@mattmary opened this Issue on December 7th 2021 Contributor

when running matomo on websites which have special characters such as japanese or chinese characters, we are not able to get them in the daily report even if we follow https://matomo.org/faq/how-to-install/faq_142/.

Expected Behavior

Having the characters displayed in the PDF report

Current Behavior

Current unicode characters are not displayed in the PDF report

Steps to Reproduce (for Bugs)

Create a webpage with a title in chinese or in japanese.
Check in the report that these characters are well displayed in the page title report.
create a daily report with page titles.
check the generated pdf that the characters for the chinese page are not well displayed.

@froger-me commented on December 8th 2021

In app/core/ReportRenderer.php line 24:

const DEFAULT_REPORT_FONT_FAMILY = 'dejavusans';
  • The font family dejavusans doesn't support CJK glyphs
  • The font family is shipped with tcpdf ; no font family shipped with tcpdf in app/vendor/tecnickcom/tcpdf/fonts support CJK glyphs unless it's set according to Matomo chosen locale (no support for mixed content)
  • changing dejavusans to droidsansfallback in app/core/ReportRenderer.php line 24 and adding the corresponding font files to app/vendor/tecnickcom/tcpdf/fonts result in reports showing CJK glyphs properly

Because no TTF or OTF font can accommodate all 100k+ glyphs (limit 64k), a suggestion would be to allow font selection/upload for different language combinations in settings to override the automatically chosen font per language.

@tsteur commented on December 8th 2021 Member

I'll mark this for now as a regression as I think this used to work?

@froger-me commented on December 8th 2021

Note in case there are discrepancies between different shipped versions: I am using Matomo Analytics - Ethical Stats. Powerful Insights. v4.5.0 for WordPress
The website's language is English, but with mixed language content.

@peterhashair commented on January 5th 2022 Contributor

I guess, to view the report with the correct format need to change the Matomo language to the same language that the report has. how to change language is here. https://matomo.org/faq/how-to/how-can-i-change-the-language-of-matomo-as-a-user/
it actually take the last param in the URL as laguage in my case is language=en

It seems like this is not a regression. More like a FAQ. Or we can add a config variable in the config.ini.php called report_lang @tsteur any suggestion?

@froger-me commented on January 6th 2022

What about when a user has mixed language content? The base language is English, but the content is bilingual, primarily English with some Chinese characters, maybe, for example? The interface language would need to be English for the user (so language=en seems right), but the Chinese characters supported nonetheless to handle displaying the content, like the behavior in a browser?

@peterhashair commented on January 6th 2022 Contributor

@froger-me, right, good question. I guess we should import the https://matomo.org/faq/how-to-install/faq_142/ fonts here to tcpdf Lib, that should fix it. I will give a try, see if that works.

@peterhashair commented on January 11th 2022 Contributor

@@froger-me Update this issue, it should be in the next release very soon, it use the unifont.ttf, https://matomo.org/faq/how-to-install/faq_142/ If you like a different font, I guess just need to overwrite unifont.ttf

@froger-me commented on January 11th 2022

Great, thanks for the update!

As a suggestion for possible further improvement - how difficult would it be to set the font in config.ini.php so that it doesn't break at update? I specifically have the WordPress plugin in mind, which I think replaces all the files entirely during plugin update (but not the config.ini.php because it's located outside of the plugin directory) - but that may be a valid concern for matomo as a whole.

It seems only 3 files would need what it looks like minor changes in a few places - calling the Piwik\Config\Config::getInstance(), test for the existence of the config in the file, and if not, use a default one:


I might be totally off base here, so please ignore if it's irrelevant.

@peterhashair commented on January 11th 2022 Contributor

@froger-me that make sense to me, set up fonts in the config will be easier, but I can't make the judgment call. Can I copy the request to another tickets category as an enhancement (new feature), which I am sure the management team will pick up,

@froger-me commented on January 11th 2022

Can I copy the request to another tickets category as an enhancement (new feature), which I am sure the management team will pick up

Absolutely :) !

This Issue was closed on January 11th 2022
Powered by GitHub Issue Mirror