diff options
| author | Patrick Simianer <p@simianer.de> | 2015-11-12 13:57:07 +0100 |
|---|---|---|
| committer | Patrick Simianer <p@simianer.de> | 2015-11-12 13:57:07 +0100 |
| commit | 5c2833c505dda0d1646b8f8c1e62abd391f0401e (patch) | |
| tree | 0259c7c2bdb531c09587a744869848d87f4bbd9e /norm_german | |
| parent | ef282dbe4fef1b0ae0c8544f0bb84ba674c68de7 (diff) | |
| parent | 8151031373c08ccd714a99f50783eafcb54d2010 (diff) | |
Merge branch 'master' of github.com:pks/scripts
Diffstat (limited to 'norm_german')
| -rwxr-xr-x | norm_german | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/norm_german b/norm_german index ef0408e..cf9c060 100755 --- a/norm_german +++ b/norm_german @@ -7,7 +7,7 @@ STDIN.set_encoding 'utf-8' STDOUT.set_encoding 'utf-8' -cfg = Trollop::options do +conf = Trollop::options do banner "norm_german < <file w/ lowercased tokens>" opt :upper, "uppercase", :type => :bool, :default => false opt :threads, "#threads", :type => :int, :default => 1, :short => '-h' @@ -18,7 +18,7 @@ end pairs_lower = [ ['ß','ss'], ['ue', 'ü'], ['ae','ä'], ['oe', 'ö'] ] pairs_upper = [ ['Ä', 'Ae'], ['Ö', 'Oe'], ['Ü', 'Ue'] ] -if cfg[:upper] +if conf[:upper] PAIRS = pairs_lower else PAIRS = pairs_lower+pairs_upper @@ -58,7 +58,7 @@ while tok = STDIN.gets token_stock << [] if !token_stock[thread_n] token_stock[thread_n] << tok.strip! counter += 1 - if token_stock[thread_n].size%cfg[:shard_size]==0 + if token_stock[thread_n].size%conf[:shard_size]==0 STDERR.write "Starting thread ##{thread_n}\n" threads << Thread.new(token_stock[thread_n]) { |tokens| th = build_partial tokens @@ -71,7 +71,7 @@ while tok = STDIN.gets else next end - if thread_n==cfg[:threads] + if thread_n==conf[:threads] threads.each { |i| i.join } token_stock.each { |i| i.clear } thread_n = 0 |