[chromium-bugs] Re: Issue 446715 in chromium: no random access for linearized PDF

[c***@googlecode.com]

Updates:
Status: Untriaged
Labels: -OS-Linux OS-All Cr-Internals-Plugins-PDF

Comment #1 on issue 446715 by ***@chromium.org: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

(No comment was entered for this change.)
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Updates:
Cc: ***@foxitsoftware.com

Comment #2 on issue 446715 by ***@foxitsoftware.com: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

(No comment was entered for this change.)
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Updates:
Cc: ***@chromium.org

Comment #3 on issue 446715 by ***@foxitsoftware.com: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

linearized pdf API are in pdfium. From what I see, there are corresponding
codes in pdfium engine that does load and renders linearized pdf.

@thestig, any advice?
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Updates:
Cc: ***@chromium.org

Comment #5 on issue 446715 by ***@chromium.org: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

Gene, do you know if random access into linearized PDFs ever worked?
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #6 on issue 446715 by ***@chromium.org: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

This issue is PDF file specific.

A bit on PDF standard:
In general, lineralized PDF does not guarantee random access to PDF pages.
First page is guaranteed to be first, however all other pages are all over
the file (according to PDF spec). So, sometimes it is necessary to download
full file before we can display a certain page.

Adobe include special hints stream into their linerilized PDFs. Format of
such stream is black box and understood only by Adobe software. This allows
Adobe to load linerilized PDFs more efficiently.

Chrome uses different approach to load parts of PDF necessary to display a
certain page. It works well for some PDFs and not very well for others.

So, to wrap it up, Chrome does have a random access functionality
implemented, unfortunately on some PDFs it does not work very well. If
there is a specific PDF that loads poorly, we can investigate why. (Prefer
this to be a real world scenario, e.g. 100+ page doc that does load
everything before displaying a certain page)
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #9 on issue 446715 by ***@gmail.com: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

There is some preliminary evidence that suggests the random access feature
works on Windows. More investigation pending.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #12 on issue 446715 by ***@chromium.org: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715#c12

The following revision refers to this bug:

https://chromium.googlesource.com/chromium/src.git/+/8032ae2156b4a5988c4bbf11a0cc4c697b594f34

commit 8032ae2156b4a5988c4bbf11a0cc4c697b594f34
Author: thestig <***@chromium.org>
Date: Mon Nov 02 23:18:36 2015

Roll PDFium f7a8b34..c89c619

https://pdfium.googlesource.com/pdfium.git/+log/f7a8b34..c89c619

BUG=446715,512445,543049,548714,549998
TBR=***@chromium.org

Review URL: https://codereview.chromium.org/1412103009

Cr-Commit-Position: refs/heads/master@{#357466}

[modify] http://crrev.com/8032ae2156b4a5988c4bbf11a0cc4c697b594f34/DEPS
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #14 on issue 446715 by ***@google.com: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

I also want to say great job on the big linearization effort. We're going
to put it to good use. Thank you.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #15 on issue 446715 by ***@foxitsoftware.com: no random access
for linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

The checking of m_bMainXRefLoadedOK should be updated after pdfium supports
hint tables. It seems that we didn’t update this part. In fact, pdfium
supports PDF progressive loading in two phases.

At beginning, pdfium only support the progressive loading without the hint
table. The procedure is like:
/Pages -> /page -> object number -> XRef table -> locate object’s offset ->
send the downloading request (offset, size) to applications.

With the hint table, the procedure is changed as below:
Linearized dictionary -> hint table (page offset table and shared object
table) -> send the downloading request (offset, size) to applications.

In the first phase, xRef table is necessary for pdfium to locate the
position of objects. But it isn’t needed for the procedure with hint table
supported.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #16 on issue 446715 by ***@foxitsoftware.com: no random access
for linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

However, we can't just simply remove visiting xRef table from the
progressive loading even after pdfium supports hint table. That's because
not all linearized PDF files have hint table information. If hint table is
not available, we need to go back to the procedure of the progressive
loading without the hint table.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #18 on issue 446715 by ***@chromium.org: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

It looks like the PDF engine requires the size of the PDF. However, with
Content-Encoding: gzip, we have no way to know the size of the PDF until we
find it in the PDF itself. So that would also require some changes.

I tried hardcoding the size of the PDF I was using for testing. With some
other changes I think got it to work by using the hint table, but most
pages required data at the very end of the PDF anyway. Same with another
PDF I tried (both PDFs were linearized using QPDF). So essentially the PDF
only loaded once all the data had been downloaded. It's very possible there
was a problem with what I did (I hacked this up quickly without knowing
fully how pdfium works).
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Updates:
Cc: ***@foxitsoftware.com ***@foxitsoftware.com

Comment #20 on issue 446715 by ***@foxitsoftware.com: no random access
for linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

+Jim from Foxit
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Updates:
Cc: ***@foxitsoftware.com

Comment #21 on issue 446715 by ***@foxitsoftware.com: no random access
for linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

Jim will work on this issue from Foxit side.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #22 on issue 446715 by ***@google.com: no random access for
linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

Jim: if this turns out to be harder than expected, please speak up. It is
helpful to have progressive rendering without seek, but probably not worth
massive amounts of effort. Thanks!
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[c***@googlecode.com]

Comment #23 on issue 446715 by ***@foxitsoftware.com: no random access
for linearized PDF
https://code.google.com/p/chromium/issues/detail?id=446715

Currently, Foxit team is working on supporting XFA on chromium. We plan to
handle this case after XFA is released. XFA is planned to release by end of
February.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings