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
def test_xslt_filename():
"""Test that checksum for xslt filename is calculated properly."""
# pylint: disable=protected-access
scraper = SchematronScraper("filename", "text/xml")
scraper._schematron_file = "tests/data/text_xml/supplementary/local.sch"
assert "76ed62" in scraper._generate_xslt_filename()
scraper._verbose = True
assert "ddb11a" in scraper._generate_xslt_filename()
scraper._extra_hash = "abc"
assert "550d66" in scraper._generate_xslt_filename()
scraper._verbose = False
assert "791b2e" in scraper._generate_xslt_filename()
def test_forced_filetype(result_dict, filetype, evaluate_scraper):
"""
Test using user-supplied MIME-types and versions.
"""
filetype[six.text_type("correct_mimetype")] = "text/xml"
correct = force_correct_filetype("valid_1.0_well_formed.xml", result_dict,
filetype, ["(:unav)"])
params = {
"mimetype": filetype["given_mimetype"],
"version": filetype["given_version"],
"schematron": os.path.join(ROOTPATH, "tests/data/text_xml/local.sch")
}
scraper = SchematronScraper(correct.filename, True, params)
scraper.scrape_file()
evaluate_scraper(scraper, correct)
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())
def test_parameters():
"""Test that parameters and default values work properly."""
# pylint: disable=protected-access
scraper = SchematronScraper("testsfile", "test/mimetype")
assert scraper._schematron_file is None
assert scraper._extra_hash is None
assert not scraper._verbose
assert scraper._cache
scraper = SchematronScraper("testfile", "text/xml",
params={"schematron": "schfile",
"extra_hash": "abc",
"verbose": True,
"cache": False})
assert scraper._schematron_file == "schfile"
assert scraper._extra_hash == "abc"
assert scraper._verbose
assert not scraper._cache
def test_filter_duplicate_elements():
"""Test duplicate element filtering."""
# pylint: disable=protected-access
schtest = \
b"""<svrl:schematron-output
xmlns:svrl="http://purl.oclc.org/dsdl/svrl">
<svrl:active-pattern id="id"/>
<svrl:active-pattern id="id"/>
<svrl:fired-rule context="context"/>
<svrl:fired-rule context="context"/>
<svrl:failed-assert test="test">
<svrl:text>string</svrl:text>
</svrl:failed-assert>
<svrl:failed-assert test="test 2">
<svrl:text>string</svrl:text>
</svrl:failed-assert>
<svrl:fired-rule context="context"/>
<svrl:active-pattern id="id"/>
</svrl:schematron-output>"""
scraper = SchematronScraper("filename", "text/xml")
result = scraper._filter_duplicate_elements(schtest)
assert result.count(b"<svrl:active-pattern") == 1
assert result.count(b"<svrl:fired-rule") == 1
assert result.count(b"<svrl:failed-assert") == 2
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())
def test_is_supported():
"""Test is_supported method."""
mime = "text/xml"
ver = "1.0"
assert SchematronScraper.is_supported(mime, ver, True,
{"schematron": None})
assert not SchematronScraper.is_supported(mime, ver, True)
assert SchematronScraper.is_supported(mime, None, True,
{"schematron": None})
assert not SchematronScraper.is_supported(mime, ver, False,
{"schematron": None})
assert SchematronScraper.is_supported(mime, "foo", True,
{"schematron": None})
assert not SchematronScraper.is_supported("foo", ver, True,
{"schematron": None})
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