latex2operatortree.py

from lxml import etree
import latex2mathml.converter


def mathml_to_operator_tree(node):
    """Convert MathML to an operator tree."""
    # If the node is a leaf node, return the operator as a string
    if len(node) == 0:
        return node.text
    # Otherwise, recursively build the operator tree for each child node
    operator_tree = ""
    for child in node:
        operator_tree += str(child.tag).split('}')[-1].replace('m', '') +"(" + str(mathml_to_operator_tree(child)) + ')'
    return operator_tree

def latex2tree(latex_text):
    mathml = latex2mathml.converter.convert(latex_text)
    root = etree.fromstring(mathml.encode())
    operator_tree = mathml_to_operator_tree(root[0])
    return operator_tree