def test_hitfilter_zero_score(subject_index):
orighits = ListAnalysisResult(
[AnalysisHit(uri='uri', label='label', score=0.0)],
subject_index)
hits = HitFilter()(orighits)
assert isinstance(hits, AnalysisResult)
assert len(hits) == 0
def generate_filter_batches(subjects):
filter_batches = collections.OrderedDict()
for limit in range(1, 16):
for threshold in [i * 0.05 for i in range(20)]:
hit_filter = HitFilter(limit, threshold)
batch = annif.eval.EvaluationBatch(subjects)
filter_batches[(limit, threshold)] = (hit_filter, batch)
return filter_batches
def run_analyze(project_id, limit, threshold, backend_param):
"""
Analyze a single document from standard input.
"""
project = get_project(project_id)
text = sys.stdin.read()
backend_params = parse_backend_params(backend_param)
hit_filter = HitFilter(limit, threshold)
hits = hit_filter(project.analyze(text, backend_params))
for hit in hits:
click.echo("<{}>\t{}\t{}".format(hit.uri, hit.label, hit.score))
def analyze(project_id, text, limit, threshold):
"""analyze text and return a dict with results formatted according to
Swagger spec"""
try:
project = annif.project.get_project(project_id)
except ValueError:
return project_not_found_error(project_id)
hit_filter = HitFilter(limit, threshold)
try:
result = project.analyze(text)
except AnnifException as err:
return server_error(err)
hits = hit_filter(result)
return {'results': [hit._asdict() for hit in hits]}
def run_eval(project_id, paths, limit, threshold, backend_param):
"""
Analyze documents and evaluate the result.
Compare the results of automated indexing against a gold standard. The
path may be either a TSV file with short documents or a directory with
documents in separate files.
"""
project = get_project(project_id)
backend_params = parse_backend_params(backend_param)
hit_filter = HitFilter(limit=limit, threshold=threshold)
eval_batch = annif.eval.EvaluationBatch(project.subjects)
docs = open_documents(paths)
for doc in docs.documents:
results = project.analyze(doc.text, backend_params)
hits = hit_filter(results)
eval_batch.evaluate(hits,
annif.corpus.SubjectSet((doc.uris, doc.labels)))
template = "{0:<20}\t{1}"
for metric, score in eval_batch.results().items():
click.echo(template.format(metric + ":", score))
def run_analyzedir(project_id, directory, suffix, force, limit, threshold,
backend_param):
"""
Analyze a directory with documents. Write the results in TSV files
with the given suffix.
"""
project = get_project(project_id)
backend_params = parse_backend_params(backend_param)
hit_filter = HitFilter(limit, threshold)
for docfilename, dummy_subjectfn in annif.corpus.DocumentDirectory(
directory, require_subjects=False):
with open(docfilename) as docfile:
text = docfile.read()
subjectfilename = re.sub(r'\.txt$', suffix, docfilename)
if os.path.exists(subjectfilename) and not force:
click.echo("Not overwriting {} (use --force to override)".format(
subjectfilename))
continue
with open(subjectfilename, 'w') as subjfile:
results = project.analyze(text, backend_params)
for hit in hit_filter(results):
line = "<{}>\t{}\t{}".format(hit.uri, hit.label, hit.score)
click.echo(line, file=subjfile)
def test_hitfilter_threshold(subject_index):
orighits = generate_hits(10, subject_index)
hits = HitFilter(threshold=0.5)(orighits)
assert isinstance(hits, AnalysisResult)
assert len(hits) == 2
def test_hitfilter_limit(subject_index):
orighits = generate_hits(10, subject_index)
hits = HitFilter(limit=5)(orighits)
assert isinstance(hits, AnalysisResult)
assert len(hits) == 5