1031612 - In PDF Viewer, the buggy XMP title "Untitled" overrides the document info title

Status: RESOLVED WORKSFORME

(Firefox :: PDF Viewer, defect, P4)

Description

[Vincent Lefevre]

When I open some PDF files in the builtin PDF viewer, their title is not always shown in the window name (titlebar). For instance:

  https://www.vinc17.net/research/slides/gdt2014-04.pdf

is shown as "Untitled", though both evince and pdfinfo say that the title is:

  Introduction to the GNU MPFR Library

Some PDF files do not have this problem, e.g.

  https://www.vinc17.net/research/slides/tamadi2013-10.pdf

It seems that the problem occurs on files for which I used ps2pdf then re-added the metadata with pdftk.

Comment 1

[Petruta Horea [:phorea], Desktop QA]

Mozilla/5.0 (Windows NT 6.1; WOW64; rv:34.0) Gecko/20100101 Firefox/34.0

I opened the pdf using latest Nightly 2014-07-27, opened the Inspector and search for title:
<title>Untitled - gdt2014-04.pdf</title>

I see the same title in the titlebar so the name could came from here.

Comment 2

[Vincent Lefevre]

After inspecting several PDF files, it seems that the problem occurs when in the PDF file, /Title is followed by a space, e.g.:

/Title (Introduction to the GNU MPFR Library)

The working PDF file mentioned above has:

/Title(Hardest-to-Round Cases \205 Part 2)

Comment 3

[Yury Delendik (:yury)]

The file contains XMP metadata, which overrides the original title with "Untitled" value. And that's what is used for web page/tab title. Please don't specify XML metadata if you don't wish the title be pulled from it. Not sure how you want to resolve it, but space between the /Title is not the reason. Do you want to close it as invalid or to request the removal of the XMP metadata support in the PDF viewer?

Comment 4

[Vincent Lefevre]

OK, updated the title according to the real problem.

That's annoying because it is Ghostscript (via ps2pdf) that automatically adds these XMP metadata in the back of the user. This is a serious bug in Ghostscript as according to http://www.pdflib.com/knowledge-base/xmp-metadata/ XMP metadata tends to be preferred to the conventional document info entries. I'll report a bug against Ghostscript, but unfortunately one already has many PDF files on the web with this "Untitled" title in the XMP metadata added by Ghostscript. IMHO, as a workaround to the Ghostscript bug, if there is a document info title and an XMP title "Untitled" (which will never be a correct title in practice), PDF Viewer should prefer the document info one.

Comment 5

[Yury Delendik (:yury)]

The PDF.js logic can be adjusted to ignore "Untitled" entry. See https://github.com/mozilla/pdf.js/blob/master/web/viewer.js#L1153 .

Comment 6

[Sushrut Bhalla]

I would like to work on this...

Comment 7

[Yury Delendik (:yury)]

Hi Sushrut! You can submit a pull request at https://github.com/mozilla/pdf.js (paste the pull request link here).

Comment 8

[Sang Mercado]

Sushrut, are you still working on this bug?

If not, Can I be assigned to this bug?

Comment 9

[Prasanth Vaaheeswaran]

This bug is still unassigned (In reply to Yury Delendik (:yury) from comment #7)
> Hi Sushrut! You can submit a pull request at
> https://github.com/mozilla/pdf.js (paste the pull request link here).

Hey, after reading the bug info and comments I'd really like to work on this as my first bug. Not sure however if a pull request was submitted, but since I don't see one on the github or a link posted here I'm going to assume it's still in limbo. Going to grab the pdf.js source and start testing :)

Comment 10

[Yury Delendik (:yury)]

(In reply to Sang Mercado from comment #8)
> Sushrut, are you still working on this bug?
> 
> If not, Can I be assigned to this bug?

So far nobody submitted a pull request. I estimate the patch be additional 3-5 lines of code.

Once a github pull request is opened, it will referred from here (to avoid future duplicate work).

Comment 11

[Sang Mercado]

Attachment: Requesting code review

Hopefully I did this right

Comment 12

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

The pull request has been merge, so is this bug fixed?

Comment 13

[Yury Delendik (:yury)]

Looks good on current nightly

Comment 14

[Vincent Lefevre]

I confirm that the bug no longer occurs in nightly.