xref: /freebsd/contrib/file/magic/Magdir/apache (revision c2c014f24c10f90d85126ac5fbd4d8524de32b1c)
1*40427ccaSGordon Tetlow
2*40427ccaSGordon Tetlow#------------------------------------------------------------------------------
3*40427ccaSGordon Tetlow# $File: apache,v 1.1 2017/04/11 14:52:15 christos Exp $
4*40427ccaSGordon Tetlow# apache: file(1) magic for Apache Big Data formats
5*40427ccaSGordon Tetlow
6*40427ccaSGordon Tetlow# Avro files
7*40427ccaSGordon Tetlow0	string		Obj		Apache Avro
8*40427ccaSGordon Tetlow>3	byte		x		version %d
9*40427ccaSGordon Tetlow
10*40427ccaSGordon Tetlow# ORC files
11*40427ccaSGordon Tetlow# Important information is in file footer, which we can't index to :(
12*40427ccaSGordon Tetlow0	string		ORC		Apache ORC
13*40427ccaSGordon Tetlow
14*40427ccaSGordon Tetlow# Parquet files
15*40427ccaSGordon Tetlow0	string		PAR1		Apache Parquet
16*40427ccaSGordon Tetlow
17*40427ccaSGordon Tetlow# Hive RC files
18*40427ccaSGordon Tetlow0	string		RCF		Apache Hive RC file
19*40427ccaSGordon Tetlow>3	byte		x		version %d
20*40427ccaSGordon Tetlow
21*40427ccaSGordon Tetlow# Sequence files (and the careless first version of RC file)
22*40427ccaSGordon Tetlow
23*40427ccaSGordon Tetlow0	string		SEQ
24*40427ccaSGordon Tetlow>3	byte		<6		Apache Hadoop Sequence file version %d
25*40427ccaSGordon Tetlow>3	byte		>6		Apache Hadoop Sequence file version %d
26*40427ccaSGordon Tetlow>3	byte		=6
27*40427ccaSGordon Tetlow>>5	string		org.apache.hadoop.hive.ql.io.RCFile$KeyBuffer  Apache Hive RC file version 0
28*40427ccaSGordon Tetlow>>3	default		x		Apache Hadoop Sequence file version 6
29