load_agent_schema(
         yaml.YAMLError: If schema file is invalid YAML
 
     Examples:
-        >>> # Load by short name (cached after first load)
-        >>> schema = load_agent_schema("contract-analyzer")
+        >>> # Load by short name (cached after first load) - case invariant
+        >>> schema = load_agent_schema("Contract-Analyzer")  # same as "contract-analyzer"
         >>>
         >>> # Load from custom path (not cached - custom paths may change)
         >>> schema = load_agent_schema("./my-agent.yaml")
@@ -241,11 +243,11 @@ def load_agent_schema(
         >>> # Load evaluator schema (cached)
         >>> schema = load_agent_schema("rem-lookup-correctness")
         >>>
-        >>> # Load custom user schema from database
-        >>> schema = load_agent_schema("my-custom-agent", user_id="user-123")
+        >>> # Load custom user schema from database (case invariant)
+        >>> schema = load_agent_schema("My-Agent", user_id="user-123")  # same as "my-agent"
     """
-    # Normalize the name for cache key
-    cache_key = str(schema_name_or_path).replace('agents/', '').replace('schemas/', '').replace('evaluators/', '').replace('core/', '').replace('examples/', '')
+    # Normalize the name for cache key (lowercase for case-invariant lookups)
+    cache_key = str(schema_name_or_path).replace('agents/', '').replace('schemas/', '').replace('evaluators/', '').replace('core/', '').replace('examples/', '').lower()
     if cache_key.endswith('.yaml') or cache_key.endswith('.yml'):
         cache_key = cache_key.rsplit('.', 1)[0]
 
@@ -266,13 +268,23 @@ def load_agent_schema(
         # Don't cache custom paths (they may change)
         return cast(dict[str, Any], schema)
 
-    # 2. Normalize name for package resource search
+    # 2. Normalize name for package resource search (lowercase)
     base_name = cache_key
 
-    # 3. Try custom schema paths (from registry + SCHEMA__PATHS env var)
+    # 3. Try custom schema paths (from registry + SCHEMA__PATHS env var + auto-detected)
     from ..registry import get_schema_paths
 
     custom_paths = get_schema_paths()
+
+    # Auto-detect local folders if they exist (convention over configuration)
+    auto_detect_folders = ["./agents", "./schemas", "./evaluators"]
+    for auto_folder in auto_detect_folders:
+        auto_path = Path(auto_folder)
+        if auto_path.exists() and auto_path.is_dir():
+            resolved = str(auto_path.resolve())
+            if resolved not in custom_paths:
+                custom_paths.insert(0, resolved)
+                logger.debug(f"Auto-detected schema directory: {auto_folder}")
     for custom_dir in custom_paths:
         # Try various patterns within each custom directory
         for pattern in [
@@ -359,10 +371,12 @@ async def load_agent_schema_async(
     """
     Async version of load_agent_schema for use in async contexts.
 
+    Schema names are case-invariant - "MyAgent", "myagent", "MYAGENT" all resolve to the same schema.
+
     This version accepts an existing database connection to avoid creating new connections.
 
     Args:
-        schema_name_or_path: Schema name or file path
+        schema_name_or_path: Schema name or file path (case-invariant for names)
         user_id: User ID for database schema lookup
         db: Optional existing PostgresService connection (if None, will create one)
 
@@ -375,8 +389,8 @@ async def load_agent_schema_async(
     # First try filesystem search (sync operations are fine)
     path = Path(schema_name_or_path)
 
-    # Normalize the name for cache key
-    cache_key = str(schema_name_or_path).replace('agents/', '').replace('schemas/', '').replace('evaluators/', '').replace('core/', '').replace('examples/', '')
+    # Normalize the name for cache key (lowercase for case-invariant lookups)
+    cache_key = str(schema_name_or_path).replace('agents/', '').replace('schemas/', '').replace('evaluators/', '').replace('core/', '').replace('examples/', '').lower()
     if cache_key.endswith('.yaml') or cache_key.endswith('.yml'):
         cache_key = cache_key.rsplit('.', 1)[0]
 
@@ -396,9 +410,20 @@ async def load_agent_schema_async(
 
     base_name = cache_key
 
-    # Try custom schema paths
+    # Try custom schema paths (from registry + SCHEMA__PATHS env var + auto-detected)
     from ..registry import get_schema_paths
     custom_paths = get_schema_paths()
+
+    # Auto-detect local folders if they exist (convention over configuration)
+    auto_detect_folders = ["./agents", "./schemas", "./evaluators"]
+    for auto_folder in auto_detect_folders:
+        auto_path = Path(auto_folder)
+        if auto_path.exists() and auto_path.is_dir():
+            resolved = str(auto_path.resolve())
+            if resolved not in custom_paths:
+                custom_paths.insert(0, resolved)
+                logger.debug(f"Auto-detected schema directory: {auto_folder}")
+
     for custom_dir in custom_paths:
         for pattern in [f"{base_name}.yaml", f"{base_name}.yml", f"agents/{base_name}.yaml"]:
             custom_path = Path(custom_dir) / pattern
@@ -437,7 +462,7 @@ async def load_agent_schema_async(
                 query = """
                     SELECT spec FROM schemas
                     WHERE LOWER(name) = LOWER($1)
-                    AND (user_id = $2 OR user_id = 'system')
+                    AND (user_id = $2 OR user_id = 'system' OR user_id IS NULL)
                     LIMIT 1
                 """
                 row = await db.fetchrow(query, base_name, user_id)
@@ -486,3 +511,73 @@ def validate_agent_schema(schema: dict[str, Any]) -> bool:
 
     logger.debug("Schema validation passed")
     return True
+
+
+def get_evaluator_schema_path(evaluator_name: str) -> Path | None:
+    """
+    Find the file path to an evaluator schema.
+
+    Searches standard locations for the evaluator schema YAML file:
+    - ./evaluators/{name}.yaml (local project)
+    - Custom schema paths from registry
+    - Package resources: schemas/evaluators/{name}.yaml
+
+    Args:
+        evaluator_name: Name of the evaluator (e.g., "mental-health-classifier")
+
+    Returns:
+        Path to the evaluator schema file, or None if not found
+
+    Example:
+        >>> path = get_evaluator_schema_path("mental-health-classifier")
+        >>> if path:
+        ...     print(f"Found evaluator at: {path}")
+    """
+    from ..registry import get_schema_paths
+
+    base_name = evaluator_name.lower().replace('.yaml', '').replace('.yml', '')
+
+    # 1. Try custom schema paths (from registry + auto-detected)
+    custom_paths = get_schema_paths()
+
+    # Auto-detect local folders
+    auto_detect_folders = ["./evaluators", "./schemas", "./agents"]
+    for auto_folder in auto_detect_folders:
+        auto_path = Path(auto_folder)
+        if auto_path.exists() and auto_path.is_dir():
+            resolved = str(auto_path.resolve())
+            if resolved not in custom_paths:
+                custom_paths.insert(0, resolved)
+
+    for custom_dir in custom_paths:
+        # Try various patterns within each custom directory
+        for pattern in [
+            f"{base_name}.yaml",
+            f"{base_name}.yml",
+            f"evaluators/{base_name}.yaml",
+        ]:
+            custom_path = Path(custom_dir) / pattern
+            if custom_path.exists():
+                logger.debug(f"Found evaluator schema: {custom_path}")
+                return custom_path
+
+    # 2. Try package resources
+    evaluator_search_paths = [
+        f"schemas/evaluators/{base_name}.yaml",
+        f"schemas/evaluators/rem/{base_name}.yaml",
+    ]
+
+    for search_path in evaluator_search_paths:
+        try:
+            schema_ref = importlib.resources.files("rem") / search_path
+            schema_path = Path(str(schema_ref))
+
+            if schema_path.exists():
+                logger.debug(f"Found evaluator schema in package: {schema_path}")
+                return schema_path
+        except Exception as e:
+            logger.debug(f"Could not check {search_path}: {e}")
+            continue
+
+    logger.warning(f"Evaluator schema not found: {evaluator_name}")
+    return None

rem/utils/sql_paths.py

@@ -0,0 +1,146 @@
+"""Utilities for resolving SQL file paths.
+
+Handles package SQL directory resolution and user migrations.
+
+Convention for user migrations:
+    Place custom SQL files in `./sql/migrations/` relative to your project root.
+    Files should be numbered (e.g., `100_custom_table.sql`) to control execution order.
+    Package migrations (001-099) run first, then user migrations (100+).
+"""
+
+from pathlib import Path
+from typing import List, Optional
+import importlib.resources
+
+# Convention: Default location for user-maintained migrations
+USER_SQL_DIR_CONVENTION = "sql"
+
+
+def get_package_sql_dir() -> Path:
+    """Get the SQL directory from the installed rem package.
+
+    Returns:
+        Path to the package's sql directory
+
+    Raises:
+        FileNotFoundError: If the SQL directory cannot be found
+    """
+    try:
+        # Use importlib.resources for Python 3.9+
+        sql_ref = importlib.resources.files("rem") / "sql"
+        package_sql = Path(str(sql_ref))
+        if package_sql.exists():
+            return package_sql
+    except (AttributeError, TypeError):
+        pass
+
+    # Fallback: use __file__ to find package location
+    try:
+        import rem
+        package_sql = Path(rem.__file__).parent / "sql"
+        if package_sql.exists():
+            return package_sql
+    except (ImportError, AttributeError):
+        pass
+
+    # Development fallback: check relative to cwd
+    dev_sql = Path("src/rem/sql")
+    if dev_sql.exists():
+        return dev_sql
+
+    raise FileNotFoundError(
+        "Could not locate rem SQL directory. "
+        "Ensure remdb is properly installed or run from the source directory."
+    )
+
+
+def get_package_migrations_dir() -> Path:
+    """Get the migrations directory from the installed rem package.
+
+    Returns:
+        Path to the package's migrations directory
+    """
+    return get_package_sql_dir() / "migrations"
+
+
+def get_user_sql_dir() -> Optional[Path]:
+    """Get the conventional user SQL directory if it exists.
+
+    Looks for `./sql/` relative to the current working directory.
+    This follows the convention for user-maintained migrations.
+
+    Returns:
+        Path to user sql directory if it exists, None otherwise
+    """
+    user_sql = Path.cwd() / USER_SQL_DIR_CONVENTION
+    if user_sql.exists() and user_sql.is_dir():
+        return user_sql
+    return None
+
+
+def list_package_migrations() -> List[Path]:
+    """List all migration files in the package.
+
+    Returns:
+        Sorted list of migration file paths
+    """
+    try:
+        migrations_dir = get_package_migrations_dir()
+        if migrations_dir.exists():
+            return sorted(
+                f for f in migrations_dir.glob("*.sql")
+                if f.name[0].isdigit()  # Only numbered migrations
+            )
+    except FileNotFoundError:
+        pass
+
+    return []
+
+
+def list_user_migrations() -> List[Path]:
+    """List all migration files in the user's sql/migrations directory.
+
+    Returns:
+        Sorted list of user migration file paths
+    """
+    user_sql = get_user_sql_dir()
+    if user_sql:
+        migrations_dir = user_sql / "migrations"
+        if migrations_dir.exists():
+            return sorted(
+                f for f in migrations_dir.glob("*.sql")
+                if f.name[0].isdigit()  # Only numbered migrations
+            )
+    return []
+
+
+def list_all_migrations() -> List[Path]:
+    """List all migration files from package and user directories.
+
+    Collects migrations from:
+    1. Package migrations directory
+    2. User directory (./sql/migrations/) if it exists
+
+    Files are sorted by name, so use numbered prefixes to control order:
+    - 001-099: Reserved for package migrations
+    - 100+: Recommended for user migrations
+
+    Returns:
+        Sorted list of all migration file paths (by filename)
+    """
+    all_migrations = []
+    seen_names = set()
+
+    # Package migrations first
+    for f in list_package_migrations():
+        if f.name not in seen_names:
+            all_migrations.append(f)
+            seen_names.add(f.name)
+
+    # User migrations second
+    for f in list_user_migrations():
+        if f.name not in seen_names:
+            all_migrations.append(f)
+            seen_names.add(f.name)
+
+    return sorted(all_migrations, key=lambda p: p.name)

rem/utils/vision.py

@@ -106,7 +106,7 @@ class ImageAnalyzer:
             elif provider == VisionProvider.GEMINI:
                 model = "gemini-2.0-flash-exp"
             elif provider == VisionProvider.OPENAI:
-                model = "gpt-4o"
+                model = "gpt-4.1"
 
         self.model = model
         self.base_url = base_url

rem/workers/__init__.py

@@ -1,5 +1,7 @@
 """Background workers for processing tasks."""
 
+from .db_listener import DBListener
 from .sqs_file_processor import SQSFileProcessor
+from .unlogged_maintainer import UnloggedMaintainer
 
-__all__ = ["SQSFileProcessor"]
+__all__ = ["DBListener", "SQSFileProcessor", "UnloggedMaintainer"]