postgresql - column headers are corrupted when querying with pyodbc on Ubuntu -

i'm using postgres on ubuntu , use unixodbc , pyodbc 4.0.16 access data. seem have issue related unicode. when querying db, column headers appear corrupted.

here's example:

import pyodbc  conn = pyodbc.connect("dsn=local_postgres")  conn.setdecoding(pyodbc.sql_char, encoding='utf-8') conn.setdecoding(pyodbc.sql_wchar, encoding='utf-8')  #conn.execute('create schema test') conn.execute('create table test.uni_test(column1 varchar)') conn.execute("insert test.uni_test(column1) values ('my value')")  results = conn.execute('select * test.uni_test')  print results.description columns = [column[0].decode('latin1') column in results.description] print "columns: " + str(columns) print list(results)  result: ((u'c\x00\x00\x00o\x00\x00', <type 'str'>, none, 255, 255, 0, true),) columns: [u'c\x00\x00\x00o\x00\x00'] [(u'my value', )] 

i'm not sure issue is. btw - same behavior observed on mac (el capitan).

thanks in advance, alex

u'c\x00\x00\x00o\x00\x00' first 7 bytes of 'column1' in utf-32le encoding. (the value apparently truncated @ 7 bytes because 'column1' 7 characters long.)

pyodbc received significant upgrade unicode handling 4.x version, , 1 of things developers discovered surprising variety of ways odbc drivers can mix-and-match encoding when returning values. pyodbc wiki page unicode recommends the following postgresql odbc under python 2.7

cnxn.setdecoding(pyodbc.sql_char, encoding='utf-8') cnxn.setdecoding(pyodbc.sql_wchar, encoding='utf-8') 

but in case following required

cnxn.setdecoding(pyodbc.sql_wmetadata, encoding='utf-32le')