feat: websearch, tool use, user intent, dynamic retrieval, multiple questions #1153

	name: Run Tests with Tika Server

	on:
	push:
	paths:
	- "core/**"
	pull_request:
	paths:
	- "core/**"
	workflow_dispatch:

	jobs:
	test:
	runs-on: ubuntu-latest

	services:
	tika:
	image: apache/tika
	ports:
	- 9998:9998

	steps:
	- name: 👀 Checkout code
	uses: actions/checkout@v2

	- name: 🔨 Install the latest version of rye
	uses: eifinger/setup-rye@v4
	with:
	enable-cache: true
	working-directory: backend
	- name: 🔄 Sync dependencies
	run: \|
	cd core
	UV_INDEX_STRATEGY=unsafe-first-match rye sync --no-lock

	- name: Run tests
	env:
	TIKA_URL: http://localhost:9998/tika
	OPENAI_API_KEY: this-is-a-test-key
	run: \|
	sudo apt-get update
	sudo apt-get install -y libmagic-dev poppler-utils libreoffice tesseract-ocr pandoc
	cd core
	rye run python -c "from unstructured.nlp.tokenize import download_nltk_packages; download_nltk_packages()"
	rye run python -c "import nltk;nltk.download('punkt_tab'); nltk.download('averaged_perceptron_tagger_eng')"
	rye test -p quivr-core

Provide feedback