HTML email's text representation includes too many linebreaks

[foobear]

MailFinder.email_text_body deduplicates multiple \n newlines before it converts \r\n linebreaks to \n. That way, newlines in bodies with only \r\n linebreaks are actually never deduplicated.

For example,

"<p>\r\nHello!\r\n</p>\r\n<p>\r\nBye!\r\n</p>"

results in

"Hello!\n\n\nBye!"

Relevant code:

spreewald/lib/spreewald_support/mail_finder.rb

Lines 60 to 70 in db95ff3

	def email_text_body(mail, type = '')
	body = if mail.html_part && type != 'plain-text'
	dom = Nokogiri::HTML(mail.html_part.body.to_s)
	dom.at_css('body').text.gsub(/\n\n/, "\n")
	elsif mail.text_part && type != 'HTML'
	mail.text_part.body.to_s
	else
	mail.body.to_s
	end
	body.gsub("\r\n", "\n") # The mail gem (>= 2.7.1) switched from \n to \r\n line breaks (LF to CRLF) in plain text mails.
	end

While we could just move gsub(/\n\n/, "\n") from the html_part section to the end of email_text_body and always deduplicate, this would impact plain-text bodies negatively.
Maybe use gsub(/\r?\n\r?\n/, "\n") in the html_part section instead. Or gsub(/\r?\n(?:\r?\n)/, "\n"), maybe?

Please also provide a 2.99.x backport for this to allow easier migration to 3.0.

[kratob]

I suggest using [\r\n]+ for the regexp.

[foobear]

Note that this would discard blank lines in plain-text bodies, and tests expecting those would fail.

Putting gsub(/[\r\n]+/, "\n") only in the HTML part's branch should be fine.

[FLeinzi]

Yes, @kratob means the regexp in line 63

[FLeinzi]

I have fixed this in v3.0.1.

There is also a backport in v2.99.2