csort.py

#!/usr/bin/python
from __future__ import print_function

#-----------------------------------------------------------------------
#  Imports
#-----------------------------------------------------------------------
import sys
import getopt


#-----------------------------------------------------------------------
#  Defaults
#-----------------------------------------------------------------------
DEF_SEPARATOR = '/'
VERBOSE = False


#-----------------------------------------------------------------------
#  Functions
#-----------------------------------------------------------------------

def Usage(msg=None):
    print("""

    csort -rc -k KEYSTR FILE1 [ FILE2 .. ]
 
    Sort clause file(s) and print to standard out.

    -k KEYSTR -  Sort clauses relative to one another
                 based on the value of KEYSTR
    -c        -  Sort fields within a clause
    -r        -  Sort in reverse order.

    Either -c or -k key must be specified.  Both is also OK

""")
    if msg: print (msg)
    raise SystemExit

def print_verbose(verbose,msg):
	if verbose:
		print("VERBOSE: ", msg)

def print_error(msg):
	print("ERROR: ", msg)
	sys.exit()

#  Read clauses
#    - Clauses in input file must be separated by one or more blank lines
#    - Yields one clause at a time.
#    - Each clause returned as a list of pairs,
#      consisting of the line number (starts at 1) and the line text
def read_clauses(files,lower_case=False):
    line_no = 0
    for file in files:
        clause = []
        #  Open file
        if file=="-":
            fin = sys.stdin
        else:
            fin = open(file,"r")
        for line in fin.readlines():
            line_no += 1
            line = line.strip()
            #  End of clause
            if not line:
                if clause:
                    yield clause
                clause = []
            #  Add line to clause, only if it has :
            if ':' in line:
                if lower_case: line = line.lower()
                k,value = line.split(":",1)
                clause.append( (k, line_no, value) )
        #  Close file
        if not file=="-":
            fin.close()
        #  return last clause in file
        if clause:
            yield clause
            clause = []

#  Converts clause as returned from read_clauses()
#  to a dictionary
#  - Key,value are taken from line like KEY: VALUE
#  - Specific keys can appear multiple times
#  - Values are returned as an array of strings
def clause2dict(clause):
	cdict = {}
	for recno, (lineno, linestr) in enumerate(clause):
		#  If no ':', ignore lines
		if not ':' in linestr: continue
		key, value = linestr.split(":",1)
		key = key.strip()
		if key in cdict:
			cdict[key].append( value.strip() )
		else:
			#  Store record number of first occurance of this key
			cdict[key] = [ recno, value.strip() ]
	#  Convert value lists to value tuples
	return { k:tuple(v) for k,v in cdict.items() }

def sort_clauses(keystr,clauses,reverse=False,numeric=False):
	global VERBOSE
	value2indices = []
	for clauseno, clause in  enumerate(clauses):
		#  Find value(s) for key in this clause
		nvalues = 0
		for k, lineno, value in clause:
			if k==keystr:
				single_value  = value
				single_lineno = lineno
				nvalues += 1
		if nvalues==0:
			print_error("Clause %d on line %d has no intance of key %s" % 
				(clauseno+1, clause[0][1]+1, keystr) )
		if nvalues>=2:
			print_error("Clause %d on line %d has multiple values for key %s" %
				(clauseno+1, clause[0][1]+1, keystr) )
		if numeric:
			try:
				single_value = float( single_value )
			except (TypeError,ValueError):
				print_verbose(VERBOSE, "Invalid numeric value on line %d, clause %d.  Key,value are (%s,%s).  Will user value of 0." %
					(single_lineno, clauseno+1, keystr, single_value))
				single_value = float(0)
		value2indices.append ( (single_value, clauseno) )
	value2indices.sort(reverse=reverse)
	return [ clauses[clauseno] for single_value, clauseno in value2indices ]

#  Print claues in pretty format
#   - Join strings for multi-valued records
#   - Optionally sort each clause by keys alphabetically
def print_clauses(clauses, sort_within_clause=False,reverse=False):
	for clauseno, clause in enumerate(clauses):
		if clauseno>0: print()
		#  Sort clause by key, and secondarily by original lineno
		#   - But don't print original line number though
		if sort_within_clause:
			for k, lineno, value in sorted(clause):
				print( "%s: %s" % (k, value) )
		else:
			for k, lineno, value in clause:
				print( "%s: %s" % (k, value) )


#-----------------------------------------------------------------------
#  Main
#-----------------------------------------------------------------------

def main():
	global VERBOSE

	#  Read options
	opts, fnames = getopt.getopt(sys.argv[1:], 'rcnk:v')

	if len(fnames)<1:  Usage()

	opts = dict(opts)
	reverse   = '-r' in opts
	numeric   = '-n' in opts
	VERBOSE   = '-v' in opts

	if not '-k' in opts and not '-c' in opts: Usage()

	#  Read clauses
	clauses = [ clause for clause in read_clauses(fnames) ]

	#  Order clauses according to the value
	#  associated with opts['-k'])
	if '-k' in opts:
		clauses = sort_clauses(opts['-k'], clauses, reverse=reverse, numeric=numeric)

	#  Print clauses
	print_clauses(clauses,'-c' in opts,reverse)

main()