NATIVE BUILD IMPROVEMENT

.dockerignore

@@ -21,12 +21,15 @@ test/
 # Student CLAP training (not needed in production)
 student_clap/
 
-# macOS and Linux standalone app: source trees (PyInstaller spec / nfpm config,
-# vendored redis/pg binaries, assets) and build artifacts. None of it is used by
-# the Linux container, so keep it out of the image.
+# Native standalone app (macOS/Linux/Windows): the per-platform source trees
+# (vendored redis/pg binaries, assets, nfpm config, runtime launchers), the shared
+# PyInstaller spec + build tooling, and build artifacts. None of it is used by the
+# Linux container, so keep it out of the image.
 macos/
 linux/
 windows/
+AudioMuse-AI.spec
+scripts/standalone/
 build/
 dist/
 

.github/workflows/build-linux.yml

@@ -14,7 +14,7 @@
 #
 # The small native build inputs (redis-server, the unaccent/pg_trgm contrib
 # modules) are built from source in this workflow (linux/vendor/*) rather than
-# committed, then baked into the bundle by linux/AudioMuse-AI.spec.
+# committed, then baked into the bundle by the shared AudioMuse-AI.spec.
 #
 # The ~5 GB of models are NOT in git. This workflow assembles ./model from the
 # same GitHub releases the Dockerfile/macOS build use, INCLUDING the HuggingFace
@@ -146,87 +146,10 @@ jobs:
       - name: Assemble ./model (mirrors the Dockerfile/macOS models stage)
         env:
           GH_TOKEN: ${{ github.token }}
-        run: |
-          set -euo pipefail
-          mkdir -p model
-
-          echo "==> musicnn + CLAP text models (from ${MODEL_RELEASE})"
-          gh release download "$MODEL_RELEASE" -R "$GITHUB_REPOSITORY" -D model --clobber \
-            -p musicnn_embedding.onnx \
-            -p musicnn_prediction.onnx \
-            -p clap_text_model.onnx
-
-          echo "==> DCLAP audio model (from ${DCLAP_RELEASE} in the -DCLAP repo)"
-          gh release download "$DCLAP_RELEASE" -R NeptuneHub/AudioMuse-AI-DCLAP -D model --clobber \
-            -p model_epoch_36.onnx \
-            -p model_epoch_36.onnx.data
-
-          echo "==> HuggingFace cache (roberta/bert/bart) -- HF_HOME points at model/huggingface"
-          tmp_hf="$(mktemp -d)"
-          gh release download "$MODEL_RELEASE" -R "$GITHUB_REPOSITORY" -D "$tmp_hf" --clobber \
-            -p huggingface_models.tar.gz
-          mkdir -p model/huggingface
-          tar -xzf "$tmp_hf/huggingface_models.tar.gz" -C model/huggingface
-          rm -rf "$tmp_hf"
-
-          echo "==> Trim HF cache to just the roberta-base tokenizer (~1.4 GB saved)"
-          # The app's ONLY runtime HF dependency is the roberta-base *tokenizer*
-          # (tasks/clap_analyzer.py: AutoTokenizer.from_pretrained("roberta-base")).
-          # bert-base-uncased and bart-base are never loaded by the app, and a
-          # tokenizer does not need model weights. Dropping them keeps the release
-          # assets under the 2 GB GitHub limit without removing any model the app
-          # uses. (This prunes only this bundle copy -- the shared release tarball
-          # and the Docker build are unaffected.)
-          hf="model/huggingface/hub"
-          rm -rf "$hf/models--bert-base-uncased" "$hf/models--facebook--bart-base"
-          rb="$hf/models--roberta-base"
-          if [ -d "$rb" ]; then
-            find "$rb/blobs" -type f -size +10M -delete
-            find "$rb/snapshots" \( -name "model.safetensors" -o -name "pytorch_model.bin" \) -delete
-            du -sh "$rb"
-          fi
-
-          echo "==> lyrics bundles (whisper / silero / gte)"
-          tmp="$(mktemp -d)"
-          gh release download "$MODEL_RELEASE" -R "$GITHUB_REPOSITORY" -D "$tmp" --clobber \
-            -p lyrics_model_whisper.tar.gz \
-            -p lyrics_model_silero_vad.tar.gz \
-            -p lyrics_model_gte_vnni.tar.gz
-          for t in lyrics_model_whisper lyrics_model_silero_vad lyrics_model_gte_vnni; do
-            tar -xzf "$tmp/$t.tar.gz" -C model
-          done
-          rm -rf "$tmp"
+        run: python3 scripts/standalone/assemble_model.py
 
       - name: Verify the assembled model/ is complete
-        run: |
-          set -euo pipefail
-          required=(
-            model/musicnn_embedding.onnx
-            model/musicnn_prediction.onnx
-            model/clap_text_model.onnx
-            model/model_epoch_36.onnx
-            model/model_epoch_36.onnx.data
-            model/huggingface/hub/models--roberta-base/snapshots
-            model/silero_vad.onnx
-            model/gte-multilingual-base-int8.onnx
-            model/whisper-small-onnx/encoder_model.onnx
-            model/whisper-small-onnx/decoder_model_merged.onnx
-            model/gte-multilingual-base/tokenizer.json
-          )
-          missing=0
-          for f in "${required[@]}"; do
-            # Flag a file that is missing OR a zero-byte/truncated download. The
-            # size test only applies to regular files (directory entries in the
-            # list pass on existence alone).
-            if [ ! -e "$f" ] || { [ -f "$f" ] && [ ! -s "$f" ]; }; then
-              echo "::error::Missing or empty: $f"; missing=1
-            fi
-          done
-          if [ -z "$(find model/huggingface/hub/models--roberta-base -name tokenizer.json -print -quit)" ]; then
-            echo "::error::roberta-base tokenizer.json missing after HF-cache prune"; missing=1
-          fi
-          [ "$missing" -eq 0 ] || { echo "::error::Refusing to build an incomplete bundle."; exit 1; }
-          du -sh model
+        run: python3 scripts/standalone/assemble_model.py --verify
 
       - name: Build the packages (.deb + .rpm)
         run: |
@@ -241,7 +164,7 @@ jobs:
           else
             PKG_VERSION="0.0.0"
           fi
-          PKG_VERSION="$PKG_VERSION" bash linux/build.sh
+          PKG_VERSION="$PKG_VERSION" python scripts/standalone/build.py --platform linux
         env:
           GITHUB_REF_NAME: ${{ github.ref_name }}
           GITHUB_REF_TYPE: ${{ github.ref_type }}

.github/workflows/build-macos.yml

@@ -104,85 +104,10 @@ jobs:
       - name: Assemble ./model (mirrors the Dockerfile models stage)
         env:
           GH_TOKEN: ${{ github.token }}
-        run: |
-          set -euo pipefail
-          mkdir -p model
-
-          echo "==> musicnn + CLAP text models (from ${MODEL_RELEASE})"
-          gh release download "$MODEL_RELEASE" -R "$GITHUB_REPOSITORY" -D model --clobber \
-            -p musicnn_embedding.onnx \
-            -p musicnn_prediction.onnx \
-            -p clap_text_model.onnx
-
-          echo "==> DCLAP audio model (from ${DCLAP_RELEASE} in the -DCLAP repo)"
-          gh release download "$DCLAP_RELEASE" -R NeptuneHub/AudioMuse-AI-DCLAP -D model --clobber \
-            -p model_epoch_36.onnx \
-            -p model_epoch_36.onnx.data
-
-          echo "==> HuggingFace cache (roberta/bert/bart) — HF_HOME points at model/huggingface"
-          tmp_hf="$(mktemp -d)"
-          gh release download "$MODEL_RELEASE" -R "$GITHUB_REPOSITORY" -D "$tmp_hf" --clobber \
-            -p huggingface_models.tar.gz
-          mkdir -p model/huggingface
-          tar -xzf "$tmp_hf/huggingface_models.tar.gz" -C model/huggingface
-          rm -rf "$tmp_hf"
-
-          echo "==> Trim HF cache to just the roberta-base tokenizer (~1.4 GB saved)"
-          # The macOS app's ONLY runtime HF dependency is the roberta-base *tokenizer*
-          # (tasks/clap_analyzer.py: AutoTokenizer.from_pretrained("roberta-base")).
-          # bert-base-uncased and bart-base are never loaded by the app, and a tokenizer
-          # does not need model weights. Dropping them keeps the release zip under the
-          # 2 GB GitHub release-asset limit without removing any model the app uses.
-          # Note: this prunes only the macOS bundle copy — the shared release tarball
-          # and the Docker build are unaffected.
-          hf="model/huggingface/hub"
-          rm -rf "$hf/models--bert-base-uncased" "$hf/models--facebook--bart-base"
-          rb="$hf/models--roberta-base"
-          if [ -d "$rb" ]; then
-            # Drop the ~476 MB weight blob; AutoTokenizer reads only the tiny
-            # tokenizer.json/vocab.json/merges.txt/config files (all < 2 MB).
-            find "$rb/blobs" -type f -size +10M -delete
-            find "$rb/snapshots" \( -name "model.safetensors" -o -name "pytorch_model.bin" \) -delete
-            du -sh "$rb"
-          fi
-
-          echo "==> lyrics bundles (whisper / silero / gte) — downloaded then extracted into ./model"
-          tmp="$(mktemp -d)"
-          gh release download "$MODEL_RELEASE" -R "$GITHUB_REPOSITORY" -D "$tmp" --clobber \
-            -p lyrics_model_whisper.tar.gz \
-            -p lyrics_model_silero_vad.tar.gz \
-            -p lyrics_model_gte_vnni.tar.gz
-          for t in lyrics_model_whisper lyrics_model_silero_vad lyrics_model_gte_vnni; do
-            tar -xzf "$tmp/$t.tar.gz" -C model
-          done
-          rm -rf "$tmp"
+        run: python3 scripts/standalone/assemble_model.py
 
       - name: Verify the assembled model/ is complete
-        run: |
-          set -euo pipefail
-          required=(
-            model/musicnn_embedding.onnx
-            model/musicnn_prediction.onnx
-            model/clap_text_model.onnx
-            model/model_epoch_36.onnx
-            model/model_epoch_36.onnx.data
-            model/huggingface/hub/models--roberta-base/snapshots
-            model/silero_vad.onnx
-            model/gte-multilingual-base-int8.onnx
-            model/whisper-small-onnx/encoder_model.onnx
-            model/whisper-small-onnx/decoder_model_merged.onnx
-            model/gte-multilingual-base/tokenizer.json
-          )
-          missing=0
-          for f in "${required[@]}"; do
-            if [ ! -e "$f" ]; then echo "::error::Missing or empty: $f"; missing=1; fi
-          done
-          # The roberta tokenizer files must survive the HF-cache prune above.
-          if [ -z "$(find model/huggingface/hub/models--roberta-base -name tokenizer.json -print -quit)" ]; then
-            echo "::error::roberta-base tokenizer.json missing after HF-cache prune"; missing=1
-          fi
-          [ "$missing" -eq 0 ] || { echo "::error::Refusing to build an incomplete bundle."; exit 1; }
-          du -sh model
+        run: python3 scripts/standalone/assemble_model.py --verify
 
       - name: Install Python dependencies
         run: |
@@ -196,7 +121,7 @@ jobs:
         run: |
           set -euo pipefail
           source .venv-macos/bin/activate
-          bash macos/build.sh
+          python scripts/standalone/build.py --platform macos
 
       - name: Upload zip as a workflow artifact
         uses: actions/upload-artifact@v4

.github/workflows/build-windows.yml

@@ -8,7 +8,7 @@
 #
 # The small native build inputs (redis-server.exe, the unaccent/pg_trgm contrib
 # modules) are built or downloaded in this workflow, then baked into the bundle
-# by windows/AudioMuse-AI.spec.
+# by the shared AudioMuse-AI.spec.
 #
 # The ~5 GB of models are NOT in git. This workflow assembles ./model from the
 # same GitHub releases the Dockerfile/macOS/Linux builds use, INCLUDING the
@@ -103,92 +103,22 @@ jobs:
         shell: powershell
         env:
           GH_TOKEN: ${{ github.token }}
-        run: |
-          New-Item -ItemType Directory -Force -Path model | Out-Null
-          $modelRelease = "$env:MODEL_RELEASE"
-          $dclapRelease = "$env:DCLAP_RELEASE"
-
-          Write-Host "==> musicnn + CLAP text models (from $modelRelease)"
-          gh release download $modelRelease -R $env:GITHUB_REPOSITORY -D model --clobber `
-            -p musicnn_embedding.onnx `
-            -p musicnn_prediction.onnx `
-            -p clap_text_model.onnx
-
-          Write-Host "==> DCLAP audio model (from $dclapRelease in the -DCLAP repo)"
-          gh release download $dclapRelease -R NeptuneHub/AudioMuse-AI-DCLAP -D model --clobber `
-            -p model_epoch_36.onnx `
-            -p model_epoch_36.onnx.data
-
-          Write-Host "==> HuggingFace cache (roberta/bert/bart)"
-          $tmp_hf = Join-Path $env:RUNNER_TEMP "hf_models"
-          New-Item -ItemType Directory -Force -Path $tmp_hf | Out-Null
-          gh release download $modelRelease -R $env:GITHUB_REPOSITORY -D $tmp_hf --clobber `
-            -p huggingface_models.tar.gz
-          New-Item -ItemType Directory -Force -Path model\huggingface | Out-Null
-          tar -xzf "$tmp_hf\huggingface_models.tar.gz" -C model\huggingface
-          if ($LASTEXITCODE -ne 0) { Write-Error "tar failed extracting huggingface models"; exit 1 }
-          Remove-Item -Recurse -Force $tmp_hf
-
-          Write-Host "==> Trim HF cache to just the roberta-base tokenizer (~1.4 GB saved)"
-          $hf = "model\huggingface\hub"
-          if (Test-Path "$hf\models--bert-base-uncased") { Remove-Item -Recurse -Force "$hf\models--bert-base-uncased" }
-          if (Test-Path "$hf\models--facebook--bart-base") { Remove-Item -Recurse -Force "$hf\models--facebook--bart-base" }
-          $rb = "$hf\models--roberta-base"
-          if (Test-Path $rb) {
-            Get-ChildItem -Recurse -File "$rb\blobs" | Where-Object { $_.Length -gt 10MB } | Remove-Item -Force
-            Get-ChildItem -Recurse "$rb\snapshots" -Include "model.safetensors","pytorch_model.bin" | Remove-Item -Force
-          }
-
-          Write-Host "==> lyrics bundles (whisper / silero / gte)"
-          $tmp = Join-Path $env:RUNNER_TEMP "lyrics_models"
-          New-Item -ItemType Directory -Force -Path $tmp | Out-Null
-          gh release download $modelRelease -R $env:GITHUB_REPOSITORY -D $tmp --clobber `
-            -p lyrics_model_whisper.tar.gz `
-            -p lyrics_model_silero_vad.tar.gz `
-            -p lyrics_model_gte_vnni.tar.gz
-          foreach ($t in @("lyrics_model_whisper", "lyrics_model_silero_vad", "lyrics_model_gte_vnni")) {
-            tar -xzf "$tmp\$t.tar.gz" -C model
-            if ($LASTEXITCODE -ne 0) { Write-Error "tar failed extracting $t"; exit 1 }
-          }
-          Remove-Item -Recurse -Force $tmp
+        run: python scripts/standalone/assemble_model.py
 
       - name: Verify the assembled model/ is complete
         shell: powershell
-        run: |
-          $required = @(
-            "model\musicnn_embedding.onnx",
-            "model\musicnn_prediction.onnx",
-            "model\clap_text_model.onnx",
-            "model\model_epoch_36.onnx",
-            "model\model_epoch_36.onnx.data",
-            "model\huggingface\hub\models--roberta-base\snapshots",
-            "model\silero_vad.onnx",
-            "model\gte-multilingual-base-int8.onnx",
-            "model\whisper-small-onnx\encoder_model.onnx",
-            "model\whisper-small-onnx\decoder_model_merged.onnx",
-            "model\gte-multilingual-base\tokenizer.json"
-          )
-          $missing = 0
-          foreach ($f in $required) {
-            if (-not (Test-Path $f)) {
-              Write-Error "Missing or empty: $f"
-              $missing = 1
-            }
-          }
-          if ($missing -ne 0) { Write-Error "Refusing to build an incomplete bundle."; exit 1 }
-          Write-Host "Model assembly verified."
+        run: python scripts/standalone/assemble_model.py --verify
 
       - name: Build the bundle
         shell: powershell
         run: |
-          .venv-windows\Scripts\activate
           if ($env:GITHUB_REF_TYPE -eq "tag") {
             $ver = $env:GITHUB_REF_NAME -replace '^v', ''
           } else {
             $ver = "0.0.0"
           }
           $env:PKG_VERSION = $ver
-          cmd /c windows\build.bat
+          .venv-windows\Scripts\python scripts\standalone\build.py --platform windows
         env:
           GITHUB_REF_NAME: ${{ github.ref_name }}
           GITHUB_REF_TYPE: ${{ github.ref_type }}