migration

.envdefault

@@ -1,4 +1,13 @@
+# APPLICATION PARAMETERS
 APP_LANG=fr en
 ASSETS_PATH_ON_HOST=./assets
 ASSETS_PATH_IN_CONTAINER=/app/assets
-WORKER_NUMBER=1
+
+# SERVING PARAMETERS
+SERVICE_MODE=http
+CONCURRENCY=1
+
+# MICRO-SERVICE PARAMETERS
+SERVICE_NAME=kpe
+SERVICES_BROKER=redis://172.17.0.1:6379
+BROKER_PASS=

.github/workflows/dockerhub-description.yml

@@ -0,0 +1,20 @@
+name: Update Docker Hub Description
+on:
+  push:
+    branches:
+      - master
+    paths:
+      - README.md
+      - .github/workflows/dockerhub-description.yml
+jobs:
+  dockerHubDescription:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v3
+    - name: Docker Hub Description
+      uses: peter-evans/dockerhub-description@v3
+      with:
+        username: ${{ secrets.DOCKERHUB_USERNAME }}
+        password: ${{ secrets.DOCKERHUB_PASSWORD }}
+        repository: lintoai/linto-platform-nlp-keyphrase-extraction
+        readme-filepath: ./README.md

Dockerfile

@@ -1,16 +1,20 @@
 FROM lintoai/linto-platform-nlp-core:latest
 LABEL maintainer="[email protected]"
 
-WORKDIR /app
+WORKDIR /usr/src/app
 
-COPY ./requirements.txt /app/
+COPY requirements.txt ./
 RUN pip install --no-cache-dir -r requirements.txt
 
-COPY .envdefault /app/
-COPY ./scripts /app/scripts
-COPY ./components /app/components
+COPY kpe /usr/src/app/kpe
+COPY components /usr/src/app/components
+COPY celery_app /usr/src/app/celery_app
+COPY http_server /usr/src/app/http_server
+COPY document /usr/src/app/document
+COPY docker-entrypoint.sh wait-for-it.sh healthcheck.sh ./
 
-HEALTHCHECK --interval=15s CMD curl -fs http://0.0.0.0/health || exit 1
+ENV PYTHONPATH="${PYTHONPATH}:/usr/src/app/kpe"
 
-ENTRYPOINT ["/opt/conda/bin/gunicorn", "scripts.main:app", "--worker-class", "uvicorn.workers.UvicornWorker", "--bind", "0.0.0.0:80", "--access-logfile", "-", "--error-logfile", "-"]
-CMD ["--workers", "1"]
+HEALTHCHECK CMD ./healthcheck.sh
+
+ENTRYPOINT ["./docker-entrypoint.sh"]

RELEASE.md

@@ -1,3 +1,6 @@
+# 0.2.0
+- Migration to the [template]((https://github.com/linto-ai/linto-template-microservice)) of LinTO microservices.
+
 # 0.1.0
 - Initial commit.
 - Keyphrase Extraction.

celery_app/celeryapp.py

@@ -0,0 +1,28 @@
+import os
+from celery import Celery
+
+from kpe import logger
+
+celery = Celery(__name__, include=['celery_app.tasks'])
+service_name = os.environ.get("SERVICE_NAME", "kpe")
+broker_url = os.environ.get("SERVICES_BROKER")
+if os.environ.get("BROKER_PASS", False):
+    components = broker_url.split('//')
+    broker_url = f'{components[0]}//:{os.environ.get("BROKER_PASS")}@{components[1]}'
+celery.conf.broker_url = "{}/0".format(broker_url)
+celery.conf.result_backend = "{}/1".format(broker_url)
+celery.conf.update(
+    result_expires=3600,
+    task_acks_late=True,
+    task_track_started = True)
+
+# Queues
+celery.conf.update(
+    {'task_routes': {
+        'kpe_task' : {'queue': 'kpe'},}
+    }
+)
+
+logger.info(
+    f"Celery configured for broker located at {broker_url} with service name {service_name}"
+)

celery_app/tasks.py

@@ -0,0 +1,34 @@
+import spacy
+import components
+
+from typing import Dict, List
+
+from celery_app.celeryapp import celery
+
+from kpe import logger
+from kpe.processing import LM_MAP, MODELS, get_model
+from kpe.processing.utils import get_data
+
+
+@celery.task(name="kpe_task")
+def kpe_task(lang: str, texts: List[str], component_cfg: Dict = {}):
+    """Process a batch of articles and return the Keyphrases predicted by the
+    given model. Each record in the data should have a key "text".
+    """
+
+    # Check language availability
+    if lang in LM_MAP.keys():
+        model_name = LM_MAP[lang]
+        if model_name not in MODELS.keys():
+            raise RuntimeError(f"Model {model_name} for language {lang} is not loaded.")
+        nlp = spacy.blank(lang)
+        nlp.add_pipe("kpe", config={"model": {"@misc": "get_model", "name": model_name}})
+    else:
+        raise ValueError(f"Language {lang} is not supported.")
+
+    response_body = []
+
+    for doc in nlp.pipe(texts, component_cfg=component_cfg):
+        response_body.append(get_data(doc))
+
+    return {"kpe": response_body}

docker-compose.yml

@@ -9,5 +9,4 @@ services:
       - .env
     volumes:
       - $ASSETS_PATH_ON_HOST:$ASSETS_PATH_IN_CONTAINER:ro
-    command: ["--workers", $WORKER_NUMBER]
-    runtime: nvidia
+    #runtime: nvidia

docker-entrypoint.sh

@@ -0,0 +1,32 @@
+#!/bin/bash
+set -ea
+
+echo "RUNNING SERVICE"
+
+# Launch parameters, environement variables and dependencies check
+if [ -z "$SERVICE_MODE" ]
+then
+    echo "ERROR: Must specify a serving mode: [ http | task ]"
+    exit -1
+else
+    if [ "$SERVICE_MODE" = "http" ] 
+    then
+        echo "RUNNING HTTP SERVER"
+        python http_server/ingress.py --debug
+    elif [ "$SERVICE_MODE" == "task" ]
+    then
+        if [[ -z "$SERVICES_BROKER" ]]
+        then 
+            echo "ERROR: SERVICES_BROKER variable not specified, cannot start celery worker."
+            return -1
+        fi
+        /usr/src/app/wait-for-it.sh $(echo $SERVICES_BROKER | cut -d'/' -f 3) --timeout=20 --strict -- echo " $SERVICES_BROKER (Service Broker) is up"
+        echo "RUNNING CELERY WORKER"
+        celery --app=celery_app.celeryapp worker -Ofair -n nlp_${SERVICE_NAME}_worker@%h --queues=${SERVICE_NAME} -c ${CONCURRENCY}
+    else
+        echo "ERROR: Wrong serving command: $1"
+        exit -1
+    fi
+fi
+
+echo "Service stopped"

document/swagger.yml

@@ -0,0 +1,83 @@
+openapi: 3.0.1
+info:
+  title: Keyphrase Extraction API
+  description: API to detect keyphrases in text.
+  version: 0.2.0
+
+servers:
+- url: /
+
+paths:
+  /kpe/{lang}:
+    post:
+      tags:
+      - Keyphrase Extraction API
+      summary: Perform Keyphrase Extraction
+      parameters:
+        - name: lang
+          in: path
+          required: true
+          description: Language
+          schema:
+            type: string
+      requestBody:
+        content:
+          application/json:
+            schema:
+              $ref: '#/components/schemas/request'
+      responses:
+        200:
+          description: "Job successfully finished"
+          content:
+            application/json:
+              schema:
+                $ref: '#/components/schemas/responsemodel'
+        400:
+          description: "Bad request"
+        500:
+          description: "Server error"
+
+components:
+  schemas:
+    article:
+      type: object
+      properties:
+        text:
+          type: string
+          default: This is an article.
+    request:
+      type: object
+      properties:
+        articles:
+          type: array
+          required: true
+          items:
+            $ref: '#/components/schemas/article'
+        component_cfg:
+          type: object
+
+    keyphrase:
+      type: object
+      properties:
+        text:
+          type: string
+        end:
+          score: float
+    batch:
+      type: object
+      properties:
+        text:
+          type: string
+        keyphrases:
+          type: array
+          items:
+            $ref: '#/components/schemas/keyphrase'
+
+    responsemodel:
+      type: object
+      properties:
+        ner:
+          type: array
+          items:
+            $ref: '#/components/schemas/batch'
+

healthcheck.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+
+set -eax
+
+if [ "$SERVICE_MODE" = "http" ]
+then
+    curl --fail http://localhost:80/healthcheck || exit 1
+else
+    celery --app=celery_app.celeryapp inspect ping -d ${SERVICE_NAME}_worker@$HOSTNAME || exit 1
+fi

http_server/confparser.py

@@ -0,0 +1,51 @@
+import os
+import argparse
+
+__all__ = ["createParser"]
+
+def createParser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser()
+
+    # SERVICE
+    parser.add_argument(
+        '--service_name',
+        type=str,
+        help='Service Name',
+        default=os.environ.get('SERVICE_NAME', 'kpe'))
+
+    #GUNICORN    
+    parser.add_argument(
+        '--service_port',
+        type=int,
+        help='Service port',
+        default=80)
+    parser.add_argument(
+        '--workers',
+        type=int,
+        help="Number of Gunicorn workers (default=CONCURRENCY + 1)",
+        default=int(os.environ.get('CONCURRENCY', 1)) + 1)
+
+    #SWAGGER
+    parser.add_argument(
+        '--swagger_url',
+        type=str,
+        help='Swagger interface url',
+        default='/docs')
+    parser.add_argument(
+        '--swagger_prefix',
+        type=str,
+        help='Swagger prefix',
+        default=os.environ.get('SWAGGER_PREFIX', ''))
+    parser.add_argument(
+        '--swagger_path',
+        type=str,
+        help='Swagger file path',
+        default=os.environ.get('SWAGGER_PATH', '/usr/src/app/document/swagger.yml'))
+
+    #MISC
+    parser.add_argument(
+        '--debug',
+        action='store_true',
+        help='Display debug logs')
+
+    return parser