295646 - Trailing whitespace is lost

Status: RESOLVED FIXED

(Core :: DOM: HTML Parser, defect)

Description

[Blake Kaplan (:mrbkap) (inactive)]

Whitespace at the end of a document is chopped of because nobody bothers
checking the scanner's mIncremental member and not returning kEOF if it's not
incremental (as all of the other tokens do).

Patch in a second.

Comment 1

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: patch v1

This checks if we're really at the end of the document, and if we are, doesn't
discard the trailing whitespace.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 184634 [details] [diff] [review]
patch v1

is haveCR guaranteed to have a non-bogus value even if nothing was consumed?

Comment 3

[Blake Kaplan (:mrbkap) (inactive)]

By the time we get to CWhitespaceToken::Consume(), we know that there is at
least one whitespace char in the scanner. Therefore, the Peek() in
nsScanner::ReadWhitespace() will succeed, and aHaveCR is set appropriately. I'll
change ReadWhitespace() to make sure to initialize aHaveCR in the second
iteration of this patch.

Comment 4

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: patch v2

As it turns out, nsScanner::ReadWhitespace(), *does* initialize its out
parameter. This patch also fixes CNewlineToken to not lose trailing newlines.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 184700 [details] [diff] [review]
patch v2

r+sr=bzbarsky

Comment 6

[Blake Kaplan (:mrbkap) (inactive)]

Comment on attachment 184700 [details] [diff] [review]
patch v2

This patch fixes a couple of problems with composer's Insert -> HTML, and also
prevents the loss of trailing whitespace/newlines.

Comment 7

[Mike Shaver (:shaver emeritus)]

Comment on attachment 184700 [details] [diff] [review]
patch v2

a=shaver

Comment 8

[Blake Kaplan (:mrbkap) (inactive)]

Fix checked in.