rossum-mcp 1.1.0__py3-none-any.whl → 1.1.1__py3-none-any.whl

rossum_mcp/tools/schemas/patching.py

@@ -0,0 +1,202 @@
+"""Schema patching utilities for Rossum MCP Server."""
+
+from __future__ import annotations
+
+import copy
+from typing import Literal
+
+PatchOperation = Literal["add", "update", "remove"]
+
+
+def _find_node_in_children(
+    children: list[dict], node_id: str, parent_node: dict | None = None
+) -> tuple[dict | None, int | None, list[dict] | None, dict | None]:
+    """Recursively find a node by ID in schema children.
+
+    Returns (node, index, parent_children_list, parent_node) or (None, None, None, None) if not found.
+    The parent_node is needed for multivalue's dict children where we need to modify the parent directly.
+    """
+    for i, child in enumerate(children):
+        if child.get("id") == node_id:
+            return child, i, children, parent_node
+
+        nested_children = child.get("children")
+        if nested_children:
+            if isinstance(nested_children, list):
+                result = _find_node_in_children(nested_children, node_id, child)
+                if result[0] is not None:
+                    return result
+            elif isinstance(nested_children, dict):
+                if nested_children.get("id") == node_id:
+                    return nested_children, 0, None, child
+                if "children" in nested_children:
+                    result = _find_node_in_children(nested_children["children"], node_id, nested_children)
+                    if result[0] is not None:
+                        return result
+
+    return None, None, None, None
+
+
+def _is_multivalue_node(node: dict) -> bool:
+    """Check if a node is a multivalue (has dict children or category is multivalue)."""
+    return node.get("category") == "multivalue" or ("children" in node and isinstance(node["children"], dict))
+
+
+def _find_parent_children_list(content: list[dict], parent_id: str) -> tuple[list[dict] | None, bool]:
+    """Find the children list of a parent node by its ID.
+
+    Returns (children_list, is_multivalue) tuple.
+    For multivalue nodes, returns (None, True) since they can't have children added.
+    """
+    for section in content:
+        if section.get("id") == parent_id:
+            if _is_multivalue_node(section):
+                return None, True
+            children: list[dict] = section.setdefault("children", [])
+            return children, False
+
+        section_children = section.get("children")
+        if section_children is None:
+            continue
+
+        if isinstance(section_children, list):
+            node, _, _, _ = _find_node_in_children(section_children, parent_id)
+        else:
+            if section_children.get("id") == parent_id:
+                node = section_children
+            elif "children" in section_children:
+                node, _, _, _ = _find_node_in_children(section_children.get("children", []), parent_id)
+            else:
+                node = None
+
+        if node is not None:
+            if _is_multivalue_node(node):
+                return None, True
+            if "children" in node:
+                if isinstance(node["children"], list):
+                    result: list[dict] = node["children"]
+                    return result, False
+            else:
+                node["children"] = []
+                node_children: list[dict] = node["children"]
+                return node_children, False
+
+    return None, False
+
+
+def _apply_add_operation(
+    content: list[dict], node_id: str, node_data: dict | None, parent_id: str | None, position: int | None
+) -> list[dict]:
+    if node_data is None:
+        raise ValueError("node_data is required for 'add' operation")
+    if parent_id is None:
+        raise ValueError("parent_id is required for 'add' operation")
+
+    node_data = copy.deepcopy(node_data)
+    node_data["id"] = node_id
+
+    parent_children, is_multivalue = _find_parent_children_list(content, parent_id)
+    if is_multivalue:
+        raise ValueError(
+            f"Cannot add children to multivalue '{parent_id}'. "
+            "Multivalue nodes have a single child (tuple or datapoint). "
+            "Use 'update' to replace the multivalue's children, or add to the tuple inside it."
+        )
+    if parent_children is None:
+        raise ValueError(f"Parent node '{parent_id}' not found in schema")
+
+    if position is not None and 0 <= position <= len(parent_children):
+        parent_children.insert(position, node_data)
+    else:
+        parent_children.append(node_data)
+    return content
+
+
+def _get_section_children_as_list(section: dict) -> list[dict]:
+    """Get section children as a list, handling both list and dict (multivalue) cases."""
+    children = section.get("children")
+    if children is None:
+        return []
+    if isinstance(children, list):
+        return children
+    if isinstance(children, dict):
+        return [children]
+    return []
+
+
+def _find_node_anywhere(
+    content: list[dict], node_id: str
+) -> tuple[dict | None, int | None, list[dict] | None, dict | None]:
+    """Find a node by ID anywhere in the schema content.
+
+    Returns (node, index, parent_children_list, parent_node).
+    """
+    for section in content:
+        if section.get("id") == node_id:
+            return section, None, None, None
+
+        section_children = _get_section_children_as_list(section)
+        result = _find_node_in_children(section_children, node_id, section)
+        if result[0] is not None:
+            return result
+
+    return None, None, None, None
+
+
+def _apply_update_operation(content: list[dict], node_id: str, node_data: dict | None) -> list[dict]:
+    if node_data is None:
+        raise ValueError("node_data is required for 'update' operation")
+
+    node, _, _, _ = _find_node_anywhere(content, node_id)
+
+    if node is None:
+        raise ValueError(f"Node '{node_id}' not found in schema")
+
+    node.update(node_data)
+    return content
+
+
+def _apply_remove_operation(content: list[dict], node_id: str) -> list[dict]:
+    for section in content:
+        if section.get("id") == node_id and section.get("category") == "section":
+            raise ValueError("Cannot remove a section - sections must exist")
+
+    node, idx, parent_list, parent_node = _find_node_anywhere(content, node_id)
+
+    if node is None:
+        raise ValueError(f"Node '{node_id}' not found in schema")
+
+    if idx is None and parent_list is None:
+        if node.get("category") == "section":
+            raise ValueError("Cannot remove a section - sections must exist")
+        raise ValueError(f"Cannot determine how to remove node '{node_id}'")
+
+    if parent_list is not None and idx is not None:
+        parent_list.pop(idx)
+    elif parent_node is not None:
+        if parent_node.get("category") == "multivalue":
+            raise ValueError(f"Cannot remove '{node_id}' from multivalue - remove the multivalue instead")
+        raise ValueError(f"Cannot remove '{node_id}' - unexpected parent structure")
+
+    return content
+
+
+def apply_schema_patch(
+    content: list[dict],
+    operation: PatchOperation,
+    node_id: str,
+    node_data: dict | None = None,
+    parent_id: str | None = None,
+    position: int | None = None,
+) -> list[dict]:
+    """Apply a patch operation to schema content."""
+    content = copy.deepcopy(content)
+
+    if operation == "add":
+        return _apply_add_operation(content, node_id, node_data, parent_id, position)
+    if operation == "update":
+        return _apply_update_operation(content, node_id, node_data)
+    if operation == "remove":
+        return _apply_remove_operation(content, node_id)
+
+    return content

rossum_mcp/tools/schemas/pruning.py

@@ -0,0 +1,133 @@
+"""Schema pruning utilities for Rossum MCP Server."""
+
+from __future__ import annotations
+
+import copy
+
+from rossum_mcp.tools.schemas.models import SchemaTreeNode
+
+
+def _build_tree_node(node: dict) -> SchemaTreeNode:
+    """Build a lightweight tree node from a schema node."""
+    category = node.get("category", "")
+    node_id = node.get("id", "")
+    label = node.get("label", "")
+    node_type = node.get("type") if category == "datapoint" else None
+
+    children_data = node.get("children")
+    children: list[SchemaTreeNode] | None = None
+
+    if children_data is not None:
+        if isinstance(children_data, list):
+            children = [_build_tree_node(child) for child in children_data]
+        elif isinstance(children_data, dict):
+            children = [_build_tree_node(children_data)]
+
+    return SchemaTreeNode(id=node_id, label=label, category=category, type=node_type, children=children)
+
+
+def _extract_schema_tree(content: list[dict]) -> list[dict]:
+    """Extract lightweight tree structure from schema content."""
+    return [_build_tree_node(section).to_dict() for section in content]
+
+
+def _collect_all_field_ids(content: list[dict]) -> set[str]:
+    """Collect all field IDs from schema content recursively."""
+    ids: set[str] = set()
+
+    def _traverse(node: dict) -> None:
+        node_id = node.get("id")
+        if node_id:
+            ids.add(node_id)
+        children = node.get("children")
+        if children is not None:
+            if isinstance(children, list):
+                for child in children:
+                    _traverse(child)
+            elif isinstance(children, dict):
+                _traverse(children)
+
+    for section in content:
+        _traverse(section)
+
+    return ids
+
+
+def _collect_ancestor_ids(content: list[dict], target_ids: set[str]) -> set[str]:
+    """Collect all ancestor IDs for the given target field IDs.
+
+    Returns set of IDs for all parent containers (multivalue, tuple, section) of target fields.
+    """
+    ancestors: set[str] = set()
+
+    def _find_ancestors(node: dict, path: list[str]) -> None:
+        node_id = node.get("id", "")
+        current_path = [*path, node_id] if node_id else path
+
+        if node_id in target_ids:
+            ancestors.update(current_path[:-1])
+
+        children = node.get("children")
+        if children is not None:
+            if isinstance(children, list):
+                for child in children:
+                    _find_ancestors(child, current_path)
+            elif isinstance(children, dict):
+                _find_ancestors(children, current_path)
+
+    for section in content:
+        _find_ancestors(section, [])
+
+    return ancestors
+
+
+def _remove_fields_from_content(content: list[dict], fields_to_remove: set[str]) -> tuple[list[dict], list[str]]:
+    """Remove multiple fields from schema content.
+
+    Returns (modified_content, list_of_removed_field_ids).
+    Sections cannot be removed.
+    """
+    content = copy.deepcopy(content)
+    removed: list[str] = []
+
+    def _filter_children(children: list[dict]) -> list[dict]:
+        result = []
+        for child in children:
+            child_id = child.get("id", "")
+            category = child.get("category", "")
+
+            if child_id in fields_to_remove and category != "section":
+                removed.append(child_id)
+                continue
+
+            nested = child.get("children")
+            if nested is not None:
+                if isinstance(nested, list):
+                    child["children"] = _filter_children(nested)
+                elif isinstance(nested, dict):
+                    nested_id = nested.get("id", "")
+                    if nested_id in fields_to_remove:
+                        removed.append(nested_id)
+                        removed.append(child_id)
+                        continue
+                    nested_children = nested.get("children")
+                    if isinstance(nested_children, list):
+                        filtered_nested = _filter_children(nested_children)
+                        if not filtered_nested:
+                            removed.append(nested_id)
+                            removed.append(child_id)
+                            continue
+                        nested["children"] = filtered_nested
+            result.append(child)
+        return result
+
+    for section in content:
+        section_children = section.get("children")
+        if isinstance(section_children, list):
+            section["children"] = _filter_children(section_children)
+
+    removed_sections = [s.get("id", "") for s in content if not s.get("children")]
+    removed.extend(removed_sections)
+    content = [s for s in content if s.get("children")]
+
+    return content, removed

rossum_mcp/tools/schemas/validation.py

@@ -0,0 +1,128 @@
+"""Schema validation utilities for Rossum MCP Server."""
+
+from __future__ import annotations
+
+MAX_ID_LENGTH = 50
+VALID_DATAPOINT_TYPES = {"string", "number", "date", "enum", "button"}
+VALID_UI_CONFIGURATION_TYPES = {"captured", "data", "manual", "formula", "reasoning", None}
+VALID_UI_CONFIGURATION_EDIT = {"enabled", "enabled_without_warning", "disabled"}
+
+
+class SchemaValidationError(ValueError):
+    """Raised when schema validation fails."""
+
+
+def _sanitize_ui_configuration(node: dict) -> None:
+    """Remove invalid ui_configuration.type values to prevent API errors."""
+    ui_config = node.get("ui_configuration")
+    if not isinstance(ui_config, dict):
+        return
+    if "type" in ui_config and ui_config["type"] not in VALID_UI_CONFIGURATION_TYPES:
+        del ui_config["type"]
+    if "edit" in ui_config and ui_config["edit"] not in VALID_UI_CONFIGURATION_EDIT:
+        del ui_config["edit"]
+    if not ui_config:
+        del node["ui_configuration"]
+
+
+def sanitize_schema_content(content: list[dict]) -> list[dict]:
+    """Sanitize schema content by removing invalid ui_configuration values.
+
+    Recursively traverses all nodes and removes invalid ui_configuration.type
+    values that would cause API errors (e.g., 'area', 'textarea').
+    """
+
+    def _traverse(node: dict) -> None:
+        _sanitize_ui_configuration(node)
+        children = node.get("children")
+        if children is not None:
+            if isinstance(children, list):
+                for child in children:
+                    _traverse(child)
+            elif isinstance(children, dict):
+                _traverse(children)
+
+    for section in content:
+        _traverse(section)
+    return content
+
+
+def _validate_id(node_id: str, context: str = "") -> None:
+    """Validate node ID constraints."""
+    if not node_id:
+        raise SchemaValidationError(f"Node id is required{context}")
+    if len(node_id) > MAX_ID_LENGTH:
+        raise SchemaValidationError(f"Node id '{node_id}' exceeds {MAX_ID_LENGTH} characters{context}")
+
+
+def _validate_datapoint(node: dict, context: str = "") -> None:
+    """Validate a datapoint node has required fields."""
+    if "label" not in node:
+        raise SchemaValidationError(f"Datapoint missing required 'label'{context}")
+    if "type" not in node:
+        raise SchemaValidationError(f"Datapoint missing required 'type'{context}")
+    if node["type"] not in VALID_DATAPOINT_TYPES:
+        raise SchemaValidationError(
+            f"Invalid datapoint type '{node['type']}'. Must be one of: {', '.join(VALID_DATAPOINT_TYPES)}{context}"
+        )
+
+
+def _validate_tuple(node: dict, node_id: str, context: str) -> None:
+    """Validate a tuple node."""
+    if "label" not in node:
+        raise SchemaValidationError(f"Tuple missing required 'label'{context}")
+    if "id" not in node:
+        raise SchemaValidationError(f"Tuple missing required 'id'{context}")
+    children = node.get("children", [])
+    if not isinstance(children, list):
+        raise SchemaValidationError(f"Tuple children must be a list{context}")
+    for i, child in enumerate(children):
+        child_id = child.get("id", f"index {i}")
+        _validate_node(child, f" in tuple '{node_id}' child '{child_id}'")
+        if "id" not in child:
+            raise SchemaValidationError(f"Datapoint inside tuple must have 'id'{context} child index {i}")
+
+
+def _validate_multivalue(node: dict, node_id: str, context: str) -> None:
+    """Validate a multivalue node."""
+    if "label" not in node:
+        raise SchemaValidationError(f"Multivalue missing required 'label'{context}")
+    children = node.get("children")
+    if children is None:
+        raise SchemaValidationError(f"Multivalue missing required 'children'{context}")
+    if isinstance(children, list):
+        raise SchemaValidationError(f"Multivalue 'children' must be a single object (dict), not a list{context}")
+    if isinstance(children, dict):
+        _validate_node(children, f" in multivalue '{node_id}' children")
+
+
+def _validate_section(node: dict, node_id: str, context: str) -> None:
+    """Validate a section node."""
+    if "label" not in node:
+        raise SchemaValidationError(f"Section missing required 'label'{context}")
+    if "id" not in node:
+        raise SchemaValidationError(f"Section missing required 'id'{context}")
+    children = node.get("children", [])
+    if not isinstance(children, list):
+        raise SchemaValidationError(f"Section children must be a list{context}")
+    for child in children:
+        child_id = child.get("id", "unknown")
+        _validate_node(child, f" in section '{node_id}' child '{child_id}'")
+
+
+def _validate_node(node: dict, context: str = "") -> None:
+    """Validate a schema node recursively."""
+    category = node.get("category")
+    node_id = node.get("id", "")
+
+    if node_id:
+        _validate_id(node_id, context)
+
+    if category == "datapoint":
+        _validate_datapoint(node, context)
+    elif category == "tuple":
+        _validate_tuple(node, node_id, context)
+    elif category == "multivalue":
+        _validate_multivalue(node, node_id, context)
+    elif category == "section":
+        _validate_section(node, node_id, context)

{rossum_mcp-1.1.0.dist-info → rossum_mcp-1.1.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rossum-mcp
-Version: 1.1.0
+Version: 1.1.1
 Summary: MCP server for AI-powered Rossum orchestration: document workflows, debug pipelines automatically, and configure intelligent document processing through natural language.
 Author-email: "Dan Stancl (Rossum AI)" <daniel.stancl@gmail.com>
 License: MIT

{rossum_mcp-1.1.0.dist-info → rossum_mcp-1.1.1.dist-info}/RECORD

@@ -1,4 +1,4 @@
-rossum_mcp/__init__.py,sha256=Iq0Wpiois2btEXR6bHCxUOPDJCUSSksVZescvgFn2W4,58
+rossum_mcp/__init__.py,sha256=1kziDEWWbx1ItojN48mBY_ToABnNwVmR0VANUW2bQ1s,58
 rossum_mcp/logging_config.py,sha256=OH5G6K4wH_g-m55FdRO1BYXIDtOop-lD9Ps_mTMQ8eY,4792
 rossum_mcp/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 rossum_mcp/server.py,sha256=6UViyT90CP8LouYgaxwSrSEdgJY1h0tpUjuqPNvbRKk,2338
@@ -14,12 +14,17 @@ rossum_mcp/tools/hooks.py,sha256=v1yKA8TtWQiZQkY_LYUC5lm8WRX4qMSeIkI2Z8m1OK0,128
 rossum_mcp/tools/queues.py,sha256=QtGCt6pSMDmARoa9_L1fSzWjeMGFJSQLoW_sgSoBbwA,11063
 rossum_mcp/tools/relations.py,sha256=ko1_v6meM2zCEmnhaDQt-T4uX_XJcRV4-X-Pxsu_prQ,2016
 rossum_mcp/tools/rules.py,sha256=T-yZM9xMtZNZilOdm_go-9gIF3B6pxHS_qsLI_9OsWU,8214
-rossum_mcp/tools/schemas.py,sha256=9QTV24SaTrzKUPkzYuOjL1b9vmVVPJee-crbWWF7RRo,30230
 rossum_mcp/tools/users.py,sha256=uVojYtUQs4KorQxgmmPRNJtExS4GbzyckZ1W4Y_0Qrk,3047
 rossum_mcp/tools/workspaces.py,sha256=miV8XeJxdWQctx6RuYJXmnbNmNDL0SKcG-Senq6cWk0,3338
-rossum_mcp-1.1.0.dist-info/licenses/LICENSE,sha256=5nqARgtmPvoIU-1o1az3i8Qi2WOHYIn03vD6haewvEI,1087
-rossum_mcp-1.1.0.dist-info/METADATA,sha256=ItaeeHBcKO7eY4Kpg4XU6Mu0_OOgQjxOw_4NSzEJb-Q,9644
-rossum_mcp-1.1.0.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-rossum_mcp-1.1.0.dist-info/entry_points.txt,sha256=eth2XELk0QGw9-_H3QL8PQ0OE-RDfzipbSvRy0EUc2c,54
-rossum_mcp-1.1.0.dist-info/top_level.txt,sha256=cziqyWFE89hFhCfsq60Fk6JcwslNtgP01WHIJ5plMEM,11
-rossum_mcp-1.1.0.dist-info/RECORD,,
+rossum_mcp/tools/schemas/__init__.py,sha256=lHnhEFGO5gvha8PGOeW-h7vaQLkpKVnEf2qa6QlCIUU,6192
+rossum_mcp/tools/schemas/models.py,sha256=sDFA-whdMHO0BlfeRjOh2WGfN_AJGmgZ5WhOeIL2gBo,4754
+rossum_mcp/tools/schemas/operations.py,sha256=CJyQEQlH8IG8PPMZdqWqkPAF1PwyZ_MOXHD0L7WyNKU,7099
+rossum_mcp/tools/schemas/patching.py,sha256=8Qf1wC8l2y8liVe-LXgAooJ1UrJPrzgnq1dKYQ0CPGs,7499
+rossum_mcp/tools/schemas/pruning.py,sha256=OafLiWMIiwZLVGYl-fouMbMV-7ayDnPSJ1mxGn7rg80,4718
+rossum_mcp/tools/schemas/validation.py,sha256=u9TZ91QEUIRy7RECyV-WjSt2Cz0HVxLno0SHtZgecUE,5177
+rossum_mcp-1.1.1.dist-info/licenses/LICENSE,sha256=5nqARgtmPvoIU-1o1az3i8Qi2WOHYIn03vD6haewvEI,1087
+rossum_mcp-1.1.1.dist-info/METADATA,sha256=FY7FnUt4iozU7JyFszEfHmuuCI2J1Lus3TYvtFvTEno,9644
+rossum_mcp-1.1.1.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+rossum_mcp-1.1.1.dist-info/entry_points.txt,sha256=eth2XELk0QGw9-_H3QL8PQ0OE-RDfzipbSvRy0EUc2c,54
+rossum_mcp-1.1.1.dist-info/top_level.txt,sha256=cziqyWFE89hFhCfsq60Fk6JcwslNtgP01WHIJ5plMEM,11
+rossum_mcp-1.1.1.dist-info/RECORD,,