(A) is a typical 直接話法 (direct speech) sentence, and (B) is a typical 間接話法 (indirect speech) sentence. These are common both in English and Japanese, so they should be easy.
(C) is the same as (A) except that it lacks the quotation marks. This is still perfectly valid and common in Japanese, because the quotation marks are optional even when direct speech is used. That is, (B) and (C) are apparently different, but have the same meaning! (as long as we know the context, of course)
This rule might be too confusing in English, but luckily Japanese has a lot of personal pronouns. We know 俺 almost always refers to a man and あなた is usually said by a woman. Knowing that, (B) and (C) are not really confusing.
However, (D) is funny. If there are brackets, it explicitly means direct speech is used (i.e., the content is her own words). (D) looks as if the woman said 俺.
These are all valid and natural, too. The word order is changed for some rhetorical effect (known as 転置法/hyperbaton), but they're basically the same as (A)-(C).
Now, what happens if we even omit the quotative particle と at the end?
(B3) 彼女は言った。笑っている俺を見ている方がいい。 ←your question
If I'm not mistaken, these are somewhat closer to free indirect speech and free direct speech. I would say these are slightly harder to understand because there is no と, but are still completely valid as literary, rhetoric, and/or poetic expressions. We can understand what is the quoted part even without と. Depending on the surrounding sentences, these may be too difficult to understand, but that's another story.
(A3)-(C3) look far more dramatic than plain (A)-(C). Among the three, (C3) is probably the most common and relatively easiest to understand. (B3) is related to free indirect speech and probably is the most tricky, but it looks very dramatic to me at the same time.