From: Christos Zoulas <christos@zoulas.com>
Date: Thu, 4 Jan 2018 01:21:32 +0000 (+0000)
Subject: Much better GNU message catalog detection (Joerg Jenderek)
X-Git-Tag: FILE5_33~37
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=0cb71af34ef356ff7aea8f98aa6198702b969b1f;p=file

Much better GNU message catalog detection (Joerg Jenderek)
---

diff --git a/magic/Magdir/gnu b/magic/Magdir/gnu
index 8d5c9721..0eccbe9a 100644
--- a/magic/Magdir/gnu
+++ b/magic/Magdir/gnu
@@ -1,21 +1,111 @@
 
 #------------------------------------------------------------------------------
-# $File: gnu,v 1.18 2017/03/17 21:35:28 christos Exp $
+# $File: gnu,v 1.19 2018/01/04 01:21:32 christos Exp $
 # gnu:  file(1) magic for various GNU tools
 #
 # GNU nlsutils message catalog file format
 #
 # GNU message catalog (.mo and .gmo files)
 
+# Update: Joerg Jenderek
+# URL: https://www.gnu.org/software/gettext/manual/html_node/MO-Files.html
+# Reference: ftp://ftp.gnu.org/pub/gnu/gettext/gettext-0.19.8.tar.gz/
+#	gettext-0.19.8.1/gettext-runtime/intl/gmo.h
+# Note: maybe call it like "GNU translation gettext machine object"
 0	string		\336\22\4\225	GNU message catalog (little endian),
->6	leshort		x		revision %d.
->4	leshort		>0		\b%d,
->>8	lelong		x		%d messages,
->>36	lelong		x		%d sysdep messages
->4	leshort		=0		\b%d,
->>8	lelong		x		%d messages
+#0	ulelong	0x950412DE		GNU-format message catalog data
+# TODO: write lines in such a way that code can also be called for big endian variant
+#>0	use		gettext-object
+#0	name		gettext-object
+>4	ulelong		x		revision
+!:mime	application/x-gettext-translation
+# mo extension is also used for Easeus Partition Master PE32 executable module
+# like ConvertFatToNTFS.mo
+!:ext	gmo/mo
+# only found three revision combinations 0.0 0.1 1.1 as unsigned 32-bit
+# major revision
+>4	ulelong/0xFFff	x		%u.
+# minor revision
+>4	ulelong&0x0000FFff	x	\b%u
+>>8	ulelong		x		\b, %u message
+# plural s
+>>8	ulelong		>1		\bs
+# size of hashing table
+#>20	ulelong		x		\b, %u hash
+#>20	ulelong		>1		\bes
+#>24	ulelong		x		at 0x%x
+# for revsion x.0 offset of table with originals is 1Ch if directly after header
+>4	ulelong&0x0000FFff	=0
+>>12	ulelong		!0x1C		\b, at 0x%x string table
+# but for x.1 table offset i found is 30h. That means directly after bigger header
+>4	ulelong&0x0000FFff	>0
+>>12	ulelong		!0x30		\b, at 0x%x string table
+# The following variables are only used in .mo files with minor revision >= 1
+# number of system dependent segments
+#>>28	ulelong		x		\b, %u segment
+#>>28	ulelong		>1		\bs
+# offset of table describing system dependent segments
+#>>32	ulelong		x		at 0x%x
+# number of system dependent strings pairs
+>>36	ulelong		x		\b, %u sysdep message
+>>36	ulelong		>1		\bs
+# offset of table with start offsets of original sysdep strings
+#>>40	ulelong		x		\b, at 0x%x sysdep strings
+# offset of table with start offsets of translated sysdep strings
+#>>44	ulelong		x		\b, at 0x%x sysdep translations
+# >>(44.l)	ulelong	x		0x%x chars
+# >>>&0		ulelong	x		at 0x%x
+# >>>>(&-4)	string	x		"%s"
+# string table after big header
+#>>48	ubequad		x		\b, string table 0x%llx
+#
+# 0th string length seems to be always 0
+#>(12.l)	ulelong	x		\b, %u chars
+#>>&0		ulelong	x		at 0x%x
+# if 1st string length positiv inspect offset and string
+#>(12.l+8)	ulelong	>0		\b, %u chars
+#>>&0		ulelong	x		at 0x%x
+# if 2nd string length positiv inspect offset and string
+# >(12.l+16)	ulelong	>0		\b, %u chars
+# >>&0		ulelong	x		at 0x%x
+# skip newline byte
+#>>>(&-4)	ubyte	=0x0A
+#>>>>&0		string	x		"%s"
+#>>>(&-4)	ubyte	!0x0A
+#>>>>&-1		string	x		'%s'
+# offset of table with translation strings
+#>16	ulelong		x		\b, at 0x%x translation table
+# check translation 0 length and offset
+>(16.l)		ulelong	>0
+>>&0		ulelong	x
+# translation 0 seems to be often Project-Id with name and version
+>>>(&-4)	string	x		\b, %s
+# trans. 1 with bytes >= 1 unlike icoutils-0.31.0\po\en@boldquot.gmo with 1 NL
+>(16.l+8)	ulelong	>1
+>>&0		ulelong	x
+>>>(&-4)	ubyte	!0x0A
+>>>>&-1		string	x		'%s'
+# 1 New Line like in tar-1.29\po\de.gmo
+>>>(&-4)	ubyte	=0x0A
+>>>>&0		ubyte	!0x0A
+>>>>>&-1	string	x		'%s'
+# 2nd New Line like in parted-3.1\po\de.gmo
+>>>>&0		ubyte	=0x0A
+>>>>>&0		string	x		'%s'
 
 0	string		\225\4\22\336	GNU message catalog (big endian),
+#0	ubelong	0x950412DE		GNU-format message catalog data
+!:mime	application/x-gettext-translation
+!:ext	gmo/mo
+# TODO: for big endian use same code as for little endian
+#>0	use		\^gettext-object
+# DEBUG code
+#>16	ubelong		x		\b, at 0x%x translation table
+#>(16.L)		ubelong	x		0x%x chars
+#>>&0		ubelong	x		at 0x%x
+# unexpected value HERE!
+#>>>(&-4)	ubequad	x		0x%llx
+#
 >4	beshort		x		revision %d.
 >6	beshort		>0		\b%d,
 >>8	belong		x		%d messages,
@@ -78,8 +168,6 @@
 >7	string	02		\b (frcode)
 
 # Files produced by GNU gettext
-0	long	0xDE120495		GNU-format message catalog data
-0	long	0x950412DE		GNU-format message catalog data
 
 # gettext message catalogue
 0	search/1024	\nmsgid