def test_scraper(filename, result_dict, params, evaluate_scraper):
    """
    Test scraper.

    :filename: Test file name
    :result_dict: Result dict containing test purpose, and parts of
                  expected results of stdout and stderr
    :params: schematron file as extra parameter
    """

    correct = parse_results(filename, "text/xml",
                            result_dict, True, params)
    scraper = SchematronScraper(filename=correct.filename,
                                mimetype="text/xml",
                                params=correct.params)
    scraper.scrape_file()

    evaluate_scraper(scraper, correct)

    if "verbose" in correct.params and correct.params["verbose"]:
        assert not partial_message_included("have been suppressed",
                                            scraper.messages())
    elif scraper.messages():
        assert partial_message_included("have been suppressed",
                                        scraper.messages())
Exemple #2
0
def test_no_wellformed():
    """Test scraper without well-formed check."""
    scraper = SchematronScraper("tests/data/text_xml/valid_1.0_wellformed.xml",
                                False)
    scraper.scrape_file()
    assert partial_message_included("Skipping scraper", scraper.messages())
    assert scraper.well_formed is None
Exemple #3
0
def test_scraper(filename, result_dict, params, evaluate_scraper):
    """Test scraper."""

    correct = parse_results(filename, "text/xml", result_dict, True, params)
    scraper = SchematronScraper(correct.filename, True, correct.params)
    scraper.scrape_file()
    correct.version = None
    correct.streams[0]["version"] = "(:unav)"
    correct.streams[0]["mimetype"] = "(:unav)"

    evaluate_scraper(scraper, correct)

    if "verbose" in correct.params and correct.params["verbose"]:
        assert not partial_message_included("have been suppressed",
                                            scraper.messages())
    elif scraper.messages():
        assert partial_message_included("have been suppressed",
                                        scraper.messages())
Exemple #4
0
def main(arguments=None):
    """Main loop"""
    usage = "usage: %prog [options] xml-file-path"

    parser = optparse.OptionParser(usage=usage)

    parser.add_option("-s", "--schemapath", dest="schemapath",
                      help="Path to schematron schemas",
                      metavar="PATH")

    (options, args) = parser.parse_args(arguments)

    if len(args) != 1:
        parser.error("Must give a path to an XML file as argument")

    if options.schemapath is None:
        parser.error("The -s switch is required")

    filename = args[0]

    if os.path.isdir(filename):
        filename = os.path.join(filename, 'mets.xml')

    scraper = SchematronScraper(
        filename, mimetype="text/xml",
        params={"schematron": options.schemapath})
    scraper.scrape_file()

    message_string = ensure_text(concat(scraper.messages()).strip())
    error_string = ensure_text(concat(scraper.errors()).strip())
    if message_string:
        print(message_string)
    if error_string:
        print(error_string, file=sys.stderr)

    if error_string or not scraper.well_formed:
        return 117
    return 0