Linux修改LANG后阿拉伯字符仍显示为????的问题求助

Hey Sameer, I’ve run into similar encoding headaches with non-Latin characters in shell scripts before—let’s break down the missing pieces that might be causing those ???? replacements for your Arabic text. Even though you’ve set LANG=ar_SA.utf8, there are several other critical places to check:

1. Ensure Database Client Encoding Matches the Server #

Most databases require explicit encoding configuration for client connections, separate from the system’s LANG variable:

• Oracle: Set the NLS_LANG environment variable to match your database’s character set (e.g., for UTF-8):

  export NLS_LANG=ARABIC_SAUDI_ARABIA.AL32UTF8
  

• MySQL/MariaDB: Add encoding flags to your database command or set the client variable:

  mysql --default-character-set=utf8mb4 -u your_user -p your_db -e "SELECT ..." > output.txt
  

• PostgreSQL: Use the client_encoding parameter in your query or connection string:

  psql -d your_db -c "SET client_encoding = 'UTF8'; SELECT ..." > output.txt
  

2. Explicitly Set All Locale Variables in Your KSH Script #

Just setting LANG might not be enough—some tools prioritize LC_CTYPE or LC_ALL for character handling. Add these lines at the top of your script to enforce consistent UTF-8:

export LANG=ar_SA.utf8
export LC_ALL=ar_SA.utf8
export LC_CTYPE=ar_SA.utf8

LC_ALL acts as an override for all other locale settings, so this ensures no conflicting variables are messing with encoding.

3. Verify Your Export Tool Writes UTF-8 #

Some command-line tools default to system legacy encodings unless told otherwise. For example:

• If using sqlplus, add these lines before your query to force UTF-8 output:

  SET SERVEROUTPUT ON
  SET NAMES UTF8
  SET TERMOUT OFF
  

• If redirecting output directly (e.g., your_query > output.txt), ensure the shell doesn’t perform unintended encoding conversion. You can confirm the output encoding with:

  file output.txt
  

  The result should show something like output.txt: UTF-8 Unicode text.

4. Check Database Server’s Character Set #

Double-check that your database itself is storing Arabic characters in a UTF-8 compatible set (like AL32UTF8 for Oracle, utf8mb4 for MySQL). If the server is using a legacy encoding (e.g., ISO-8859-6), even correct client settings won’t fix the export—you’d need to migrate the data to a UTF-8 collation first.

Quick Test to Isolate the Issue #

Run a simple query directly in your shell (not via script) with the same environment variables, then check the output:

export LANG=ar_SA.utf8
export NLS_LANG=ARABIC_SAUDI_ARABIA.AL32UTF8 # Adjust for your DB
sqlplus your_user/your_pass@your_db <<EOF
SELECT arabic_column FROM your_table WHERE rownum <= 1;
EOF

If this shows correct Arabic text, the problem is in your script’s environment variable inheritance—make sure the script isn’t overriding LANG or NLS_LANG somewhere.

内容的提问来源于stack exchange，提问作者Sameer