Import is very time -consuming: from unnersted.partition.pdf import partition_pdf

[peanutpaste]

t1 = time.time() from unstructured.partition.pdf import partition_pdf t2 = time.time() print(t2-t1)

I run this, it takes nearly 1 minute

my env： cpu：13th Gen Intel(R) Core(TM) i7-13700K 3.40 GHz ram：32.0 GB nvidia 4070

Is this time -consuming situation normal? Or is it wrong?

[peanutpaste]

If I disconnect the network, I can import it at a normal time. What information will this be automatically loaded from the Internet?

[scanny]

Try setting the environment variable:

$ export SCARF_NO_ANALYTICS=true

and see if that makes a difference. On some network configurations the analytics appear to take longer than desired.

It's mentioned in the README here: https://github.com/Unstructured-IO/unstructured?tab=readme-ov-file#chart_with_upwards_trend-analytics

[Lilyzzzzzz]

It takes time more than 4 minute to connect network, suggest modifying the logic in scarf_analytics function, set SCARF_NO_ANALYTICS default value to false