def main():
parser = optparse.OptionParser()
parser.add_option(
'-b', '--buffer',
dest='buffer',
type='int', default=1000000,
help='Number of lines to buffer at a time. Default: 1,000,000 lines. A buffer of 0 will attempt to use memory only.'
)
parser.add_option(
'-d', '--index_depth',
dest='index_depth',
type='int', default=3,
help='Depth to use on filebased offset indexing. Default: 3.'
)
parser.add_option(
'-p', '--keep_partial',
action='store_true',
dest='keep_partial',
default=False,
help='Keep rows in first input which are missing identifiers.')
parser.add_option(
'-u', '--keep_unmatched',
action='store_true',
dest='keep_unmatched',
default=False,
help='Keep rows in first input which are not joined with the second input.')
parser.add_option(
'-f', '--fill_options_file',
dest='fill_options_file',
type='str', default=None,
help='Fill empty columns with a values from a JSONified file.')
options, args = parser.parse_args()
fill_options = None
if options.fill_options_file is not None:
try:
fill_options = Bunch( **stringify_dictionary_keys( json.load( open( options.fill_options_file ) ) ) ) # json.load( open( options.fill_options_file ) )
except Exception as e:
print("Warning: Ignoring fill options due to json error (%s)." % e)
if fill_options is None:
fill_options = Bunch()
if 'fill_unjoined_only' not in fill_options:
fill_options.fill_unjoined_only = True
if 'file1_columns' not in fill_options:
fill_options.file1_columns = None
if 'file2_columns' not in fill_options:
fill_options.file2_columns = None
try:
filename1 = args[0]
filename2 = args[1]
column1 = int( args[2] ) - 1
column2 = int( args[3] ) - 1
out_filename = args[4]
except:
print("Error parsing command line.", file=sys.stderr)
sys.exit()
# Character for splitting fields and joining lines
split = "\t"
return join_files( filename1, column1, filename2, column2, out_filename, split, options.buffer, options.keep_unmatched, options.keep_partial, options.index_depth, fill_options=fill_options )
fill_options = None
if options.fill_options_file is not None:
try:
fill_options = Bunch(
**stringify_dictionary_keys(json.load(open(options.fill_options_file)))
) # json.load( open( options.fill_options_file ) )
except Exception, e:
print "Warning: Ignoring fill options due to json error (%s)." % e
if fill_options is None:
fill_options = Bunch()
if "fill_unjoined_only" not in fill_options:
fill_options.fill_unjoined_only = True
if "file1_columns" not in fill_options:
fill_options.file1_columns = None
if "file2_columns" not in fill_options:
fill_options.file2_columns = None
try:
filename1 = args[0]
filename2 = args[1]
column1 = int(args[2]) - 1
column2 = int(args[3]) - 1
out_filename = args[4]
except:
print >> sys.stderr, "Error parsing command line."
sys.exit()
# Character for splitting fields and joining lines
split = "\t"
return join_files(
fill_options = None
if options.fill_options_file is not None:
try:
fill_options = Bunch(**stringify_dictionary_keys(
json.load(open(options.fill_options_file))
)) # json.load( open( options.fill_options_file ) )
except Exception, e:
print "Warning: Ignoring fill options due to json error (%s)." % e
if fill_options is None:
fill_options = Bunch()
if 'fill_unjoined_only' not in fill_options:
fill_options.fill_unjoined_only = True
if 'file1_columns' not in fill_options:
fill_options.file1_columns = None
if 'file2_columns' not in fill_options:
fill_options.file2_columns = None
try:
filename1 = args[0]
filename2 = args[1]
column1 = int(args[2]) - 1
column2 = int(args[3]) - 1
out_filename = args[4]
except:
print >> sys.stderr, "Error parsing command line."
sys.exit()
# Character for splitting fields and joining lines
split = "\t"
return join_files(filename1,
def main():
parser = optparse.OptionParser()
parser.add_option(
'-b', '--buffer',
dest='buffer',
type='int', default=1000000,
help='Number of lines to buffer at a time. Default: 1,000,000 lines. A buffer of 0 will attempt to use memory only.'
)
parser.add_option(
'-d', '--index_depth',
dest='index_depth',
type='int', default=3,
help='Depth to use on filebased offset indexing. Default: 3.'
)
parser.add_option(
'-p', '--keep_partial',
action='store_true',
dest='keep_partial',
default=False,
help='Keep rows in first input which are missing identifiers.')
parser.add_option(
'-u', '--keep_unmatched',
action='store_true',
dest='keep_unmatched',
default=False,
help='Keep rows in first input which are not joined with the second input.')
parser.add_option(
'-f', '--fill_options_file',
dest='fill_options_file',
type='str', default=None,
help='Fill empty columns with a values from a JSONified file.')
parser.add_option(
'-H', '--keep_headers',
action='store_true',
dest='keep_headers',
default=False,
help='Keep the headers')
options, args = parser.parse_args()
fill_options = None
if options.fill_options_file is not None:
try:
fill_options = Bunch(**stringify_dictionary_keys(json.load(open(options.fill_options_file)))) # json.load( open( options.fill_options_file ) )
except Exception as e:
print("Warning: Ignoring fill options due to json error (%s)." % e)
if fill_options is None:
fill_options = Bunch()
if 'fill_unjoined_only' not in fill_options:
fill_options.fill_unjoined_only = True
if 'file1_columns' not in fill_options:
fill_options.file1_columns = None
if 'file2_columns' not in fill_options:
fill_options.file2_columns = None
try:
filename1 = args[0]
filename2 = args[1]
column1 = int(args[2]) - 1
column2 = int(args[3]) - 1
out_filename = args[4]
except Exception:
print("Error parsing command line.", file=sys.stderr)
sys.exit()
# Character for splitting fields and joining lines
split = "\t"
return join_files(filename1, column1, filename2, column2, out_filename, split, options.buffer, options.keep_unmatched, options.keep_partial, options.keep_headers, options.index_depth, fill_options=fill_options)
