Hi michalzwierz,
You could use Get Fulltext processor for automatically grabbing the whole content from the source page. Next step, you should use Cut Introtext processor to truncate 300 words from the output po[] fulltext of Get Fulltext processor. Then with Original Source processor, you could add the link of the source at the bottom of that 300 words.
In case Get Fulltext processor can not detect automatically the whole content from the source, I think you should try with Parser Code feature of HTML Parser add-on ( https://thimpress.com/product/html-parser-processor-for-pipes/ ) .
Best Regards!