1442995 - Weird characters only in Firefox inline PDF viewer

Status: VERIFIED FIXED

(Firefox :: PDF Viewer, defect)

Description

[Anjanesh Lekshminarayanan]

Attachment: Sample line

User Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:58.0) Gecko/20100101 Firefox/58.0
Build ID: 20180208175844

Steps to reproduce:

I'm using a 3rd party software from phpdocx to convert data from a MySQL database and convert them to PDF.


Actual results:

If I open the PDF in Firefox there are some wierd characters in the text.  


Expected results:

Other readers (standalone and inline-browser-readers) don't show them. It's just Firefox's PDF reader. There's no unicode characters in the text.

Comment 1

[Gingerbread Man]

(In reply to Anjanesh Lekshminarayanan from comment #0)
> I'm using a 3rd party software from phpdocx to convert data from a MySQL
> database and convert them to PDF.

Please attach one such file.

> There's no unicode characters in the text.

The screenshot suggests the exact opposite. E000 and E001 are Private Use Area characters; what they correspond to depends on the font used. It seems other readers are using an appropriate font to render ff and ſt ligatures respectively. Meanwhile Firefox is using a font that lacks those characters, hence the placeholder Unicode codeboxes.

Comment 2

[Anjanesh Lekshminarayanan]

Attachment: Another screenshot

I don't think I can attach a PDF here in the public forum.
I've attached a screenshot of another section here.

Comment 3

[Gingerbread Man]

(In reply to Anjanesh Lekshminarayanan from comment #2)
> Created attachment 8955923 [details]
> Another screenshot

No new information there. E002 codebox shows up instead of ligature ti.

Comment 4

[Anjanesh Lekshminarayanan]

Can I email the PDF to you ?

Comment 5

[Anjanesh Lekshminarayanan]

Attachment: Sample PDF

Attached : Sample PDF

Comment 6

[Gingerbread Man]

(In reply to Anjanesh Lekshminarayanan from comment #5)
> Created attachment 8956110 [details]

Thank you. From what I can see in Inspector's code view, there actually aren't any ligatures, despite what I suspected earlier.

I was able to reproduce the issue in Firefox 58.0.2 but not the latest Nightly. I tracked down the bug fix using mozregression:
https://hg.mozilla.org/integration/mozilla-inbound/json-pushes?changeset=318cb586911e4a74d24c61eb24c2769b5a76a154&full=1

Comment 7

[Cristian Baica [:cbaica], Release Desktop QA]

I was able to reproduce the issue on Linux 14.04 x86_x64 using Fx58.0.2 and the sample PDF that was provided by Anjanesh.
Tested on the same OS using Fx60.0b6 and the same PDF file and I can confirm that the issue was fixed. Also tested on the latest nightly build obtaining the same result as on beta.