Impala
Impalaistheopensource,nativeanalyticdatabaseforApacheHadoop.
|
#include <hdfs-scanner.h>
Public Attributes | |
char * | start |
int | len |
Static Public Attributes | |
static const char * | LLVM_CLASS_NAME = "struct.impala::FieldLocation" |
Intermediate structure used for two pass parsing approach. In the first pass, the FieldLocation structs are filled out and contain where all the fields start and their lengths. In the second pass, the FieldLocation is used to write out the slots. We want to keep this struct as small as possible.
Definition at line 52 of file hdfs-scanner.h.
int impala::FieldLocation::len |
Encodes the length and whether or not this fields needs to be unescaped. If len < 0, then the field needs to be unescaped.
Definition at line 59 of file hdfs-scanner.h.
Referenced by impala::DelimitedTextParser::AddColumn(), impala::HdfsTextScanner::CopyBoundaryField(), impala::HdfsScanner::WriteCompleteTuple(), and impala::HdfsTextScanner::WritePartialTuple().
|
static |
Definition at line 61 of file hdfs-scanner.h.
Referenced by impala::HdfsScanner::CodegenWriteCompleteTuple().
char* impala::FieldLocation::start |
start of field. This is set to NULL for FieldLocations that are past the end of the row in the file. E.g. the file only had 3 cols but the table has 10. These then get written out as NULL slots.
Definition at line 56 of file hdfs-scanner.h.
Referenced by impala::DelimitedTextParser::AddColumn(), impala::HdfsTextScanner::CopyBoundaryField(), and impala::HdfsTextScanner::WritePartialTuple().