544462 - HTML5 Offline features - Asterisk (*) not supported in network section of manifest file

Status: RESOLVED FIXED

(Core :: Networking: Cache, defect)

Description

[Aymeric Vitte]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; fr; rv:1.9.1.7) Gecko/20091221 Firefox/3.5.7
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; fr; rv:1.9.1.7) Gecko/20091221 Firefox/3.5.7

As per HTML5 specifications, putting * in the network section of manifest file should allow the browser to access all resources on the internet.

It is not working with Firefox, up to the latest 3.7 build

There is a workaround putting http://* instead of *

To test it go to (please send a valid user agent that the system can understand as being firefox >= 3.5):

http://www.blimpme.com/mobile/indexo.php

the manifest file uses here http://* in network section

Choose a tag, you will see some results appear

Clear the offline data and go to :

http://www.blimpme.com/mobile/indexo.php?f=offline2

the manifest file uses here * in network section

Refresh the page

Choose a tag, as you can see nothing happens, the browser does not send any longer requests outside


Reproducible: Always

Comment 1

[Honza Bambas (:mayhemer)]

Confirming, the spec allows this.

Comment 2

[Honza Bambas (:mayhemer)]

-> me

Comment 4

[Honza Bambas (:mayhemer)]

This is important part of the spec, we should ship this in Fx4.

Comment 5

[Doug Turner (:dougt)]

love to have, but not blocking fennec.

Comment 6

[Johnny Stenback (:jst)]

Not blocking, but I'd approve a safe patch!

Comment 7

[Honza Bambas (:mayhemer)]

Attachment: v1

- very very simple patch
- no API modifications
- has a test

Comment 8

[Honza Bambas (:mayhemer)]

Attachment: v1 [Check in comment 14]

qref

Comment 9

[Johnny Stenback (:jst)]

Comment on attachment 488697 [details] [diff] [review]
v1 [Check in comment 14]

- In nsOfflineManifestItem::HandleManifestLine():

     case PARSE_BYPASS_ENTRIES: {
+        if (line[0] == '*' && (line.Length() == 1 || line[1] == ' ' || line[1] == '\t'))

This condition looks a bit off, or maybe I'm just not clear on what the format is here. But '*' alone on a line, or followed by ' ' or '\t', followed by *anything*, should be accepted here?

Comment 10

[Honza Bambas (:mayhemer)]

(In reply to comment #9)
> But '*' alone on a line, or followed by ' ' or '\t', followed by
> *anything*, should be accepted here?

The spec parse a line of the manifest by splitting it to tokes separated by 0x20, 0x09.  Before that all trailing spaces are cut.  Tokes then are in an array.  The spec says then:

'If mode is "online whitelist": If the first item in _tokens_ is a U+002A
ASTERISK character (*), then set online whitelist wildcard flag to open and jump back to the step labeled "start of line"'

That means to ignore the rest of the line.  The spec is trying to be forward compatible when parsing the manifest.  Unknown is silently ignored.

Better would be to split the line to tokes as the spec does, but we don't have a good parser for it.

Comment 11

[Johnny Stenback (:jst)]

Comment on attachment 488697 [details] [diff] [review]
v1 [Check in comment 14]

Ok, thanks for the explanation!

r=jst assuming jupiter.jpg is licensed such that we can include it here.

Comment 12

[Honza Bambas (:mayhemer)]

(In reply to comment #11)
> r=jst assuming jupiter.jpg is licensed...

It is my own astro photography ;)

Comment 13

[Honza Bambas (:mayhemer)]

Comment on attachment 488697 [details] [diff] [review]
v1 [Check in comment 14]

This is a very simple patch, backward compatible, making us more conforming the HTML5 spec.

Comment 14

[Honza Bambas (:mayhemer)]

Comment on attachment 488697 [details] [diff] [review]
v1 [Check in comment 14]

http://hg.mozilla.org/mozilla-central/rev/bb5091f2e6a1