@BeforeClass
public static void setUpBeforeClass() throws Exception {
cluster = MiniCluster.buildCluster();
pc = new PigContext(ExecType.LOCAL, new Properties());
pcMR = new PigContext(ExecType.MAPREDUCE, cluster.getProperties());
pc.connect();
}
private void checkParsedConstContent(
PigServer pigServer, PigContext pigContext, String query, String expectedContent)
throws Exception {
pigContext.connect();
LogicalPlan lp = Util.buildLp(pigServer, query + "store B into 'output';");
// Digging down the tree
Operator load = lp.getSources().get(0);
Operator filter = lp.getSuccessors(load).get(0);
LogicalExpressionPlan comparisonPlan = ((LOFilter) filter).getFilterPlan();
List<Operator> comparisonPlanRoots = comparisonPlan.getSinks();
Operator compRootOne = comparisonPlanRoots.get(0);
Operator compRootTwo = comparisonPlanRoots.get(1);
// Here is the actual check logic
if (compRootOne instanceof ConstantExpression) {
assertTrue(
"Must be equal",
((String) ((ConstantExpression) compRootOne).getValue()).equals(expectedContent));
}
// If not left, it must be right.
else {
assertTrue(
"Must be equal",
((String) ((ConstantExpression) compRootTwo).getValue()).equals(expectedContent));
}
}
{
try {
pigContext.connect();
} catch (ExecException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
static {
pc = new PigContext();
try {
pc.connect();
} catch (ExecException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
r = new Random(SEED);
}
public static void oneTimeSetUp() throws Exception {
cluster = MiniGenericCluster.buildCluster();
pc = new PigContext(cluster.getExecType(), cluster.getProperties());
try {
pc.connect();
} catch (ExecException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
GenPhyOp.setPc(pc);
Util.copyFromLocalToCluster(cluster, "test/org/apache/pig/test/data/passwd", "/passwd");
}
@Test
public void testParserWithEscapeCharacters() throws Exception {
// All the needed variables
PigContext pigContext = new PigContext(ExecType.LOCAL, new Properties());
PigServer pigServer = new PigServer(pigContext);
pigContext.connect();
String tempFile = this.prepareTempFile();
String query = String.format("A = LOAD '%s' ;", Util.encodeEscape(tempFile));
// Start the real parsing job
{
// Initial statement
Util.buildLp(pigServer, query);
}
{
// Normal condition
String q = query + "B = filter A by $0 eq 'This is a test string' ;";
checkParsedConstContent(pigServer, pigContext, q, "This is a test string");
}
{
// single-quote condition
String q = query + "B = filter A by $0 eq 'This is a test \\'string' ;";
checkParsedConstContent(pigServer, pigContext, q, "This is a test 'string");
}
{
// escaping dot
// the reason we have 4 backslashes below is we really want to put two backslashes but
// since this is to be represented in a Java String, we escape each backslash with one more
// backslash - hence 4. In a pig script in a file, this would be
// \\.string
String q = query + "B = filter A by $0 eq 'This is a test \\\\.string' ;";
checkParsedConstContent(pigServer, pigContext, q, "This is a test \\.string");
}
{
// newline condition
String q = query + "B = filter A by $0 eq 'This is a test \\nstring' ;";
checkParsedConstContent(pigServer, pigContext, q, "This is a test \nstring");
}
{
// Unicode
String q = query + "B = filter A by $0 eq 'This is a test \\uD30C\\uC774string' ;";
checkParsedConstContent(pigServer, pigContext, q, "This is a test \uD30C\uC774string");
}
}
