[GRASS-user] ps.map character encoding problems

[Glynn Clements]

Frank Broniewski wrote:

> I have trouble with the ps.map module and character encoding in the legend. I 
> have french characters (é,è and others) as a layer label. The layer itself 
> was imported from a postgis database, which has a unicode encoding. As far as 
> I understand the manual for ps.map, you need your mapping instructions file 
> encoded to ISO-8859-1. At the end of the manual is the appropriate command 
> listed. so far so good.
> But I use a condition for my vector layer, which has the layer name as where 
> clause term. Converting my mapping instructions file to ISO-8859-1 gives me 
> another error, with the result that the layer cannot be drawn and ps.map 
> exits. Here's the error:
> 
> ps.map -e input=print.txt output=print.ps
> PS-PAINT: scale set to 1 : 360512.
> PS-PAINT: reading raster file <dgm in PERMANENT> ...
> PS-PAINT: reading vector points file <sites in PERMANENT> ...DBMI-Postgres 
> driver error:
> Cannot select:
> SELECT id FROM sites WHERE epoch = 'M�solithique'
> ERROR:  invalid byte sequence for encoding "UTF8": 0xe9736f
> HINT:  This error can also happen if the byte sequence does not match the 
> encoding expected by the server, which is controlled by "client_encoding".
> 
> Any help is greatly appreciated

ps.map itself doesn't care about encodings; it's all just bytes.

Any text which is meant to appear in the output is embedded verbatim
in the PostScript file. The prolog automatically re-encodes any fonts
to ISOLatin1Encoding (PostScript's default encoding isn't much use, as
nothing other than PostScript uses it), so any text needs to use that
encoding.

Similarly, any strings which are passed to library functions etc are
passed without any conversions.

In this particular case, it should suffice to use ISO-8859-1 for
displayed text and UTF-8 for the WHERE clause. The reference to
client_encoding suggests that it may be possible for the DBMS to
perform encoding conversions, but I don't know how, or if it's even
possible from GRASS.

It would theoretically be possible to modify ps.map to convert
displayed text from a user-defined encoding to ISO-8859-1. However,
this could require a significant amount of work. As ps.map will be
superseded in 7.x, I'm not sure that it's worth the effort.

-- 
Glynn Clements <glynn at gclements.plus.com>