Importing a CSV file with Chinese column headers will cause the labeling text parsing error.

[CraigMChen]

Describe the bug
Here is a test csv with Chinese column headers:

第一列,第二列,第三列
a1,b1,c1
a2,b2,c2
a3,b3,c3

Import the csv file and select the text classification template as project lable config.

In Labeling Interface page, UI preview cannot load the example text correctly; instead, it mistakenly treats the $第一列 as a string.

The same goes for the Labeling page.

To Reproduce
Steps to reproduce the behavior:

1. Create a project.
2. Import a csv file with Chinese column headers.
3. Select a text template for the project.
4. Go to labeling page.
5. All labeling text is a same string like $第一列.

Expected behavior
The displayed text should be the text in the corresponding row of the specified column in the csv file.

Environment (please complete the following information):

• OS: Ubuntu
• Label Studio Version 1.15.0

[heidi-humansignal]

Hello,

Thank you for contacting Label Studio,

Please be informed that when importing a CSV file for a text classification project, Label Studio expects the key for the text field to be a valid identifier (typically, a Latin-based name like “text”). In your case, the header is “第一列”, so when you reference it in the labeling configuration (i.e. using $第一列), the system does not replace it with the cell value and instead displays it literally.

Therefore, could you please apply the following and let us know if it worked out :

Rename the CSV header:
Change your CSV so that the text column header uses a supported name (like “text”). For example, update your CSV as follows:

text,第二列,第三列 a1,b1,c1 a2,b2,c2 a3,b3,c3

Then, update your labeling configuration to reference the “text” field (e.g., <Text name="text" value="$text"/>).

Comment by Oussama Assili
Workflow Run