def test_description_transform2():
INPUT = {"metadata": {"mods": {"note": {"#text": "A description"}}}}
EXPECTED = {"description": "A description"}
resp, content = _get_server_response(json.dumps(INPUT), provider="HARVARD")
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content)["sourceResource"])
def test_geocode_set_name_by_feature():
"""Should set the name property to the smallest available feature value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"country": "Canada",
"city": "Bananas"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
'coordinates': '62.8329086304, -95.9133224487',
'country': 'Canada',
'name': 'Bananas',
'state': 'Nunavut',
"city": "Bananas"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_contentdm_identify_object_usc():
"""
Should add a thumbnail URL made of the source URL.
"""
INPUT = {
u"something": "x",
u"somethink": "y",
u"originalRecord":
{"handle":
["aaa", "http://some.url/cdm/ref/12345"]
},
u"left": "right now!"
}
EXPECTED = {
u"something": "x",
u"somethink": "y",
u"originalRecord": {
"handle":
["aaa", "http://some.url/cdm/ref/12345"]
},
u"object": ("http://some.url/utils/getthumbnail/12345"),
u"admin": {u"object_status": 0},
u"left": "right now!"
}
url = contentdm_url("False")
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
print_error_log()
assert str(resp.status).startswith("2")
result = json.loads(content)
assert_same_jsons(EXPECTED, result)
def test_remove_type():
"""Should remove type"""
INPUT = {"id": "123", "sourceResource": {"type": "bananas"}}
EXPECTED = {"id": "123", "sourceResource": {}}
resp, content = _get_server_response(json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_not_remove_valid_date_dict():
INPUT = {
"sourceResource": {
"date":
{
"begin": "1959-01-01",
"end": "1959-01-01",
"displayDate": "1959-01-01"
},
}
}
EXPECTED_OUTPUT = {
"sourceResource": {
"date":
{
"begin": "1959-01-01",
"end": "1959-01-01",
"displayDate": "1959-01-01"
},
}
}
resp, content = _get_server_response(INPUT)
print_error_log()
assert resp["status"].startswith("2")
assert_same_jsons(EXPECTED_OUTPUT, content)
def test_prop_doesnt_exist():
"""Should return original JSON when prop doesn't exist."""
x = {"aaa": "BBB"}
r, c = _get_server_response(json.dumps(x), 'aaa%2Fddd')
pinfo(x, r, c)
print_error_log()
assert_same_jsons(c, x)
def test_geocode_do_not_skip_united_states():
"""Should geocode when name value is 'United States' is followed by a '-'
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {"name": "United States--California"}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"coordinates": "37.25022, -119.75126",
"country": "United States",
"name": "United States--California",
"state": "California"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_unicode():
"""Should handle unicode values
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": u"États-Unis"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": u"États-Unis"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_set_name_state():
"""Should set the name property to the state value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"state": "California"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
'coordinates': '37.25022, -119.75126',
"country": "United States",
"state": "California",
"name": "California"
}]
}
}
url = server() + "geocode"
resp,content = H.request(url,"POST",body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_set_name_by_feature():
"""Should set the name property to the smallest available feature value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"country": "Canada",
"city": "Bananas"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
'coordinates': '62.8329086304, -95.9133224487',
'country': 'Canada',
'name': 'Bananas',
'state': 'Nunavut',
"city": "Bananas"
}]
}
}
url = server() + "geocode"
resp,content = H.request(url,"POST",body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_set_name_region():
"""Should set the name property to the region value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"region": "Ecuador"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [
{
"region": "Ecuador",
"name": "Ecuador",
"country": "Ecuador"
}
]
}
}
url = server() + "geocode"
resp,content = H.request(url,"POST",body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_set_name_county():
"""Should set the name property to the county value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"county": "Los Angeles County",
"country": "Bananas"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [
{
"county": "Los Angeles County",
"country": "Bananas",
"name": "Los Angeles County",
"state": "California",
#uses bing because geonames wants to match country values
"coordinates": "33.9934997559, -118.29750824"
}
]
}
}
url = server() + "geocode"
resp,content = H.request(url,"POST",body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_set_name_city():
"""Should set the name property to the city value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"city": "Los Angeles",
"state": "California"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [
{
"coordinates": '34.05223, -118.24368',
"city": "Los Angeles",
'county': 'Los Angeles County',
"state": "California",
"country": "United States",
"name": "Los Angeles"
}
]
}
}
url = server() + "geocode"
resp,content = H.request(url,"POST",body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_exclude_coordinates_from_countries():
"""Should not include coordinates or smaller administrative units in
country enhancements
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": "Greece"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"country": "Greece",
"name": "Greece"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_exclude_coordinates_from_countries():
"""Should not include coordinates or smaller administrative units in
country enhancements
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {"name": "Greece"}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"country": "Greece",
"name": "Greece"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_geonames_name_search_context():
"""Should find a place name, only if matching other data.
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": "Portland",
"state": "Maine"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"county": "Cumberland County",
"country": "United States",
"state": "Maine",
"name": "Portland",
"coordinates": "43.66147, -70.25533"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_geonames_name_search():
"""Should find a place name.
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {"name": "Portland, OR"}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"county": "Multnomah County",
"country": "United States",
"state": "Oregon",
"name": "Portland, OR",
"coordinates": "45.52345, -122.67621"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_cleanup_enrich_then_lookup1():
"""Should produce both name and iso639_3 language fields"""
INPUT = ["en", "English", ["eng"], ["English"], ["en", "English"]]
EXPECTED = {
"sourceResource": {
"language": [{
"name": "English",
"iso639_3": "eng"
}]
}
}
for i in range(len(INPUT)):
input = {"sourceResource": {"language": INPUT[i]}}
url = server() + "cleanup_language"
resp, content = H.request(url, "POST", json.dumps(input))
assert resp.status == 200
url = server() + "enrich_language"
resp, content = H.request(url, "POST", content)
assert resp.status == 200
url = server() + "lookup?prop=sourceResource%2Flanguage%2Fname" + \
"&target=sourceResource%2Flanguage%2Fname&substitution=iso639_3"
resp, content = H.request(url, "POST", content)
assert resp.status == 200
url = server() + "lookup?prop=sourceResource%2Flanguage%2Fname" + \
"&target=sourceResource%2Flanguage%2Fiso639_3" + \
"&substitution=iso639_3&inverse=True"
resp, content = H.request(url, "POST", content)
assert resp.status == 200
assert_same_jsons(content, EXPECTED)
def test_geocode_geonames_name_search_context():
"""Should find a place name, only if matching other data.
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": "Portland",
"state": "Maine"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"county": "Cumberland County",
"country": "United States",
"state": "Maine",
"name": "Portland",
"coordinates": "43.66147, -70.25533"
}
]}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_uiuc_cleanup_spatial_name1():
INPUT = {
"sourceResource": {
"spatial": [{
"state": "US-CA",
"name": "California"
}, {
"state": "US-CA",
"name": "\n\n\n \nCalifornia"
}]
}
}
EXPECTED = {
"sourceResource": {
"spatial": [{
"state": "US-CA",
"name": "California"
}, {
"state": "US-CA",
"name": "California"
}]
}
}
resp, content = _get_server_response(json.dumps(INPUT))
assert resp["status"] == "200"
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_geonames_name_search_failure():
"""Shouldn't fall down when nothing is returned.
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": "1234567"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": "1234567"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_set_type_from_physical_format3():
"""Should not set type"""
INPUT = {"sourceResource": {"format": ["format"]}}
resp, content = _get_server_response(json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(INPUT, content)
def test_geocode_unicode():
"""Should handle unicode values
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": u"États-Unis"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": u"États-Unis"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_suppress_creator_value_for_string():
"""Should remove creator value if the is 'creator'."""
INPUT = {
"sourceResource": {
"creator": "creator",
"aaa": [
"Fisker, Kay--Architect--Danish--Male",
"Fisker, Kay--Architect--Danish--Male",
"bbb",
"ccc"
]
}
}
EXPECTED_OUTPUT = {
"sourceResource": {
"aaa": [
"Fisker, Kay--Architect--Danish--Male",
"Fisker, Kay--Architect--Danish--Male",
"bbb",
"ccc"
]
}
}
resp, content = _get_server_response(json.dumps(INPUT))
print_error_log()
assert resp["status"].startswith("2")
assert_same_jsons(EXPECTED_OUTPUT, content)
def test_georgia_set_spec_type():
"""Should set the specType to include Book, Government Document,
and Serial
"""
INPUT = {
"sourceResource": {
"type": [
"something Books",
" GOVERNMENT something",
" Something periodicals",
" Another periodicals"
]
}
}
EXPECTED = {
"sourceResource": {
"type": [
"something Books",
" GOVERNMENT something",
" Something periodicals",
" Another periodicals"
],
"specType": [
"Book",
"Government Document",
"Serial"
]
}
}
resp, content = _get_server_response(json.dumps(INPUT))
print_error_log()
assert resp["status"] == "200"
assert_same_jsons(EXPECTED, json.loads(content))
def test_strip_html():
"""'strip_html' strips HTML tags and entities recursively"""
request_data = {
'a': {
'b': [' <i>string</i> <b>one</b> \n \t', 'string < two ']
},
'c': ' \n <p>string three</p>',
'd': {},
'e': 1,
'f': '1 film negative: b&w ;'
}
expected_result = {
'a': {
'b': [u'string one', u'string < two']
},
'c': u'string three',
'd': {}, # unaltered
'e': 1, # unaltered
'f': '1 film negative: b&w ;' #unaltered
}
url = server() + 'strip_html'
resp_meta, resp_body = H.request(url,
'POST',
body=json.dumps(request_data))
assert resp_meta.status == 200
assert_same_jsons(expected_result, resp_body)
def test_populating_publisher_field():
INPUT = {
"freetext": {
"publisher": [
{
"#text": "Glossary of Coins and Currency Terms",
"@label": "Publication title"
},
{
"#text": "http://americanhistory.si.edu/coins/glossary.cfm",
"@label": "Publication URL"
},
{
"#text": "xx",
"@label": "Publisher"
}
],
}
}
EXPECTED_PUBLISHER = {
"publisher": ["xx"]
}
resp, content = _get_server_response(INPUT)
print_error_log()
assert resp["status"].startswith("2")
CONTENT = json.loads(content)
pinfo(content)
assert_same_jsons(EXPECTED_PUBLISHER, CONTENT["sourceResource"])
def test_geocode_set_name_coordinates():
"""Should set the name property to the lowest hierarchy value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"coordinates": "37.7771186829, -122.419639587",
"city": "Bananas"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [
{
"coordinates": "37.7771186829, -122.419639587",
"city": "Bananas",
"state": "California",
"name": "Bananas",
"county": "San Francisco County",
"country": "United States"
}
]
}
}
url = server() + "geocode"
resp,content = H.request(url,"POST",body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_type_has_priority():
INPUT = {
"id": "123",
"sourceResource": {
"type": [{
"text": "text",
"attrib": {}
}, {
"text": "text",
"attrib": {
"q": "mods"
}
}, {
"text": "postcards",
"attrib": {
"q": "genreform"
}
}],
"format":
"postcard"
}
}
EXPECTED = {
"id": "123",
"sourceResource": {
"type": "text",
"format": "postcard"
}
}
resp, content = _get_server_response(json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_do_not_skip_united_states():
"""Should geocode when name value is 'United States' is followed by a '-'
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": "United States--California"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"coordinates": "37.25022, -119.75126",
"country": "United States",
"name": "United States--California",
"state": "California"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_remove_type():
"""Should remove type"""
INPUT = {"id": "123", "sourceResource": {"type": "bananas"}}
EXPECTED = {"id": "123", "sourceResource": {}}
resp, content = _get_server_response(json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_geonames_name_search():
"""Should find a place name.
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": "Portland, OR"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"county": "Multnomah County",
"country": "United States",
"state": "Oregon",
"name": "Portland, OR",
"coordinates": "45.52345, -122.67621"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_default_type():
"""Should set type to default value"""
INPUT = {"id": "123", "sourceResource": {"type": "bananas"}}
EXPECTED = {"id": "123", "sourceResource": {"type": "image"}}
resp, content = _get_server_response(json.dumps(INPUT), default="image")
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_geonames_name_search_failure():
"""Shouldn't fall down when nothing is returned.
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"name": "1234567"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": "1234567"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_iso639_3_substitution():
"""
Should add "name" to language
"""
INPUT = {
"sourceResource": {
"language": [
{"iso639_3": "eng"},
{"iso639_3": "ara"}
]
}
}
EXPECTED = {
"sourceResource": {
"language": [
{"iso639_3": "eng", "name": "English"},
{"iso639_3": "ara", "name": "Arabic"}
]
}
}
prop = "sourceResource%2Flanguage%2Fiso639_3"
targ = "sourceResource%2Flanguage%2Fname"
subs = "iso639_3"
resp, content = _get_server_response(json.dumps(INPUT), prop, targ, subs)
print_error_log()
assert resp.status == 200
assert_same_jsons(content, EXPECTED)
def test_substitute_with_list_of_dictionaries():
"""
Should convert all dicts in a list.
"""
data = {
"xxx": "yyy",
"aaa": {
"bbb": "ccc",
"xxx": [
{"eee": "aaa"},
{"xxx": "eee"},
{"eee": "bbb"}
]
}
}
INPUT = json.dumps(data)
data["aaa"]["xxx"] = [
{"eee": "AAA222"},
{"xxx": "eee"},
{"eee": "BBB222"},
]
EXPECTED_OUTPUT = json.dumps(data)
resp, content = _get_server_response(INPUT, "aaa/xxx/eee", "aaa/xxx/eee", "test2")
print_error_log()
pinfo(resp, content)
assert resp.status == 200
assert_same_jsons(content, EXPECTED_OUTPUT)
def test_substitution_with_deleting_missing_values():
data = {
"xxx": "yyy",
"aaa": {
"bbb": "ccc",
"xxx": [
{"eee": "aaa"},
{"xxx": "eee"},
{"eee": "bbb"},
{"eee": "doesnt exist"},
{"eee": "doesnt exist"}
]
},
}
INPUT = json.dumps(data)
data["aaa"]["xxx"] = [
{"eee": "AAA222"},
{"xxx": "eee"},
{"eee": "BBB222"},
{ },
{ }
]
EXPECTED_OUTPUT = json.dumps(data)
resp, content = _get_server_response(INPUT, "aaa/xxx/eee", "aaa/xxx/eee", "test2", None, True)
assert resp.status == 200
assert_same_jsons(content, EXPECTED_OUTPUT)
def test_geocode_coordinate_provided():
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [
{
"name": "42.358631134, -71.0567016602"
}
]
},
"creator": "David"
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [
{
"state": "Massachusetts",
"country": "United States",
"name": "42.358631134, -71.0567016602",
"coordinates": "42.358631134, -71.0567016602"
}
]
},
"creator": "David"
}
url = server() + "geocode"
resp,content = H.request(url,"POST",body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(json.loads(content), EXPECTED)
def test_substitution_using_scdl_format_dict():
formats = \
("Pamphlets", "Pamphlets"), \
("Pamphlet", "Pamphlets"), \
("pamphlets", "Pamphlets"), \
("Manuscripts", "Manuscripts"), \
("Manuscript", "Manuscripts"), \
("manuscripts", "Manuscripts"), \
("Photograph", "Photographs"), \
("Photographs", "Photographs"), \
("Photograph", "Photographs") \
data = {
"xxx": "yyy",
"aaa": ""
}
for f in formats:
data["aaa"] = f[0]
INPUT = json.dumps(data)
data["aaa"] = f[1]
EXPECTED_OUTPUT = json.dumps(data)
print "Checking: %s" + repr(f)
resp, content = _get_server_response(INPUT, "aaa", "aaa", "scdl_fix_format", None, False)
print_error_log()
assert resp.status == 200
assert_same_jsons(EXPECTED_OUTPUT, content)
def test_removing_multiple_invalid_date_dicts():
INPUT = {
"sourceResource": {
"date": [
{
"begin": "1959-01-01",
"end": "1959-01-01",
"displayDate": "1959-01-01"
},
{
"begin": "2008-11-18",
"end": "2008-11-18",
"displayDate": "2008-11-18"
},
{
"begin": "1959-01-01",
"end": None,
"displayDate": "1959-01-01"
},
{
"begin": "2008-11-18",
"end": "",
"displayDate": "2008-11-18"
},
{
"begin": None,
"end": "1959-01-01",
"displayDate": "1959-01-01"
},
{
"begin": "",
"end": "2008-11-18",
"displayDate": "2008-11-18"
},
"aaa",
"bbb",
"ccc",
]
}
}
EXPECTED_OUTPUT = {
"sourceResource": {
"date": [
{
"begin": "1959-01-01",
"end": "1959-01-01",
"displayDate": "1959-01-01",
},
{
"begin": "2008-11-18",
"end": "2008-11-18",
"displayDate": "2008-11-18"
},
]
}
}
resp, content = _get_server_response(INPUT, "sourceResource/date")
print_error_log()
assert resp["status"].startswith("2")
assert_same_jsons(EXPECTED_OUTPUT, content)
def test_geocode():
"""
Simple geocode
"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": "Bakersfield, CA"
}]
},
"creator": "David"
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": "Bakersfield, CA",
"state": "California",
"county": "Kern County",
"country": "United States",
"coordinates": "35.37329, -119.01871"
}]
},
"creator": "David"
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_changing_values():
INPUT = [
"hello there", "123", ". hi ", ". hi", " . hi there ",
"a banana", "''.more complicated....",
'""""....even more complicated....."\'""""', "hello there;;",
";;hello there;;", "aaa--bbb", "aaa --bbb", "aaa-- bbb", "aaa -- bbb",
"aaa -- bbb -- ccc - - ddd -- ", ["aaa", "bbb"],
[" - aaa", " bbb -- "], "aaa bbb -- ccc",
"...,,,;;;;..,;'''\t\t\t aaa -- bbb ccc ddd;;;..;,,,,,;;;..... \t ",
"aaa -- bbb ccc\t\t\t\t\tddd",
" aaa -- bbb\t\t \t\t ccc\t\t\t ", "\t\taaa\tbbb\t\t",
""".. \t\t
sss ddd
\t\t .. """
]
EXPECTED = [
"hello there", "123", "hi", "hi", "hi there", "a banana",
"more complicated", 'even more complicated', "hello there",
"hello there", "aaa--bbb", "aaa--bbb", "aaa--bbb",
"aaa--bbb", "aaa--bbb--ccc - - ddd--", ["aaa", "bbb"],
["- aaa", "bbb--"], "aaa bbb--ccc", "aaa--bbb ccc ddd",
"aaa--bbb ccc ddd", "aaa--bbb ccc", "aaa\tbbb", """sss ddd"""
]
for i in xrange(0, len(INPUT)):
data = {}
data["aaa"] = {"bbb": INPUT[i]}
r, c = _get_server_response(json.dumps(data), 'aaa%2Fbbb')
exp = {}
exp["aaa"] = {"bbb": EXPECTED[i]}
print_error_log()
assert_same_jsons(exp, c)
def test_geocode_coordinate_provided2():
"""Should use coordinates provided in the coordinates property"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": "United States--Massachussetts",
"coordinates": "42.358631134, -71.0567016602"
}]
},
"creator": "David"
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"state": "Massachusetts",
"country": "United States",
"name": "United States--Massachussetts",
"coordinates": "42.358631134, -71.0567016602"
}]
},
"creator": "David"
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_list_of_properties():
"""Should process all properties."""
props = ["aaa", "bbb/ccc", "ddd/eee/fff"]
INPUT = {
"aaa": "....a -- b....",
"bbb": {
"ccc": ["aaa", "bbb", "'''''x''''"]
},
"ddd": {
"eee": {
"fff": ["a", "b", "...d..."]
}
}
}
EXPECTED = {
"aaa": "a--b",
"bbb": {
"ccc": ["aaa", "bbb", "x"]
},
"ddd": {
"eee": {
"fff": ["a", "b", "d"]
}
}
}
r, c = _get_server_response(json.dumps(INPUT), props)
assert r['status'] == '200'
assert_same_jsons(EXPECTED, c)
def test_geocode_with_existing_props():
"""Should not update existing fields since "state" value exists"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"name": "United States--Massachussetts",
"coordinates": "42.358631134, -71.0567016602",
"country": "Bananas",
"state": "Apples"
}]
},
"creator": "David"
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"state": "Apples",
"country": "Bananas",
"name": "United States--Massachussetts",
"coordinates": "42.358631134, -71.0567016602"
}]
},
"creator": "David"
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_set_type_from_physical_format4():
"""Should set type"""
INPUT = {
"sourceResource": {
"format": [
"Textile", "frame", "Photographs", "still image", "prints",
"magazines", "letters", "Sound Recording", "Audio",
"online Collections", "finding aid", "Online Exhibit"
]
}
}
EXPECTED = {
"sourceResource": {
"format": [
"Textile", "frame", "Photographs", "still image", "prints",
"magazines", "letters", "Sound Recording", "Audio",
"online Collections", "finding aid", "Online Exhibit"
],
"type": [
"physical object", "image", "text", "sound", "collection",
"interactive resource"
]
}
}
resp, content = _get_server_response(json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, content)
def test_geocode_set_name_coordinates():
"""Should set the name property to the lowest hierarchy value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"coordinates": "37.7771186829, -122.419639587",
"city": "Bananas"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"coordinates": "37.7771186829, -122.419639587",
"city": "Bananas",
"state": "California",
"name": "Bananas",
"county": "San Francisco County",
"country": "United States"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_contentdm_identify_object_with_download():
"""
Should add a thumbnail URL made of the source URL.
"""
INPUT = {
u"something": "x",
u"somethink": "y",
u"originalRecord": {
"handle": ["aaa", "http://repository.clemson.edu/u?/scp,104"]
},
u"left": "right now!"
}
EXPECTED = {
u"something": "x",
u"somethink": "y",
u"originalRecord": {
"handle": ["aaa", "http://repository.clemson.edu/u?/scp,104"]
},
u"object": ("http://repository.clemson.edu/cgi-bin/" +
"thumbnail.exe?CISOROOT=/scp&CISOPTR=104"),
u"admin": {u"object_status": 1},
u"left": "right now!"
}
url = contentdm_url("True")
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert str(resp.status).startswith("2")
result = json.loads(content)
assert_same_jsons(EXPECTED, result)
def test_geocode_set_name_city():
"""Should set the name property to the city value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"city": "Los Angeles",
"state": "California"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"coordinates": '34.05223, -118.24368',
"city": "Los Angeles",
'county': 'Los Angeles County',
"state": "California",
"country": "United States",
"name": "Los Angeles"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_cleanup_value_for_list():
INPUT = {
"aaa": {
"bbb": [
"Fisker, Kay--Architect--Danish--Male",
"Fisker, Kay--Architect--Danish--Male",
"bbb",
"ccc"
]
}
}
EXPECTED_OUTPUT = {
"aaa": {
"bbb": [
"Fisker, Kay",
"Fisker, Kay",
"bbb",
"ccc"
]
}
}
resp, content = _get_server_response(json.dumps(INPUT), "aaa/bbb")
print_error_log()
assert resp["status"].startswith("2")
assert_same_jsons(EXPECTED_OUTPUT, content)
def test_geocode_set_name_county():
"""Should set the name property to the county value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"county": "Los Angeles County",
"country": "Bananas"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"county": "Los Angeles County",
"country": "Bananas",
"name": "Los Angeles County",
"state": "California",
#uses bing because geonames wants to match country values
"coordinates": "33.9934997559, -118.29750824"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_default_type():
"""Should set type to default value"""
INPUT = {"id": "123", "sourceResource": {"type": "bananas"}}
EXPECTED = {"id": "123", "sourceResource": {"type": "image"}}
resp, content = _get_server_response(json.dumps(INPUT), default="image")
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_geocode_set_name_region():
"""Should set the name property to the region value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"region": "Ecuador"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
"region": "Ecuador",
"name": "Ecuador",
"country": "Ecuador"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_range_with_brackets():
"""Should transform date range with brackets."""
ranges = [
("1960-05-01 - 1960-05-15", "1960-05-01 - 1960-05-15"),
("[ 1960-05-01 - 1960-05-15 ]", "1960-05-01 - 1960-05-15"),
("[1960-05-01 - 1960-05-15]", "1960-05-01 - 1960-05-15"),
("[1960-05-01 / 1960-05-15]", "1960-05-01 / 1960-05-15"),
("[1960-05-01/1960-05-15]", "1960-05-01/1960-05-15"),
]
for r in ranges:
INPUT = {"date": r[0]}
EXPECTED = {
u'date': {
u'begin': u'1960-05-01',
u'end': u'1960-05-15',
"displayDate": r[1]
}
}
url = server() + "enrich_earliest_date?prop=date"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert str(resp.status).startswith("2")
print_error_log()
assert_same_jsons(EXPECTED, content)
def test_geocode_set_name_state():
"""Should set the name property to the state value"""
INPUT = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": {
"state": "California"
}
}
}
EXPECTED = {
"id": "12345",
"_id": "12345",
"sourceResource": {
"spatial": [{
'coordinates': '37.25022, -119.75126',
"country": "United States",
"state": "California",
"name": "California"
}]
}
}
url = server() + "geocode"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert resp.status == 200
assert_same_jsons(EXPECTED, json.loads(content))
def test_dedup_value1():
"""Should remove duplicate values"""
props = "subject,spatial,description"
INPUT = {
"subject": [
"This is a subject",
"This is a subject.",
" this is a SuBject . ",
"This is another subject (1780).",
"This is another subject 1780",
" thiS IS anOther subject (1780)"
],
"spatial": ["North Carolina", "New York"],
"description": "A description"
}
EXPECTED = {
"subject": [
"This is a subject",
"This is another subject (1780)."
],
"spatial": ["North Carolina", "New York"],
"description": "A description"
}
resp, content = _get_server_response(json.dumps(INPUT), props)
assert resp.status == 200
assert_same_jsons(EXPECTED, content)
def test_range_with_brackets():
"""Should transform date range with brackets."""
ranges = [
("1960-05-01 - 1960-05-15", "1960-05-01 - 1960-05-15"),
("[ 1960-05-01 - 1960-05-15 ]", "1960-05-01 - 1960-05-15"),
("[1960-05-01 - 1960-05-15]", "1960-05-01 - 1960-05-15"),
("[1960-05-01 / 1960-05-15]", "1960-05-01 / 1960-05-15"),
("[1960-05-01/1960-05-15]", "1960-05-01/1960-05-15"),
]
for r in ranges:
INPUT = {"date": r[0]}
EXPECTED = {
u'date' : {
u'begin' : u'1960-05-01',
u'end' : u'1960-05-15',
"displayDate" : r[1]
}
}
url = server() + "enrich_earliest_date?prop=date"
resp, content = H.request(url, "POST", body=json.dumps(INPUT))
assert str(resp.status).startswith("2")
print_error_log()
assert_same_jsons(EXPECTED, content)
