unoconv转换XLS到CSV时的段错误及错误处理失效问题排查

I’ve run into this exact issue before—when unoconv crashes with a segmentation fault, the basic $? check can behave unexpectedly, and sometimes it feels like it’s not triggering at all. Let’s break down why this happens and fix it properly.

Why Your Original Script Isn’t Working #

First, segmentation faults are signal-based terminations. When a program dies from SIGSEGV (the signal for a segfault), the shell returns an exit code of 128 + signal number—for SIGSEGV, that’s 128 + 11 = 139. Your [ $? -ne 0 ] should technically catch this, but two common issues might be blocking it:

1. You’re overwriting $?: If there’s any command (like an echo or ls) between the unoconv call and the if check, $? will get updated to that command’s exit code (usually 0, which breaks your check).
2. Batch processing multiple files: If $FILES contains multiple files, unoconv might process some successfully before crashing, and in some cases, it could return an exit code that doesn’t reflect the segfault.

Fixed Script with Proper Error Handling #

Here’s a robust version that avoids these issues, captures segfault-specific errors, and logs details for debugging:

# Capture conversion errors to a log file (adjust path as needed)
ERROR_LOG="./conversion_errors.log"

# If processing multiple files, loop through them one by one (better for debugging)
for FILE in $FILES; do
    echo "Starting conversion for: $FILE"
    # Run unoconv and check success directly in the if condition
    if unoconv -f csv "$FILE" 2>> "$ERROR_LOG"; then
        echo "Successfully converted $FILE to CSV"
    else
        EXIT_CODE=$?
        # Check if the exit code indicates a signal termination (like segfault)
        if [ "$EXIT_CODE" -gt 128 ]; then
            SIGNAL_NUM=$((EXIT_CODE - 128))
            echo "ERROR: Segmentation fault (signal $SIGNAL_NUM) occurred while converting $FILE" | tee -a "$ERROR_LOG"
        else
            echo "ERROR: Conversion failed for $FILE with exit code $EXIT_CODE" | tee -a "$ERROR_LOG"
        fi
    fi
done

Key Improvements #

• Direct command check: Wrapping unoconv directly in the if statement avoids accidentally overwriting $? with other commands.
• Segfault detection: Explicitly checks for exit codes above 128 (the marker for signal-based terminations) to call out segfaults specifically.
• Per-file processing: Looping through individual files lets you pinpoint exactly which Excel file is causing the crash, which is critical for debugging.
• Error logging: Captures all stderr output and error messages to a log file, so you can review what went wrong later.

Bonus Troubleshooting Tips #

1. Test problematic files individually: If you know which file is crashing, run unoconv -f csv bad_file.xls; echo $?—you should see 139 as the exit code, confirming the segfault.
2. Update unoconv/LibreOffice: Segfaults are often bugs in the software. Updating to the latest version of unoconv and its underlying LibreOffice/OpenOffice stack might resolve the crash entirely.
3. Check file corruption: Some malformed Excel files can trigger segfaults. Try opening the problematic file in LibreOffice manually to see if it throws errors.

内容的提问来源于stack exchange，提问作者user2058738