tools/anonym/joincsv.py

#!/usr/bin/env python
#
# (c) 2008 by Intevation GmbH
# author: Sascha L. Teichmann ([email protected])
#
import sys
import codecs

SEP = '\t'

def main():
    csvs = []
    max_cols = -1;
    max_idx  = -1;
    for idx, arg in enumerate(sys.argv[1:]):
        print >> sys.stderr, "file: %s" % arg
        f = None
        try:
            f = codecs.open(arg, "r", "latin1")
            csv = []
            for line in f:
                line = line.replace('\r', '').replace('\n', '')
                if not line: continue
                line = line.split(SEP)
                l = len(line)
                line[0] = line[0].replace('#', '', 1)
                if l > max_cols:
                    max_cols = l
                    max_idx = idx
                csv.append(line)
            csvs.append(csv)
        finally:
            if f:
                try: f.close()
                except: pass

    print >> sys.stderr, "max_cols: %d" % max_cols
    print >> sys.stderr, "max_idx: %d" % max_idx

    maps = []
    for i in csvs:
        maps.append([-1] * max_cols)

    master = csvs[max_idx]

    for j, m in enumerate(master[0]):
        for i in range(len(csvs)):
            try:
                idx = csvs[i][0].index(m)
                maps[i][j] = idx
            except ValueError:
                pass
    
    Writer = codecs.getwriter("latin-1")
    f = Writer(sys.stdout)
    f.write("#%s\r\n" % SEP.join(master[0]))

    for row, csv in enumerate(csvs):
        map = maps[row]
        for c in csv[1:]:
            line = []
            for j in xrange(max_cols):
                idx = map[j]
                if idx >= 0:
                    line.append(c[idx])
                else:
                    line.append(SEP)
            f.write("%s\r\n" % SEP.join(line))

if __name__ == "__main__":
    main()
1	teichmann	236	#!/usr/bin/env python
2			#
3			# (c) 2008 by Intevation GmbH
4			# author: Sascha L. Teichmann ([email protected])
5			#
6			import sys
7			import codecs
8
9			SEP = '\t'
10
11			def main():
12			csvs = []
13			max_cols = -1;
14			max_idx = -1;
15			for idx, arg in enumerate(sys.argv[1:]):
16			print >> sys.stderr, "file: %s" % arg
17			f = None
18			try:
19			f = codecs.open(arg, "r", "latin1")
20			csv = []
21			for line in f:
22			line = line.replace('\r', '').replace('\n', '')
23			if not line: continue
24			line = line.split(SEP)
25			l = len(line)
26			line[0] = line[0].replace('#', '', 1)
27			if l > max_cols:
28			max_cols = l
29			max_idx = idx
30			csv.append(line)
31			csvs.append(csv)
32			finally:
33			if f:
34			try: f.close()
35			except: pass
36
37			print >> sys.stderr, "max_cols: %d" % max_cols
38			print >> sys.stderr, "max_idx: %d" % max_idx
39
40			maps = []
41			for i in csvs:
42			maps.append([-1] * max_cols)
43
44			master = csvs[max_idx]
45
46			for j, m in enumerate(master[0]):
47			for i in range(len(csvs)):
48			try:
49			idx = csvs[i][0].index(m)
50			maps[i][j] = idx
51			except ValueError:
52			pass
53
54			Writer = codecs.getwriter("latin-1")
55			f = Writer(sys.stdout)
56			f.write("#%s\r\n" % SEP.join(master[0]))
57
58			for row, csv in enumerate(csvs):
59			map = maps[row]
60			for c in csv[1:]:
61			line = []
62			for j in xrange(max_cols):
63			idx = map[j]
64			if idx >= 0:
65			line.append(c[idx])
66			else:
67			line.append(SEP)
68			f.write("%s\r\n" % SEP.join(line))
69
70			if __name__ == "__main__":
71			main()