public static void main(final String[] arg)
throws IOException, JSAPException, NoSuchMethodException {
final SimpleJSAP jsap =
new SimpleJSAP(
BloomFilter.class.getName(),
"Creates a Bloom filter reading from standard input a newline-separated list of terms.",
new Parameter[] {
new FlaggedOption(
"bufferSize",
IntSizeStringParser.getParser(),
"64Ki",
JSAP.NOT_REQUIRED,
'b',
"buffer-size",
"The size of the I/O buffer used to read terms."),
new FlaggedOption(
"encoding",
ForNameStringParser.getParser(Charset.class),
"UTF-8",
JSAP.NOT_REQUIRED,
'e',
"encoding",
"The term file encoding."),
new UnflaggedOption(
"bloomFilter",
JSAP.STRING_PARSER,
JSAP.NO_DEFAULT,
JSAP.REQUIRED,
JSAP.NOT_GREEDY,
"The filename for the serialised front-coded list."),
new UnflaggedOption(
"size",
JSAP.INTSIZE_PARSER,
JSAP.NO_DEFAULT,
JSAP.REQUIRED,
JSAP.NOT_GREEDY,
"The size of the filter (i.e., the expected number of elements in the filter; usually, the number of terms)."),
new UnflaggedOption(
"precision",
JSAP.INTEGER_PARSER,
JSAP.NO_DEFAULT,
JSAP.REQUIRED,
JSAP.NOT_GREEDY,
"The precision of the filter.")
});
JSAPResult jsapResult = jsap.parse(arg);
if (jsap.messagePrinted()) return;
final int bufferSize = jsapResult.getInt("bufferSize");
final String filterName = jsapResult.getString("bloomFilter");
final Charset encoding = (Charset) jsapResult.getObject("encoding");
BloomFilter filter = new BloomFilter(jsapResult.getInt("size"), jsapResult.getInt("precision"));
final ProgressLogger pl = new ProgressLogger();
pl.itemsName = "terms";
pl.start("Reading terms...");
MutableString s = new MutableString();
FastBufferedReader reader =
new FastBufferedReader(new InputStreamReader(System.in, encoding), bufferSize);
while (reader.readLine(s) != null) {
filter.add(s);
pl.lightUpdate();
}
pl.done();
BinIO.storeObject(filter, filterName);
}
