def rank_texts(cls):
"""
Get total citation counts and ranks for texts.
Returns: list
"""
count = fn.Count(Citation.id)
query = (
Text.select(Text, count)
.join(Citation)
.where(Text.display == True)
.where(Text.valid == True)
.group_by(Text.id)
.order_by(Text.id)
.naive()
)
counts = [t.count for t in query]
# Compute dense-rank ratios.
dense_ranks = rankdata(counts, "dense")
top = max(dense_ranks)
scores = [float(r / top) for r in dense_ranks]
# Compute overall ranks (#1 is most frequent).
max_ranks = rankdata(counts, "max")
top = max(max_ranks)
ranks = [int(top - r + 1) for r in max_ranks]
return [dict(zip(["text", "rank", "score"], t)) for t in zip(query, ranks, scores)]
def test_require_title_and_author(title, author, mock_hlom):
"""
Skip records that don't have a query-able title and author.
"""
mock_hlom.add_marc(title=title, author=author)
Text.ingest_hlom()
assert Text.select().count() == 0
def test_require_title_and_author(title, author, mock_hlom):
"""
Skip records that don't have a query-able title and author.
"""
mock_hlom.add_marc(title=title, author=author)
Text.ingest_hlom()
assert Text.select().count() == 0
def test_require_title_and_author(title, author, mock_jstor):
"""
Skip records that don't have a query-able title and author.
"""
mock_jstor.add_article(article_title=title, author=author)
Text.ingest_jstor()
assert Text.select().count() == 0
def test_set_multiple_authors(mock_jstor):
mock_jstor.add_article(author=[
('David W.', 'McClure'),
('Kara G.', 'Weisman'),
])
Text.ingest_jstor()
assert Text.select().first().authors == [
'McClure, David W.',
'Weisman, Kara G.',
]
def test_load_multiple(mock_hlom):
"""
Text.ingest_hlom() should ingest multiple records.
"""
# 100 records.
for i in range(100):
mock_hlom.add_marc()
Text.ingest_hlom()
# 100 rows.
assert Text.select().count() == 100
def test_load_multiple(mock_jstor):
"""
Text.ingest_jstor() should ingest multiple records.
"""
# 100 records.
for i in range(100):
mock_jstor.add_article()
Text.ingest_jstor()
# 100 rows.
assert Text.select().count() == 100
def test_load_multiple(mock_jstor):
"""
Text.ingest_jstor() should ingest multiple records.
"""
# 100 records.
for i in range(100):
mock_jstor.add_article()
Text.ingest_jstor()
# 100 rows.
assert Text.select().count() == 100
def test_load_multiple(mock_hlom):
"""
Text.ingest_hlom() should ingest multiple records.
"""
# 100 records.
for i in range(100):
mock_hlom.add_marc()
Text.ingest_hlom()
# 100 rows.
assert Text.select().count() == 100
def test_set_url(mock_jstor):
mock_jstor.add_article(url='http://test.org')
Text.ingest_jstor()
assert Text.select().first().url == 'http://test.org'
def test_set_pagination(mock_jstor):
mock_jstor.add_article(fpage=200, lpage=300)
Text.ingest_jstor()
assert Text.select().first().pagination == '200-300'
def test_set_corpus(mock_hlom):
mock_hlom.add_marc()
Text.ingest_hlom()
assert Text.select().first().corpus == 'hlom'
def test_set_publisher(mock_hlom):
mock_hlom.add_marc(publisher='Chicago Press')
Text.ingest_hlom()
assert Text.select().first().publisher == 'Chicago Press'
def test_set_publisher(mock_hlom):
mock_hlom.add_marc(publisher='Chicago Press')
Text.ingest_hlom()
assert Text.select().first().publisher == 'Chicago Press'
def test_set_surname(mock_hlom):
mock_hlom.add_marc(author='McClure, David W.')
Text.ingest_hlom()
assert Text.select().first().surname == 'McClure'
def test_set_identifier(mock_hlom):
mock_hlom.add_marc(control_number='001')
Text.ingest_hlom()
assert Text.select().first().identifier == '001'
def test_set_journal_identifier(mock_jstor):
mock_jstor.add_article(journal_id='criticalinquiry')
Text.ingest_jstor()
assert Text.select().first().journal_identifier == 'criticalinquiry'
def test_set_authors(mock_hlom):
mock_hlom.add_marc(author='David W. McClure')
Text.ingest_hlom()
assert Text.select().first().authors == ['David W. McClure']
def test_set_pagination(mock_jstor):
mock_jstor.add_article(fpage=200, lpage=300)
Text.ingest_jstor()
assert Text.select().first().pagination == '200-300'
def test_set_issue_number(mock_jstor):
mock_jstor.add_article(issue_number=10)
Text.ingest_jstor()
assert Text.select().first().issue_number == '10'
def test_set_corpus(mock_jstor):
mock_jstor.add_article()
Text.ingest_jstor()
assert Text.select().first().corpus == 'jstor'
def test_set_date(mock_hlom):
mock_hlom.add_marc(pubyear='1987')
Text.ingest_hlom()
assert Text.select().first().date == '1987'
def test_set_journal_title(mock_jstor):
mock_jstor.add_article(journal_title='Critical Inquiry')
Text.ingest_jstor()
assert Text.select().first().journal_title == 'Critical Inquiry'
def test_set_title(mock_hlom):
mock_hlom.add_marc(title='Book Title')
Text.ingest_hlom()
assert Text.select().first().title == 'Book Title'
def test_set_issue_volume(mock_jstor):
mock_jstor.add_article(issue_volume=200)
Text.ingest_jstor()
assert Text.select().first().issue_volume == '200'
def test_set_identifier(mock_jstor):
mock_jstor.add_article(article_id='001')
Text.ingest_jstor()
assert Text.select().first().identifier == '001'
def test_set_title(mock_hlom):
mock_hlom.add_marc(title='Book Title')
Text.ingest_hlom()
assert Text.select().first().title == 'Book Title'
def test_set_title(mock_jstor):
mock_jstor.add_article(article_title='Article Title')
Text.ingest_jstor()
assert Text.select().first().title == 'Article Title'
def test_set_authors(mock_hlom):
mock_hlom.add_marc(author='David W. McClure')
Text.ingest_hlom()
assert Text.select().first().authors == ['David W. McClure']
def test_set_surname(mock_jstor):
mock_jstor.add_article(author=[('David W.', 'McClure')])
Text.ingest_jstor()
assert Text.select().first().surname == 'McClure'
def test_set_date(mock_hlom):
mock_hlom.add_marc(pubyear='1987')
Text.ingest_hlom()
assert Text.select().first().date == '1987'
def test_set_single_author(mock_jstor):
mock_jstor.add_article(author=[('David W.', 'McClure')])
Text.ingest_jstor()
assert Text.select().first().authors == ['McClure, David W.']
def test_set_corpus(mock_hlom):
mock_hlom.add_marc()
Text.ingest_hlom()
assert Text.select().first().corpus == 'hlom'
def test_set_identifier(mock_hlom):
mock_hlom.add_marc(control_number='001')
Text.ingest_hlom()
assert Text.select().first().identifier == '001'
def test_set_issue_number(mock_jstor):
mock_jstor.add_article(issue_number=10)
Text.ingest_jstor()
assert Text.select().first().issue_number == '10'
def test_set_publisher(mock_jstor):
mock_jstor.add_article(publisher_name='Chicago Journals')
Text.ingest_jstor()
assert Text.select().first().publisher == 'Chicago Journals'
def test_set_url(mock_jstor):
mock_jstor.add_article(url='http://test.org')
Text.ingest_jstor()
assert Text.select().first().url == 'http://test.org'
def test_set_date(mock_jstor):
mock_jstor.add_article(pub_year=1987, pub_month=6, pub_day=25)
Text.ingest_jstor()
assert Text.select().first().date == '1987-06-25'
def test_set_identifier(mock_jstor):
mock_jstor.add_article(article_id='001')
Text.ingest_jstor()
assert Text.select().first().identifier == '001'
def test_set_surname(mock_hlom):
mock_hlom.add_marc(author='McClure, David W.')
Text.ingest_hlom()
assert Text.select().first().surname == 'McClure'