dictee-test-rules

#!/bin/bash
# dictee-test-rules — Test what Parakeet hears and how rules transform it
# Usage: dictee-test-rules [--loop] [--wav file.wav]
#   No args:  record once (Enter to stop), show raw + processed
#   --loop:   record repeatedly (Enter to stop each, Ctrl+C to quit)
#   --wav:    skip recording, transcribe an existing WAV file

set -euo pipefail

# ── Config ──────────────────────────────────────────────────────────────
CONF="$HOME/.config/dictee.conf"
if [ -f "$CONF" ]; then
    # shellcheck disable=SC1090
    source "$CONF"
fi
LANG_SOURCE="${DICTEE_LANG_SOURCE:-fr}"
TMPWAV="/tmp/dictee-test-recording.wav"

# ── Colors ──────────────────────────────────────────────────────────────
RED='\033[0;31m'
GREEN='\033[0;32m'
BLUE='\033[0;34m'
YELLOW='\033[1;33m'
CYAN='\033[0;36m'
BOLD='\033[1m'
DIM='\033[2m'
NC='\033[0m'

# ── Helpers ─────────────────────────────────────────────────────────────
cleanup() {
    rm -f "$TMPWAV"
    # Kill pw-record if still running
    if [ -n "${PW_PID:-}" ] && kill -0 "$PW_PID" 2>/dev/null; then
        kill "$PW_PID" 2>/dev/null
        wait "$PW_PID" 2>/dev/null
    fi
}
trap cleanup EXIT

record_audio() {
    echo -e "${DIM}Appuyez sur Entrée pour commencer l'enregistrement...${NC}"
    read -r
    pw-record --rate 16000 --channels 1 --format s16 "$TMPWAV" &
    PW_PID=$!
    sleep 0.3
    echo -e "${RED}● Parlez maintenant...${NC} ${DIM}(Entrée pour arrêter)${NC}"
    read -r
    kill "$PW_PID" 2>/dev/null
    wait "$PW_PID" 2>/dev/null || true
    PW_PID=
}

transcribe() {
    local wav="$1"
    transcribe-client "$wav" 2>/dev/null
}

postprocess() {
    local raw="$1"
    echo "$raw" | DICTEE_LANG_SOURCE="$LANG_SOURCE" dictee-postprocess 2>/dev/null
}

show_result() {
    local raw="$1"
    local processed="$2"

    echo -e "${CYAN}──────────────────────────────────────${NC}"
    echo -e "${YELLOW}RAW${NC}        ${BOLD}$raw${NC}"

    # If raw contains non-latin characters, decide based on whether
    # processing already transformed it.
    if echo "$raw" | grep -qP '[А-Яа-яЁё]'; then
        if [ "$raw" != "$processed" ]; then
            # An existing rule already matched — no new rule needed.
            echo -e "${RED}CYRILLIQUE${NC} ${GREEN}✓ déjà couvert par une règle existante${NC}"
        else
            # No rule matched — propose new ones (with \n and ", " variants).
            local pattern
            pattern=$(echo "$raw" | sed 's/[.?!,]*$//' | sed 's/[]\/$*.^[]/\\&/g')
            echo -e "${RED}CYRILLIQUE${NC} ${DIM}Parakeet a détecté du cyrillique — aucune règle existante ne matche${NC}"
            echo -e "${DIM}Règles candidates (choisis selon l'intention) :${NC}"
            echo -e "  ${DIM}→ pour « virgule » :${NC} ${GREEN}[fr] /^[,.\\s]*${pattern}[,.\\s]*/, /igm${NC}"
            echo -e "  ${DIM}→ pour « à la ligne » :${NC} ${GREEN}[fr] /^[,.\\s]*${pattern}[,.\\s]*/\\n/igm${NC}"
        fi
    fi

    # Show processed with visible control chars
    local display
    display=$(printf '%s' "$processed" | sed ':a;N;$!ba;s/\n/\\n/g;s/\t/\\t/g')
    if [ "$raw" = "$processed" ]; then
        echo -e "${BLUE}PROCESSED${NC}  ${DIM}(inchangé)${NC}"
    else
        echo -e "${GREEN}PROCESSED${NC}  ${BOLD}$display${NC}"
    fi
    echo
}

# ── Main ────────────────────────────────────────────────────────────────
echo -e "${BOLD}dictee-test-rules${NC} — Langue: ${CYAN}$LANG_SOURCE${NC}"
echo -e "${DIM}Testez ce que Parakeet détecte et comment les règles le transforment.${NC}"
echo

LOOP=false
WAV_FILE=

while [ $# -gt 0 ]; do
    case "$1" in
        --loop) LOOP=true; shift ;;
        --wav)  WAV_FILE="$2"; shift 2 ;;
        *)      echo "Usage: $0 [--loop] [--wav file.wav]"; exit 1 ;;
    esac
done

if [ -n "$WAV_FILE" ]; then
    # Single file mode
    raw=$(transcribe "$WAV_FILE")
    processed=$(postprocess "$raw")
    show_result "$raw" "$processed"
    exit 0
fi

while true; do
    record_audio

    if [ ! -f "$TMPWAV" ]; then
        echo -e "${RED}Pas d'audio enregistré.${NC}"
        continue
    fi

    raw=$(transcribe "$TMPWAV")
    if [ -z "$raw" ]; then
        echo -e "${DIM}(silence détecté)${NC}"
        echo
    else
        processed=$(postprocess "$raw")
        show_result "$raw" "$processed"
    fi

    rm -f "$TMPWAV"

    if [ "$LOOP" = false ]; then
        break
    fi
    echo -e "${DIM}Prêt pour le prochain enregistrement... (Ctrl+C pour quitter)${NC}"
done