[PATCH] Recoding shapefile that has "8859x" in the cpg file is not correct.

[akaginch]

If GDAL is linked with the iconv library, recoding a shapefile dataset that has "8859x" in the cpg file is not correct.

I saw ogrshapelayer.cpp and tested iconv in the shell as below.

$ iconv -f UTF-8 -t ISO88591 test.txt
iconv: conversion to ISO88591 unsupported

$ iconv -f UTF-8 -t ISO8859-1 test.txt
Hello!

$ iconv -f UTF-8 -t ISO-8859-1 test.txt
Hello!

In addition, CPL_ENC_ISO8859_1 is defined as below(cpl_string.h).

#define CPL_ENC_ISO8859_1  "ISO-8859-1"

So, I made a patch.

--- ogrshapelayer_org.cpp
+++ ogrshapelayer.cpp
@@ -280,7 +280,7 @@
     }
     if( EQUALN(pszCodePage,"8859",4) )
     {
-        osEncoding.Printf( "ISO%s", pszCodePage );
+        osEncoding.Printf( "ISO-%.4s-%s", pszCodePage, pszCodePage + 4 );
         return osEncoding;
     }
     if( EQUALN(pszCodePage,"UTF-8",5) )

[Even Rouault]

Seems reasonable, except that we should probably validate more the string. For example, I'm wondering if someone couldn't write 8859-1 in the .cpg (although this does not seem to be documented in ​http://resources.arcgis.com/fr/content/kbase?fa=articleShow&d=21106), in which case the suggested formatting wouldn't work

[akaginch]

Yes, it's a user-friendly idea. I rewrite the code as below.

osEncoding.Printf( "ISO-%.4s-%s", pszCodePage, pszCodePage + 4 + ( *(pszCodePage + 4) == '-' ? 1 : 0 ) );

[Even Rouault]

r24680 /trunk/gdal/ogr/ogrsf_frmts/shape/ogrshapelayer.cpp: Shape: correctly deal with .cpg files containing 8859xx string (#4743)

r24681 /branches/1.9/gdal/ogr/ogrsf_frmts/shape/ogrshapelayer.cpp: Shape: correctly deal with .cpg files containing 8859xx string (#4743)