XML: Token-Saving Inspection Tools (High Priority)

[jwesleye]

Overview

Add memory-efficient inspection tools for XML files to help agents avoid loading entire documents into context, similar to the CSV and JSON token-saving tools.

Motivation

XML files can be massive (thousands of elements, deeply nested). Agents often need to inspect structure and extract specific data without parsing entire documents, which wastes tokens and may exceed context windows.

Proposed Functions

High Priority - Inspection Tools

• get_xml_structure - Get element hierarchy/schema without loading content
• count_xml_elements - Count elements by tag name without loading content
• get_xml_element_at_path - Extract specific element by XPath
• get_xml_attributes - List attributes for element type without loading content
• search_xml_tags - Find all paths containing tags matching pattern

High Priority - Selective Extraction

• select_xml_elements - Get only specific elements by tag name
• filter_xml_elements - Filter elements by attribute/text criteria (6 operators)
• preview_xml_elements - Get first N elements of a specific type
• slice_xml_elements - Get range of elements (pagination support)

Medium Priority - Analysis

• get_xml_namespace_info - List all namespaces without loading content
• get_xml_element_stats - Statistics for element type (count, attributes, depth)
• validate_xml_structure_simple - Quick validation without full parse

Design Principles

• Google ADK compliant (JSON-serializable types, no defaults)
• @strands_tool decorator
• Memory-efficient (streaming/iterative parsing where possible)
• XPath support for path notation
• Consistent with CSV/JSON token-saving patterns
• Process data without loading entire documents

Related

• Related to issue XML: XPath and Stream Processing (High Priority) #44 (XPath and Stream Processing)
• Related to issue XML: Diff, Merge, and Advanced Validation (High Priority) #45 (Diff, Merge, Advanced Validation)
• Complements existing xml/parsing.py functions
• Similar pattern to JSON: get_json_value_at_path, get_json_structure, etc.

Module

xml/parsing.py or new xml/inspection.py

[jwesleye]

Implemented all 12 XML token-saving inspection tools in commit 8ef337c.

Added functions:

• get_xml_structure - Hierarchy without content
• count_xml_elements - Count by tag name
• get_xml_element_at_path - Extract by XPath
• get_xml_attributes - List unique attributes
• search_xml_tags - Find tags by pattern
• select_xml_elements - Get specific tag types only
• filter_xml_elements - Filter by attribute/text (6 operators)
• preview_xml_elements - First N elements preview
• slice_xml_elements - Pagination support
• get_xml_namespace_info - Extract namespace mappings
• get_xml_element_stats - Element statistics
• validate_xml_structure_simple - Quick validation

All functions use defusedxml when available for security, follow Google ADK standards, and include @strands_tool decorators. XML module now has 18 parsing functions (6 original + 12 new).