xml_cdata_perf.py

import time
import xml.etree.ElementTree as ET

# Create test data
ITERATIONS = 10000
TEXT_WITH_SPECIAL_CHARS = "This is a test with <, >, &, and other symbols. " * 1000
TEXT_ESCAPED = TEXT_WITH_SPECIAL_CHARS.replace("&", "&amp;").replace("<", "&lt;").replace(">", "&gt;")
PLAIN_TEXT = "This is a test with only plain text and no special characters. " * 1000

def generate_xml_with_cdata(text):
    return f"<root><![CDATA[{text}]]></root>"

def generate_xml_with_escaped(text):
    return f"<root>{text}</root>"

def parse_xml(xml_string):
    return ET.fromstring(xml_string).text

if __name__ == "__main__":
    # Generate XML strings
    xml_cdata = generate_xml_with_cdata(TEXT_WITH_SPECIAL_CHARS)
    xml_escaped = generate_xml_with_escaped(TEXT_ESCAPED)

    # Benchmark CDATA with special characters
    start_cdata = time.perf_counter()
    for _ in range(ITERATIONS):
        parse_xml(xml_cdata)
    end_cdata = time.perf_counter()

    # Benchmark escaped text with special characters
    start_escaped = time.perf_counter()
    for _ in range(ITERATIONS):
        parse_xml(xml_escaped)
    end_escaped = time.perf_counter()

    # Generate XML strings with plain text
    xml_cdata_plain = generate_xml_with_cdata(PLAIN_TEXT)
    xml_plain = generate_xml_with_escaped(PLAIN_TEXT)

    # Benchmark CDATA with plain text
    start_cdata_plain = time.perf_counter()
    for _ in range(ITERATIONS):
        parse_xml(xml_cdata_plain)
    end_cdata_plain = time.perf_counter()

    # Benchmark plain text (no escaping)
    start_plain = time.perf_counter()
    for _ in range(ITERATIONS):
        parse_xml(xml_plain)
    end_plain = time.perf_counter()

    # Print results
    print(f"Parsing performance ({ITERATIONS} iterations):\n")
    print(f"CDATA (with special chars):     {end_cdata - start_cdata:.4f} seconds")
    print(f"Escaped text:                   {end_escaped - start_escaped:.4f} seconds")
    print(f"CDATA (plain text):             {end_cdata_plain - start_cdata_plain:.4f} seconds")
    print(f"Plain text (no escaping):       {end_plain - start_plain:.4f} seconds")

Parsing performance (10000 iterations):

CDATA (with special chars):     1.4372 seconds
Escaped text:                   2.2830 seconds
CDATA (plain text):             1.3776 seconds
Plain text (no escaping):       0.4864 seconds