466067 - gloda should convert messages with only HTML bodies to text for full-text indexing

Status: RESOLVED FIXED

(MailNews Core :: Database, defect, P2)

Description

[Andrew Sutherland [:asuth] (he/him)]

If a message only has a text/html and no text/plain, it won't get full-text indexed.  We should use the html to text conversion support available to create a text representation.

Marking blocking because people will probably be sad and confused if they can't search their HTML emails.

Comment 1

[David :Bienvenu]

moving to b2

Comment 2

[Andrew Sutherland [:asuth] (he/him)]

Attachment: v1 kludge expose nsMsgDBFolder::ConvertMsgSnippetToPlainText

None of the interfaces we need are scriptable; they're all hand-rolled (.h, not .idl) XPCOM implementations.  The implementations themselves are in mozilla/, making conversion unpalatable/unlikely.  I fear the 'best' course of action may be to create an XPCOM wrapper.  This patch simply exposes the nsMsgDBFolder implementation of a conversion wrapper function used by GetMsgTextFromStream.  It's a kludge, but it works.  Input desired/requested.

Comment 3

[David :Bienvenu]

Comment on attachment 362557 [details] [diff] [review]
v1 kludge expose nsMsgDBFolder::ConvertMsgSnippetToPlainText

+nsresult nsMsgDBFolder::ConvertMsgSnippetToPlainText(const nsAString& aMessageText,

should be NS_IMETHODIMP

other than that, seems OK, and definitely needed. Yeah, I remember not being able to get at this stuff in js when I did the spotlight indexer

Comment 4

[Andrew Sutherland [:asuth] (he/him)]

pushed with change http://hg.mozilla.org/comm-central/rev/4b77887894f1