def ensure_database_pressed(filepath: str,
return_not_raise: bool = False) -> List[str]:
""" Ensures that the given HMMer database exists and that the hmmpress
generated files aren't out of date.
Arguments:
filepath: the path to the HMMer database
return_not_raise: whether to catch errors and return their messages as strings
Returns:
any encountered error messages, will never be populated without return_not_raise == True
"""
components = [
"{}{}".format(filepath, ext)
for ext in ['.h3f', '.h3i', '.h3m', '.h3p']
]
if path.is_outdated(components, filepath):
logging.info("%s components missing or obsolete, re-pressing database",
filepath)
result = subprocessing.run_hmmpress(filepath)
if not result.successful():
msg = "Failed to hmmpress {!r}: {}".format(filepath, result.stderr)
if not return_not_raise:
raise RuntimeError(msg)
return [msg]
return []
def check_diamond_files(definition_file: str,
fasta_file: str,
db_file: str,
logging_only: bool = False) -> List[str]:
""" Check if the database files exist in the right version.
Arguments:
definition_file: the path to a database metadata file
fasta_file: the path to a proteins fasta file
db_file: the path to the diamond databse file
logging_only: return a list of errors messages instead of raising errors
Returns:
a list of error strings
"""
failure_messages: List[str] = []
if path.locate_file(definition_file) is None:
failure_messages.append(
"Failed to locate cluster definition file: {!r}".format(
definition_file))
regen_message = ""
if path.locate_file(fasta_file) is None:
failure_messages.append(
"Failed to locate cluster proteins: {!r}".format(fasta_file))
if not logging_only:
raise FileNotFoundError(failure_messages[-1])
elif path.locate_file(db_file) is None:
regen_message = f"could not find diamond database: {db_file}"
elif not check_diamond_db_compatible(db_file):
regen_message = f"incompatible diamond database version: {db_file}"
elif path.is_outdated(db_file, fasta_file):
regen_message = f"diamond database outdated: {db_file}"
if regen_message:
try:
logging.debug("%s, regenerating", regen_message)
run_diamond_makedb(db_file, fasta_file)
except RuntimeError:
if not logging_only:
raise
failure_messages.append(
"Failed to regenerate diamond database %r" % db_file)
if failure_messages:
failure_messages.append(
f"with diamond executable: {get_config().executables.diamond}")
return failure_messages
def check_clusterblast_files(definition_file: str,
fasta_file: str,
db_file: str,
logging_only: bool = False) -> List[str]:
""" Check if the clusterblast files exist in the right version.
Arguments:
definition_file: the path to the cluster definition TSV file
fasta_file: the path to the cluster proteins fasta file
db_file: the path to the diamond databse file
Returns:
A list of error strings the way `check_prereqs` does
"""
failure_messages: List[str] = []
if path.locate_file(definition_file) is None:
failure_messages.append(
"Failed to locate cluster definition file: {!r}".format(
definition_file))
regen_message = ""
if path.locate_file(fasta_file) is None:
failure_messages.append(
"Failed to locate cluster proteins: {!r}".format(fasta_file))
elif path.locate_file(db_file) is None:
regen_message = "could not find diamond database: %s" % db_file
elif not check_diamond_db_compatible(db_file):
regen_message = "incompatible diamond database version: %s" % db_file
elif path.is_outdated(db_file, fasta_file):
regen_message = "diamond database outdated: %s" % db_file
if regen_message:
try:
logging.debug("%s, regenerating", regen_message)
subprocessing.run_diamond_makedb(db_file, fasta_file)
except RuntimeError:
if not logging_only:
raise
failure_messages.append(
"Failed to regenerate diamond database %r" % db_file)
if failure_messages:
failure_messages.append("with diamond executable: %s" %
get_config().executables.diamond)
return failure_messages
def prepare_data(_logging_only: bool = False) -> List[str]:
""" Rebuild any dynamically buildable data """
flavours = ["bacteria", "fungi", "plants"]
with path.changed_directory(path.get_full_path(__file__, "css")):
built_files = [os.path.abspath("%s.css" % flavour) for flavour in flavours]
if path.is_outdated(built_files, glob.glob("*.scss")):
logging.info("CSS files out of date, rebuilding")
for flavour in flavours:
target = "%s.css" % flavour
assert os.path.exists(flavour + ".scss"), flavour
result = scss.Compiler(output_style="expanded").compile(flavour + ".scss")
assert result
with open(target, "w") as out:
out.write(result)
return []
