Skip to main content
Question

Regarding Font Recognition on FME Flow

  • November 12, 2025
  • 1 reply
  • 44 views

sei
Contributor
Forum|alt.badge.img+3

Hello everyone,

I hope you are doing well.

I encountered an issue when trying to polygonize text in DWG data that contains Japanese characters using the Transformer: TextStroker. The text was displayed as tofu (blank squares).

When I tested the same process on FME Form 2025.1.1 (Windows 11), the text was polygonized correctly without any problem.
If anyone has insight into possible misconfigurations or additional steps needed for proper handling on Linux, I would greatly appreciate your advice.

Environment:

  • OS: Linux Ubuntu

  • FME Flow: 2025.1.1

  • Target language: Japanese

  • Desired font: Noto Sans CJK JP

Notes:
Font recognition status on OS:

>(base) ubuntu:/usr/share/fonts/opentype/$ fc-list
>/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc: Noto Sans CJK JP:style=Regular
>/usr/share/fonts/truetype/noto/NotoSansCJK-Regular.ttc: Noto Sans Mono CJK JP:style=Regular

I’ve attached a sample workspace and test files for reference.
Thank you in advance for your time and assistance.

 

1 reply

sei
Contributor
Forum|alt.badge.img+3
  • Author
  • Contributor
  • November 26, 2025

After further investigation, I confirmed that the issue was resolved in this file as well as in another test file.

It appears that the text was not completely unrecognized; rather, some characters were partially rendered as tofu. This led me to suspect that unexpected control characters or invalid data might have been included in the stored text strings.

Therefore, I applied a StringReplacer with the regular expression [\x00-\x1F\x7F] to automatically remove any potentially problematic control characters, and confirmed that this eliminated the tofu output.

Please note that this is only one possible cause. Similar issues may still occur depending on the data, so I would appreciate it if you could regard this simply as one example case.