summaryrefslogtreecommitdiff
path: root/ng
diff options
context:
space:
mode:
Diffstat (limited to 'ng')
-rwxr-xr-xng44
1 files changed, 12 insertions, 32 deletions
diff --git a/ng b/ng
index de314b8..dbc59eb 100755
--- a/ng
+++ b/ng
@@ -1,39 +1,19 @@
#!/usr/bin/env ruby
-def ngrams_it(s, n, fix=false)
- a = s.strip.split
- a.each_with_index { |tok, i|
- tok.strip!
- 0.upto([n-1, a.size-i-1].min) { |m|
- yield a[i..i+m] if !(fix||(a[i..i+m].size>n))
- }
- }
-end
-
-def main(n, fix, sep)
- STDIN.set_encoding 'utf-8'
- STDOUT.set_encoding 'utf-8'
- while line = STDIN.gets
- a = []
- ngrams_it(line, n, fix) {|ng| a << ng.join(' ')}
- a.reject! {|i| i.strip.size==0 }
- puts a.join sep if a.size > 0
- end
-end
+require 'nlp_ruby'
+require 'trollop'
-def usage
- STDERR.write "./ng [-n <n>] [--fix] [--separator <s>] < <one number per line>\n"
- exit 1
+cfg = Trollop::options do
+ banner "ng < <input>"
+ opt :n, "n for Ngrams", :type => :int, :default => 4
+ opt :fix, "Don't output lower order Ngrams.", :type => :bool, :default => false
+ opt :separator, "separte ngrams of a line by this string", :type => :string, :default => "\n"
end
-if __FILE__ == $0
- require 'trollop'
- opts = Trollop::options do
- opt :n, "Ngrams", :type => :int, :default => 4
- opt :fix, "Don't output lower order Ngrams.", :type => :bool, :default => false
- opt :separator, "separte ngrams of a line by this string", :type => :string, :default => "\n"
- end
- usage if not [0,2,4,6].include? ARGV.size
- main(opts[:n], opts[:fix], opts[:separator])
+while line = STDIN.gets
+ a = []
+ ngrams(line, cfg[:n], cfg[:fix]) { |ng| a << ng.join(' ') }
+ a.reject! { |i| i.strip.size==0 }
+ puts a.join cfg[:separator] if a.size>0
end