def test_directory_hash(self):
"""Test that we can read a hashcode for a directory."""
test_dir = "testdata/thaga_janakari_gene_ds/hg19"
hasher = Hasher()
self.assertTrue(
hasher.create_hashcode_for_dir(
test_dir) == "cd04c3299a21ce5fbe1bb9dc0d73d907",
"Hashed directory did not match ground truth. (" +
hasher.create_hashcode_for_dir(test_dir) + ") for path: " +
os.path.abspath(test_dir))
def get_hashcode(self):
""" Since this class can change annotation values depending on certain state attributes (e.g. tx-mode), we need the
hashcode to change. The super class hashcode attribute is treated like an initial hashcode here.
In other words, hashcode is not a simple attribute for this datasource class.
:return: hashcode including state information
"""
hasher = Hasher()
attrs_relevant_for_caching = [self.hashcode, self.get_tx_mode(), str(self._custom_canonical_txs)]
for attr in attrs_relevant_for_caching:
hasher.update(attr)
return Hasher.md5_hash(hasher.hexdigest())
def create_datasource_md5_file(datasource_dir):
"""datasource_dir should be the /db_dir/ds_name/genome_build.
For example,
create_datasource_md5_file("/home/user/my_db_dir/gaf/hg19")
"""
if datasource_dir.endswith('/'):
datasource_dir = datasource_dir[:-1]
md5_filename = os.path.abspath(datasource_dir) + ".md5"
logging.getLogger(__name__).info("md5 being written to: " + os.path.abspath(md5_filename))
hasher = Hasher()
hashcode = hasher.create_hashcode_for_dir(datasource_dir)
fp = file(md5_filename, "w")
fp.write(hashcode)
fp.close()
def get_hashcode(self):
"""The GAF datasource has to adjust its key based on the internal tx mode. set_hashcode sends in
an initial hashcode, which is then adjusted by tx-mode
"""
hasher = Hasher()
hasher.update(self.hashcode)
hasher.update(self.get_tx_mode())
return hasher.hexdigest()
def get_hashcode(self):
"""The GAF datasource has to adjust its key based on the internal tx mode. set_hashcode sends in
an initial hashcode, which is then adjusted by tx-mode
"""
hasher = Hasher()
hasher.update(self.hashcode)
hasher.update(self.get_tx_mode())
return hasher.hexdigest()
def create_db_dir_key(self):
"""Create the db_dir_key for this annotation configuration. Requires the datasources."""
self.logger.info("Generating db-dir key from datasources...")
hasher = Hasher()
for ds in self._datasources:
self.logger.info(ds.title + " " + ds.version + " md5: " + ds.get_hashcode())
hasher.update(ds.get_hashcode())
db_dir_key = Hasher.md5_hash(hasher.hexdigest())
self.logger.info("Final db-dir md5: " + db_dir_key)
return db_dir_key
def get_hashcode(self):
""" Since this class can change annotation values depending on certain state attributes (e.g. tx-mode), we need the
hashcode to change. The super class hashcode attribute is treated like an initial hashcode here.
In other words, hashcode is not a simple attribute for this datasource class.
:return: hashcode including state information
"""
hasher = Hasher()
attrs_relevant_for_caching = [self.hashcode, self.get_tx_mode(), str(self._custom_canonical_txs)]
for attr in attrs_relevant_for_caching:
hasher.update(attr)
return Hasher.md5_hash(hasher.hexdigest())
def create_db_dir_key_simple(self):
"""Create the db_dir_key for this annotation configuration. Requires the datasources."""
db_dir_key = Hasher.md5_hash(self.createHeaderString(False))
return db_dir_key
def test_simple_hash(self):
"""Test that the single md5 call (static) functions correctly."""
guess = Hasher.md5_hash("blah\n")
self.assertTrue(guess == "0d599f0ec05c3bda8c3b8a68c32a1b47")
def create_db_dir_key_simple(self):
"""Create the db_dir_key for this annotation configuration. Requires the datasources."""
db_dir_key = Hasher.md5_hash(self.createHeaderString(False))
return db_dir_key
def test_simple_hash(self):
"""Test that the single md5 call (static) functions correctly."""
guess = Hasher.md5_hash("blah\n")
self.assertTrue(guess == "0d599f0ec05c3bda8c3b8a68c32a1b47")
def test_directory_hash(self):
"""Test that we can read a hashcode for a directory."""
test_dir = "testdata/thaga_janakari_gene_ds/hg19"
hasher = Hasher()
self.assertTrue(hasher.create_hashcode_for_dir(test_dir)=="cd04c3299a21ce5fbe1bb9dc0d73d907", "Hashed directory did not match ground truth. (" + hasher.create_hashcode_for_dir(test_dir) + ") for path: " + os.path.abspath(test_dir))
