Merge multiple XML files into one. Two nodes are the same if they have the same names and attributes.

XML merger.py

#! /usr/bin/python3
# SPDX-AI-Disclosure: ai-generated
# SPDX-AI-Model: claude-haiku-4.5
# SPDX-AI-Provider: Anthropic
# SPDX-AI-Scope: Implementation only; human debugged and tested.

"""
Merge multiple XML files into one.
Two nodes are the same if they have the same names and attributes.
"""

from typing import Optional
import xml.etree.ElementTree as ET
import copy
from os import path
import sys


def elements_match(elem1: ET.Element, elem2: ET.Element) -> bool:
    """Check if two elements have the same tag name and attributes."""
    if elem1.tag != elem2.tag:
        return False
    return elem1.attrib == elem2.attrib


def merge_element(target: ET.Element, source: ET.Element) -> None:
    """Recursively merge source element's children into target element.

    For each child in source:
    - If a matching child (same tag and attributes) exists in target, recursively merge
    - Otherwise, deep copy and append the source child to target
    """
    # print(f"Iterating source: {source.tag} {source.attrib}", file=sys.stderr)
    for source_child in source:
        # Look for a matching child in target with same tag and attributes
        matching_child = None
        for target_child in target:
            if elements_match(source_child, target_child):
                matching_child = target_child
                break

        # print(f"Child element {'not found' if matching_child is None else 'found'}: {source_child.tag} {source_child.attrib}", file=sys.stderr)
        if matching_child is not None:
            # Recursively merge the matching child
            merge_element(matching_child, source_child)
        else:
            # No matching child found, deep copy and append
            target.append(copy.deepcopy(source_child))


def merge_xml(files: list[str]) -> Optional[ET.ElementTree]:
    if len(files) == 0:
        return None
    if not path.exists(files[0]):
        print(f"File {files[0]} does not exist, skipping.", file=sys.stderr)
        return merge_xml(files[1:])
    tree = ET.parse(files[0])
    root = tree.getroot()

    for file in files[1:]:
        if not path.exists(file):
            print(f"File {file} does not exist, skipping.", file=sys.stderr)
            continue
        tree_to_merge = ET.parse(file)
        root_to_merge = tree_to_merge.getroot()
        # Only merge if roots have the same tag name and attributes
        if elements_match(root, root_to_merge):
            merge_element(root, root_to_merge)
        else:
            print(f"Cannot merge {file}: different root: {root.tag}, {root.attrib}", file=sys.stderr)

    return tree


if __name__ == "__main__":
    if len(sys.argv) < 2:
        print("Usage: {} input_files...".format(sys.argv[0]), file=sys.stderr)
        raise SystemExit(1)
    input_files = sys.argv[1:]
    merged_tree = merge_xml(input_files)
    if merged_tree is not None:
        merged_tree.write(sys.stdout.buffer, xml_declaration=True, encoding="utf-8")

Example Testcase

a.xml

<?xml version='1.0' encoding='utf-8'?>
<coverage version="1">
  <file path="hello_world.c">
    <lineToCover lineNumber="24" covered="true"/>
    <lineToCover lineNumber="26" covered="true"/>
  </file>
  <file path="foo.c">
    <lineToCover lineNumber="15" covered="true"/>
    <lineToCover lineNumber="18" covered="true"/>
  </file>
</coverage>

b.xml

<?xml version='1.0' encoding='utf-8'?>
<coverage version="1">
  <file path="hello_world.c">
    <lineToCover lineNumber="28" covered="false"/>
    <lineToCover lineNumber="30" covered="false"/>
  </file>
  <file path="foo.c">
    <lineToCover lineNumber="20" covered="false"/>
    <lineToCover lineNumber="22" covered="true"/>
  </file>
</coverage>

Output

<?xml version='1.0' encoding='utf-8'?>
<coverage version="1">
  <file path="hello_world.c">
    <lineToCover lineNumber="24" covered="true" />
    <lineToCover lineNumber="26" covered="true" />
  <lineToCover lineNumber="28" covered="false" />
    <lineToCover lineNumber="30" covered="false" />
  </file>
  <file path="foo.c">
    <lineToCover lineNumber="15" covered="true" />
    <lineToCover lineNumber="18" covered="true" />
  <lineToCover lineNumber="20" covered="false" />
    <lineToCover lineNumber="22" covered="true" />
  </file>
</coverage>