Commit 3c125252 authored by Steije van Schelt's avatar Steije van Schelt Committed by Mark Canterbury
Browse files

Check for untyped elements

parent 0b44d8e5

testing/xml/lint_xml.py

+49 −9
Original line number Diff line number Diff line
@@ -7,6 +7,8 @@ from colorama import Fore, Style 


import argparse



from lxml import etree



colorama.init()



ignore_paths = [Path(x) for x in ["testing/deps"]]
@@ -16,6 +18,15 @@ ignore_files = [str(Path("104000/schema/TS_104_000.xsd")), 
                str(Path("103221-2/TS_103_221_02_Configuration.xsd")),

                str(Path("103707/TS_103_707.xsd"))]



XS = "{http://www.w3.org/2001/XMLSchema}"

etree_parser = etree.XMLParser(

    resolve_entities=False,

    no_network=True,

    dtd_validation=False,

    load_dtd=False,

    huge_tree=False,

)





def print_colorized_diff_line(line: str):

    if line.startswith("-"):
@@ -35,23 +46,52 @@ def run_linter(file: Path) -> str: 
    linted_xml = completed.stdout

    return linted_xml



def get_linting_errors(file: Path) -> str | None:

def get_errors(file: Path) -> str:

    errors = get_linting_errors(file)

    errors += get_type_errors(file)



    return errors



def get_linting_errors(file: Path) -> str:

    linted_xml = run_linter(file)

    orig_xml = file.read_text(encoding="utf8")

    diff = list(unified_diff(orig_xml.splitlines(), linted_xml.splitlines()))

    if len(diff) == 0:

        return None

        return ""

    else:

        return "\n".join([print_colorized_diff_line(d) for d in diff])



def get_type_errors(file: Path) -> str:

    tree = etree.parse(file, parser=etree_parser)

    root = tree.getroot()



    errors = []



    for el in root.iter(f"{XS}element"):

        has_ref = el.get("ref") is not None

        has_type_attr = el.get("type") is not None

        has_inline_type = (

            el.find(f"{XS}simpleType") is not None

            or el.find(f"{XS}complexType") is not None

        )



        if not has_ref and not has_type_attr and not has_inline_type:

            errors.append(f"Line {el.sourceline} has untyped element {el.get('name')}")



    if len(errors) > 0:

        return "\n".join(errors)



    return ""



def prettify_file(file: Path):

    linted_xml = run_linter(file)

    linting_errors = get_linting_errors(file)

    if linting_errors:



    errors = get_errors(file)

    if errors:

        file.write_text(linted_xml, encoding="utf8")

        return True

    else:

        print("Linter returned no errors - nothing to do!")



    print("Check returned no errors - nothing to do!")

    return False
@@ -95,7 +135,7 @@ if __name__ == "__main__": 
           (str(f) in ignore_files):

            print(f"(Ignoring {f})")

            continue

        new_errors = get_linting_errors(f)

        new_errors = get_errors(f)

        if new_errors:

            files_with_errors += 1

            print(f"❌ {str(f)}")

testing/xml/requirements.txt

+1 −0
Original line number Diff line number Diff line 
colorama

lxml