From cdb3c1100ee1117afedbc70437317acaf7c7c1d3 Mon Sep 17 00:00:00 2001 From: roryhaung Date: Wed, 16 Oct 2024 20:05:03 +0800 Subject: [PATCH] test: Add scrape_graph test --- tests/graphs/scrape_graph_test.py | 50 +++++++++++++++++++++++++++++++ 1 file changed, 50 insertions(+) create mode 100644 tests/graphs/scrape_graph_test.py diff --git a/tests/graphs/scrape_graph_test.py b/tests/graphs/scrape_graph_test.py new file mode 100644 index 00000000..00d3f4fb --- /dev/null +++ b/tests/graphs/scrape_graph_test.py @@ -0,0 +1,50 @@ +""" +Module for testing the scrape graph class +""" + +import os +import pytest +import pandas as pd +from dotenv import load_dotenv +from scrapegraphai.graphs import ScrapeGraph +from scrapegraphai.utils import prettify_exec_info + +load_dotenv() + +@pytest.fixture +def graph_config(): + """Configuration of the graph""" + openai_key = os.getenv("OPENAI_APIKEY") + return { + "llm": { + "api_key": openai_key, + "model": "openai/gpt-3.5-turbo", + }, + "verbose": True, + "headless": False, + } + +def test_scraping_pipeline(graph_config): + """Start of the scraping pipeline""" + scrape_graph = ScrapeGraph( + source="https://perinim.github.io/projects/", + config=graph_config, + ) + + result = scrape_graph.run() + + assert result is not None + assert isinstance(result, list) + +def test_get_execution_info(graph_config): + """Get the execution info""" + scrape_graph = ScrapeGraph( + source="https://perinim.github.io/projects/", + config=graph_config, + ) + + scrape_graph.run() + + graph_exec_info = scrape_graph.get_execution_info() + + assert graph_exec_info is not None