build-deb.sh

#!/bin/bash
set -e

cd "$(dirname "$0")"

# Packaging deps manifest audit — fails the build before any work if the
# Depends/Recommends lines below diverge from packaging/dependencies.yaml.
# Prevents silent regressions (cf. ffmpeg removed in d4d1fff "fix v1.1.1").
python3 packaging/audit-deps.py

VERSION="1.4.0~beta"
PKG_DIR="pkg/dictee"

# Final artefacts go in .dev/dist/ (gitignored), keeping the repo root clean.
DIST_DIR=".dev/dist"
mkdir -p "$DIST_DIR"

DOTOOL_REPO="https://git.sr.ht/~geb/dotool"
DOTOOL_DIR="/tmp/dotool-build"

# shellcheck disable=SC1091
source ./build-common.sh

echo "========================================"
echo "  Building dictee $VERSION"
echo "========================================"
echo ""
echo "Build dependencies:"
echo "  sudo apt install golang-go scdoc libxkbcommon-dev"
echo ""

# Wrappers Python / shell, configs, locales, VERSION, assets.
# Shared with build-rpm.sh / build-tar.sh via build-common.sh — single
# source of truth, no more drift between scripts.
dict_prepare_pkg_dir

# Build dotool (keyboard input tool)
build_dotool() {
    echo "=== Building dotool ==="

    # Vérifier les dépendances de compilation
    local missing=()
    command -v go >/dev/null || missing+=("golang-go")
    command -v scdoc >/dev/null || missing+=("scdoc")
    dpkg -s libxkbcommon-dev >/dev/null 2>&1 || missing+=("libxkbcommon-dev")
    if [ ${#missing[@]} -gt 0 ]; then
        echo "Dépendances manquantes pour compiler dotool : ${missing[*]}"
        echo "  sudo apt install ${missing[*]}"
        exit 1
    fi

    if [ ! -f "$DOTOOL_DIR/dotool" ]; then
        rm -rf "$DOTOOL_DIR"
        git clone "$DOTOOL_REPO" "$DOTOOL_DIR"
        (cd "$DOTOOL_DIR" && ./build.sh)
    else
        echo "dotool already built, skipping"
    fi

    # Install into package tree
    cp "$DOTOOL_DIR/dotool" "$PKG_DIR/usr/bin/"
    cp "$DOTOOL_DIR/dotoold" "$PKG_DIR/usr/bin/"
    mkdir -p "$PKG_DIR/etc/udev/rules.d"
    cp "$DOTOOL_DIR/80-dotool.rules" "$PKG_DIR/etc/udev/rules.d/"
    echo "dotool built and staged"
    echo ""
}

build_dotool

# Plasmoid: now built by dict_prepare_pkg_dir (build-common.sh) so all
# builders (deb / rpm / tar) embed the same just-built artifact. We just
# expose a copy in $DIST_DIR/ for upload-as-release-asset convenience.
expose_plasmoid_in_dist() {
    if [ -f "$PKG_DIR/usr/share/dictee/dictee.plasmoid" ]; then
        mkdir -p "$DIST_DIR"
        cp "$PKG_DIR/usr/share/dictee/dictee.plasmoid" "$DIST_DIR/dictee.plasmoid"
        echo "Plasmoid exposed in $DIST_DIR/"
    fi
}

expose_plasmoid_in_dist

# Build CUDA version
build_cuda() {
    echo "=== [CUDA] Compiling binaries with GPU support ==="
    # Use a dedicated staging dir so the CUDA binaries + libonnxruntime libs
    # don't get overwritten by the subsequent CPU build (which would leave
    # pkg/dictee/usr/bin/transcribe-daemon as a CPU binary and silently
    # downgrade `/usr/bin/transcribe-daemon` dev symlinks). Snapshot the
    # common staging (pkg/dictee, populated by dict_prepare_pkg_dir +
    # build_dotool) so we inherit all .py wrappers, configs, locales,
    # systemd units, etc. — only the binaries + control + libs differ.
    local PKG_DIR="pkg/dictee-cuda"
    rm -rf "$PKG_DIR"
    cp -a pkg/dictee "$PKG_DIR"

    # CRITICAL: --no-default-features disables ort-defaults (static linking)
    # load-dynamic enables runtime loading of libonnxruntime.so for CUDA
    cargo build --release --no-default-features --features "cuda,sortformer,load-dynamic" \
        --bin transcribe \
        --bin transcribe-daemon \
        --bin transcribe-client \
        --bin transcribe-diarize \
        --bin transcribe-stream-diarize \
        --bin transcribe-diarize-batch \
        --bin diarize-only

    # Update control file for CUDA
    cat > "$PKG_DIR/DEBIAN/control" << 'EOF'
Package: dictee-cuda
Version: 1.4.0~beta
Section: sound
Priority: optional
Architecture: amd64
Depends: python3, python3-venv, python3-pip, python3-evdev, pulseaudio-utils, pipewire | alsa-utils, libnotify-bin, python3-pyqt6, python3-pyqt6.qtmultimedia, python3-pyqt6.qtsvg, sox, ffmpeg
Recommends: wl-clipboard, xclip | xsel, curl, translate-shell, python3-numpy, docker.io, python3-gi, gir1.2-ayatanaappindicator3-0.1, qt6-gtk-platformtheme, pkexec, libkf6config-bin, wireplumber
Suggests: gnome-shell-extension-appindicator
Conflicts: dictee-cpu
Provides: dictee
Maintainer: rcspam <rcspams@gmail.com>
Description: Fast speech-to-text with NVIDIA Parakeet (CUDA GPU version)
 A daemon-based speech recognition system using NVIDIA's Parakeet TDT model.
 This version uses CUDA for GPU acceleration (NVIDIA GPUs only).
 Falls back to CPU if CUDA is not available.
 .
 Features:
  - GPU-accelerated transcription via CUDA
  - Low-latency daemon mode with preloaded model
  - Push-to-talk dictation with dotool integration
  - Speaker diarization with Sortformer (who speaks when)
EOF

    # Copy binaries
    cp target/release/transcribe "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-daemon "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-client "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-diarize "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-stream-diarize "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-diarize-batch "$PKG_DIR/usr/bin/"
    cp target/release/diarize-only "$PKG_DIR/usr/bin/"

    # ONNX Runtime CUDA libs (load-dynamic: libonnxruntime.so not in target/release)
    echo "=== Copying CUDA ONNX Runtime libs ==="
    mkdir -p "$PKG_DIR/usr/lib/dictee"

    # Search paths for libonnxruntime.so (not produced by load-dynamic build)
    ORT_LIB=""
    for candidate in \
        "target/release/libonnxruntime.so" \
        "onnxruntime-linux-x64-gpu-*/lib/libonnxruntime.so" \
        "$HOME/.cache/ort.pyke.io/dfbin/*/libonnxruntime.so"; do
        # shellcheck disable=SC2086
        for f in $candidate; do
            if [ -f "$f" ]; then
                ORT_LIB="$f"
                break 2
            fi
        done
    done
    if [ -z "$ORT_LIB" ]; then
        echo "ERROR: libonnxruntime.so not found! Download ONNX Runtime GPU from:"
        echo "  https://github.com/microsoft/onnxruntime/releases"
        echo "Extract it in the project root (onnxruntime-linux-x64-gpu-*/lib/)"
        exit 1
    fi
    echo "Using libonnxruntime.so from: $ORT_LIB"
    cp -L "$ORT_LIB" "$PKG_DIR/usr/lib/dictee/"

    # Provider libs from the same directory as libonnxruntime.so.
    # ORT_DIR was historically undefined — fall back to dirname(ORT_LIB).
    ORT_LIB_DIR="$(dirname "$ORT_LIB")"
    _missing_providers=""
    for lib in libonnxruntime_providers_cuda.so libonnxruntime_providers_shared.so; do
        if [ -f "$ORT_LIB_DIR/$lib" ]; then
            cp -L "$ORT_LIB_DIR/$lib" "$PKG_DIR/usr/lib/dictee/"
            echo "  $lib ← $ORT_LIB_DIR/$lib"
        else
            _missing_providers="$_missing_providers $lib"
        fi
    done
    if [ -n "$_missing_providers" ]; then
        echo "ERROR: CUDA providers missing in $ORT_LIB_DIR:$_missing_providers" >&2
        echo "The CUDA package will silently fall back to CPU at runtime." >&2
        exit 1
    fi

    # CUDA runtime libs (cudart, cublas, cudnn, cufft, curand, nvrtc) are
    # NOT bundled here. They are pip-installed at postinst time into
    # /opt/dictee/cuda-venv and symlinked into /usr/lib/dictee/ so this
    # package stays portable on any distro without requiring the NVIDIA
    # repo. See pkg/dictee/DEBIAN/postinst.

    # ld.so.conf.d entry so the dynamic linker finds them
    mkdir -p "$PKG_DIR/etc/ld.so.conf.d"
    echo "/usr/lib/dictee" > "$PKG_DIR/etc/ld.so.conf.d/dictee.conf"

    chmod 755 "$PKG_DIR/usr/bin/"*
    chmod 755 "$PKG_DIR/DEBIAN/postinst"
    chmod 755 "$PKG_DIR/DEBIAN/postrm"

    # Compress man pages
    gzip -9 -f "$PKG_DIR/usr/share/man/man1/"*.1 2>/dev/null || true
    gzip -9 -f "$PKG_DIR/usr/share/man/fr/man1/"*.1 2>/dev/null || true

    dpkg-deb --build "$PKG_DIR" "$DIST_DIR/dictee-cuda_${VERSION}_amd64.deb"

    # Decompress for next build
    gunzip "$PKG_DIR/usr/share/man/man1/"*.gz 2>/dev/null || true
    gunzip "$PKG_DIR/usr/share/man/fr/man1/"*.gz 2>/dev/null || true

    # The universal CUDA tarball is now built by ./build-tar.sh, which
    # runs its own CUDA build — no need to stash binaries or libs from
    # this script anymore.

    # Cleanup CUDA libs for CPU build (keep shared scripts/modules).
    # Glob *.so* catches libcudart.so.12 / libcufft.so.11 too (the plain
    # *.so only matches libonnxruntime.so, leaving 278 MB of CUDA libs
    # behind inside the CPU .deb).
    rm -f "$PKG_DIR/usr/lib/dictee/"*.so \
          "$PKG_DIR/usr/lib/dictee/"*.so.* \
          "$PKG_DIR/etc/ld.so.conf.d/dictee.conf"
    echo "Built: $DIST_DIR/dictee-cuda_${VERSION}_amd64.deb"
}

# Build CPU version
build_cpu() {
    echo ""
    echo "=== [CPU] Compiling binaries for CPU-only ==="
    # Dedicated staging dir (cf. build_cuda comment). Snapshot the common
    # staging — only the binaries + control differ from the CUDA flavor.
    local PKG_DIR="pkg/dictee-cpu"
    rm -rf "$PKG_DIR"
    cp -a pkg/dictee "$PKG_DIR"

    cargo build --release --features "sortformer" \
        --bin transcribe \
        --bin transcribe-daemon \
        --bin transcribe-client \
        --bin transcribe-diarize \
        --bin transcribe-stream-diarize \
        --bin transcribe-diarize-batch \
        --bin diarize-only

    # Update control file for CPU
    cat > "$PKG_DIR/DEBIAN/control" << 'EOF'
Package: dictee-cpu
Version: 1.4.0~beta
Section: sound
Priority: optional
Architecture: amd64
Depends: python3, python3-venv, python3-evdev, pulseaudio-utils, pipewire | alsa-utils, libnotify-bin, python3-pyqt6, python3-pyqt6.qtmultimedia, python3-pyqt6.qtsvg, sox, ffmpeg
Recommends: wl-clipboard, xclip | xsel, curl, translate-shell, python3-numpy, docker.io, python3-gi, gir1.2-ayatanaappindicator3-0.1, qt6-gtk-platformtheme, pkexec, libkf6config-bin, wireplumber
Suggests: gnome-shell-extension-appindicator
Conflicts: dictee-cuda
Provides: dictee
Maintainer: rcspam <rcspams@gmail.com>
Description: Fast speech-to-text with NVIDIA Parakeet (CPU version)
 A daemon-based speech recognition system using NVIDIA's Parakeet TDT model.
 This version runs on CPU only (works on any computer, slower than GPU).
 .
 Features:
  - CPU-based transcription (no GPU required)
  - Low-latency daemon mode with preloaded model
  - Push-to-talk dictation with dotool integration
  - Speaker diarization with Sortformer (who speaks when)
EOF

    # Copy binaries
    cp target/release/transcribe "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-daemon "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-client "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-diarize "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-stream-diarize "$PKG_DIR/usr/bin/"
    cp target/release/transcribe-diarize-batch "$PKG_DIR/usr/bin/"
    cp target/release/diarize-only "$PKG_DIR/usr/bin/"

    chmod 755 "$PKG_DIR/usr/bin/"*

    # Compress man pages
    gzip -9 -f "$PKG_DIR/usr/share/man/man1/"*.1 2>/dev/null || true
    gzip -9 -f "$PKG_DIR/usr/share/man/fr/man1/"*.1 2>/dev/null || true

    dpkg-deb --build "$PKG_DIR" "$DIST_DIR/dictee-cpu_${VERSION}_amd64.deb"
    echo "Built: $DIST_DIR/dictee-cpu_${VERSION}_amd64.deb"

    # Decompress for potential next build
    gunzip "$PKG_DIR/usr/share/man/man1/"*.gz 2>/dev/null || true
    gunzip "$PKG_DIR/usr/share/man/fr/man1/"*.gz 2>/dev/null || true
}

# Build standalone dictee-plasmoid .deb (Architecture: all, depends on dictee).
# Rebuild guarantees pkg/dictee-plasmoid/ is refreshed from source on every run
# so the .deb content always matches the just-built dictee.plasmoid.
build_plasmoid_deb() {
    echo ""
    echo "=== [PLASMOID] Building standalone dictee-plasmoid .deb ==="
    local PP="pkg/dictee-plasmoid"

    if [ ! -f "$DIST_DIR/dictee.plasmoid" ]; then
        echo "ERROR: $DIST_DIR/dictee.plasmoid not built (build_plasmoid must run first)." >&2
        exit 1
    fi

    # /usr/share/dictee/dictee.plasmoid is shipped by dictee-cuda/cpu (used by
    # their postinst for `kpackagetool6 -i`). NOT shipped here to avoid the
    # dpkg file-conflict observed in v1.3.4 ("trying to overwrite '/usr/share/
    # dictee/dictee.plasmoid', which is also in package dictee-cuda"). The
    # extracted tree below is enough for plasmashell auto-discovery.
    rm -rf "$PP/usr/share/dictee"

    # Refresh the extracted plasmoid tree under /usr/share/plasma/plasmoids/
    local PLASMA_DIR="$PP/usr/share/plasma/plasmoids/com.github.rcspam.dictee"
    rm -rf "$PLASMA_DIR"
    mkdir -p "$PLASMA_DIR"
    unzip -q -o "$DIST_DIR/dictee.plasmoid" -d "$PLASMA_DIR"

    # Refresh locale .mo files from plasmoid source (system-wide install path)
    for _lang in fr de es it pt uk; do
        local _mo="plasmoid/package/contents/locale/$_lang/LC_MESSAGES/plasma_applet_com.github.rcspam.dictee.mo"
        if [ -f "$_mo" ]; then
            mkdir -p "$PP/usr/share/locale/$_lang/LC_MESSAGES"
            cp "$_mo" "$PP/usr/share/locale/$_lang/LC_MESSAGES/"
        fi
    done

    # Regenerate control file with current $VERSION
    cat > "$PP/DEBIAN/control" << EOF
Package: dictee-plasmoid
Version: ${VERSION}
Section: kde
Priority: optional
Architecture: all
Depends: dictee
Recommends: python3-numpy, pulseaudio-utils
Maintainer: rcspam <rcspams@gmail.com>
Description: KDE Plasma 6 widget for dictee voice dictation
 A native KDE Plasma 6 widget for dictee voice dictation.
 Displays real-time audio visualization during recording, daemon status,
 and provides quick controls (dictate, translate, meeting).
 .
 Five animation styles with configurable sensitivity and color gradients.
EOF

    chmod 755 "$PP/DEBIAN/postinst" "$PP/DEBIAN/postrm" 2>/dev/null || true

    dpkg-deb --build "$PP" "$DIST_DIR/dictee-plasmoid_${VERSION}_all.deb"
    echo "Built: $DIST_DIR/dictee-plasmoid_${VERSION}_all.deb"
    echo ""
}

# Build tar.gz (non-Debian)
# The universal CUDA tarball is now built by ./build-tar.sh — that
# script does its own forced CUDA build (with hard guards) so the
# archive can never silently end up with CPU binaries again.
# Run `./build-tar.sh` separately when you need the tarball.

# Build all Debian variants
# Order matters: CPU first, CUDA last → cargo build for CUDA finishes the
# pass, so target/release/ ends with CUDA-build binaries. That way the dev
# host symlinks /usr/bin/transcribe-{daemon,…} → target/release/* keep
# working in GPU mode after the script returns.
build_cpu
build_cuda
build_plasmoid_deb

# Safeguard: the build MUST produce cuda + cpu + plasmoid .deb. If any is
# missing, someone removed or broke a build step — fail loud instead of
# shipping an incomplete release. Do not remove this check.
_expected_debs=(
    "$DIST_DIR/dictee-cuda_${VERSION}_amd64.deb"
    "$DIST_DIR/dictee-cpu_${VERSION}_amd64.deb"
    "$DIST_DIR/dictee-plasmoid_${VERSION}_all.deb"
)
_missing=""
for _d in "${_expected_debs[@]}"; do
    [ -f "$_d" ] || _missing="$_missing $_d"
done
if [ -n "$_missing" ]; then
    echo "" >&2
    echo "ERROR: expected .deb packages missing:$_missing" >&2
    echo "Every build must produce cuda + cpu + plasmoid .deb." >&2
    exit 1
fi

echo ""
echo "========================================"
echo "  Build complete!"
echo "========================================"
echo ""
echo "Packages created in $DIST_DIR/ :"
echo "  - dictee-cuda_${VERSION}_amd64.deb     (Debian/Ubuntu, NVIDIA GPU)"
echo "  - dictee-cpu_${VERSION}_amd64.deb      (Debian/Ubuntu, CPU)"
echo "  - dictee-plasmoid_${VERSION}_all.deb   (Debian/Ubuntu, plasmoid only)"
echo "  - dictee-${VERSION}_amd64.tar.gz       (Autres distributions)"
echo ""
echo "Install (.deb):"
echo "  sudo dpkg -i $DIST_DIR/dictee-{cuda,cpu}_${VERSION}_amd64.deb"
echo "  sudo dpkg -i $DIST_DIR/dictee-plasmoid_${VERSION}_all.deb  # plasmoid only"
echo "  sudo apt-get install -f  # if dependencies missing"
echo ""
echo "Install (.tar.gz):"
echo "  tar xzf $DIST_DIR/dictee-${VERSION}_amd64.tar.gz"
echo "  cd dictee-${VERSION}"
echo "  sudo ./install.sh"
echo ""
echo "Uninstall:"
echo "  sudo dpkg -r dictee-{cuda,cpu,plasmoid}"
echo "  sudo ./uninstall.sh  # tar.gz"