23 lines
650 B
Ruby
23 lines
650 B
Ruby
|
class Sentencepiece < Formula
|
||
|
|
desc "Unsupervised text tokenizer and detokenizer"
|
||
|
|
homepage "https://github.com/google/sentencepiece"
|
||
|
|
url "https://github.com/google/sentencepiece/archive/v0.1.84.tar.gz"
|
||
|
|
sha256 "7f5505bd46aa5103ad235d61c269932c5922e68597eb484a9d27faed6af1419f"
|
||
|
|
|
||
|
|
depends_on "cmake" => :build
|
||
|
|
|
||
|
|
def install
|
||
|
|
mkdir "build" do
|
||
|
|
system "cmake", "..", *std_cmake_args
|
||
|
|
system "make"
|
||
|
|
system "make", "install"
|
||
|
|
end
|
||
|
|
pkgshare.install "data"
|
||
|
|
end
|
||
|
|
|
||
|
|
test do
|
||
|
|
cp (pkgshare/"data/botchan.txt"), testpath
|
||
|
|
system "#{bin}/spm_train", "--input=botchan.txt", "--model_prefix=m", "--vocab_size=1000"
|
||
|
|
end
|
||
|
|
end
|