func writeData(out io.Writer, instances []TrainingInstance) {
writer := input.NewTrainDataWriter(out)
for _, instance := range instances {
err := writer.Write(instance.Y, instance.X)
common.ExitIfError("Error writing data: ", err)
}
}
func normalizeData(n *input.Normalizer, inFilename string, outFilename string) error {
outF, err := os.Create(outFilename)
if err != nil {
return err
}
defer outF.Close()
trainDataWriter := input.NewTrainDataWriter(outF)
return processData(inFilename, func(y int, x input.InputVector) error {
n.Normalize(x)
return trainDataWriter.Write(y, x)
})
}
func main() {
flag.Parse()
if flag.NArg() != 3 {
flag.Usage()
os.Exit(1)
}
config := common.ReadConfigOrExit(flag.Arg(0))
transitionSystem, ok := common.TransitionSystems[config.Parser.System]
if !ok {
log.Fatalf("Unknown transition system: %s", config.Parser.System)
}
oracleConstructor, ok := common.Oracles[config.Parser.System]
if !ok {
log.Fatalf("Unknown transition system: %s", config.Parser.System)
}
log.Printf("Transition system: %s", config.Parser.System)
ilas := common.ReadIlasOrExit(config.Parser.Inputs)
var labelNumberer *system.LabelNumberer
if config.Parser.Transitions != "" {
if _, err := os.Stat(config.Parser.Transitions); err == nil {
log.Printf("Transitions filename %s exists, reusing...", config.Parser.Transitions)
labelNumberer = common.ReadTransitionsOrExit(config.Parser.Transitions, transitionSystem)
}
}
instanceWriter, err := os.Create(flag.Arg(2))
common.ExitIfError("Cannot open instance file for writing:", err)
defer instanceWriter.Close()
trainDataWriter := input.NewTrainDataWriter(instanceWriter)
layerEmbeddings := common.MustReadAllEmbeddings(config.Embeddings)
realizer := input.NewInputVectorRealizer(ilas, layerEmbeddings, nil)
var collector *common.WritingCollector
if labelNumberer == nil {
collector = common.NewWritingCollector(realizer, trainDataWriter)
} else {
collector = common.NewWritingCollectorWithLabelNumberer(realizer, labelNumberer, trainDataWriter)
}
trainer := system.NewGreedyTrainer(transitionSystem, collector)
f, err := os.Open(flag.Arg(1))
common.ExitIfError("Cannot open training data:", err)
defer f.Close()
log.Println("Creating training instances...")
common.ProcessData(f, func(s []conllx.Token) error {
goldDependencies, err := system.SentenceToDependencies(s)
if err != nil {
return fmt.Errorf("Cannot extract dependencies: %s", err.Error())
}
trainer.Parse(s, oracleConstructor(goldDependencies))
return nil
})
if err != nil {
common.ExitIfError("Cannot process data:", err)
}
if config.Parser.Transitions != "" {
if _, err := os.Stat(config.Parser.Transitions); err != nil {
writeTransitions(transitionSystem, collector.LabelNumberer(), config.Parser.Transitions)
}
}
}