tooluniverse 0.2.0__py3-none-any.whl → 1.0.1__py3-none-any.whl

tooluniverse/tool_registry.py

@@ -0,0 +1,416 @@
+"""Simplified tool registry for automatic tool discovery and registration."""
+
+import importlib
+import pkgutil
+import os
+import json
+import glob
+import logging
+from typing import Dict
+
+# Initialize logger for this module
+logger = logging.getLogger("ToolRegistry")
+
+# Global registries
+_tool_registry = {}
+_config_registry = {}
+_lazy_registry: Dict[str, str] = {}  # Maps tool names to module names
+_discovery_completed = False
+_lazy_cache = {}
+
+
+def register_tool(tool_type_name=None, config=None):
+    """
+    Decorator to automatically register tool classes and their configs.
+
+    Usage:
+        @register_tool('CustomToolName', config={...})
+        class MyTool:
+            pass
+    """
+
+    def decorator(cls):
+        name = tool_type_name or cls.__name__
+        _tool_registry[name] = cls
+
+        if config:
+            _config_registry[name] = config
+            logger.info(f"Registered tool with config: {name}")
+        else:
+            logger.debug(f"Registered tool: {name} -> {cls.__name__}")
+
+        return cls
+
+    return decorator
+
+
+def register_external_tool(tool_name, tool_class):
+    """Allow external registration of tool classes."""
+    _tool_registry[tool_name] = tool_class
+    logger.info(f"Externally registered tool: {tool_name}")
+
+
+def register_config(tool_type_name, config):
+    """Register a config for a tool type."""
+    _config_registry[tool_type_name] = config
+    logger.info(f"Registered config for: {tool_type_name}")
+
+
+def get_tool_registry():
+    """Get a copy of the current tool registry."""
+    return _tool_registry.copy()
+
+
+def get_config_registry():
+    """Get a copy of the current config registry."""
+    return _config_registry.copy()
+
+
+def lazy_import_tool(tool_name):
+    """
+    Lazily import a tool by name without importing all tool modules.
+    Only imports the specific module containing the requested tool.
+    """
+    global _tool_registry, _lazy_registry, _lazy_cache
+
+    # If tool is already in registry, return it
+    if tool_name in _tool_registry:
+        return _tool_registry[tool_name]
+
+    # If we have a lazy mapping for this tool, import its module
+    if tool_name in _lazy_registry:
+        module_name = _lazy_registry[tool_name]
+
+        # Ensure we have the full module path
+        if not module_name.startswith("tooluniverse."):
+            full_module_name = f"tooluniverse.{module_name}"
+        else:
+            full_module_name = module_name
+
+        # Only import if we haven't cached this module yet
+        if full_module_name not in _lazy_cache:
+            try:
+                logger.debug(
+                    f"Lazy importing module: {full_module_name} for tool: {tool_name}"
+                )
+                module = importlib.import_module(full_module_name)
+                _lazy_cache[full_module_name] = module
+                logger.debug(f"Successfully imported module: {full_module_name}")
+
+                # Check if the tool is now in the registry
+                if tool_name in _tool_registry:
+                    logger.debug(f"Successfully lazy-loaded tool: {tool_name}")
+                    return _tool_registry[tool_name]
+                else:
+                    logger.warning(
+                        f"Tool {tool_name} not found in module {full_module_name} after import"
+                    )
+
+            except ImportError as e:
+                logger.warning(f"Failed to lazy import {full_module_name}: {e}")
+                # Remove this bad mapping so we don't try again
+                del _lazy_registry[tool_name]
+        else:
+            # Module was already imported, check if tool is now available
+            if tool_name in _tool_registry:
+                return _tool_registry[tool_name]
+            else:
+                logger.warning(
+                    f"Tool {tool_name} not found in already imported module {full_module_name}"
+                )
+
+    # If still not found after lazy loading attempt, return None
+    # Don't fall back to full discovery as that defeats the purpose of lazy loading
+    logger.debug(f"Tool {tool_name} not found in lazy registry")
+    return None
+
+
+def build_lazy_registry(package_name=None):
+    """
+    Build a mapping of tool names to module names using config files and naming patterns.
+    This is truly lazy - it doesn't import any modules, just creates the mapping.
+    """
+    global _lazy_registry
+
+    if package_name is None:
+        package_name = "tooluniverse"
+
+    try:
+        package = importlib.import_module(package_name)
+        package_path = package.__path__
+    except (ImportError, AttributeError):
+        logger.warning(f"Could not import package {package_name}")
+        return {}
+
+    logger.debug(f"Building lazy registry for package: {package_name}")
+
+    # Strategy 1: Parse config files for accurate mappings WITHOUT importing modules
+    config_mappings = _discover_from_configs()
+    config_count = 0
+
+    for module_name, tool_classes in config_mappings.items():
+        # Don't verify module exists by importing - just trust the mapping
+        # The actual import will happen when the tool is first requested
+        for tool_class in tool_classes:
+            if tool_class not in _lazy_registry:
+                _lazy_registry[tool_class] = module_name
+        config_count += len(tool_classes)
+
+    # Strategy 2: Pattern-based fallback for modules without configs
+    pattern_count = 0
+    for _importer, modname, _ispkg in pkgutil.iter_modules(package_path):
+        if "_tool" in modname and modname not in [m for m in config_mappings.keys()]:
+            # Simple pattern: module_tool -> ModuleTool, ModuleRESTTool
+            base_name = modname.replace("_tool", "").replace("_", "")
+            potential_names = [
+                f"{base_name.title()}Tool",
+                f"{base_name.title()}RESTTool",
+                f"{base_name.upper()}Tool",
+            ]
+
+            for tool_name in potential_names:
+                if tool_name not in _lazy_registry:
+                    _lazy_registry[tool_name] = modname
+                    pattern_count += 1
+
+    logger.info(
+        f"Built lazy registry: {config_count} from configs, {pattern_count} from patterns (no modules imported)"
+    )
+    return _lazy_registry.copy()
+
+
+def _discover_from_configs():
+    """
+    Fully dynamic config file discovery - no hardcoded mappings.
+    Automatically discovers config-to-module mappings by:
+    1. Finding all JSON config files
+    2. Finding all Python tool modules
+    3. Smart matching between config names and module names
+    """
+    # Get the data directory path relative to tooluniverse module
+    try:
+        import tooluniverse
+
+        package_dir = os.path.dirname(tooluniverse.__file__)
+        data_dir = os.path.join(package_dir, "data")
+    except ImportError:
+        # Fallback: assume we're in the right directory structure
+        current_dir = os.path.dirname(os.path.abspath(__file__))
+        data_dir = os.path.join(current_dir, "data")
+
+    if not os.path.exists(data_dir):
+        logger.warning(f"Data directory not found: {data_dir}")
+        return {}
+
+    # Step 1: Get all available tool modules
+    available_modules = _get_available_tool_modules()
+    logger.debug(f"Found {len(available_modules)} tool modules: {available_modules}")
+
+    tool_mapping = {}
+
+    try:
+        for json_file in glob.glob(os.path.join(data_dir, "*.json")):
+            try:
+                config_name = os.path.basename(json_file).replace(".json", "")
+
+                # Step 2: Smart matching to find the best module for this config
+                module_name = _smart_match_config_to_module(
+                    config_name, available_modules
+                )
+
+                if not module_name:
+                    logger.debug(f"No module match found for config: {config_name}")
+                    continue
+
+                # Step 3: Extract tool types from config
+                with open(json_file, "r", encoding="utf-8") as f:
+                    config_data = json.load(f)
+
+                tool_types = set()
+                if isinstance(config_data, list):
+                    for tool_config in config_data:
+                        if isinstance(tool_config, dict) and "type" in tool_config:
+                            tool_types.add(tool_config["type"])
+                elif isinstance(config_data, dict) and "type" in config_data:
+                    tool_types.add(config_data["type"])
+
+                if tool_types and module_name:
+                    if module_name not in tool_mapping:
+                        tool_mapping[module_name] = []
+                    tool_mapping[module_name].extend(list(tool_types))
+                    logger.debug(
+                        f"Dynamic mapping: {config_name} -> {module_name} -> {tool_types}"
+                    )
+
+            except Exception as e:
+                logger.debug(f"Skipped config file {json_file}: {e}")
+                continue
+    except Exception as e:
+        logger.warning(f"Error reading config files: {e}")
+
+    logger.debug(f"Dynamically discovered {len(tool_mapping)} modules from configs")
+    return tool_mapping
+
+
+def _get_available_tool_modules():
+    """
+    Get all available tool modules by scanning the tooluniverse package.
+    """
+    try:
+        import tooluniverse
+
+        package_path = tooluniverse.__path__
+    except ImportError:
+        logger.warning("Cannot import tooluniverse package")
+        return []
+
+    modules = []
+    for _importer, modname, _ispkg in pkgutil.iter_modules(package_path):
+        if "_tool" in modname or modname in [
+            "compose_tool",
+            "agentic_tool",
+        ]:  # Include compose_tool and agentic_tool
+            modules.append(modname)
+
+    return modules
+
+
+def _smart_match_config_to_module(config_name, available_modules):
+    """
+    Smart matching algorithm to find the best module for a config file.
+    Uses multiple strategies in order of preference.
+    """
+    # Strategy 1: Direct name matching
+    # "chembl_tools" -> "chem_tool"
+    if config_name.endswith("_tools"):
+        candidate = config_name.replace("_tools", "_tool")
+        if candidate in available_modules:
+            return candidate
+
+    # Strategy 2: Exact match
+    # "chem_tool" -> "chem_tool"
+    if config_name in available_modules:
+        return config_name
+
+    # Strategy 3: Fuzzy matching based on keywords
+    # Extract key parts from config name and match with modules
+    config_parts = set(config_name.replace("_", " ").split())
+
+    best_match = None
+    best_score = 0
+
+    for module in available_modules:
+        module_parts = set(module.replace("_", " ").split())
+
+        # Calculate similarity score
+        common_parts = config_parts & module_parts
+        if common_parts:
+            score = len(common_parts) / max(len(config_parts), len(module_parts))
+            if score > best_score:
+                best_score = score
+                best_match = module
+
+    # Only return match if score is reasonably high
+    if best_score > 0.3:  # At least 30% similarity
+        return best_match
+
+    # Strategy 4: Pattern-based matching for known patterns
+    patterns = [
+        # FDA patterns
+        ("fda", "openfda_tool"),
+        ("clinicaltrials", "ctg_tool"),
+        ("clinical_trials", "ctg_tool"),
+        ("opentargets", "graphql_tool"),
+        ("monarch", "restful_tool"),
+        ("url_fetch", "url_tool"),
+        ("europe_pmc", "europe_pmc_tool"),
+        ("semantic_scholar", "semantic_scholar_tool"),
+        # ChEMBL pattern
+        ("chembl", "chem_tool"),
+    ]
+
+    for pattern, module in patterns:
+        if pattern in config_name and module in available_modules:
+            return module
+
+    return None
+
+
+def auto_discover_tools(package_name=None, lazy=True):
+    """
+    Automatically discover and import all tool modules.
+    If lazy=True, only builds the mapping without importing any modules.
+    If lazy=False, imports all tool modules immediately.
+    """
+    global _discovery_completed
+
+    if package_name is None:
+        package_name = "tooluniverse"
+
+    # In lazy mode, just build the registry without importing anything
+    if lazy:
+        if not _lazy_registry:
+            build_lazy_registry(package_name)
+            logger.debug(
+                f"Lazy discovery complete. Registry contains {len(_lazy_registry)} tool mappings (no modules imported)"
+            )
+        return _tool_registry.copy()
+
+    # Return cached registry if full discovery already done
+    if _discovery_completed:
+        return _tool_registry.copy()
+
+    try:
+        package = importlib.import_module(package_name)
+        package_path = package.__path__
+    except (ImportError, AttributeError):
+        logger.warning(f"Could not import package {package_name}")
+        return _tool_registry.copy()
+
+    logger.info(
+        f"Auto-discovering tools in package: {package_name} (lazy={lazy}) - importing ALL modules"
+    )
+
+    # Import all tool modules (non-lazy mode)
+    imported_count = 0
+    for _importer, modname, _ispkg in pkgutil.iter_modules(package_path):
+        if "_tool" in modname or modname in ["compose_tool", "agentic_tool"]:
+            try:
+                importlib.import_module(f"{package_name}.{modname}")
+                logger.debug(f"Imported tool module: {modname}")
+                imported_count += 1
+            except ImportError as e:
+                logger.warning(f"Could not import {modname}: {e}")
+
+    _discovery_completed = True
+    logger.info(
+        f"Full discovery complete. Imported {imported_count} modules, registered {len(_tool_registry)} tools"
+    )
+    return _tool_registry.copy()
+
+
+def get_tool_class_lazy(tool_name):
+    """
+    Get a tool class by name, using lazy loading if possible.
+    Only imports the specific module needed, not all modules.
+    """
+    # First try lazy import
+    tool_class = lazy_import_tool(tool_name)
+    if tool_class:
+        return tool_class
+
+    # If lazy loading fails and we haven't done full discovery yet,
+    # check if the tool exists in the current registry
+    if tool_name in _tool_registry:
+        return _tool_registry[tool_name]
+
+    # As a last resort, if full discovery hasn't been done, do it
+    # But this should be rare with a properly configured lazy registry
+    if not _discovery_completed:
+        logger.warning(
+            f"Tool {tool_name} not found in lazy registry, falling back to full discovery"
+        )
+        auto_discover_tools(lazy=False)
+        return _tool_registry.get(tool_name)
+
+    return None

tooluniverse/uniprot_tool.py

@@ -0,0 +1,155 @@
+import requests
+from typing import Any, Dict
+from .base_tool import BaseTool
+from .tool_registry import register_tool
+
+
+@register_tool("UniProtRESTTool")
+class UniProtRESTTool(BaseTool):
+    def __init__(self, tool_config: Dict):
+        super().__init__(tool_config)
+        self.endpoint = tool_config["fields"]["endpoint"]
+        self.extract_path = tool_config["fields"].get("extract_path")
+        self.timeout = 15  # Increase timeout for large entries
+
+    def _build_url(self, args: Dict[str, Any]) -> str:
+        url = self.endpoint
+        for k, v in args.items():
+            url = url.replace(f"{{{k}}}", str(v))
+        return url
+
+    def _extract_data(self, data: Dict, extract_path: str) -> Any:
+        """Custom data extraction with support for filtering"""
+
+        # Handle specific UniProt extraction patterns
+        if extract_path == "comments[?(@.commentType=='FUNCTION')].texts[*].value":
+            # Extract function comments
+            result = []
+            for comment in data.get("comments", []):
+                if comment.get("commentType") == "FUNCTION":
+                    for text in comment.get("texts", []):
+                        if "value" in text:
+                            result.append(text["value"])
+            return result
+
+        elif (
+            extract_path
+            == "comments[?(@.commentType=='SUBCELLULAR LOCATION')].subcellularLocations[*].location.value"
+        ):
+            # Extract subcellular locations
+            result = []
+            for comment in data.get("comments", []):
+                if comment.get("commentType") == "SUBCELLULAR LOCATION":
+                    for location in comment.get("subcellularLocations", []):
+                        if "location" in location and "value" in location["location"]:
+                            result.append(location["location"]["value"])
+            return result
+
+        elif extract_path == "features[?(@.type=='VARIANT')]":
+            # Extract variant features (correct type is "Natural variant")
+            result = []
+            for feature in data.get("features", []):
+                if feature.get("type") == "Natural variant":
+                    result.append(feature)
+            return result
+
+        elif (
+            extract_path
+            == "features[?(@.type=='MODIFIED RESIDUE' || @.type=='SIGNAL')]"
+        ):
+            # Extract PTM and signal features (correct types are "Modified residue" and "Signal")
+            result = []
+            for feature in data.get("features", []):
+                if feature.get("type") in ["Modified residue", "Signal"]:
+                    result.append(feature)
+            return result
+
+        elif (
+            extract_path
+            == "comments[?(@.commentType=='ALTERNATIVE PRODUCTS')].isoforms[*].isoformIds[*]"
+        ):
+            # Extract isoform IDs
+            result = []
+            for comment in data.get("comments", []):
+                if comment.get("commentType") == "ALTERNATIVE PRODUCTS":
+                    for isoform in comment.get("isoforms", []):
+                        for isoform_id in isoform.get("isoformIds", []):
+                            result.append(isoform_id)
+            return result
+
+        # For simple paths, use jsonpath_ng
+        try:
+            from jsonpath_ng import parse
+
+            expr = parse(extract_path)
+            matches = expr.find(data)
+            extracted_data = [m.value for m in matches]
+
+            # Return single item if only one match, otherwise return list
+            if len(extracted_data) == 0:
+                return {"error": f"No data found for JSONPath: {extract_path}"}
+            elif len(extracted_data) == 1:
+                return extracted_data[0]
+            else:
+                return extracted_data
+
+        except ImportError:
+            return {"error": "jsonpath_ng library is required for data extraction"}
+        except Exception as e:
+            return {
+                "error": f"Failed to extract UniProt fields using JSONPath '{extract_path}': {e}"
+            }
+
+    def run(self, arguments: Dict[str, Any]) -> Any:
+        # Build URL
+        url = self._build_url(arguments)
+        try:
+            resp = requests.get(url, timeout=self.timeout)
+            if resp.status_code != 200:
+                return {
+                    "error": f"UniProt API returned status code: {resp.status_code}",
+                    "detail": resp.text,
+                }
+            data = resp.json()
+        except requests.exceptions.Timeout:
+            return {"error": "Request to UniProt API timed out"}
+        except requests.exceptions.RequestException as e:
+            return {"error": f"Request to UniProt API failed: {e}"}
+        except ValueError as e:
+            return {"error": f"Failed to parse JSON response: {e}"}
+
+        # If extract_path is configured, extract the corresponding subset
+        if self.extract_path:
+            result = self._extract_data(data, self.extract_path)
+
+            # Handle empty results
+            if isinstance(result, list) and len(result) == 0:
+                return {"error": f"No data found for path: {self.extract_path}"}
+            elif isinstance(result, dict) and "error" in result:
+                return result
+
+            return result
+
+        return data
+
+    # Method bindings for backward compatibility
+    def get_entry_by_accession(self, accession: str) -> Any:
+        return self.run({"accession": accession})
+
+    def get_function_by_accession(self, accession: str) -> Any:
+        return self.run({"accession": accession})
+
+    def get_names_taxonomy_by_accession(self, accession: str) -> Any:
+        return self.run({"accession": accession})
+
+    def get_subcellular_location_by_accession(self, accession: str) -> Any:
+        return self.run({"accession": accession})
+
+    def get_disease_variants_by_accession(self, accession: str) -> Any:
+        return self.run({"accession": accession})
+
+    def get_ptm_processing_by_accession(self, accession: str) -> Any:
+        return self.run({"accession": accession})
+
+    def get_sequence_isoforms_by_accession(self, accession: str) -> Any:
+        return self.run({"accession": accession})