Bakefile

#!/bin/bash

VENDOR="vendor"

# If this folder does not exist, the build will still
# work, but the eSpeak engine in /pronounce will not function.
ESPEAK="$HOME/projects/exp/zkcnik-tts/dist"

build() {
  setup
  vendor
  precache
}

setup() {
  bun i
  vendor
}

vendor() {
  mkdir -p $VENDOR

  echo "==> Copying prebuilt browser bundles"

  cp node_modules/instant.page/instantpage.js                                       $VENDOR/instant-page.js
  cp node_modules/@jaames/iro/dist/iro.js                                           $VENDOR/iro.js
  cp node_modules/cronstrue/dist/cronstrue.js                                       $VENDOR/cronstrue.js
  cp node_modules/marked/lib/marked.umd.js                                          $VENDOR/marked.js
  cp node_modules/prettier/standalone.js                                            $VENDOR/prettier.js
  cp node_modules/prettier/plugins/babel.js                                         $VENDOR/prettier-babel.js
  cp node_modules/prettier/plugins/estree.js                                        $VENDOR/prettier-estree.js
  cp node_modules/prettier/plugins/postcss.js                                       $VENDOR/prettier-postcss.js
  cp node_modules/prettier/plugins/html.js                                          $VENDOR/prettier-html.js
  cp node_modules/prettier/plugins/markdown.js                                      $VENDOR/prettier-markdown.js
  cp node_modules/terser/dist/bundle.min.js                                         $VENDOR/terser.js
  cp node_modules/csso/dist/csso.js                                                 $VENDOR/csso.js
  cp node_modules/html-minifier-terser/dist/htmlminifier.umd.bundle.min.js          $VENDOR/html-minifier.js
  cp -r node_modules/uuid/dist/esm-browser                                          $VENDOR/uuid
  cp node_modules/@huggingface/transformers/dist/ort-wasm-simd-threaded.jsep.mjs    $VENDOR/
  cp node_modules/@huggingface/transformers/dist/ort-wasm-simd-threaded.jsep.wasm   $VENDOR/

  if [ -d "$ESPEAK" ]; then
    echo "==> Copying espeak wasm bundle"

    mkdir -p $VENDOR/espeakng
    cp $ESPEAK/espeak.js $ESPEAK/espeak.wasm $ESPEAK/espeak.data $VENDOR/espeakng/
  fi

  echo "==> Building other browser bundles"

  bun build --target browser --format esm node_modules/@huggingface/transformers/dist/transformers.web.js --outfile $VENDOR/transformers.js >/dev/null

  echo "==> Downloading image models"

  # Keep this in sync with `dtype` in remove.html and translate.html
  ONNX=model_quantized.onnx    # q8,   ~44MB
  
  # Better models:
  # ONNX=model_fp16.onnx       # fp16, ~88MB
  # ONNX=model.onnx            # fp32, ~176MB

  mkdir -p $VENDOR/models/briaai/RMBG-1.4/onnx

  fetch https://huggingface.co/briaai/RMBG-1.4/resolve/main/config.json \
    $VENDOR/models/briaai/RMBG-1.4/config.json
  fetch "https://huggingface.co/briaai/RMBG-1.4/resolve/main/onnx/$ONNX" \
    "$VENDOR/models/briaai/RMBG-1.4/onnx/$ONNX"

  cat > $VENDOR/models/briaai/RMBG-1.4/preprocessor_config.json <<'EOF'
{
  "feature_extractor_type": "ImageFeatureExtractor",
  "do_normalize": true,
  "do_pad": false,
  "do_rescale": true,
  "do_resize": true,
  "image_mean": [0.5, 0.5, 0.5],
  "image_std": [1, 1, 1],
  "resample": 2,
  "rescale_factor": 0.00392156862745098,
  "size": { "width": 1024, "height": 1024 }
}
EOF

  echo "==> Downloading translation models"

  for model in Xenova/opus-mt-en-nl Xenova/opus-mt-nl-en; do
    mkdir -p "$VENDOR/models/$model/onnx"

    for file in config.json tokenizer.json tokenizer_config.json \
                onnx/encoder_model_quantized.onnx \
                onnx/decoder_model_merged_quantized.onnx; do
      fetch "https://huggingface.co/$model/resolve/main/$file" \
        "$VENDOR/models/$model/$file"
    done
  done
}

serve() {
  waiter --dev
}

clean() {
  rm -rf node_modules
  rm -rf $VENDOR
}

precache() {
  {
    find . -maxdepth 1 -name "*.html"
    find . -maxdepth 1 -name "*.css"
    find . -maxdepth 1 -name "*.json"
    find . -maxdepth 1 -name "*.js" ! -name "sw.js"
    find ./$VENDOR -type f
  } | sort > .listing.tmp

  CACHE_HASH=$(xargs cat < .listing.tmp | openssl dgst -sha1 | awk '{print $NF}' | cut -c1-8)

  { printf "  '/',\n"; sed "s|^\./|/|; s|.*|  '&',|" .listing.tmp; } > .precache.tmp

  awk -v hash="tools-${CACHE_HASH}" '
    /^const CACHE_KEY = / { q=sprintf("%c", 39); print "const CACHE_KEY = " q hash q ";"; next }
    /^const PRECACHE = \[/ { print; in_block=1; while ((getline line < ".precache.tmp") > 0) print line; next }
    /^\];/ && in_block { in_block=0 }
    !in_block { print }
  ' sw.js > .sw.tmp && mv .sw.tmp sw.js

  rm .listing.tmp .precache.tmp
}

deploy() {
  lift deploy
}

fetch() {
  if [ -f "$2" ]; then
    echo "Skipped $2"
  else
    curl -fL --progress-bar "$1" -o "$2"
  fi
}