Lib/xml/dom/minidom.py
cpython 3.14 @ ab2d84fe1023/Lib/xml/dom/minidom.py
xml.dom.minidom is CPython's built-in DOM implementation. It covers the core of DOM Level 2 Core: element, attribute, text, CDATA section, comment, processing instruction, entity reference, document type, and document nodes. The interface is intentionally minimal. There is no XPath, no CSS selector, and no HTML5-specific extension. The goal is a correct, readable implementation that handles real-world XML without external dependencies.
Parsing is handled by xml.dom.expatbuilder, which drives the expat C extension. The two public entry points, parse(file) and parseString(string), delegate immediately to that builder and return a Document node. Every node in the resulting tree is a Python object that holds its children in a plain list. Serialization walks that list recursively. No lazy loading or streaming is used: the entire document lives in memory as Python objects.
The file is large (roughly 1700 lines) because every node type is represented by its own class, and each class carries serialization, cloning, and namespace methods inline. The design mirrors the W3C interface names closely, which makes the W3C DOM specification a reliable companion when reading the code.
Map
| Lines | Symbol | Role | gopy |
|---|---|---|---|
| 1-80 | Node base class | Common tree operations: appendChild, removeChild, cloneNode, normalize | — |
| 81-280 | Document | Root node, factory methods for all other node types, getElementsByTagName | — |
| 281-500 | Element | Attribute storage, getAttribute, setAttribute, getElementsByTagName | — |
| 501-620 | Attr | Attribute node, value, owner element back-pointer | — |
| 621-750 | Text, CDATASection, Comment | Character-data nodes, splitText on Text | — |
| 751-900 | NodeList, NamedNodeMap | Sequence and map wrappers over Python lists/dicts | — |
| 901-1200 | _write_data, _serialize_* helpers | Recursive serialization to a writer object | — |
| 1201-1500 | Node.toxml, Node.toprettyxml | Public serialization API with indent and encoding options | — |
| 1501-1700 | parse, parseString | Entry points, call into expatbuilder | — |
Reading
Node base class and tree operations
Node defines the structural API shared by every node type. appendChild, insertBefore, replaceChild, and removeChild each update both the childNodes list and the parentNode back-pointer atomically. cloneNode(deep) copies the node and, when deep is true, recursively copies the entire subtree. normalize merges adjacent Text nodes, which expat can produce when CDATA sections are involved.
Document and factory methods
Document is the root node and acts as a factory for every other node type. Methods like createElement, createTextNode, createComment, and createAttributeNS all return nodes whose ownerDocument is set to the calling Document. This enforces the DOM invariant that a node can only belong to one document. importNode and adoptNode transfer nodes between documents.
Element and attribute handling
Element stores attributes in a NamedNodeMap keyed by (namespaceURI, localName) pairs. getAttribute and getAttributeNS look up attribute values directly from that map. setAttribute replaces or inserts an Attr node. Namespace-unaware and namespace-aware methods coexist: the unaware ones pass None as the namespace URI and rely on the NamedNodeMap to handle the lookup gracefully.
Serialization
toxml(encoding=None, standalone=None) writes the node and its descendants to a string or bytes object. toprettyxml(indent='\t', newl='\n', encoding=None) adds indentation by walking the tree and inserting text nodes. Both methods delegate to internal _serialize_* functions that dispatch on node type. The _write_data helper escapes &, <, >, and quotes as required by the XML specification.
# Round-trip example
from xml.dom.minidom import parseString
doc = parseString('<root><child attr="v">text</child></root>')
print(doc.toprettyxml(indent=' '))
gopy mirror
Not yet ported.