After the Enron debacle / scandal in the USA, a dataset of 600,000 emails of 158 employees was made public by the Federal Energy Regulatory Commission. It was later bought by MIT and processed, with some redactions and deletions of attachments. Versions of that dataset are still available at the US Library of Congress and at  https://www.cs.cmu.edu/~./enron/. 
Wikipedia has a good summary as well, at https://en.wikipedia.org/wiki/Enron_Corpus. 

Various subsets of this dataset can be found on the internet, including at Github, HuggingFace and Kaggle. An often-used subset was produced by researchers at the Greek Institute of Informatics and
Telecommunications, as described in their paper [Metsis]. The authors of this paper selected six Enron employees with large email volumes, and produced the dataset in order to analyze and test various kinds of spam filters, including several Naïve Bayes versions. 
My CSV file contains this particular dataset, containing 33,716 emails, of which 17,171 are spam. I included the Subject concatenated with Contents, as well as the original file names in a separate column.

[Metsis] Metsis, V., Androutsopoulos, I., & Paliouras. G., Spam filtering with naive bayes-which naive bayes? 
Proceedings of the 3rd Conference  on Email and Anti-Spam (CEAS 2006), Mountain View, CA, USA, 2006.