Calling Grep inside Java gives incorrect results while calling grep in shell gives correct results.

Question

I've got a problem where calling grep from inside java gives incorrect results, as compared to the results from calling grep on the same file in the shell.

My grep command (called both in Java and in bash. I escaped the slash in Java accordingly):

/bin/grep -vP --regexp='^[0-9]+\t.*' /usr/local/apache-tomcat-6.0.18/work/Catalina/localhost/saccitic/237482319867147879_1271411421

Java Code:

String filepath = "/path/to/file";
String options = "P";
String grepparams = "^[0-9]+\\t.*";
String greppath = "/bin/";

String[] localeArray = new String[] {
    "LANG=",
    "LC_COLLATE=C",
    "LC_CTYPE=UTF-8",
    "LC_MESSAGES=C",
    "LC_MONETARY=C",
    "LC_NUMERIC=C",
    "LC_TIME=C",
    "LC_ALL="
};

options = "v"+options; //Assign optional params

if (options.contains("P")) {
    grepparams = "\'"+grepparams+"\'"; //Quote the regex expression if -P flag is used
} else {
    options = "E"+options; //equivalent to calling egrep
}

proc = sysRuntime.exec(greppath+"/grep -"+options+" --regexp="+grepparams+" "+filepath, localeArray);
System.out.println(greppath+"/grep -"+options+" --regexp="+grepparams+" "+filepath);
inStream = proc.getInputStream();

The command is supposed to match and discard strings like these:

85295371616 Hi Mr Lee, please be informed that...

My input file is this:

85aaa234567 Hi Ms Chan, please be informed that...
85292vx5678 Hi Mrs Ng, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85aaa234567 Hi Ms Chan, please be informed that...
85292vx5678 Hi Mrs Ng, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
8~!95371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
852&^*&1616 Hi Mr Lee, please be informed that...
8529537Ax16 Hi Mr Lee, please be informed that...
85====ppq16 Hi Mr Lee, please be informed that...
85291234783 a3283784428349247233834728482984723333
85219299222

The commands works when I call it from inside bash (Results below):

85aaa234567 Hi Ms Chan, please be informed that...
85292vx5678 Hi Mrs Ng, please be informed that...
85aaa234567 Hi Ms Chan, please be informed that...
85292vx5678 Hi Mrs Ng, please be informed that...
8~!95371616 Hi Mr Lee, please be informed that...
852&^*&1616 Hi Mr Lee, please be informed that...
8529537Ax16 Hi Mr Lee, please be informed that...
85====ppq16 Hi Mr Lee, please be informed that...
85219299222

However, when I call grep again inside java, I get the entire file (Results below):

85aaa234567 Hi Ms Chan, please be informed that...
85292vx5678 Hi Mrs Ng, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85aaa234567 Hi Ms Chan, please be informed that...
85292vx5678 Hi Mrs Ng, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
8~!95371616 Hi Mr Lee, please be informed that...
85295371616 Hi Mr Lee, please be informed that...
852&^*&1616 Hi Mr Lee, please be informed that...
8529537Ax16 Hi Mr Lee, please be informed that...
85====ppq16 Hi Mr Lee, please be informed that...
85291234783 a3283784428349247233834728482984723333
85219299222

What could be the problem that will cause the grep called by Java to return incorrect results? I tried passing local information via the environment string array in runtime.exec, but nothing seems to change. Am I passing in the locale information incorrectly, or is the problem something else entirely?

Answer 1

The single quotes are used by the shell, and not seen by grep. You should not add them in java.

Also, you should use the exec(String[], String[]) method instead of concatenating the parameters, so you have control over the separation of the parameters.

And i agree with the other comments saying you should do this in pure java instead of starting grep in a separate process.