[voicerss] Add LRU cache

Signed-off-by: Laurent Garnier <[email protected]>

bundles/org.openhab.voice.voicerss/src/main/java/org/openhab/voice/voicerss/internal/VoiceRSSAudioStream.java

@@ -12,26 +12,58 @@
  */
 package org.openhab.voice.voicerss.internal;
 
-import java.io.File;
+import java.io.IOException;
+import java.io.InputStream;
 
 import org.eclipse.jdt.annotation.NonNullByDefault;
 import org.openhab.core.audio.AudioException;
 import org.openhab.core.audio.AudioFormat;
 import org.openhab.core.audio.AudioStream;
-import org.openhab.core.audio.FileAudioStream;
+import org.openhab.core.audio.FixedLengthAudioStream;
 
 /**
  * Implementation of the {@link AudioStream} interface for the
- * {@link VoiceRSSTTSService}. It simply uses a {@link FileAudioStream} which is
+ * {@link VoiceRSSTTSService}. It simply uses a {@link FixedLengthAudioStream} which is
  * doing all the necessary work, e.g. supporting MP3 and WAV files with fixed
  * stream length.
  *
- * @author Jochen Hiller - Initial contribution and API
+ * @author Laurent Garnier - Initial contribution
  */
 @NonNullByDefault
-class VoiceRSSAudioStream extends FileAudioStream {
+class VoiceRSSAudioStream extends FixedLengthAudioStream {
 
-    public VoiceRSSAudioStream(File audioFile, AudioFormat format) throws AudioException {
-        super(audioFile, format);
+    private InputStream inputStream;
+    private long length;
+    private AudioFormat format;
+
+    public VoiceRSSAudioStream(InputStream inputStream, long length, AudioFormat format) {
+        this.inputStream = inputStream;
+        this.length = length;
+        this.format = format;
+    }
+
+    @Override
+    public long length() {
+        return length;
+    }
+
+    @Override
+    public AudioFormat getFormat() {
+        return format;
+    }
+
+    @Override
+    public InputStream getClonedStream() throws AudioException {
+        throw new AudioException("getClonedStream not supported");
+    }
+
+    @Override
+    public int read() throws IOException {
+        return inputStream.read();
+    }
+
+    @Override
+    public void close() throws IOException {
+        inputStream.close();
     }
 }

bundles/org.openhab.voice.voicerss/src/main/java/org/openhab/voice/voicerss/internal/VoiceRSSFileAudioStream.java

@@ -0,0 +1,37 @@
+/**
+ * Copyright (c) 2010-2023 Contributors to the openHAB project
+ *
+ * See the NOTICE file(s) distributed with this work for additional
+ * information.
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ */
+package org.openhab.voice.voicerss.internal;
+
+import java.io.File;
+
+import org.eclipse.jdt.annotation.NonNullByDefault;
+import org.openhab.core.audio.AudioException;
+import org.openhab.core.audio.AudioFormat;
+import org.openhab.core.audio.AudioStream;
+import org.openhab.core.audio.FileAudioStream;
+
+/**
+ * Implementation of the {@link AudioStream} interface for the
+ * {@link VoiceRSSTTSService}. It simply uses a {@link FileAudioStream} which is
+ * doing all the necessary work, e.g. supporting MP3 and WAV files with fixed
+ * stream length.
+ *
+ * @author Jochen Hiller - Initial contribution and API
+ */
+@NonNullByDefault
+class VoiceRSSFileAudioStream extends FileAudioStream {
+
+    public VoiceRSSFileAudioStream(File audioFile, AudioFormat format) throws AudioException {
+        super(audioFile, format);
+    }
+}

bundles/org.openhab.voice.voicerss/src/main/java/org/openhab/voice/voicerss/internal/VoiceRSSTTSService.java

@@ -27,14 +27,18 @@
 import org.openhab.core.audio.AudioFormat;
 import org.openhab.core.audio.AudioStream;
 import org.openhab.core.config.core.ConfigurableService;
+import org.openhab.core.voice.AbstractCachedTTSService;
+import org.openhab.core.voice.TTSCache;
 import org.openhab.core.voice.TTSException;
 import org.openhab.core.voice.TTSService;
 import org.openhab.core.voice.Voice;
 import org.openhab.voice.voicerss.internal.cloudapi.CachedVoiceRSSCloudImpl;
+import org.openhab.voice.voicerss.internal.cloudapi.SizedInputStream;
 import org.osgi.framework.Constants;
 import org.osgi.service.component.annotations.Activate;
 import org.osgi.service.component.annotations.Component;
 import org.osgi.service.component.annotations.Modified;
+import org.osgi.service.component.annotations.Reference;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -45,9 +49,10 @@
  * @author Laurent Garnier - add support for OGG and AAC audio formats
  */
 @NonNullByDefault
-@Component(configurationPid = "org.openhab.voicerss", property = Constants.SERVICE_PID + "=org.openhab.voicerss")
+@Component(service = TTSService.class, configurationPid = "org.openhab.voicerss", property = Constants.SERVICE_PID
+        + "=org.openhab.voicerss")
 @ConfigurableService(category = "voice", label = "VoiceRSS Text-to-Speech", description_uri = "voice:voicerss")
-public class VoiceRSSTTSService implements TTSService {
+public class VoiceRSSTTSService extends AbstractCachedTTSService {
 
     /** Cache folder name is below userdata/voicerss/cache. */
     private static final String CACHE_FOLDER_NAME = "voicerss" + File.separator + "cache";
@@ -87,6 +92,11 @@ public class VoiceRSSTTSService implements TTSService {
      */
     private @Nullable Set<AudioFormat> audioFormats;
 
+    @Activate
+    public VoiceRSSTTSService(final @Reference TTSCache ttsCache) {
+        super(ttsCache);
+    }
+
     /**
      * DS activate, with access to ConfigAdmin
      */
@@ -130,6 +140,43 @@ public AudioStream synthesize(String text, Voice voice, AudioFormat requestedFor
         if (voiceRssCloud == null) {
             throw new TTSException("The service is not correctly initialized");
         }
+        // trim text
+        String trimmedText = text.trim();
+        if (trimmedText.isEmpty()) {
+            throw new TTSException("The passed text is empty");
+        }
+        Set<Voice> localVoices = voices;
+        if (localVoices == null || !localVoices.contains(voice)) {
+            throw new TTSException("The passed voice is unsupported");
+        }
+
+        // If one predefined cache entry for given text, locale, voice, codec and format exists,
+        // create the input from this file stream and return it.
+        try {
+            File cacheAudioFile = voiceRssCloud.getTextToSpeechInCache(trimmedText, voice.getLocale().toLanguageTag(),
+                    voice.getLabel(), getApiAudioCodec(requestedFormat), getApiAudioFormat(requestedFormat));
+            if (cacheAudioFile != null) {
+                logger.debug("Use cache entry '{}'", cacheAudioFile.getName());
+                return new VoiceRSSFileAudioStream(cacheAudioFile, requestedFormat);
+            }
+        } catch (AudioException ex) {
+            throw new TTSException("Could not create AudioStream: " + ex.getMessage(), ex);
+        } catch (IOException ex) {
+            throw new TTSException("Could not read from VoiceRSS service: " + ex.getMessage(), ex);
+        }
+
+        // If no predefined cache entry exists, use the common TTS cache mechanism from core framework
+        logger.debug("Use common TTS cache mechanism");
+        return super.synthesize(text, voice, requestedFormat);
+    }
+
+    @Override
+    public AudioStream synthesizeForCache(String text, Voice voice, AudioFormat requestedFormat) throws TTSException {
+        logger.debug("synthesizeForCache '{}' for voice '{}' in format {}", text, voice.getUID(), requestedFormat);
+        CachedVoiceRSSCloudImpl voiceRssCloud = voiceRssImpl;
+        if (voiceRssCloud == null) {
+            throw new TTSException("The service is not correctly initialized");
+        }
         // Validate known api key
         String key = apiKey;
         if (key == null) {
@@ -145,14 +192,10 @@ public AudioStream synthesize(String text, Voice voice, AudioFormat requestedFor
             throw new TTSException("The passed voice is unsupported");
         }
 
-        // now create the input stream for given text, locale, voice, codec and format.
         try {
-            File cacheAudioFile = voiceRssCloud.getTextToSpeechAsFile(key, trimmedText,
-                    voice.getLocale().toLanguageTag(), voice.getLabel(), getApiAudioCodec(requestedFormat),
-                    getApiAudioFormat(requestedFormat));
-            return new VoiceRSSAudioStream(cacheAudioFile, requestedFormat);
-        } catch (AudioException ex) {
-            throw new TTSException("Could not create AudioStream: " + ex.getMessage(), ex);
+            SizedInputStream input = voiceRssCloud.getTextToSpeech(key, trimmedText, voice.getLocale().toLanguageTag(),
+                    voice.getLabel(), getApiAudioCodec(requestedFormat), getApiAudioFormat(requestedFormat));
+            return new VoiceRSSAudioStream(input.getStream(), input.getLength(), requestedFormat);
         } catch (IOException ex) {
             throw new TTSException("Could not read from VoiceRSS service: " + ex.getMessage(), ex);
         }

bundles/org.openhab.voice.voicerss/src/main/java/org/openhab/voice/voicerss/internal/cloudapi/CachedVoiceRSSCloudImpl.java

@@ -69,7 +69,7 @@ public File getTextToSpeechAsFile(String apiKey, String text, String locale, Str
         }
 
         // if not in cache, get audio data and put to cache
-        try (InputStream is = super.getTextToSpeech(apiKey, text, locale, voice, audioCodec, audioFormat);
+        try (InputStream is = super.getTextToSpeech(apiKey, text, locale, voice, audioCodec, audioFormat).getStream();
                 FileOutputStream fos = new FileOutputStream(audioFileInCache)) {
             copyStream(is, fos);
             // write text to file for transparency too
@@ -83,6 +83,16 @@ public File getTextToSpeechAsFile(String apiKey, String text, String locale, Str
         }
     }
 
+    public @Nullable File getTextToSpeechInCache(String text, String locale, String voice, String audioCodec,
+            String audioFormat) throws IOException {
+        String fileNameInCache = getUniqueFilenameForText(text, locale, voice, audioFormat);
+        if (fileNameInCache == null) {
+            throw new IOException("Could not infer cache file name");
+        }
+        File audioFileInCache = new File(cacheFolder, fileNameInCache + "." + audioCodec.toLowerCase());
+        return audioFileInCache.exists() ? audioFileInCache : null;
+    }
+
     /**
      * Gets a unique filename for a give text, by creating a MD5 hash of it. It
      * will be preceded by the locale and suffixed by the format if it is not the

bundles/org.openhab.voice.voicerss/src/main/java/org/openhab/voice/voicerss/internal/cloudapi/SizedInputStream.java

@@ -0,0 +1,42 @@
+/**
+ * Copyright (c) 2010-2023 Contributors to the openHAB project
+ *
+ * See the NOTICE file(s) distributed with this work for additional
+ * information.
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ */
+package org.openhab.voice.voicerss.internal.cloudapi;
+
+import java.io.InputStream;
+
+import org.eclipse.jdt.annotation.NonNullByDefault;
+
+/**
+ * Class to store an InputStream and its length
+ *
+ * @author Laurent Garnier - Initial contribution
+ */
+@NonNullByDefault
+public class SizedInputStream {
+
+    private InputStream stream;
+    private long length;
+
+    public SizedInputStream(InputStream stream, long length) {
+        this.stream = stream;
+        this.length = length;
+    }
+
+    public InputStream getStream() {
+        return stream;
+    }
+
+    public long getLength() {
+        return length;
+    }
+}

bundles/org.openhab.voice.voicerss/src/main/java/org/openhab/voice/voicerss/internal/cloudapi/VoiceRSSCloudAPI.java

@@ -13,7 +13,6 @@
 package org.openhab.voice.voicerss.internal.cloudapi;
 
 import java.io.IOException;
-import java.io.InputStream;
 import java.util.Locale;
 import java.util.Set;
 
@@ -74,11 +73,11 @@ public interface VoiceRSSCloudAPI {
      *            the audio codec to use
      * @param audioFormat
      *            the audio format to use
-     * @return an InputStream to the audio data in specified format
+     * @return a SizedInputStream to the audio data in specified format
      * @throws IOException
      *             will be raised if the audio data can not be retrieved from
      *             cloud service
      */
-    InputStream getTextToSpeech(String apiKey, String text, String locale, String voice, String audioCodec,
+    SizedInputStream getTextToSpeech(String apiKey, String text, String locale, String voice, String audioCodec,
             String audioFormat) throws IOException;
 }

bundles/org.openhab.voice.voicerss/src/main/java/org/openhab/voice/voicerss/internal/cloudapi/VoiceRSSCloudImpl.java

@@ -215,7 +215,7 @@ public Set<String> getAvailableVoices(Locale locale) {
      * dependencies.
      */
     @Override
-    public InputStream getTextToSpeech(String apiKey, String text, String locale, String voice, String audioCodec,
+    public SizedInputStream getTextToSpeech(String apiKey, String text, String locale, String voice, String audioCodec,
             String audioFormat) throws IOException {
         String url = createURL(apiKey, text, locale, voice, audioCodec, audioFormat);
         if (logging) {
@@ -243,9 +243,12 @@ public InputStream getTextToSpeech(String apiKey, String text, String locale, St
             }
         }
         String contentType = connection.getHeaderField("Content-Type");
+        long length = connection.getContentLengthLong();
         InputStream is = connection.getInputStream();
         // check if content type is text/plain, then we have an error
-        if (contentType.contains("text/plain")) {
+        if (length < 0) {
+            throw new IOException("Could not read audio content, missing header Content-Length");
+        } else if (contentType.contains("text/plain")) {
             byte[] bytes = new byte[256];
             is.read(bytes, 0, 256);
             // close before throwing an exception
@@ -259,7 +262,7 @@ public InputStream getTextToSpeech(String apiKey, String text, String locale, St
             throw new IOException(
                     "Could not read audio content, service returned an error: " + new String(bytes, "UTF-8"));
         } else {
-            return is;
+            return new SizedInputStream(is, length);
         }
     }
 

bundles/org.openhab.voice.voicerss/src/test/java/org/openhab/voice/voicerss/internal/VoiceRSSTTSServiceTest.java

@@ -18,12 +18,16 @@
 import static org.openhab.core.audio.AudioFormat.*;
 import static org.openhab.voice.voicerss.internal.CompatibleAudioFormatMatcher.compatibleAudioFormat;
 
+import java.util.HashMap;
+import java.util.Map;
 import java.util.Set;
 
 import org.junit.jupiter.api.BeforeEach;
 import org.junit.jupiter.api.Test;
 import org.openhab.core.audio.AudioFormat;
+import org.openhab.core.storage.StorageService;
 import org.openhab.core.voice.TTSService;
+import org.openhab.core.voice.internal.cache.TTSLRUCacheImpl;
 
 /**
  * Tests for {@link VoiceRSSTTSService}.
@@ -43,14 +47,19 @@ public class VoiceRSSTTSServiceTest {
     private static final AudioFormat WAV_48KHZ_16BIT = new AudioFormat(AudioFormat.CONTAINER_WAVE,
             AudioFormat.CODEC_PCM_SIGNED, false, 16, null, 48_000L);
 
+    private StorageService storageService;
+
     /**
      * The {@link VoiceRSSTTSService} under test.
      */
     private TTSService ttsService;
 
     @BeforeEach
     public void setUp() {
-        final VoiceRSSTTSService ttsService = new VoiceRSSTTSService();
+        Map<String, Object> config = new HashMap<>();
+        config.put("enableCacheTTS", false);
+        TTSLRUCacheImpl voiceLRUCache = new TTSLRUCacheImpl(storageService, config);
+        final VoiceRSSTTSService ttsService = new VoiceRSSTTSService(voiceLRUCache);
         ttsService.activate(null);
 
         this.ttsService = ttsService;