from langchain_community.document_loaders import WebBaseLoader

MANUAL_URLS = [
    "https://www.tracwater.com",
    "https://www.tracwater.com/tracwater-sensors-tws",
    "https://www.tracwater.com/wirelessgroundwatersensors",
    "https://www.tracwater.com/wireless-transient-pressure-monitor",
    "https://www.tracwater.com/ijinus-wireless-pressure-and-flow-meters",
    "https://aquaanalytics.com.au/7-trends-for-smart-water-technology-in-australia-and-new-zealand/",
    "https://www.infoasaservice.com.au/home",
    "https://www.tracwater.com/ijinus-wireless-radar-level-sensor",
    "https://www.tracnet.com.au",
    "https://www.tracwater.com/wireless-underground-sewer-monitor"
]

loader = WebBaseLoader(MANUAL_URLS)
documents = loader.load()

print(len(documents))
print(documents[0].page_content[:500])
