Merge pull request #554 from deeppavlov/dev

Release v1.10.0

.env

@@ -32,4 +32,5 @@ INFILLING_SERVICE_URL=http://infilling:8106/respond
 DIALOGPT_CONTINUE_SERVICE_URL=http://dialogpt:8125/continue
 PROMPT_STORYGPT_SERVICE_URL=http://prompt-storygpt:8127/respond
 STORYGPT_SERVICE_URL=http://storygpt:8126/respond
-FILE_SERVER_URL=http://files:3000
+FILE_SERVER_URL=http://files:3000
+SUMMARIZATION_SERVICE_URL=http://dialog-summarizer:8059/respond_batch

.env_ru

@@ -21,3 +21,4 @@ BADLIST_ANNOTATOR_URL=http://badlisted-words-ru:8018/badlisted_words_batch
 DP_WIKIDATA_URL=http://wiki-parser-ru:8077/model
 DP_ENTITY_LINKING_URL=http://entity-linking-ru:8075/model
 FILE_SERVER_URL=http://files:3000
+SUMMARIZATION_SERVICE_URL=http://rut5-summarizer:8060/respond_batch

annotators/summarization_annotator/Dockerfile

@@ -0,0 +1,9 @@
+FROM python:3.7.4
+
+ARG SUMMARIZATION_REQUEST_TIMEOUT
+ENV SUMMARIZATION_REQUEST_TIMEOUT ${SUMMARIZATION_REQUEST_TIMEOUT}
+
+COPY ${WORK_DIR}/requirements.txt /src/requirements.txt
+RUN pip install -r /src/requirements.txt
+COPY ${WORK_DIR} /src
+WORKDIR /src

annotators/summarization_annotator/requirements.txt

@@ -0,0 +1,7 @@
+sentry-sdk[flask]==0.14.1
+flask==1.1.1
+itsdangerous==2.0.1
+gunicorn==19.9.0
+requests==2.22.0
+jinja2<=3.0.3
+Werkzeug<=2.0.3

annotators/summarization_annotator/server.py

@@ -0,0 +1,75 @@
+import logging
+import time
+from os import getenv
+
+import sentry_sdk
+import requests
+from flask import Flask, jsonify, request
+
+
+sentry_sdk.init(getenv("SENTRY_DSN"))
+logging.basicConfig(format="%(asctime)s - %(name)s - %(levelname)s - %(message)s", level=logging.INFO)
+logger = logging.getLogger(__name__)
+app = Flask(__name__)
+
+SUMMARIZATION_REQUEST_TIMEOUT = int(getenv("SUMMARIZATION_REQUEST_TIMEOUT"))
+SUMMARIZATION_SERVICE_URL = getenv("SUMMARIZATION_SERVICE_URL")
+logger.info(f"summarization-annotator considered summarizer: {SUMMARIZATION_SERVICE_URL}")
+
+
+def get_summary(dialog):
+    summary = ""
+    if len(dialog) != 11:
+        logger.debug(
+            f"summarization-annotator is not ready to summarize dialog as the length of unsummarized dialog is "
+            f"{len(dialog)} != 11"
+        )
+        return summary
+
+    logger.debug("summarization-annotator is ready to summarize dialog as the length of unsummarized dialog is 11")
+    dialog = dialog[:6]
+    for i in range(len(dialog)):
+        if i % 2 == 0:
+            dialog[i] = "User: " + dialog[i]
+        else:
+            dialog[i] = "Bot: " + dialog[i]
+    dialog = ["\n".join(dialog)]
+    logger.debug(f"summarization-annotator will summarize this: {dialog}")
+
+    try:
+        summary = requests.post(
+            SUMMARIZATION_SERVICE_URL,
+            json={"sentences": dialog},
+            timeout=SUMMARIZATION_REQUEST_TIMEOUT,
+        ).json()[0]["batch"][0]
+    except Exception as exc:
+        logger.exception(exc)
+        sentry_sdk.capture_exception(exc)
+
+    return summary
+
+
+@app.route("/respond", methods=["POST"])
+def respond():
+    start_time = time.time()
+    dialogs_batch = request.json.get("dialogs", [])
+    summaries_batch = request.json.get("previous_summaries", [])
+    summarization_attribute = []
+
+    for dialog, prev_summary in zip(dialogs_batch, summaries_batch):
+        logger.debug(f"summarization-annotator received dialog: {dialog}")
+        logger.debug(f"summarization-annotator received previous summary: {[prev_summary]}")
+        result = prev_summary
+        new_summary = get_summary(dialog)
+        if new_summary:
+            result = f"{result} {new_summary}".strip()
+        summarization_attribute.append({"bot_attributes": {"summarized_dialog": result}})
+        logger.info(f"summarization-annotator output: {summarization_attribute}")
+
+    total_time = time.time() - start_time
+    logger.info(f"summarization-annotator exec time: {total_time:.2f}s")
+    return jsonify(summarization_attribute)
+
+
+if __name__ == "__main__":
+    app.run(debug=False, host="0.0.0.0", port=8058)

annotators/summarization_annotator/service_configs/summarization-annotator/environment.yml

@@ -0,0 +1,4 @@
+SERVICE_PORT: 8058
+SERVICE_NAME: summarization_annotator
+SUMMARIZATION_REQUEST_TIMEOUT: 10
+FLASK_APP: server

annotators/summarization_annotator/service_configs/summarization-annotator/service.yml

@@ -0,0 +1,26 @@
+name: summarization-annotator
+endpoints:
+- respond
+compose:
+  env_file:
+  - .env
+  build:
+    args:
+      SERVICE_PORT: 8058
+      SERVICE_NAME: summarization_annotator
+      SUMMARIZATION_REQUEST_TIMEOUT: 10
+    context: ./annotators/summarization_annotator/
+  command: flask run -h 0.0.0.0 -p 8058
+  environment:
+  - FLASK_APP=server
+  deploy:
+    resources:
+      limits:
+        memory: 256M
+      reservations:
+        memory: 256M
+  volumes:
+  - ./annotators/summarization_annotator:/src
+  ports:
+  - 8058:8058
+proxy: null

annotators/summarization_annotator/test.py

@@ -0,0 +1,70 @@
+import requests
+from os import getenv
+
+
+SUMMARIZATION_SERVICE_URL = getenv("SUMMARIZATION_SERVICE_URL")
+
+
+def test_skill():
+    url = "http://0.0.0.0:8058/respond"
+
+    if SUMMARIZATION_SERVICE_URL == "http://dialog-summarizer:8059/respond_batch":
+        input_data = {
+            "dialogs": [
+                [
+                    "Hi, my name is Mark!",
+                    "Good morning, Mark! How can I assist you today?",
+                    "Let's talk about cooking.",
+                    "Sure! What is your favourite type of cuisine to cook or experiment with in the " "kitchen?",
+                    "I like a wide range of cooking styles, such as Italian, Chinese, French and many " "more.",
+                    "May I recommend you any Italian dish?",
+                    "No. Better tell me what do you have in mind?",
+                    "I've recently found a couple easy and healthy meals. How about cooking quinoa with "
+                    "turkey and broccoli?",
+                    "That sounds like a healthy and tasty meal! Quinoa is a great source of protein, and "
+                    "when paired with lean turkey and broccoli, it's a well-rounded and balanced meal.",
+                    "I am glad for you! I listened to my favorite music all day. "
+                    "Such a great thing you know! Has anything extraordinary happened today?",
+                    "I can tell you more about what made your day great or we can just chat?" "I'm happy to listen!",
+                ]
+            ],
+            "previous_summaries": [""],
+        }
+
+        desired_output = [
+            "Bot wants to know what is Mark's favorite type of cuisine to cook. Mark likes Italian, "
+            "Chinese, French and many other cooking styles."
+        ]
+    else:
+        input_data = {
+            "dialogs": [
+                [
+                    "Привет! У тебя есть хобби?",
+                    "Мое хобби — кулинария.",
+                    "Здорово! А ты любишь готовить?",
+                    "Ага, я могу отлично приготовить разные блюда.",
+                    "Ты собираешь кулинарные рецепты?",
+                    "Да, уже есть большая коллекция.",
+                    "А какая национальная кухня тебе нравится?",
+                    "Конечно, русская.",
+                    "Русские блюда очень оригинальные, вкусные и полезные.",
+                    "А что ты любишь готовить больше всего?",
+                    "Я люблю готовить мясные блюда. Так что приглашаю в гости!",
+                ]
+            ],
+            "previous_summaries": [""],
+        }
+
+        desired_output = [
+            "У тебя есть хобби — кулинария, а у тебя есть большая коллекция кулинарных рецептов. Bot: Я "
+            "собираю кулинарные рецепты, собираю кулинарные рецепты, собираю кулинарные рецепты."
+        ]
+
+    result = requests.post(url, json=input_data).json()
+
+    assert result == [{"bot_attributes": {"summarized_dialog": desired_output[0]}}]
+    print("SUCCESS!")
+
+
+if __name__ == "__main__":
+    test_skill()

annotators/summarization_annotator/test.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+
+python test.py

assistant_dists/dream/dev.yml

@@ -141,6 +141,16 @@ services:
       - "./common:/src/common"
     ports:
       - 8170:8170
+  summarization-annotator:
+    volumes:
+      - "./annotators/summarization_annotator:/src"
+    ports:
+      - 8058:8058
+  dialog-summarizer:
+    volumes:
+      - "./services/dialog_summarizer:/src"
+    ports:
+      - 8059:8059
   openai-api-chatgpt-16k:
     volumes:
       - "./services/openai_api_lm:/src"

assistant_dists/dream/docker-compose.override.yml

@@ -8,7 +8,7 @@ services:
           combined-classification:8087, fact-retrieval:8100, entity-detection:8103,
           sentence-ranker:8128, property-extraction:8136, prompt-selector:8135, openai-api-chatgpt:8145,
           dff-dream-persona-chatgpt-prompted-skill:8137, dff-dream-faq-prompted-skill:8170,
-          openai-api-chatgpt-16k:8167"
+          openai-api-chatgpt-16k:8167, summarization-annotator:8058, dialog-summarizer:8059"
       WAIT_HOSTS_TIMEOUT: ${WAIT_TIMEOUT:-1000}
       HIGH_PRIORITY_INTENTS: 1
       RESTRICTION_FOR_SENSITIVE_CASE: 1
@@ -445,4 +445,41 @@ services:
         reservations:
           memory: 100M
 
+  summarization-annotator:
+    env_file: [ .env ]
+    build:
+      args:
+        SERVICE_PORT: 8058
+        SERVICE_NAME: summarization_annotator
+        SUMMARIZATION_REQUEST_TIMEOUT: 10
+      context: ./annotators/summarization_annotator/
+    command: flask run -h 0.0.0.0 -p 8058
+    environment:
+      - FLASK_APP=server
+    deploy:
+      resources:
+        limits:
+          memory: 256M
+        reservations:
+          memory: 256M
+
+  dialog-summarizer:
+    env_file: [ .env ]
+    build:
+      args:
+        SERVICE_PORT: 8059
+        SERVICE_NAME: dialog_summarizer
+        PRETRAINED_MODEL_NAME: "knkarthick/MEETING_SUMMARY"
+      context: ./services/dialog_summarizer/
+    command: flask run -h 0.0.0.0 -p 8059
+    environment:
+      - CUDA_VISIBLE_DEVICES=0
+      - FLASK_APP=server
+    deploy:
+      resources:
+        limits:
+          memory: 4G
+        reservations:
+          memory: 4G
+
 version: '3.7'

assistant_dists/dream/pipeline_conf.json

@@ -263,6 +263,24 @@
                     "component": "components/PbLNvh4hrvs47rPaf2bfYQ.yml",
                     "service": "annotators/combined_classification/service_configs/combined-classification"
                 }
+            },
+            "summarization_annotator": {
+                "connector": {
+                    "protocol": "http",
+                    "timeout": 10.0,
+                    "url": "http://summarization-annotator:8058/respond"
+                },
+                "dialog_formatter": "state_formatters.dp_formatters:summarization_annotator_formatter",
+                "response_formatter": "state_formatters.dp_formatters:simple_formatter_service",
+                "previous_services": [
+                    "annotators.spelling_preprocessing"
+                ],
+                "state_manager_method": "update_attributes",
+                "is_enabled": true,
+                "source": {
+                    "component": "components/riRfdGz86P51B9bL7fO6JR.yml",
+                    "service": "annotators/summarization_annotator/service_configs/summarization-annotator"
+                }
             }
         },
         "response_annotators": {

assistant_dists/dream_alexa/docker-compose.override.yml

@@ -24,6 +24,10 @@ services:
       HIGH_PRIORITY_INTENTS: 1
       RESTRICTION_FOR_SENSITIVE_CASE: 1
       ALWAYS_TURN_ON_ALL_SKILLS: 0
+      ENABLE_NP_QUESTIONS: 1
+      ENABLE_SWITCH_TOPIC: 1
+      ENABLE_LINK_QUESTIONS: 1
+      ENABLE_NP_FACTS: 1
       LANGUAGE: EN
       FALLBACK_FILE: fallbacks_dream_en.json
 

assistant_dists/dream_reasoning/docker-compose.override.yml

@@ -118,9 +118,10 @@ services:
         GENERATIVE_TIMEOUT: 120
         N_UTTERANCES_CONTEXT: 1
         ENVVARS_TO_SEND: OPENAI_API_KEY,GOOGLE_CSE_ID,GOOGLE_API_KEY,OPENWEATHERMAP_API_KEY,NEWS_API_KEY,WOLFRAMALPHA_APP_ID
+        TIME_SLEEP: 5
       context: .
       dockerfile: ./skills/dff_reasoning_skill/Dockerfile
-    command: gunicorn --workers=1 server:app -b 0.0.0.0:8169 --reload
+    command: gunicorn --workers=1 server:app -b 0.0.0.0:8169 --timeout 600
     deploy:
       resources:
         limits:

assistant_dists/dream_robot_prompted/cpu.yml

@@ -0,0 +1,10 @@
+version: '3.7'
+services:
+  combined-classification:
+    environment:
+      DEVICE: cpu
+      CUDA_VISIBLE_DEVICES: ""
+  sentence-ranker:
+    environment:
+      DEVICE: cpu
+      CUDA_VISIBLE_DEVICES: ""

assistant_dists/dream_robot_prompted/db_conf.json

@@ -0,0 +1,6 @@
+{
+    "host": "DB_HOST",
+    "port": "DB_PORT",
+    "name": "DB_NAME",
+    "env": true
+}

assistant_dists/dream_robot_prompted/dev.yml

@@ -0,0 +1,50 @@
+# С такими volumes удобно дебажить, не нужно пересобирать контейнер каждый раз при изменении кода
+services:
+  agent:
+    volumes:
+      - ".:/dp-agent"
+    ports:
+      - 4242:4242
+  sentseg:
+    volumes:
+      - "./annotators/SentSeg:/src"
+    ports:
+      - 8011:8011
+  ranking-based-response-selector:
+    volumes:
+      - "./response_selectors/ranking_based_response_selector:/src"
+      - "./common:/src/common"
+    ports:
+      - 8002:8002
+  combined-classification:
+    volumes:
+      - "./common:/src/common"
+      - "./annotators/combined_classification:/src"
+    ports:
+      - 8087:8087
+  sentence-ranker:
+    volumes:
+      - "./services/sentence_ranker:/src"
+      - "~/.deeppavlov/cache:/root/.cache"
+    ports:
+      - 8128:8128
+  prompt-selector:
+    volumes:
+      - "./annotators/prompt_selector:/src"
+      - "./common:/src/common"
+    ports:
+      - 8135:8135
+  openai-api-chatgpt:
+    volumes:
+      - "./services/openai_api_lm:/src"
+      - "./common:/src/common"
+    ports:
+      - 8145:8145
+  dff-robot-prompted-skill:
+    volumes:
+      - "./skills/dff_template_prompted_skill:/src"
+      - "./common:/src/common"
+    ports:
+      - 8179:8179
+
+version: "3.7"