def list_bibliographies(database, count):
"""
List all the document sets.
"""
bibliography_sets = Bibliography.list(database, count)
click.echo(
tabulate.tabulate(
[
[
bibset.eid[:8],
bibset.description,
bibset.modified.strftime('%b %d, %Y, %I:%M%p'),
len(bibset.documents)
]
for bibset in bibliography_sets
],
headers=[
'Identifier', 'Description', 'Updated at', 'Docs count',
],
tablefmt='rst',
)
)
total = Bibliography.count(database)
if count >= total:
click.echo('Showing all the document sets.')
else:
click.echo(
'Showing {count} out of {total} document sets.'
.format(count=count, total=total)
)
def create(database, **kwargs):
"""
Populates the condor database with information from the given files
kind parameter indicates what type of files you're working with.
"""
verbose = kwargs.pop('verbose', False)
if verbose:
file_names = '\n'.join(
f.name for f in kwargs.get('files', [])
if hasattr(f, 'name')
)
kind = kwargs.get('kind')
click.echo(f'I\'m looking for {kind} records in these files:')
click.echo(f'{file_names}')
kwargs['show_progress_bar'] = verbose
_bibliography = Bibliography.from_files(**kwargs)
database.add(_bibliography)
database.flush()
click.echo(f'I\'m writing to {_bibliography.eid}')
click.echo('And... I\'m done')
click.echo('The database contains {} records'.format(
Document.count(database, _bibliography.eid)
))
def test_matrix_creation_endpoint(client, session):
bib = Bibliography(eid='123', description='lorem')
session.add(bib)
session.flush()
response = client.post('/matrix',
json={
'bibliography': '123',
'fields': ['title', 'description'],
})
assert response.status_code == 200
assert response.json().get('bibliography_eid') == '123'
def test_matrix_creation_endpoint_fails(client, session):
bib = Bibliography(eid='123', description='lorem')
session.add(bib)
session.flush()
response = client.post('/matrix',
json={
'bibliography': '1234',
'fields': ['title', 'description'],
})
assert response.status_code == 404
assert 'message' in response.json()
def get_bibliography(eid, session: CondorSession) -> Response:
"""
List the bibliography that has the specified eid from the database.
"""
bibliography = Bibliography.find_by_eid(session, eid)
if not bibliography:
return Response(
{'message': 'The especified eid is not found on database'},
status=404,
)
return Response(sc.Bibliography(bibliography))
def test_single_bibliography_existing(client, session):
# Given an existing bibliography
bib = Bibliography(eid='123', description='lorem')
session.add(bib)
session.commit()
# When I ask for it
res = client.get('/bibliography/123')
# Then I get a successful response
assert res.status_code == 200
# And the response is non empty
assert res.json().get('eid') == '123'
assert res.json().get('description') == 'lorem'
def test_bibliography_endpoint_when_many_bibliographies(client, session):
# Create a bibliography
bib = Bibliography(eid='345', description='lorem')
session.add(bib)
session.commit()
# When I request the list bibliography
res = client.get('/bibliography')
# Then I get a successful response
assert res.status_code == 200
# The response is non empty
assert len(res.json()) == 1
# And has bibliographies
assert any(b.get('eid') == '345' for b in res.json())
def test_document_endpoint_actually_returns_documents(client, session):
# Given some records matching records in the database
bib = Bibliography(eid='123', description='lorem')
session.add(bib)
session.flush()
doc = Document(eid='345',
bibliography_eid='123',
title='lorem',
keywords='{}',
description='lorem',
language='english',
hash='alksdjksjdf')
session.add(doc)
session.commit()
# When I request the documents with the given bibliography eid
res = client.get('/document?bibliography=123')
# Then I receive a successful response
assert res.status_code == 200
# And the response is non empty
assert len(res.json()) > 0
def test_individual_matrix_endpoint_existing(client, session):
# Given some records matching records in the database
bib = Bibliography(eid='123', description='lorem')
session.add(bib)
session.flush()
mat = TermDocumentMatrix(
eid='345',
bibliography_eid='123',
bibliography_options='',
processing_options='',
term_list_path='',
matrix_path='',
)
session.add(mat)
session.commit()
# When I request the matrix with the eid
res = client.get('/matrix/345')
# Then I receive a successful response
assert res.status_code == 200
# And the response is non empty
assert len(res.json()) > 0
def create_matrix(descriptor: sc.MatrixDescriptor,
session: CondorSession) -> Response:
"""
Create a term document matrix.
"""
bibliography = Bibliography.find_by_eid(session,
descriptor.get('bibliography'))
if bibliography is None:
return Response(
{
'message':
'The especified bibliography eid is not found on database'
},
status=404,
)
td_matrix = TermDocumentMatrix.from_bibliography_set(
bibliography,
regularise=descriptor.get('regularise'),
fields=descriptor.get('fields'))
session.add(td_matrix)
session.commit()
return Response(sc.Matrix(td_matrix))
def test_matrix_endpoint_when_there_are_matrix(client, session):
# Given some records matching records in the database
bib = Bibliography(eid='123', description='lorem')
session.add(bib)
session.flush()
mat = TermDocumentMatrix(
eid='345',
bibliography_eid='123',
bibliography_options='',
processing_options='',
term_list_path='',
matrix_path='',
)
session.add(mat)
session.commit()
# When I request the documents with the given bibliography eid
res = client.get('/matrix')
# Then I receive a successful response
assert res.status_code == 200
# The response is non empty
assert len(res.json()) == 1
# And has matrix
assert any(m.get('eid') == '345' for m in res.json())
def get_all_bibliographies(session: CondorSession) -> List[sc.Bibliography]:
"""
List all bibliographies in the database.
"""
return [sc.Bibliography(bib) for bib in Bibliography.list(session)]