regex word filtering

src/util.jl

@@ -1,12 +1,13 @@
-function read_from_file(vocab_path::String, min_freq::Int64=0, stopwords::Set{String}=Set{String}())
+function read_from_file(vocab_path::String, min_freq::Int64=0, stopwords::Set{String}=Set{String}(); 
+		regex::Regex=r"")
 	fin = open(vocab_path)
 	freqs = Array(Int64, 0)
 	id2word = Array(String, 0)
 	while !eof(fin)
 		try
 			word, freq = split(readline(fin))
 			freq_num = int64(freq)
-			if freq_num < min_freq || word in stopwords continue end
+			if freq_num < min_freq || word in stopwords || !ismatch(regex, word) continue end
 			push!(id2word, word)
 			push!(freqs, freq_num)
 		catch e

train.jl

@@ -71,6 +71,10 @@ s = ArgParseSettings()
     help = "minimal probability of a meaning to contribute into gradients"
     arg_type = Float64
     default = 1e-10
+  "--regex"
+    help = "ignore words not matching provided regex"
+    arg_type = String
+    default = ""
 end
 
 args = parse_args(ARGS, s)
@@ -88,7 +92,7 @@ end
 
 print("Building dictionary... ")
 vm, dict = read_from_file(args["dict"], args["dim"], args["prototypes"],
-  args["min-freq"], args["remove-top-k"], stopwords)
+  args["min-freq"], args["remove-top-k"], stopwords; regex=Regex(args["regex"]))
 println("Done!")
 
 vm.alpha = args["alpha"]