PyT (Python Transformer) is a tool designed to simplify file processing routine. Basically, it is a tool that runs a user specified piece of Python code for each line in the file. PyT can be used as an alternative to standard Unix tools such as awk, sed, grep, etc.

Just clone the repository and make a symlink:

$ git clone git://github.com/mroizner/pyt.git ~/pyt
$ mkdir -p ~/bin
$ ln -s ~/pyt/pyt.py ~/bin/pyt

If ~/bin is not in your $PATH you can add it:

$ echo 'export PATH+=\"~/bin:\$PATH\"' >> ~/.bashrc
$ . ~/.bashrc

Suppose you have a file "input.txt" that you want to be processed:

1
1	1
1	2	1
1	3	3	1
1	4	6	4	1

The simple way of using PyT is just specifying any Python code you want to be called for each line. The line itself will be passed to the code as a variable line. The index of the line (0-based) will be passed as a variable index.

If you want output number of characters on each line you can type:

$ pyt 'print len(line)' input.txt
1
3
5
7
9

You can write multi-line code:

$ pyt '
if index % 2 == 0:
  line = line.replace("\t", ", ")
  print len(line)
' input.txt
1
7
13

If you have a tsv-file you can use tab-separated values for each line in a variable records by specifying --split (-s):

$ pyt -s 'print sum(int(x) for x in records)' input.txt
1
2
4
8
16

If you want to print something for each line you can use --print (-p) option. In the split-mode, it prints tab-joined records variable. Otherwise it prints just the line variable.

$ pyt -s -p '
if len(records) > 1:
  records[0], records[1] = records[1], records[0]
' input.txt
1
1	1
2	1	1
3	1	3	1
4	1	6	4	1

As always, if you want to save the output to some file you can do it by specifying the file name. You can also read input data from the standard input rather than from a file, just omit the input file name. If the output file name is the same as the input file name the output will be written to a temp file and then the temp file will be renamed to the output file.

$ cp input.txt output.txt
$ pyt -s 'print sum(int(x) for x in records)' output.txt output.txt
$ cat output.txt
1
2
4
8
16

By default, \n on the end of every line is omitted. You can use --nostrip (-S) option to keep it:

$ pyt -S 'print len(line)' input.txt
2
4
6
8
10

There some cases when you need to do something more than just to execute some simple code for every line. For example, you might need to import some modules. For such cases, you can use --extended (-e) option. In this mode, your code is executed only once for the whole input data (rather than for every line) and all the input is passed to your code as a varaible input. You can iterate over input lines with it:

$ pyt -e '
S = 0
for line in input:
  tokens = line.split()
  s = sum(int(x) for x in tokens)
  S += s
  print math.log(s)  # math is already imported
print S
' input.txt
0.0
0.69314718056
1.38629436112
2.07944154168
2.77258872224
31

If you use --nostrip option together with this mode, input will represent the input stream (with all the useful methods such as read) rather than just a generator of stripped lines.