From e26434979adc33bd949566ba7bf02dff64e80a3e Mon Sep 17 00:00:00 2001 From: Chris Dyer Date: Tue, 2 Oct 2012 00:19:43 -0400 Subject: cdec cleanup, remove bayesian stuff, parsing stuff --- gi/morf-segmentation/linestripper.py | 40 ------------------------------------ 1 file changed, 40 deletions(-) delete mode 100755 gi/morf-segmentation/linestripper.py (limited to 'gi/morf-segmentation/linestripper.py') diff --git a/gi/morf-segmentation/linestripper.py b/gi/morf-segmentation/linestripper.py deleted file mode 100755 index 04e9044a..00000000 --- a/gi/morf-segmentation/linestripper.py +++ /dev/null @@ -1,40 +0,0 @@ -#!/usr/bin/python - -import sys - -#linestripper file file maxlen [numlines] - -if len(sys.argv) < 3: - print "linestripper file1 file2 maxlen [numlines]" - print " outputs subset of file1 to stdout, ..of file2 to stderr" - sys.exit(1) - - -f1 = open(sys.argv[1],'r') -f2 = open(sys.argv[2],'r') - -maxlen=int(sys.argv[3]) -numlines = 0 - -if len(sys.argv) > 4: - numlines = int(sys.argv[4]) - -count=0 -for line1 in f1: - line2 = f2.readline() - - w1 = len(line1.strip().split()) - w2 = len(line2.strip().split()) - - if w1 <= maxlen and w2 <= maxlen: - count = count + 1 - sys.stdout.write(line1) - sys.stderr.write(line2) - - if numlines > 0 and count >= numlines: - break - -f1.close() -f2.close() - - -- cgit v1.2.3