problem

Uploading very large files in the Task Import wizard can crash the browser before the upload starts. The frontend previously read the complete selected file with FileReader.readAsText() to detect XLIFF language metadata. For files around 1 GB or larger this can exhaust browser memory.

solution

Limit frontend language detection to a small initial file chunk instead of reading the complete file. Do not rely on file extensions or File.type, because customers may rename files and browser MIME detection is unreliable. Instead, sniff only the bounded prefix for XLIFF-like content, such as an <xliff> root or XLIFF-style source-language / target-language attributes. Keep the actual upload unchanged by submitting the original File object via FormData.