#!/usr/bin/env bash

# Prepare script for MLS English ASR recipe in icefall
export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
set -eou pipefail

stage=-1
stop_stage=100

# Configuration for BPE tokenizer
vocab_sizes=(500)

# Directory where dataset will be downloaded
dl_dir=$PWD/download

. shared/parse_options.sh || exit 1

# All files generated by this script are saved in "data/".
mkdir -p data/manifests data/fbank data/audio data/lang

log() {
    local fname=${BASH_SOURCE[1]##*/}
    echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${LINENO}:${FUNCNAME[1]}) $*"
}

log "Starting MLS English data preparation"

# Stage 0: Download corpus
if [ $stage -le 0 ] && [ $stop_stage -ge 0 ]; then
  log "Stage 0: Download MLS English dataset"
  if [ ! -d $dl_dir/mls_english ]; then
    git clone https://huggingface.co/datasets/parler-tts/mls_eng \
      $dl_dir/mls_english || {
        log "Failed to download MLS English dataset"; exit 1; }
  fi
fi

# Stage 1: Compute fbank & emit manifests
if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
  log "Stage 1: Compute & validate MLS English fbank"
  # we already did `mkdir -p data/manifests data/fbank data/audio` above

  if [ ! -e data/fbank/.mls_eng-fbank.done ]; then
    python local/compute_fbank_mls_english.py \
      --manifest-dir data/manifests \
      --audio-dir    data/audio \
      --dl-dir       $dl_dir/mls_english \
      --fbank-dir    data/fbank

    # Validate each split’s manifest
    for split in train dev test; do
      python local/validate_manifest.py \
        --manifest data/manifests/mls_eng_cuts_${split}.jsonl.gz
    done

    touch data/fbank/.mls_eng-fbank.done
    log "fbank + manifest generation complete."
  else
    log "Skipping: fbank already done (data/fbank/.mls_eng-fbank.done exists)."
  fi
fi


# Stage 2: Prepare transcript for BPE
if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
  log "Stage 2: Generate transcript for BPE"
  if [ ! -f data/lang/transcript.txt ]; then
    ./local/utils/generate_transcript.py --lang-dir data/lang
  fi
fi

# Stage 3: Train BPE models
if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
  log "Stage 3: Train BPE models"
  for vocab_size in "${vocab_sizes[@]}"; do
    bpe_dir=data/lang_bpe_${vocab_size}
    mkdir -p $bpe_dir
    if [ ! -f $bpe_dir/bpe.model ]; then
      ./local/train_bpe_model.py \
        --lang-dir  $bpe_dir \
        --vocab-size $vocab_size \
        --transcript data/lang/transcript.txt
    fi
  done
fi

log "MLS English data preparation completed successfully"