feat(listener): add local intents via api

local actions for the speaker, such as bluetooth pair, pause, volume
currently in English, Spanish and Catalan (few basic examples)

this processing is done via API since it already loads into memory,
otherwise running a new Python script would use more start time.

api/intents.py

@@ -0,0 +1,57 @@
+from flask import Blueprint, request, jsonify
+import yaml
+import os
+
+intents = Blueprint('intents', __name__)
+
+intents_file = os.path.join(os.path.dirname(__file__), 'intents.yaml')
+intents_data = yaml.load(open(intents_file, 'r'), Loader=yaml.FullLoader)
+
+def remove_accents(input: str) -> str:
+  characters = {
+    'á': 'a', 'é': 'e', 'í': 'i', 'ó': 'o', 'ú': 'u',
+    'à': 'a', 'è': 'e', 'ì': 'i', 'ò': 'o', 'ù': 'u',
+  }
+
+  for char, repl in characters.items():
+    input = input.replace(char, repl)
+  return input
+
+@intents.post('/intent')
+def get_intents():
+  text = None
+  if request.is_json:
+    data = request.get_json()
+    text = data.get('text', '').strip()
+  else:
+    text = request.form.get('text', '').strip()
+  if not text:
+    return jsonify({'error': 'No text provided'}), 400
+
+  accepts_json = request.headers.get('Accept') == 'application/json'
+
+  text = text.lower()
+  text = remove_accents(text)
+  for char in ['.', ',', '?', '!', ':', ';']:
+    text = text.replace(char, '')
+
+  for word, replacements in intents_data.get('replaces', {}).items():
+    for replacement in replacements:
+      if text == replacement:
+        text = word
+        break
+      elif f' {replacement} ' in text:
+        text = text.replace(f' {replacement} ', f' {word} ')
+      elif f'{replacement} ' in text:
+        text = text.replace(f'{replacement} ', f'{word} ')
+      elif text.endswith(replacement):
+        text = text[:len(text)-len(replacement)] + word
+
+  for entry in intents_data['intents']:
+    if text in entry['sentences']:
+      if accepts_json:
+        return jsonify({'intent': entry['action']})
+      return entry['action'] + '\n'
+  if accepts_json:
+    return jsonify({'intent': ''}), 404
+  return "", 404

api/intents.yaml

@@ -0,0 +1,94 @@
+replaces:
+  bluetooth:
+  - bluetooh
+  - el blog
+  - blog
+  - brutus
+  - bruto
+  - lotus
+  - amb l'atom
+  desconecta:
+  - desconec
+  - desco
+  - desconnect
+  - desconnecte
+
+intents:
+- action: BluetoothPair
+  sentences:
+  - pair bluetooth
+  - pair bluetooth device
+  - connect bluetooth
+  - connect bluetooth device
+  - connect to my phone
+  # -- 
+  - pareja bluetooth
+  - en pareja bluetooth
+  - empareja bluetooth
+  - empareja dispositivo bluetooth
+  - empareja mi movil
+  - conecta bluetooth
+  - conectate a mi movil
+  - conecta dispositivo bluetooth
+  - busca bluetooth
+  - buscar bluetooth
+  # --
+  - connectar bluetooth
+  - emparella bluetooth
+  - en parella bluetooth
+  - amb parella bluetooth
+- action: BluetoothDisconnect
+  sentences:
+  - disconnect
+  - disconnect bluetooth
+  # --
+  - desconecta
+  - desconectar
+  - desconectate
+  - desconecta bluetooth
+  - desconectar bluetooth
+  - desconecta mi movil
+  # --
+  - desconnecta
+  - desconnectar
+  - desconnectat
+- action: Pause
+  sentences:
+  - pause
+  - stop
+  - pausa
+  - pausar
+  - paus
+  - para
+  - para la musica
+- action: Next
+  sentences:
+  - next
+  - siguiente
+  - siguiente canción
+  - siguiente pista
+  - pasa a la siguiente
+- action: VolumeUp
+  sentences:
+  - volume up
+  - sube
+  - sube volumen
+  - subir
+  - subir volumen
+  - puja
+  - puja volum
+  - pujar
+  - pujar volum
+- action: VolumeDown
+  sentences:
+  - volume down
+  - baja
+  - baja volumen
+  - baja la voz
+  - bajar
+  - bajar volumen
+  - baix
+  - baixa
+  - baixa volum
+  - baixar
+  - baixar volum

api/main.py

@@ -12,10 +12,14 @@
 from utils import get_ip_address, get_wifi_mac_address, get_bt_mac_address, get_device_id, get_uptime, get_load_avg, get_memory_usage, get_volume, set_volume
 import const
 
+from intents import intents
+
 hostname = os.uname()[1]
 speaker_ip = get_ip_address('wlan0')
 app = Flask(__name__)
 
+app.register_blueprint(intents)
+
 config = ConfigManager(const.config_listener)
 config_tts = ConfigManager(const.config_tts)
 system_version = ConfigUci(const.mico_version)

api/requirements.txt

@@ -1,6 +1,6 @@
 Flask>=3
 Flask-APScheduler==1.13.1
-Flask-MQTT==1.2.1
+#Flask-MQTT==1.2.1
 
 requests>2,<3
 certifi>=2024
@@ -14,3 +14,5 @@ wyoming==1.6.0
 #numpy<2
 
 pyring-buffer
+
+pyyaml

packages/porcupine/config/launcher

@@ -23,6 +23,7 @@ RECORDING_CHANNEL=1
 VOLUME_THRESHOLD=10
 VOLUME_DURING_STT=10
 SOX_SILENCE_ARGS="1 0.2 1% 0.5 1.2 1%"
+API_AVAILABLE=$(/etc/init.d/api status >/dev/null && echo 1 || echo 0)
 
 CONFIG_FILE=/data/listener
 
@@ -141,6 +142,29 @@ get_stt_settings(){
   #echo "end $(date)"
 }
 
+intent_run(){
+  if [ "$1" = "BluetoothPair" ] ; then
+    if bluetoothctl list | grep -q .; then
+      /bin/bluetooth_pair &
+    fi
+  elif [ "$1" = "BluetoothDisconnect" ] ; then
+    BT_STATUS=$(timeout 1 bluetoothctl player.show | grep -i status | awk '{print $2}')
+    if [ -n "$BT_STATUS" ]; then
+      timeout 8 bluetoothctl disconnect &
+      miplay sound shutdown
+    fi
+  elif [ "$1" = "Pause" ] ; then
+    /bin/play_button
+  elif [ "$1" = "VolumeUp" ] ; then
+    # HACK! We are restoring volume after STT
+    #/bin/volume +10
+    SAVED_VOL=$((SAVED_VOL + 10))
+  elif [ "$1" = "VolumeDown" ] ; then
+    #/bin/volume -15
+    SAVED_VOL=$((SAVED_VOL - 15))
+  fi
+}
+
 which arecord &>/dev/null && {
   RECORD_COMMAND="arecord -N -D$MIC -d $TIME -f S16_LE -c $RECORDING_CHANNEL -r $RECORDING_RATE -"
 }
@@ -196,6 +220,8 @@ log "activated"
 ubus send wakeword '{"name": "'${WORD}'"}'
 
 ERROR_LISTENER=0
+INTENT=
+INTENT_SUCCESS=
 SAVED_VOL=`current_volume`
 # lower volume EXCEPT for notifications
 if [ "$SAVED_VOL" -gt ${VOLUME_THRESHOLD} ]; then
@@ -246,6 +272,17 @@ if [ "${STT_SUCCESS}" = 1 ]; then
     # NOTE: if empty text received, there may be some error with recording or other.
     miplay sound notice
   else
+    if [ "${API_AVAILABLE}" = "1" ]; then
+      # Attempt to check for local intents before asking HA
+      INTENT=$(curl -fs -XPOST -F "text=${STT_TEXT}" localhost/intent)
+      INTENT_SUCCESS=$?
+      if [ "$INTENT_SUCCESS" -eq 0 ] && [ -n "$INTENT" ]; then
+        echo "intent found: $INTENT"
+        log "intent: ${INTENT}"
+        intent_run $INTENT
+      fi
+    fi
+    if [ "${INTENT_SUCCESS}" != 0 ]; then
     CONVERSATION_RESPONSE=$(curl \
       -H "Authorization: Bearer ${HA_TOKEN}" \
       -H "Content-Type: application/json" \
@@ -258,6 +295,7 @@ if [ "${STT_SUCCESS}" = 1 ]; then
     fi
     TTS_TEXT=$(echo "${CONVERSATION_RESPONSE}" | jq -r .response.speech.plain.speech)
     ${SPEAK} "${TTS_TEXT}"
+    fi # INTENT_SUCCESS
   fi
 else
   log "error"