def test_queryindex(self):
res = [{
'label': 'Doc #1',
'uri': 'http://example.org/doc1',
'text': 'matching doc 1'
}, {
'label': 'Doc #2',
'uri': 'http://example.org/doc2',
'text': 'matching doc 2'
}]
pager = None
config = {
'connect.return_value':
Mock(**{'query.return_value': (res, pager)})
}
printmock = MagicMock()
with patch('ferenda.devel.FulltextIndex', **config):
with patch('builtins.print', printmock):
d = Devel()
d.config = LayeredConfig(
Defaults({
'indextype': 'a',
'indexlocation': 'b'
}))
d.queryindex("doc")
want = """
Doc #1 (http://example.org/doc1): matching doc 1
Doc #2 (http://example.org/doc2): matching doc 2
""".strip()
got = "\n".join([x[1][0] for x in printmock.mock_calls])
self.maxDiff = None
self.assertEqual(want, got)
def test_select(self):
uri = "http://example.org/doc"
with open("testselecttemplate.rq", "wb") as fp:
fp.write("""PREFIX dcterms: <http://purl.org/dc/terms/>
SELECT ?p ?o
WHERE { <%(uri)s> ?p ?o . }
""".encode())
result = """
[
{
"p": "http://purl.org/dc/terms/title",
"o": "Document title"
},
{
"p": "http://purl.org/dc/terms/identifier",
"o": "Document ID"
}
]""".lstrip().encode("utf-8")
config = {
'connect.return_value': Mock(**{'select.return_value': result})
}
printmock = MagicMock()
with patch('ferenda.devel.TripleStore', **config):
with patch('builtins.print', printmock):
d = Devel()
d.config = LayeredConfig(
Defaults({
'storetype': 'a',
'storelocation': 'b',
'storerepository': 'c'
}))
d.select("testselecttemplate.rq", uri)
want = """
# Constructing the following from b, repository c, type a
# PREFIX dcterms: <http://purl.org/dc/terms/>
#
# SELECT ?p ?o
# WHERE { <http://example.org/doc> ?p ?o . }
#
[
{
"p": "http://purl.org/dc/terms/title",
"o": "Document title"
},
{
"p": "http://purl.org/dc/terms/identifier",
"o": "Document ID"
}
]
# Selected in 0.001s
""".strip()
got = "\n".join([x[1][0] for x in printmock.mock_calls])
self.maxDiff = None
self.assertEqual(self.mask_time(want), self.mask_time(got))
os.unlink("testselecttemplate.rq")
def test_select(self):
uri = "http://example.org/doc"
with open("testselecttemplate.rq", "wb") as fp:
fp.write(
"""PREFIX dct: <http://purl.org/dc/terms/>
SELECT ?p ?o
WHERE { <%(uri)s> ?p ?o . }
""".encode()
)
result = """
[
{
"p": "http://purl.org/dc/terms/title",
"o": "Document title"
},
{
"p": "http://purl.org/dc/terms/identifier",
"o": "Document ID"
}
]""".lstrip().encode(
"utf-8"
)
config = {"connect.return_value": Mock(**{"select.return_value": result})}
printmock = MagicMock()
with patch("ferenda.devel.TripleStore", **config):
with patch(builtins + ".print", printmock):
d = Devel()
d.config = LayeredConfig({"storetype": "a", "storelocation": "b", "storerepository": "c"})
d.select("testselecttemplate.rq", uri)
want = """
# Constructing the following from b, repository c, type a
# PREFIX dct: <http://purl.org/dc/terms/>
#
# SELECT ?p ?o
# WHERE { <http://example.org/doc> ?p ?o . }
#
[
{
"p": "http://purl.org/dc/terms/title",
"o": "Document title"
},
{
"p": "http://purl.org/dc/terms/identifier",
"o": "Document ID"
}
]
# Selected in 0.001s
""".strip()
got = "\n".join([x[1][0] for x in printmock.mock_calls])
self.maxDiff = None
self.assertEqual(self.mask_time(want), self.mask_time(got))
os.unlink("testselecttemplate.rq")
def test_dumpstore(self):
d = Devel()
d.config = Mock()
# only test that Triplestore is called correctly, mock any
# calls to any real database
config = {"connect.return_value": Mock(**{"get_serialized.return_value": b"[fake store content]"})}
printmock = MagicMock()
with patch("ferenda.devel.TripleStore", **config):
with patch(builtins + ".print", printmock):
d.dumpstore(format="trix")
want = "[fake store content]"
printmock.assert_has_calls([call(want)])
def test_construct(self):
uri = "http://example.org/doc"
with open("testconstructtemplate.rq", "wb") as fp:
fp.write("""PREFIX dcterms: <http://purl.org/dc/terms/>
CONSTRUCT { ?s ?p ?o . }
WHERE { ?s ?p ?o .
<%(uri)s> ?p ?o . }
""".encode())
g = Graph()
g.bind("dcterms", str(DCTERMS))
g.add((URIRef(uri), DCTERMS.title, Literal("Document title")))
config = {
'connect.return_value': Mock(**{'construct.return_value': g})
}
printmock = MagicMock()
with patch('ferenda.devel.TripleStore', **config):
with patch('builtins.print', printmock):
d = Devel()
d.config = LayeredConfig(
Defaults({
'storetype': 'a',
'storelocation': 'b',
'storerepository': 'c'
}))
d.construct("testconstructtemplate.rq", uri)
want = """
# Constructing the following from b, repository c, type a
# PREFIX dcterms: <http://purl.org/dc/terms/>
#
# CONSTRUCT { ?s ?p ?o . }
# WHERE { ?s ?p ?o .
# <http://example.org/doc> ?p ?o . }
#
@prefix dcterms: <http://purl.org/dc/terms/> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xml: <http://www.w3.org/XML/1998/namespace> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
<http://example.org/doc> dcterms:title "Document title" .
# 1 triples constructed in 0.001s
""".strip()
got = "\n".join([x[1][0] for x in printmock.mock_calls])
self.maxDiff = None
self.assertEqual(self.mask_time(want), self.mask_time(got))
os.unlink("testconstructtemplate.rq")
def test_construct(self):
uri = "http://example.org/doc"
with open("testconstructtemplate.rq", "wb") as fp:
fp.write("""PREFIX dcterms: <http://purl.org/dc/terms/>
CONSTRUCT { ?s ?p ?o . }
WHERE { ?s ?p ?o .
<%(uri)s> ?p ?o . }
""".encode())
g = Graph()
g.bind("dcterms", str(DCTERMS))
g.add((URIRef(uri),
DCTERMS.title,
Literal("Document title")))
config = {'connect.return_value':
Mock(**{'construct.return_value': g})}
printmock = MagicMock()
with patch('ferenda.devel.TripleStore', **config):
with patch('builtins.print', printmock):
d = Devel()
d.config = LayeredConfig(Defaults({'storetype': 'a',
'storelocation': 'b',
'storerepository': 'c'}))
d.construct("testconstructtemplate.rq", uri)
want = """
# Constructing the following from b, repository c, type a
# PREFIX dcterms: <http://purl.org/dc/terms/>
#
# CONSTRUCT { ?s ?p ?o . }
# WHERE { ?s ?p ?o .
# <http://example.org/doc> ?p ?o . }
#
@prefix dcterms: <http://purl.org/dc/terms/> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix xml: <http://www.w3.org/XML/1998/namespace> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
<http://example.org/doc> dcterms:title "Document title" .
# 1 triples constructed in 0.001s
""".strip()
got = "\n".join([x[1][0] for x in printmock.mock_calls])
self.maxDiff = None
self.assertEqual(self.mask_time(want),
self.mask_time(got))
os.unlink("testconstructtemplate.rq")
def test_dumpstore(self):
d = Devel()
d.config = Mock()
# only test that Triplestore is called correctly, mock any
# calls to any real database
config = {
'connect.return_value':
Mock(**{'get_serialized.return_value': b'[fake store content]'})
}
printmock = MagicMock()
with patch('ferenda.devel.TripleStore', **config):
with patch('builtins.print', printmock):
d.dumpstore(format="trix")
want = "[fake store content]"
printmock.assert_has_calls([call(want)])
def test_queryindex(self):
res = [
{"identifier": "Doc #1", "about": "http://example.org/doc1", "text": "matching doc 1"},
{"identifier": "Doc #2", "about": "http://example.org/doc2", "text": "matching doc 2"},
]
config = {"connect.return_value": Mock(**{"query.return_value": res})}
printmock = MagicMock()
with patch("ferenda.devel.FulltextIndex", **config):
with patch(builtins + ".print", printmock):
d = Devel()
d.config = LayeredConfig({"indextype": "a", "indexlocation": "b"})
d.queryindex("doc")
want = """
Doc #1 (http://example.org/doc1): matching doc 1
Doc #2 (http://example.org/doc2): matching doc 2
""".strip()
got = "\n".join([x[1][0] for x in printmock.mock_calls])
self.maxDiff = None
self.assertEqual(want, got)
def test_queryindex(self):
res = [{'label': 'Doc #1',
'uri': 'http://example.org/doc1',
'text': 'matching doc 1'},
{'label': 'Doc #2',
'uri': 'http://example.org/doc2',
'text': 'matching doc 2'}]
pager = None
config = {'connect.return_value': Mock(**{'query.return_value': (res, pager)})}
printmock = MagicMock()
with patch('ferenda.devel.FulltextIndex', **config):
with patch('builtins.print', printmock):
d = Devel()
d.config = LayeredConfig(Defaults({'indextype': 'a',
'indexlocation': 'b'}))
d.queryindex("doc")
want = """
Doc #1 (http://example.org/doc1): matching doc 1
Doc #2 (http://example.org/doc2): matching doc 2
""".strip()
got = "\n".join([x[1][0] for x in printmock.mock_calls])
self.maxDiff = None
self.assertEqual(want, got)
def test_mkpatch(self):
tempdir = tempfile.mkdtemp()
basefile = "1"
# Test 1: A repo which do not use any intermediate files. In
# this case, the user edits the downloaded file, then runs
# mkpatch, which saves the edited file, re-downloads the file,
# and computes the diff.
store = DocumentStore(tempdir + "/base")
downloaded_path = store.downloaded_path(basefile)
def my_download_single(self):
# this function simulates downloading
with open(downloaded_path, "wb") as fp:
fp.write(
"""This is a file.
It has been downloaded.
""".encode()
)
repo = DocumentRepository(datadir=tempdir)
with repo.store.open_downloaded(basefile, "wb") as fp:
fp.write(
"""This is a file.
It has been patched.
""".encode()
)
d = Devel()
globalconf = LayeredConfig(
{
"datadir": tempdir,
"patchdir": tempdir,
"devel": {"class": "ferenda.Devel"},
"base": {"class": "ferenda.DocumentRepository"},
},
cascade=True,
)
d.config = globalconf.devel
with patch("ferenda.DocumentRepository.download_single") as mock:
mock.side_effect = my_download_single
patchpath = d.mkpatch("base", basefile, "Example patch")
patchcontent = util.readfile(patchpath)
self.assertIn("Example patch", patchcontent)
self.assertIn("@@ -1,2 +1,2 @@", patchcontent)
self.assertIn("-It has been downloaded.", patchcontent)
self.assertIn("+It has been patched.", patchcontent)
# test 2: Same, but with a multi-line desc
with repo.store.open_downloaded(basefile, "wb") as fp:
fp.write(
"""This is a file.
It has been patched.
""".encode()
)
longdesc = """A longer comment
spanning
several lines"""
with patch("ferenda.DocumentRepository.download_single") as mock:
mock.side_effect = my_download_single
patchpath = d.mkpatch("base", basefile, longdesc)
patchcontent = util.readfile(patchpath)
desccontent = util.readfile(patchpath.replace(".patch", ".desc"))
self.assertEqual(longdesc, desccontent)
self.assertFalse("A longer comment" in patchcontent)
self.assertIn("@@ -1,2 +1,2 @@", patchcontent)
self.assertIn("-It has been downloaded.", patchcontent)
self.assertIn("+It has been patched.", patchcontent)
# test 3: If intermediate file exists, patch that one
intermediate_path = store.intermediate_path(basefile)
util.ensure_dir(intermediate_path)
with open(intermediate_path, "wb") as fp:
fp.write(
"""This is a intermediate file.
It has been patched.
""".encode()
)
intermediate_path = store.intermediate_path(basefile)
def my_parse(self, basefile=None):
# this function simulates downloading
with open(intermediate_path, "wb") as fp:
fp.write(
"""This is a intermediate file.
It has been processed.
""".encode()
)
with patch("ferenda.DocumentRepository.parse") as mock:
mock.side_effect = my_parse
patchpath = d.mkpatch("base", basefile, "Example patch")
patchcontent = util.readfile(patchpath)
self.assertIn("@@ -1,2 +1,2 @@ Example patch", patchcontent)
self.assertIn(" This is a intermediate file", patchcontent)
self.assertIn("-It has been processed.", patchcontent)
self.assertIn("+It has been patched.", patchcontent)
