def test_rebuild_cache(self):
"""classifier - test rebuilding cache."""
from invenio.modules.classifier import reader
info = reader._get_ontology(self.taxonomy_name)
self.assertTrue(info[0])
cache = reader._get_cache_path(info[0])
if os.path.exists(cache):
ctime = os.stat(cache)[stat.ST_CTIME]
else:
ctime = -1
time.sleep(0.5) # sleep a bit for timing issues
rex = reader.get_regular_expressions(
self.taxonomy_name, rebuild=True)
self.assertTrue(os.path.exists(cache))
ntime = os.stat(cache)[stat.ST_CTIME]
self.assertTrue((ntime > ctime))
self.assertEqual(len(rex[0]) + len(rex[1]), 63)
def test_cache_accessibility(self):
"""classifier - test cache accessibility/writability"""
from flask import current_app
from invenio.modules.classifier.registry import taxonomies
from invenio.modules.classifier import reader
from invenio.modules.classifier.errors import TaxonomyError
# we will do tests with a copy of test taxonomy, in case anything goes
# wrong...
orig_name, orig_taxonomy_path, orig_taxonomy_url = reader._get_ontology(
self.taxonomy_name)
taxonomy_name = self.taxonomy_name + '.copy'
taxonomy_path = os.path.join(
current_app.config['CFG_TMPDIR'], taxonomy_name + '.rdf')
shutil.copy(orig_taxonomy_path, taxonomy_path)
taxonomies[taxonomy_name] = taxonomy_path
assert(os.path.exists(taxonomy_path))
name, taxonomy_path, taxonomy_url = reader._get_ontology(
taxonomy_name)
cache = reader._get_cache_path(
os.path.basename(taxonomy_path))
assert name
if os.path.exists(cache):
os.remove(cache)
reader.get_regular_expressions(
taxonomy_name, rebuild=True, no_cache=False)
assert(os.path.exists(cache))
# set cache unreadable
os.chmod(cache, 000)
self.assertRaises(
TaxonomyError,
reader.get_regular_expressions,
taxonomy_name, rebuild=False, no_cache=False
)
# set cache unreadable and test writing
os.chmod(cache, 000)
self.assertRaises(
TaxonomyError,
reader.get_regular_expressions,
taxonomy_name, rebuild=True, no_cache=False
)
# set cache readable and test writing
os.chmod(cache, 600)
self.assertRaises(
TaxonomyError,
reader.get_regular_expressions,
taxonomy_name, rebuild=True, no_cache=False
)
# set cache writable only
os.chmod(cache, 200)
reader.get_regular_expressions(
taxonomy_name, rebuild=True, no_cache=False)
reader.get_regular_expressions(
taxonomy_name, rebuild=False, no_cache=False)
# set cache readable/writable but corrupted (must rebuild itself)
os.chmod(cache, 600)
os.remove(cache)
open(cache, 'w').close()
reader.get_regular_expressions(
taxonomy_name, rebuild=False, no_cache=False)
# set cache readable/writable but corrupted (must rebuild itself)
open(cache, 'w').close()
try:
os.rename(taxonomy_path, taxonomy_path + 'x')
open(taxonomy_path, 'w').close()
self.assertRaises(
TaxonomyError,
reader.get_regular_expressions,
taxonomy_name, rebuild=False, no_cache=False
)
finally:
os.rename(taxonomy_path + 'x', taxonomy_path)
# make cache ok, but corrupt source
reader.get_regular_expressions(
taxonomy_name, rebuild=True, no_cache=False)
try:
os.rename(taxonomy_path, taxonomy_path + 'x')
open(taxonomy_path, 'w').close()
time.sleep(.1)
# touch the taxonomy to be older
os.utime(cache, (time.time() + 100, time.time() + 100))
reader.get_regular_expressions(
taxonomy_name, rebuild=False, no_cache=False)
finally:
os.rename(taxonomy_path + 'x', taxonomy_path)
# make cache ok (but old), and corrupt source
reader.get_regular_expressions(
taxonomy_name, rebuild=True, no_cache=False)
try:
os.rename(taxonomy_path, taxonomy_path + 'x')
open(taxonomy_path, 'w').close()
self.assertRaises(
TaxonomyError,
reader.get_regular_expressions,
taxonomy_name, rebuild=False, no_cache=False
)
finally:
os.rename(taxonomy_path + 'x', taxonomy_path)
name, taxonomy_path, taxonomy_url = reader._get_ontology(
taxonomy_name)
cache = reader._get_cache_path(name)
os.remove(taxonomy_path)
os.remove(cache)