def test_checksums(schematron_fx, algorithm):
    """Test that the controlled vocabulary for messageDigestAlgorithm work.

    :schematron_fx: Schematron compile fixture
    :algorithm: Valid algorithm name or 'xxx' as invalid arbitrary name.
    """
    xml = '''<mets:techMD xmlns:mets="%(mets)s" xmlns:premis="%(premis)s"
               xmlns:xsi="%(xsi)s">
               <mets:mdWrap MDTYPE="PREMIS:OBJECT" MDTYPEVERSION="2.3">
                 <mets:xmlData>
                   <premis:object xsi:type="premis:file">
                     <premis:objectCharacteristics><premis:fixity>
                     <premis:messageDigestAlgorithm/>
                     </premis:fixity></premis:objectCharacteristics>
                   </premis:object>
                 </mets:xmlData>
               </mets:mdWrap>
             </mets:techMD>''' % NAMESPACES
    (mets, root) = parse_xml_string(xml)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')
    elem_handler = find_element(root, 'messageDigestAlgorithm', 'premis')
    elem_handler.text = algorithm
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    if algorithm == 'xxx':
        assert svrl.count(SVRL_FAILED) == 2
    else:
        assert svrl.count(SVRL_FAILED) == 1
def test_pronom_codes(schematron_fx, fileformat, pronom):
    """Test that checks for PRONOM code matching to MIME type is done.

    :schematron_fx: Schematron compile fixture
    :fileformat: File format as MIME type
    :pronom: Allowed PRONOM codes for the file format
    """
    xml = '''<mets:techMD xmlns:mets="%(mets)s" xmlns:premis="%(premis)s">
               <mets:mdWrap MDTYPEVERSION="2.3"><mets:xmlData><premis:object>
                 <premis:objectCharacteristics>
                 <premis:format><premis:formatDesignation><premis:formatName/>
                   </premis:formatDesignation><premis:formatRegistry>
                   <premis:formatRegistryKey/></premis:formatRegistry>
               </premis:format></premis:objectCharacteristics>
               </premis:object></mets:xmlData></mets:mdWrap>
             </mets:techMD>''' % NAMESPACES
    (mets, root) = parse_xml_string(xml)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')
    elem_handler = find_element(root, 'formatName', 'premis')
    elem_handler.text = fileformat
    elem_handler = find_element(root, 'formatRegistryKey', 'premis')

    # All acceptable PRONOM codes work
    for pronomcode in pronom:
        elem_handler.text = pronomcode
        svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
        assert svrl.count(SVRL_FAILED) == 0

    # Arbitrary PRONOM code fails
    elem_handler.text = 'xxx'
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 1
def test_identifier_value_object(schematron_fx, nonempty):
    """PREMIS identifier elements can not be empty. Test that a value in the
    element is required.

    :schematron_fx: Schematron compile fixture
    :nonempty: element that requires a value
    """
    xml = '''<mets:techMD xmlns:mets="%(mets)s" xmlns:premis="%(premis)s">
               <mets:mdWrap MDTYPE="PREMIS:OBJECT" MDTYPEVERSION="2.3">
               <mets:xmlData><premis:object><premis:objectIdentifier/>
               </premis:object></mets:xmlData></mets:mdWrap>
             </mets:techMD>''' % NAMESPACES
    (mets, root) = parse_xml_string(xml)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')
    elem_handler = find_element(root, 'objectIdentifier', 'premis')
    elem_handler = set_element(elem_handler, nonempty, 'premis')

    # Empty identifier element fails
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 1

    # Arbitrary value added
    elem_handler.text = 'xxx'
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 0
Exemplo n.º 4
0
def test_identifier_value_premis_event_agent(schematron_fx, premisroot, mdtype,
                                             nonempty):
    """PREMIS identifier elements can not be empty. Test that a value in the
    element is required.

    :schematron_fx: Schematron compile fixture
    :premisroot: PREMIS root element name
    :mdtype: MDTYPE
    :nonempty: element that requires a value
    """
    xml = '''<mets:digiprovMD xmlns:mets="%(mets)s">
               <mets:mdWrap MDTYPEVERSION="2.3"><mets:xmlData/></mets:mdWrap>
             </mets:digiprovMD>''' % NAMESPACES
    (mets, root) = parse_xml_string(xml)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')
    elem_handler = find_element(root, 'mdWrap', 'mets')
    set_attribute(elem_handler, 'MDTYPE', 'mets', mdtype)
    elem_handler = find_element(elem_handler, 'xmlData', 'mets')
    for elem in premisroot:
        elem_handler = set_element(elem_handler, elem, 'premis')
    elem_handler = set_element(elem_handler, nonempty, 'premis')

    # Empty identifier element fails
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 1

    # Arbitrary value added
    elem_handler.text = 'xxx'
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 0
def test_disallowed_attribute_premis_rightsmd(schematron_fx):
    """Test that attributes 'authority', 'authorityURI', 'valueURI' in
    PREMIS 2.3 are disallowed in PREMIS 2.2.

    :schematron_fx: Schematron compile fixture
    """
    xml = '''<mets:rightsMD xmlns:mets="%(mets)s" xmlns:premis="%(premis)s">
               <mets:mdWrap MDTYPE="PREMIS:RIGHTS" MDTYPEVERSION="2.3">
               <mets:xmlData><premis:rights><premis:rightsStatement/>
               </premis:rights></mets:xmlData></mets:mdWrap>
             </mets:rightsMD>''' % NAMESPACES
    (mets, root) = parse_xml_string(xml)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')
    elem_handler = find_element(root, 'rightsStatement', 'premis')
    elem_handler = set_element(elem_handler, 'xxx', 'premis')
    for disallowed in ['authority', 'authorityURI', 'valueURI']:
        set_attribute(elem_handler, disallowed, 'premis', 'default')

    # Works in 2.3
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 0

    # Disallowed in 2.2
    elem_wrap = find_element(root, 'mdWrap', 'mets')
    set_attribute(elem_wrap, 'MDTYPEVERSION', 'mets', '2.2')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 3

    # 2.2 works, if attributes removed
    elem_handler.clear()
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 0
Exemplo n.º 6
0
def test_disallowed_attribute_premis_digiprov(schematron_fx, premisroot,
                                              mdtype):
    """Test that attributes 'authority', 'authorityURI', 'valueURI' in
    PREMIS 2.3 are disallowed in PREMIS 2.2.

    :schematron_fx: Schematron compile fixture
    :premisroot: Root element in PREMIS.
    :mdtype: MDTYPE value
    """
    xml = '''<mets:digiprovMD xmlns:mets="%(mets)s">
               <mets:mdWrap MDTYPEVERSION="2.3"><mets:xmlData/></mets:mdWrap>
             </mets:digiprovMD>''' % NAMESPACES
    (mets, root) = parse_xml_string(xml)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')
    elem_wrap = find_element(root, 'mdWrap', 'mets')
    set_attribute(elem_wrap, 'MDTYPE', 'mets', mdtype)
    elem_handler = find_element(elem_wrap, 'xmlData', 'mets')
    elem_handler = set_element(elem_handler, premisroot, 'premis')
    elem_handler = set_element(elem_handler, 'xxx', 'premis')
    for disallowed in ['authority', 'authorityURI', 'valueURI']:
        set_attribute(elem_handler, disallowed, 'premis', 'default')

    # Works in 2.3
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 0

    # Disallowed in 2.2
    set_attribute(elem_wrap, 'MDTYPEVERSION', 'mets', '2.2')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 3

    # 2.2 works, if attributes removed
    elem_handler.clear()
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 0
Exemplo n.º 7
0
def test_addml_headerlevel_version(schematron_fx):
    """Test element 'headerLevel' with ADDML versions 8.2 and 8.3. It is
    a new element in 8.3 (forbidden in 8.2).

    :schematron_fx: Schematron compile fixture
    """
    xml = '''<addml:addml><addml:dataset><addml:reference/><addml:flatFiles>
               <addml:flatFileDefinitions><addml:flatFileDefinition>
                 <addml:recordDefinitions><addml:recordDefinition>
                   <addml:headerLevel/>
                 </addml:recordDefinition></addml:recordDefinitions>
               </addml:flatFileDefinition></addml:flatFileDefinitions>
             </addml:flatFiles></addml:dataset></addml:addml>''' \
             % NAMESPACES
    (mets, root) = parse_xml_string(METS_HEAD+xml+METS_TAIL)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')

    # Element 'headerLevel' is forbidden in 8.2
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets, params=False)
    assert svrl.count(SVRL_FIRED) == 2
    assert svrl.count(SVRL_FAILED) == 1

    # No checks with version 8.3
    elem_handler = find_element(root, 'mdWrap', 'mets')
    set_attribute(elem_handler, 'MDTYPEVERSION', 'mets', '8.3')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets, params=False)
    assert svrl.count(SVRL_FIRED) == 0
    assert svrl.count(SVRL_FAILED) == 0
Exemplo n.º 8
0
def test_addml_reference_version(schematron_fx):
    """Test element 'reference' with ADDML versions 8.2 and 8.3. It is required
    only in 8.2.

    :schematron_fx: Schematron compile fixture
    """
    xml = METS_HEAD+'''<addml:addml><addml:dataset/></addml:addml>'''+METS_TAIL

    # Element 'reference' is required in 8.2, but missing
    (mets, root) = parse_xml_string(xml)
    (mets, root) = add_containers(root, 'mets:mets/mets:amdSec')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FIRED) == 1
    assert svrl.count(SVRL_FAILED) == 1

    # Element 'reference' is required in 8.2, and it exists
    elem_handler = find_element(root, 'dataset', 'addml')
    set_element(elem_handler, 'reference', 'addml')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets)
    assert svrl.count(SVRL_FAILED) == 0

    # No checks with version 8.3
    elem_handler = find_element(root, 'mdWrap', 'mets')
    set_attribute(elem_handler, 'MDTYPEVERSION', 'mets', '8.3')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=mets, params=False)
    assert svrl.count(SVRL_FIRED) == 0
    assert svrl.count(SVRL_FAILED) == 0
def test_container():
    """Test container addition
    """
    (_, root
     ) = parse_xml_string('<mets:foo xmlns:mets="http://www.loc.gov/METS/"/>')
    (_, root) = add_containers(root, 'mets:mets/mets:amdSec')
    xml = b'<mets:mets xmlns:mets="http://www.loc.gov/METS/">' \
        b'<mets:amdSec><mets:foo/></mets:amdSec></mets:mets>'
    assert ET.tostring(root) == xml
def prepare_xml(context_element, version):
    """The EAD3 metadata is prepared here for all the tests below.

    :context_element: Context element, where the test is done
    :version: EAD3 version to be used.
    """
    xml = '''<mets:dmdSec xmlns:mets="%(mets)s" xmlns:ead3="%(ead3)s">
               <mets:mdWrap MDTYPE="EAD3" MDTYPEVERSION="1.1.0">
               <mets:xmlData><ead3:ead/></mets:xmlData></mets:mdWrap>
             </mets:dmdSec>''' % NAMESPACES
    (ead3, root) = parse_xml_string(xml)
    (ead3, root) = add_containers(root, 'mets:mets')
    elem_handler = find_element(root, 'ead', 'ead3')
    if context_element is not None:
        elem_context = set_element(elem_handler, context_element, 'ead3')
    elem_version = find_element(root, 'mdWrap', 'mets')
    set_attribute(elem_version, 'MDTYPEVERSION', 'mets', version)
    return (ead3, elem_context, elem_version)
def test_ead3_extension(schematron_fx, xml, failures):
    """Test the rule which checks the content of the <objectxmlwrap> element in
       EAD3. It may contain any elements, but only outside EAD3 namespace.

    :schematron_fx: Schematron compile fixture
    :xml: XML to be validated
    :failures: Number of Schematron failures to expect
    """
    (xmltree, root) = parse_xml_string(xml)
    (xmltree, mets_root) = add_containers(
        root, 'mets:mets/mets:dmdSec/mets:mdWrap/mets:xmlData/ead3:ead')
    mdwrap = find_element(mets_root, 'mdWrap', 'mets')
    set_attribute(mdwrap, 'MDTYPEVERSION', 'mets', '1.0.0')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=xmltree, params=False)
    assert svrl.count(SVRL_FIRED) == 1
    assert svrl.count(SVRL_FAILED) == failures

    set_attribute(mdwrap, 'MDTYPEVERSION', 'mets', '1.1.0')
    svrl = schematron_fx(schematronfile=SCHFILE, xmltree=xmltree, params=False)
    assert svrl.count(SVRL_FIRED) == 0
Exemplo n.º 12
0
def prepare_xml(context_element, version):
    """The MODS metadata is prepared here for all the tests below.

    :context_element: Context element, where the test is done
    :version: MODS version to be used.
    """
    xml = '''<mets:dmdSec xmlns:mets="%(mets)s" xmlns:mods="%(mods)s">
               <mets:mdWrap MDTYPE="MODS" MDTYPEVERSION="3.7">
               <mets:xmlData><mods:mods/></mets:xmlData></mets:mdWrap>
             </mets:dmdSec>''' % NAMESPACES
    (mods, root) = parse_xml_string(xml)
    (mods, root) = add_containers(root, 'mets:mets')
    elem_handler = find_element(root, 'mods', 'mods')
    if context_element is not None:
        elem_context = set_element(elem_handler, context_element, 'mods')
    else:
        elem_context = elem_handler
    elem_version = find_element(root, 'mdWrap', 'mets')
    set_attribute(elem_version, 'MDTYPEVERSION', 'mets', version)
    return (mods, elem_context, elem_version)