jentic-openapi-traverse 1.0.0a29__py3-none-any.whl

jentic/apitools/openapi/traverse/datamodels/low/__init__.py

@@ -0,0 +1,14 @@
+"""Low-level OpenAPI datamodel traversal."""
+
+from .merge import merge_visitors
+from .path import NodePath
+from .traversal import BREAK, DataModelLowVisitor, traverse
+
+
+__all__ = [
+    "DataModelLowVisitor",
+    "traverse",
+    "BREAK",
+    "NodePath",
+    "merge_visitors",
+]

jentic/apitools/openapi/traverse/datamodels/low/introspection.py

@@ -0,0 +1,138 @@
+"""Utilities for introspecting datamodel fields."""
+
+from dataclasses import is_dataclass
+from typing import Any
+
+from jentic.apitools.openapi.datamodels.low.fields import fixed_fields, patterned_fields
+from jentic.apitools.openapi.datamodels.low.sources import (
+    FieldSource,
+    KeySource,
+    ValueSource,
+)
+
+
+__all__ = [
+    "get_traversable_fields",
+    "unwrap_value",
+    "is_datamodel_node",
+    "get_yaml_field_name",
+]
+
+
+def get_yaml_field_name(node_class: type, python_field_name: str) -> str:
+    """
+    Convert Python field name to YAML field name using metadata.
+
+    Args:
+        node_class: Datamodel class type
+        python_field_name: Python attribute name (e.g., "external_docs")
+
+    Returns:
+        YAML field name from metadata, or python_field_name if not found
+        (e.g., "externalDocs")
+    """
+    fixed = fixed_fields(node_class)
+    patterned = patterned_fields(node_class)
+
+    # Try fixed fields first, then patterned
+    field_obj = fixed.get(python_field_name) or patterned.get(python_field_name)
+    if field_obj:
+        return field_obj.metadata.get("yaml_name", python_field_name)
+
+    return python_field_name
+
+
+# Cache of field names to check per class type
+# {Info: ["title", "description", "contact", ...], Operation: [...], ...}
+_FIELD_NAMES_CACHE: dict[type, list[str]] = {}
+
+
+def get_traversable_fields(node):
+    """
+    Get all fields that should be traversed in a datamodel node.
+
+    Uses field metadata (fixed_field, patterned_field) to identify OpenAPI
+    specification fields. This leverages the explicit field marking system
+    from the datamodels package.
+
+    Caches field names per class type for performance.
+
+    Args:
+        node: Datamodel node (dataclass instance)
+
+    Returns:
+        List of (field_name, field_value) tuples
+    """
+    if not is_dataclass(node):
+        return []
+
+    node_class = type(node)
+
+    # Get or compute field names for this class
+    if node_class not in _FIELD_NAMES_CACHE:
+        # Get all OpenAPI spec fields (fixed + patterned)
+        fixed = fixed_fields(node_class)
+        patterned = patterned_fields(node_class)
+        # Combine and extract field names
+        field_names = list(fixed.keys()) + list(patterned.keys())
+        _FIELD_NAMES_CACHE[node_class] = field_names
+
+    # Use cached field names
+    result = []
+    for field_name in _FIELD_NAMES_CACHE[node_class]:
+        value = getattr(node, field_name, None)
+
+        # Skip None values
+        if value is None:
+            continue
+
+        # Check if it's traversable
+        unwrapped = unwrap_value(value)
+
+        # Skip scalar primitives
+        if isinstance(unwrapped, (str, int, float, bool, type(None))):
+            continue
+
+        result.append((field_name, value))
+
+    return result
+
+
+def unwrap_value(value: Any) -> Any:
+    """
+    Unwrap FieldSource/ValueSource/KeySource to get actual value.
+
+    Wrapper types (FieldSource, ValueSource, KeySource) are used to preserve
+    source location information for field values. This function extracts the
+    actual value from these wrappers.
+
+    This excludes datamodel nodes like Example which may have .value field
+    but are not wrapper types.
+
+    Args:
+        value: Potentially wrapped value
+
+    Returns:
+        Unwrapped value, or original if not wrapped
+    """
+    # Check if it's a wrapper type
+    if isinstance(value, (FieldSource, ValueSource, KeySource)):
+        return value.value
+    return value
+
+
+def is_datamodel_node(value):
+    """
+    Check if value is a low-level datamodel object.
+
+    Low-level datamodels are distinguished by having a root_node field,
+    which contains the YAML source location information. This excludes
+    wrapper types (FieldSource, ValueSource, KeySource) and other dataclasses.
+
+    Args:
+        value: Value to check
+
+    Returns:
+        True if it's a low-level datamodel node, False otherwise
+    """
+    return is_dataclass(value) and hasattr(value, "root_node")

jentic/apitools/openapi/traverse/datamodels/low/merge.py

@@ -0,0 +1,111 @@
+"""Visitor merging utilities (ApiDOM pattern)."""
+
+from .path import NodePath
+from .traversal import BREAK
+
+
+__all__ = ["merge_visitors"]
+
+
+def merge_visitors(*visitors) -> object:
+    """
+    Merge multiple visitors into one composite visitor (ApiDOM semantics).
+
+    Each visitor maintains independent state:
+    - If visitor[i] returns False, only that visitor skips children (resumes when leaving)
+    - If visitor[i] returns BREAK, only that visitor stops permanently
+    - Other visitors continue normally
+
+    This matches ApiDOM's per-visitor control model where each visitor can
+    independently skip subtrees or stop without affecting other visitors.
+
+    Args:
+        *visitors: Visitor objects (with visit_* methods)
+
+    Returns:
+        A new visitor object that runs all visitors with independent state
+
+    Example:
+        security_check = SecurityCheckVisitor()
+        counter = OperationCounterVisitor()
+        validator = SchemaValidatorVisitor()
+
+        # Each visitor can skip/break independently
+        merged = merge_visitors(security_check, counter, validator)
+        traverse(doc, merged)
+
+        # If security_check.visit_PathItem returns False:
+        # - security_check skips PathItem's children
+        # - counter and validator still visit children normally
+    """
+
+    class MergedVisitor:
+        """Composite visitor with per-visitor state tracking (ApiDOM pattern)."""
+
+        def __init__(self, visitors):
+            self.visitors = visitors
+            # State per visitor: None = active, NodePath = skipping, BREAK = stopped
+            self._skipping_state: list[NodePath | object | None] = [None] * len(visitors)
+
+        def _is_active(self, visitor_idx):
+            """Check if visitor is active (not skipping or stopped)."""
+            return self._skipping_state[visitor_idx] is None
+
+        def _is_skipping_node(self, visitor_idx, path):
+            """Check if we're leaving the exact node this visitor skipped."""
+            state = self._skipping_state[visitor_idx]
+            if state is None or state is BREAK:
+                return False
+            # At this point, state must be a NodePath
+            # Compare node identity (not equality)
+            assert isinstance(state, NodePath)
+            return state.node is path.node
+
+        def __getattr__(self, name):
+            """
+            Dynamically handle visit_* method calls.
+
+            Maintains per-visitor state for skip/break control.
+
+            Args:
+                name: Method name being called
+
+            Returns:
+                Callable that merges visitor results with state tracking
+
+            Raises:
+                AttributeError: If not a visit_* method
+            """
+            if not name.startswith("visit"):
+                raise AttributeError(f"'{self.__class__.__name__}' has no attribute '{name}'")
+
+            # Determine if this is a leave hook
+            is_leave_hook = name.startswith("visit_leave")
+
+            def merged_visit_method(path: NodePath):
+                for i, visitor in enumerate(self.visitors):
+                    if is_leave_hook:
+                        # Leave hook: only call if visitor is active
+                        if self._is_active(i) and hasattr(visitor, name):
+                            result = getattr(visitor, name)(path)
+                            if result is BREAK:
+                                self._skipping_state[i] = BREAK  # Stop this visitor
+                        # Resume visitor if leaving the skipped node (don't call leave hook)
+                        elif self._is_skipping_node(i, path):
+                            self._skipping_state[i] = None  # Resume for next nodes
+                    else:
+                        # Enter/visit hook: only call if visitor is active
+                        if self._is_active(i) and hasattr(visitor, name):
+                            result = getattr(visitor, name)(path)
+
+                            if result is BREAK:
+                                self._skipping_state[i] = BREAK  # Stop this visitor
+                            elif result is False:
+                                self._skipping_state[i] = path  # Skip descendants
+
+                # Never return BREAK or False - let traversal continue for all visitors
+                return None
+
+            return merged_visit_method
+
+    return MergedVisitor(visitors)

jentic/apitools/openapi/traverse/datamodels/low/path.py

@@ -0,0 +1,169 @@
+"""NodePath context for traversal."""
+
+from dataclasses import dataclass
+from typing import Any, Literal
+
+from jsonpointer import JsonPointer
+
+from .introspection import get_yaml_field_name
+
+
+__all__ = ["NodePath"]
+
+
+@dataclass(frozen=True, slots=True)
+class NodePath:
+    """
+    Context for a node during traversal.
+
+    Provides access to node, parent, ancestry, and path formatting.
+    Supports Babel-style sub-traversal via traverse() method.
+    """
+
+    node: Any  # Current datamodel object
+    parent_path: "NodePath | None"  # Reference to parent's NodePath (chain)
+    parent_field: str | None  # Field name in parent
+    parent_key: str | int | None  # Key if parent field is list/dict
+
+    @property
+    def parent(self) -> Any | None:
+        """Get parent node. Computed from parent_path for convenience."""
+        return self.parent_path.node if self.parent_path else None
+
+    @property
+    def ancestors(self) -> tuple[Any, ...]:
+        """Get ancestor nodes from root to parent. Computed for convenience."""
+        result = []
+        current = self.parent_path
+        while current is not None:
+            result.append(current.node)
+            current = current.parent_path
+        result.reverse()  # Root first
+        return tuple(result)
+
+    def create_child(
+        self, node: Any, parent_field: str | None, parent_key: str | int | None
+    ) -> "NodePath":
+        """
+        Create a child NodePath from this path.
+
+        Helper for creating child paths during traversal.
+
+        Args:
+            node: Child node
+            parent_field: Field name in current node (None for dict items to avoid duplicates)
+            parent_key: Key if field is list/dict
+
+        Returns:
+            New NodePath for the child
+        """
+        return NodePath(
+            node=node,
+            parent_path=self,
+            parent_field=parent_field,
+            parent_key=parent_key,
+        )
+
+    def traverse(self, visitor) -> None:
+        """
+        Traverse from this node as root (Babel pattern).
+
+        Allows convenient sub-traversal with a different visitor.
+
+        Args:
+            visitor: Visitor object with visit_* methods
+
+        Example:
+            class PathItemVisitor:
+                def visit_PathItem(self, path):
+                    # Only traverse GET operations
+                    if path.node.get:
+                        operation_visitor = OperationOnlyVisitor()
+                        get_path = path.create_child(
+                            node=path.node.get.value,
+                            parent_field="get",
+                            parent_key=None
+                        )
+                        get_path.traverse(operation_visitor)
+                    return False  # Skip automatic traversal
+        """
+        from .traversal import traverse
+
+        traverse(self.node, visitor)
+
+    def format_path(
+        self, *, path_format: Literal["jsonpointer", "jsonpath"] = "jsonpointer"
+    ) -> str:
+        """
+        Format path as RFC 6901 JSON Pointer or RFC 9535 Normalized JSONPath.
+
+        Args:
+            path_format: Output format - "jsonpointer" (default) or "jsonpath"
+
+        Returns:
+            JSONPointer string like "/paths/~1pets/get/responses/200"
+            or Normalized JSONPath like "$['paths']['/pets']['get']['responses']['200']"
+
+        Examples (jsonpointer):
+            "" (root)
+            "/info"
+            "/paths/~1pets/get"
+            "/paths/~1users~1{id}/parameters/0"
+            "/components/schemas/User/properties/name"
+
+        Examples (jsonpath):
+            "$" (root)
+            "$['info']"
+            "$['paths']['/pets']['get']"
+            "$['paths']['/users/{id}']['parameters'][0]"
+            "$['components']['schemas']['User']['properties']['name']"
+        """
+        # Root node
+        if self.parent_path is None:
+            return "$" if path_format == "jsonpath" else ""
+
+        # Walk back collecting all segments
+        segments: list[str | int] = []
+        current = self
+        while current.parent_path is not None:
+            # Add in reverse order (key first, then field) because we'll reverse the list
+            # This ensures field comes before key in the final path
+            if current.parent_key is not None:
+                segments.append(current.parent_key)
+            if current.parent_field:
+                # Convert Python field name to YAML name for output
+                parent_class = type(current.parent_path.node)
+                yaml_name = get_yaml_field_name(parent_class, current.parent_field)
+                segments.append(yaml_name)
+            current = current.parent_path
+
+        segments.reverse()  # Root to leaf order
+
+        if path_format == "jsonpath":
+            # RFC 9535 Normalized JSONPath: $['field'][index]['key']
+            result = ["$"]
+            for segment in segments:
+                if isinstance(segment, int):
+                    # Array index: $[0]
+                    result.append(f"[{segment}]")
+                else:
+                    # Member name: $['field']
+                    # Escape single quotes in the string
+                    escaped = str(segment).replace("'", "\\'")
+                    result.append(f"['{escaped}']")
+            return "".join(result)
+
+        # RFC 6901 JSON Pointer
+        return JsonPointer.from_parts(segments).path
+
+    def get_root(self) -> Any:
+        """
+        Get the root node of the tree.
+
+        Returns:
+            Root datamodel object
+        """
+        current = self
+        while current.parent_path is not None:
+            current = current.parent_path
+        return current.node

jentic/apitools/openapi/traverse/datamodels/low/traversal.py

@@ -0,0 +1,245 @@
+"""Core traversal functionality for low-level OpenAPI datamodels."""
+
+from jentic.apitools.openapi.datamodels.low.fields import patterned_fields
+
+from .introspection import get_traversable_fields, is_datamodel_node, unwrap_value
+from .path import NodePath
+
+
+__all__ = ["DataModelLowVisitor", "traverse", "BREAK"]
+
+
+class _BreakType: ...
+
+
+BREAK = _BreakType()
+
+
+class DataModelLowVisitor:
+    """
+    Optional base class for OpenAPI datamodel visitors.
+
+    You don't need to inherit from this class - just implement visit_* methods.
+    Inheritance is optional and provides no functionality - use for organizational purposes only.
+
+    Visitor Method Signatures:
+        Generic hooks (fire for ALL nodes):
+        - visit_enter(path: NodePath) -> None | False | BREAK
+        - visit_leave(path: NodePath) -> None | False | BREAK
+
+        Class-specific hooks (fire for matching node types):
+        - visit_ClassName(path: NodePath) -> None | False | BREAK
+        - visit_enter_ClassName(path: NodePath) -> None | False | BREAK
+        - visit_leave_ClassName(path: NodePath) -> None | False | BREAK
+
+    Dispatch Order:
+        1. visit_enter(path) - generic enter
+        2. visit_enter_ClassName(path) - specific enter
+        3. visit_ClassName(path) - main visitor
+        4. [child traversal - automatic unless False returned]
+        5. visit_leave_ClassName(path) - specific leave
+        6. visit_leave(path) - generic leave
+
+    Return Values:
+        - None: Continue traversal normally (children visited automatically)
+        - False: Skip visiting children of this node
+        - BREAK: Stop entire traversal immediately
+
+    Example (with enter/leave hooks):
+        class MyVisitor(DataModelLowVisitor):  # Optional inheritance
+            def visit_enter_Operation(self, path):
+                print(f"Entering: {path.format_path()}")
+
+            def visit_leave_Operation(self, path):
+                print(f"Leaving: {path.format_path()}")
+
+        class SimpleVisitor:  # No inheritance (duck typing works too)
+            def visit_Operation(self, path):
+                print(f"Found: {path.format_path()}")
+                # Children automatically visited
+    """
+
+    pass
+
+
+def traverse(root, visitor) -> None:
+    """
+    Traverse OpenAPI datamodel tree using visitor pattern.
+
+    The visitor can be any object with visit_* methods (duck typing).
+    Optionally inherit from DataModelLowVisitor for organizational purposes.
+
+    Children are automatically traversed unless a visitor method returns False.
+    Use enter/leave hooks for pre/post traversal logic.
+
+    Args:
+        root: Root datamodel object (OpenAPI30, OpenAPI31, or any datamodel node)
+        visitor: Object with visit_* methods
+
+    Example:
+        # Using enter/leave hooks
+        class MyVisitor(DataModelLowVisitor):
+            def visit_enter_Operation(self, path):
+                print(f"Entering operation: {path.format_path()}")
+
+            def visit_leave_Operation(self, path):
+                print(f"Leaving operation: {path.format_path()}")
+
+        # Simple visitor (duck typing - no inheritance needed)
+        class SimpleVisitor:
+            def visit_Operation(self, path):
+                print(f"Found operation: {path.format_path()}")
+                # Children automatically visited
+
+        doc = parser.parse(..., return_type=DataModelLow)
+        traverse(doc, MyVisitor())
+        traverse(doc, SimpleVisitor())
+    """
+    # Create initial root path
+    initial_path = NodePath(
+        node=root,
+        parent_path=None,
+        parent_field=None,
+        parent_key=None,
+    )
+
+    # Start traversal
+    _visit_node(visitor, initial_path)
+
+
+def _default_traverse_children(visitor, path: NodePath) -> _BreakType | None:
+    """
+    Internal child traversal logic.
+
+    Iterates through traversable fields and visits datamodel children.
+    Called automatically during traversal.
+
+    Args:
+        visitor: Visitor object with visit_* methods
+        path: Current node path
+
+    Returns:
+        BREAK to stop traversal, None otherwise
+    """
+    # Get all traversable fields
+    for field_name, field_value in get_traversable_fields(path.node):
+        unwrapped = unwrap_value(field_value)
+
+        # Handle single datamodel nodes
+        if is_datamodel_node(unwrapped):
+            child_path = path.create_child(node=unwrapped, parent_field=field_name, parent_key=None)
+            result = _visit_node(visitor, child_path)
+            if result is BREAK:
+                return BREAK
+
+        # Handle lists
+        elif isinstance(unwrapped, list):
+            for idx, item in enumerate(unwrapped):
+                if is_datamodel_node(item):
+                    child_path = path.create_child(
+                        node=item, parent_field=field_name, parent_key=idx
+                    )
+                    result = _visit_node(visitor, child_path)
+                    if result is BREAK:
+                        return BREAK
+
+        # Handle dicts
+        elif isinstance(unwrapped, dict):
+            # Check if this field is a patterned field
+            # Patterned fields (like Paths.paths, Components.schemas) should not
+            # add their field name to the path when iterating dict items
+            patterned_field_names = patterned_fields(type(path.node))
+            is_patterned = field_name in patterned_field_names
+
+            for key, value in unwrapped.items():
+                unwrapped_key = unwrap_value(key)
+                unwrapped_value = unwrap_value(value)
+
+                if is_datamodel_node(unwrapped_value):
+                    # Dict keys should be str after unwrapping (field names, paths, status codes, etc.)
+                    assert isinstance(unwrapped_key, (str, int)), (
+                        f"Expected str or int key, got {type(unwrapped_key)}"
+                    )
+                    # For patterned fields, don't include the field name in the path
+                    # (e.g., Paths.paths is patterned, so /paths/{path-key} not /paths/paths/{path-key})
+                    dict_field_name: str | None = None if is_patterned else field_name
+                    child_path = path.create_child(
+                        node=unwrapped_value,
+                        parent_field=dict_field_name,
+                        parent_key=unwrapped_key,
+                    )
+                    result = _visit_node(visitor, child_path)
+                    if result is BREAK:
+                        return BREAK
+
+    return None
+
+
+def _visit_node(visitor, path: NodePath) -> _BreakType | None:
+    """
+    Visit a single node with the visitor.
+
+    Handles enter/main/leave dispatch and control flow.
+    Duck typed - works with any object that has visit_* methods.
+
+    Args:
+        visitor: Visitor object with visit_* methods
+        path: Current node path
+
+    Returns:
+        BREAK to stop traversal, None otherwise
+    """
+    node_class = path.node.__class__.__name__
+
+    # Generic enter hook: visit_enter (fires for ALL nodes)
+    if hasattr(visitor, "visit_enter"):
+        result = visitor.visit_enter(path)
+        if result is BREAK:
+            return BREAK
+        if result is False:
+            return None  # Skip children, but continue traversal
+
+    # Try enter hook: visit_enter_ClassName
+    enter_method = f"visit_enter_{node_class}"
+    if hasattr(visitor, enter_method):
+        result = getattr(visitor, enter_method)(path)
+        if result is BREAK:
+            return BREAK
+        if result is False:
+            return None  # Skip children, but continue traversal
+
+    # Try main visitor: visit_ClassName
+    visit_method = f"visit_{node_class}"
+    skip_auto_traverse = False
+
+    if hasattr(visitor, visit_method):
+        result = getattr(visitor, visit_method)(path)
+        # Only care about BREAK and False:
+        # - BREAK: stop entire traversal
+        # - False: skip children of this node
+        # - Any other value (None, True, etc.): continue normally
+        if result is BREAK:
+            return BREAK
+        if result is False:
+            skip_auto_traverse = True
+
+    # Automatic child traversal (unless explicitly skipped)
+    if not skip_auto_traverse:
+        result = _default_traverse_children(visitor, path)
+        if result is BREAK:
+            return BREAK
+
+    # Try leave hook: visit_leave_ClassName
+    leave_method = f"visit_leave_{node_class}"
+    if hasattr(visitor, leave_method):
+        result = getattr(visitor, leave_method)(path)
+        if result is BREAK:
+            return BREAK
+
+    # Generic leave hook: visit_leave (fires for ALL nodes)
+    if hasattr(visitor, "visit_leave"):
+        result = visitor.visit_leave(path)
+        if result is BREAK:
+            return BREAK
+
+    return None

jentic/apitools/openapi/traverse/json/__init__.py

@@ -0,0 +1,20 @@
+"""Generic JSON tree traversal utilities."""
+
+from jentic.apitools.openapi.traverse.json.traversal import (
+    JSONContainer,
+    JSONPath,
+    JSONValue,
+    PathSeg,
+    TraversalNode,
+    traverse,
+)
+
+
+__all__ = [
+    "JSONContainer",
+    "JSONPath",
+    "JSONValue",
+    "PathSeg",
+    "TraversalNode",
+    "traverse",
+]