Depending on the nature of your .txt documents, you don't have to use one of the complicated regexes that attempt to validate email addresses. You're not trying to validate anything. You're just trying to grab what's already there. Generally speaking, a regex to grab what's already there can be much simpler than a regex that needs to validate input.
An important question is whether your .txt documents contain @ signs that are not part of an email address you want to extract.
This regex handles your first two requirements:
w+@[w.-]+|{(?:w+, *)+w+}@[w.-]+
Or if you want to allow any sequence of non-space characters containing an @ sign, plus your second requirement (which has spaces):
S+@S+|{(?:w+, *)+w+}@[w.-]+
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…