distribusi-verse/verse/search/search_index.py

import os
from whoosh.fields import *
from whoosh.index import create_in, open_dir
from whoosh.qparser import QueryParser
from models.distribusi_model import Distribusis
from models.distribusi_file_model import DistribusiFiles
import flask_apscheduler


SCRIPT_DIR = os.path.dirname(__file__)
SEARCH_DATA_DIR = os.path.abspath(os.path.join(SCRIPT_DIR, "searchdata"))


def init_search_index(APP):
    scheduler = flask_apscheduler.APScheduler()
    scheduler.api_enabled = False
    scheduler.init_app(APP)
    scheduler.start()

    schema = Schema(
        title=TEXT(stored=True), path=ID(stored=True), content=TEXT
    )
    ix = create_in(SEARCH_DATA_DIR, schema)
    writer = ix.writer()
    index_distribusis(APP, writer)
    index_distribusi_files(APP, writer)
    writer.commit(optimize=True)

    @scheduler.task("interval", id="update", minutes=60)
    def update_search_index():
        ix = open_dir(SEARCH_DATA_DIR)
        update_writer = ix.writer()
        index_distribusis(APP, update_writer)
        index_distribusi_files(APP, update_writer)
        update_writer.commit(optimize=True)


def index_distribusis(APP, writer):
    distribusis = _visible_distribusis(APP)
    for distribusi in distribusis:
        writer.add_document(
            title=distribusi.distribusiname,
            path="/a",
            content=distribusi.description,
        )


def index_distribusi_files(APP, writer):
    with APP.app_context():
        for distribusi_file in DistribusiFiles.query.all():
            APP.logger.info(
                f"adding distribusi file {distribusi_file.path} to search index"
            )
            writer.add_document(
                title=distribusi_file.path,
                path="/b",
                content=distribusi_file.description,
            )


def _visible_distribusis(APP):
    with APP.app_context():
        distribusis = Distribusis.query.filter(
            Distribusis.visible.isnot(False)
        ).all()
        return distribusis