def chunk_input(self, input_files, chunksize):
"""Chunk input files into pieces for performance and parallelism."""
part_lists = [] # Lists of partial files
known_nlines = None
part_suffix = ""
chunk_nlines = chunksize * 2
for input_file in input_files:
# Count number of lines in the file
nlines = int(
command.execute_with_output("wc -l %s" %
input_file).strip().split()[0])
# Number of lines should be the same in paired files
if known_nlines is not None:
msg = "Mismatched line counts in supposedly paired files: {}".format(
input_files)
assert nlines == known_nlines, msg
known_nlines = nlines
# Set number of pieces and names
numparts = (nlines + chunk_nlines - 1) // chunk_nlines
ndigits = len(str(numparts - 1))
part_suffix = "-chunksize-%d-numparts-%d-part-" % (chunksize,
numparts)
out_prefix_base = os.path.basename(input_file) + part_suffix
out_prefix = os.path.join(self.chunks_result_dir_local,
out_prefix_base)
# Split large file into smaller named pieces
command.execute("split -a %d --numeric-suffixes -l %d %s %s" %
(ndigits, chunk_nlines, input_file, out_prefix))
command.execute_with_retries(
f"aws s3 sync --only-show-errors {self.chunks_result_dir_local}/ {self.chunks_result_dir_s3}/ --exclude '*' --include '{out_prefix_base}*'"
)
# Get the partial file names
partial_files = []
paths = command.execute_with_output(
"ls %s*" % out_prefix).rstrip().split("\n")
for pf in paths:
partial_files.append(os.path.basename(pf))
# Check that the partial files match our expected chunking pattern
pattern = "{:0%dd}" % ndigits
expected_partial_files = [(out_prefix_base + pattern.format(i))
for i in range(numparts)]
msg = "something went wrong with chunking: {} != {}".format(
partial_files, expected_partial_files)
assert expected_partial_files == partial_files, msg
part_lists.append(partial_files)
# Ex: [["input_R1.fasta-part-1", "input_R2.fasta-part-1"],
# ["input_R1.fasta-part-2", "input_R2.fasta-part-2"],
# ["input_R1.fasta-part-3", "input_R2.fasta-part-3"],...]
input_chunks = [list(part) for part in zip(*part_lists)]
return part_suffix, input_chunks
def delete_tag_with_retries(instance_id, tag_key):
command.execute_with_retries(
command_patterns.SingleCommand(
cmd="aws",
args=[
"ec2",
"delete-tags",
"--resources",
instance_id,
"--tags",
f"Key={tag_key}"
]
)
)
def chunk_input(self, input_files, chunksize):
"""Chunk input files into pieces for performance and parallelism."""
part_lists = [] # Lists of partial files
known_nlines = None
part_suffix = ""
chunk_nlines = chunksize * 2
for input_file in input_files:
# Count number of lines in the file
cmd_output = command.execute_with_output(
command_patterns.SingleCommand(cmd="wc",
args=["-l", input_file]))
nlines = int(cmd_output.strip().split()[0])
# Number of lines should be the same in paired files
if known_nlines is not None:
msg = "Mismatched line counts in supposedly paired files: {}".format(
input_files)
assert nlines == known_nlines, msg
known_nlines = nlines
# Set number of pieces and names
numparts = (nlines + chunk_nlines - 1) // chunk_nlines
ndigits = len(str(numparts - 1))
part_suffix = "-chunksize-%d-numparts-%d-part-" % (chunksize,
numparts)
out_prefix_base = os.path.basename(input_file) + part_suffix
out_prefix = os.path.join(self.chunks_result_dir_local,
out_prefix_base)
# Split large file into smaller named pieces
command.execute(
command_patterns.SingleCommand(cmd="split",
args=[
"-a", ndigits,
"--numeric-suffixes", "-l",
chunk_nlines, input_file,
out_prefix
]))
command.execute_with_retries(
command_patterns.SingleCommand(
cmd="aws",
args=[
"s3", "sync", "--only-show-errors",
os.path.join(self.chunks_result_dir_local, ""),
os.path.join(self.chunks_result_dir_s3, ""),
"--exclude", "*", "--include", out_prefix_base + "*"
]))
# Get the partial file names
partial_files = []
paths = command.glob(glob_pattern=out_prefix + "*",
strip_folder_names=True)
partial_files.extend(paths)
# Check that the partial files match our expected chunking pattern
pattern = "{:0%dd}" % ndigits
expected_partial_files = [(out_prefix_base + pattern.format(i))
for i in range(numparts)]
msg = "something went wrong with chunking: {} != {}".format(
partial_files, expected_partial_files)
assert expected_partial_files == partial_files, msg
part_lists.append(partial_files)
# Ex: [["input_R1.fasta-part-1", "input_R2.fasta-part-1"],
# ["input_R1.fasta-part-2", "input_R2.fasta-part-2"],
# ["input_R1.fasta-part-3", "input_R2.fasta-part-3"],...]
input_chunks = [list(part) for part in zip(*part_lists)]
return part_suffix, input_chunks
def delete_tag_with_retries(instance_iD, tag_key):
command.execute_with_retries(f"aws ec2 delete-tags --resources {instance_iD} --tags Key={tag_key}")
