Japanese text in vertical writing mode rendered with glyphs for horizontal writing

[mandel59]

Attach (recommended) or Link to PDF file here:

https://www.mofa.go.jp/mofaj/gaiko/treaty/pdfs/treaty159_4a.pdf (via https://www.mofa.go.jp/mofaj/gaiko/treaty/treaty159_4.html)

Configuration:

• Web browser and its version: Firefox 80.0.1
• Operating system and its version: Windows 10 Home (Version 10.0.19041 build 19041)
• PDF.js version: 2.6.123
• Is a browser extension: yes (built-in)

Steps to reproduce the problem: open the PDF and jump to page 2.

What is the expected behavior? (add screenshot)

What went wrong? (add screenshot)

Glyphs ー, ュ and 、 are in horizontal form. (cf. Requirements for Japanese Text Layout 3.1.1 Differences in Vertical and Horizontal Composition in Use of Punctuation Marks )

Link to a viewer (if hosted on a site other than mozilla.github.io/pdf.js or as Firefox/Chrome extension): n/a

[THausherr]

It's also in page 1. The font is named "Ryumin-regular-Identity-H" and the encoding is "Identity-H". I suspect that Adobe itself notices that this is vertical writing and then displays that glyph differently.

1 g
/GS2 gs
0 841 m
0 841 l
f
q
  1 i
  0 841 595 -841 re
  0 841 m
  W
  n
  0.06 840.96 594.96 -840.96 re
  W
  n
  BT
    /G1 1 Tf
    20 0 0 20 287.64 681.67 Tm
    0 0 0 1 k
    0 Tc
    0 Tw
    (\003\261) Tj
    0 -1 TD
    (\003\240) Tj
    T*
    (\003\314) Tj
    T*
    (\036\323) Tj      % this is 0x1ED3 = 7891d
    T*
    (\015\\) Tj
    T*
    (\010Q) Tj
    T*
    (\003t) Tj
    T*
    (\006\024) Tj
    T*
    (\003b) Tj
    T*
    (\003\224) Tj
    T*
    (\011\332) Tj
    T*
    (\016\377) Tj
  ET
Q

[mandel59]

Japanese fonts are often CID-keyed and basing on Adobe-Japan1. CID+7891 of Adobe-Japan1 is for vertical prolonged sound mark.

https://raw.githubusercontent.com/adobe-type-tools/Adobe-Japan1/master/Adobe-Japan1-7.pdf

[mandel59]

Here is the font information of the document.

Type: Type 1 (CID)
Encoding: Identity-H
Actual Font: KozMinPr6N-Regular
Actual Font Type: Type 1 (CID)

[mandel59]

I mean, Adobe Acrobat Reader should not care whether it is vertical writing or not (because a Japanese document often includes both horizontal and vertical parts, such as tate-chu-yoko, so a Japanese font has both forms of glyphs.)
Ryumin (a well-known Japanese serif font) just falls back to KozMinPr6N, using CID.

[THausherr]

Thank you both. I found that font (A-OTF-RyuminPro-Regular.otf) but PDF.js still has the problem, despite restarting firefox. (However PDFBox now renders correctly, thus proving your argument)

[brendandahl]

Looks like we should set writing-mode: vertical-rl; on the canvas when the font file isn't in the PDF and we're drawing vertical text. This seems to work correctly in firefox, but in chrome all glyphs are rotated.
Firefox:

Chrome: