#!/bin/bash

# Loop over languages
for L in de cs hu et mt
do
  # Loop over alignment methods that yielded sentence pairs
  for ALIGN in hunalign vecalign bleualign-nmt bleualign-smt
  do
    # Raw sentence alignment file that contains URLs
    echo "wget http://www.statmt.org/paracrawl-benchmarks/paracrawl-benchmark.en-$L.$ALIGN.xz"
    # Deduplicated version of above
    echo "wget http://www.statmt.org/paracrawl-benchmarks/paracrawl-benchmark.en-$L.$ALIGN.dedup.xz"
    # Loop over baseline filtering methods
    for FILTER in zipporah bicleaner laser
    do
      echo "wget http://www.statmt.org/paracrawl-benchmarks/paracrawl-benchmark.en-$L.$ALIGN.dedup.$FILTER.xz"
    done
  done
done