|
| 1 | +""" |
| 2 | +Module for testing the smart scraper class |
| 3 | +""" |
| 4 | + |
| 5 | +import os |
| 6 | +import pytest |
| 7 | +import pandas as pd |
| 8 | +from dotenv import load_dotenv |
| 9 | +from scrapegraphai.graphs import SmartScraperMultiParseConcatFirstGraph |
| 10 | +from scrapegraphai.utils import prettify_exec_info |
| 11 | + |
| 12 | +load_dotenv() |
| 13 | + |
| 14 | +@pytest.fixture |
| 15 | +def graph_config(): |
| 16 | + """Configuration of the graph""" |
| 17 | + openai_key = os.getenv("OPENAI_APIKEY") |
| 18 | + |
| 19 | + return { |
| 20 | + "llm": { |
| 21 | + "api_key": openai_key, |
| 22 | + "model": "openai/gpt-3.5-turbo", |
| 23 | + }, |
| 24 | + "verbose": True, |
| 25 | + "headless": False, |
| 26 | + } |
| 27 | + |
| 28 | +def test_scraping_pipeline(graph_config): |
| 29 | + """Start of the scraping pipeline""" |
| 30 | + smart_scraper_multi_parse_concat_first_graph = SmartScraperMultiParseConcatFirstGraph( |
| 31 | + prompt="Who is Marco Perini?", |
| 32 | + source= [ |
| 33 | + "https://perinim.github.io/", |
| 34 | + "https://perinim.github.io/cv/" |
| 35 | + ], |
| 36 | + config=graph_config, |
| 37 | + ) |
| 38 | + |
| 39 | + result = smart_scraper_multi_parse_concat_first_graph.run() |
| 40 | + |
| 41 | + assert result is not None |
| 42 | + assert isinstance(result, dict) |
| 43 | + |
| 44 | +def test_get_execution_info(graph_config): |
| 45 | + """Get the execution info""" |
| 46 | + smart_scraper_multi_parse_concat_first_graph = SmartScraperMultiParseConcatFirstGraph( |
| 47 | + prompt="Who is Marco Perini?", |
| 48 | + source= [ |
| 49 | + "https://perinim.github.io/", |
| 50 | + "https://perinim.github.io/cv/" |
| 51 | + ], |
| 52 | + config=graph_config, |
| 53 | + ) |
| 54 | + |
| 55 | + smart_scraper_multi_parse_concat_first_graph.run() |
| 56 | + |
| 57 | + graph_exec_info = smart_scraper_multi_parse_concat_first_graph.get_execution_info() |
| 58 | + |
| 59 | + assert graph_exec_info is not None |
0 commit comments