Mapserver+PostGIS+Charset encoding problems

[Marc Compte]

Hi,

Sorry if this has already been dealt with before but I searched the 
archives and couldn't find an answer for this case.

I have a mapserver getting data from a PostGIS database. The database 
charset encoding is defined as LATIN1 and the text in the tables 
displays all fine and good in the pgAdmin as well as in my browser when 
I just get a list of them with php.

When I set that particular field as a LABEL in the map-file the 
resulting map shows the special characters wrong (mostly accents is what 
I have to deal with). I guess that'd be normal since I wasn't specifying 
any charset encoding yet. So I specified this attribute in my file ... 
but now the funny thing comes.

When I set my layer to use ENCODING 'LATIN1' (same as my db) the map 
just shows a different set of strange characters, but still not the 
right glyphs to show.

If I try using UTF-8 in my LABEL then it starts working some of the 
accents out but not all of them. Special chars that use the accent ´ 
appear perfectly (such as á, é, í, ó and ú), and so does the ¨ on top of 
vowels. But the ` accent does not seem to work.

I have been struggling against this problem for a few days already. Now 
I noticed that using the UTF-8 gives me back a set of characters that 
ressembles very much that of LATIN2.

An 'à' (E0) in my database produces an 'r' with accent (´r).
An 'è' (E8) in my database produces a 'c' with the ^ symbol on top (and 
inverted)
... ì (EC) => 'e' with ^ inverted on top
... ò (F2) => 'n' with ^ inverted on top
... ù (F9) => 'u' with a º on top

This leaves me very puzzled since I don't think I ever used or made any 
reference to Latin2 charset. I am getting very confused about the whole 
thing ... I have a db in Latin1, I try to convert it to UTF-8 and what I 
get is Latin2 ... does this make sense to anyone? I don't know what else 
to try, I made a replica of my table in a UTF-8 db, I have tried using  
"options='-c client_encoding=LATIN1'" in my connection string ... and 
always end up getting the same thing.

By the way, using  "options='-c client_encoding=LATIN1'" in my 
connection string produces exactly the same result as specifying 
ENCODING 'UTF-8' in the LABEL of my map file, which seems also a bit weird.

Does anyone have any idea on what could I try next or where to look? I 
am running out of ideas and I'm very close to go to the database and 
strip all the accents out ...

Thanks,

Marc Compte