|
| 1 | +""" |
| 2 | +Basic example of scraping pipeline using SmartScraper |
| 3 | +""" |
| 4 | + |
| 5 | +from scrapegraphai.graphs import SmartScraperGraph |
| 6 | +from scrapegraphai.utils import prettify_exec_info |
| 7 | + |
| 8 | +# ************************************************ |
| 9 | +# Define the configuration for the graph |
| 10 | +# ************************************************ |
| 11 | +schema= """ |
| 12 | + { |
| 13 | + "Projects": [ |
| 14 | + "Project #": |
| 15 | + { |
| 16 | + "title": "...", |
| 17 | + "description": "...", |
| 18 | + }, |
| 19 | + "Project #": |
| 20 | + { |
| 21 | + "title": "...", |
| 22 | + "description": "...", |
| 23 | + } |
| 24 | + ] |
| 25 | + } |
| 26 | +""" |
| 27 | + |
| 28 | +# ************************************************ |
| 29 | +# Define the configuration for the graph |
| 30 | +# ********************************************* |
| 31 | + |
| 32 | +graph_config = { |
| 33 | + "llm": { |
| 34 | + "api_key": "***************************", |
| 35 | + "model": "oneapi/qwen-turbo", |
| 36 | + "base_url": "http://127.0.0.1:3000/v1", # 设置 OneAPI URL |
| 37 | + }, |
| 38 | + "embeddings": { |
| 39 | + "model": "ollama/nomic-embed-text", |
| 40 | + "base_url": "http://127.0.0.1:11434", # 设置 Ollama URL |
| 41 | + } |
| 42 | +} |
| 43 | + |
| 44 | +# ************************************************ |
| 45 | +# Create the SmartScraperGraph instance and run it |
| 46 | +# ************************************************ |
| 47 | + |
| 48 | +smart_scraper_graph = SmartScraperGraph( |
| 49 | + prompt="该网站为XXXXX,请提取出标题、发布时间、发布来源以及内容摘要,并以中文回答。", |
| 50 | + # 也可以使用已下载的 HTML 代码的字符串 |
| 51 | + source="http://XXXX", |
| 52 | + schema=schema, |
| 53 | + config=graph_config |
| 54 | +) |
| 55 | + |
| 56 | +# ************************************************ |
| 57 | +# Get graph execution info |
| 58 | +# ************************************************ |
| 59 | +result = smart_scraper_graph.run() |
| 60 | +print(result) |
| 61 | +print(prettify_exec_info(result)) |
0 commit comments