Fixed multibyte handling as reported by .

diff --git a/src/interfaces/ecpg/ChangeLog b/src/interfaces/ecpg/ChangeLog
index 36c670d45f6b0219f89251862b4ff5e65f1eff20..64535e5dbce0f60941dbe0d9f5ab076384a5d909 100644 (file)
--- a/src/interfaces/ecpg/ChangeLog
+++ b/src/interfaces/ecpg/ChangeLog
@@ -2165,5 +2165,9 @@ Fr 2. Feb 09:53:48 CET 2007
 
    - Cleaned up va_list handling. Hopefully this now works on all archs.
    - Applied Magnus Hagander's patch to take away some compiler warnings.
+
+Su 11. Feb 16:09:31 CET 2007
+
+   - Fixed multibyte handling as reported by .
    - Set ecpg library version to 5.3.
    - Set ecpg version to 4.3.1.

diff --git a/src/interfaces/ecpg/ecpglib/execute.c b/src/interfaces/ecpg/ecpglib/execute.c
index 2331250bba3349db1548d1f128290bfae8ac013b..bf72920bdcc3497dc4558d1aeaa7ca0171d4efac 100644 (file)
--- a/src/interfaces/ecpg/ecpglib/execute.c
+++ b/src/interfaces/ecpg/ecpglib/execute.c
@@ -1,4 +1,4 @@
-/* $PostgreSQL: pgsql/src/interfaces/ecpg/ecpglib/execute.c,v 1.63 2007/02/02 08:58:23 meskes Exp $ */
+/* $PostgreSQL: pgsql/src/interfaces/ecpg/ecpglib/execute.c,v 1.64 2007/02/11 15:18:17 meskes Exp $ */
 
 /*
  * The aim is to get a simpler inteface to the database routines.
@@ -38,9 +38,11 @@
 static char *
 quote_postgres(char *arg, bool quote, int lineno)
 {
-   char       *res;
-   int         i,
-               ri = 0;
+   char    *res;
+   int error;
+   size_t  length;
+   size_t  escaped_len;
+   size_t  buffer_len;
 
    /*
     * if quote is false we just need to store things in a descriptor they
@@ -50,29 +52,35 @@ quote_postgres(char *arg, bool quote, int lineno)
        return res = ECPGstrdup(arg, lineno);
    else
    {
-       res = (char *) ECPGalloc(2 * strlen(arg) + 3, lineno);
+       length = strlen(arg);
+       buffer_len = 2 * length + 1;
+       res = (char *) ECPGalloc(buffer_len + 3, lineno);
        if (!res)
            return (res);
 
-       /*
-        * We don't know if the target database is using
-        * standard_conforming_strings, so we always use E'' strings.
-        */
-       if (strchr(arg, '\\') != NULL)
-           res[ri++] = ESCAPE_STRING_SYNTAX;
-
-       res[ri++] = '\'';
-
-       for (i = 0; arg[i]; i++, ri++)
+       error = 0;
+       escaped_len = PQescapeString(res+1, arg, buffer_len);
+       if (error)
        {
-           if (SQL_STR_DOUBLE(arg[i], true))
-               res[ri++] = arg[i];
-           res[ri] = arg[i];
+           ECPGfree(res);
+           return NULL;
+       }
+       if (length == escaped_len)
+       {
+           res[0] = res[escaped_len+1] = '\'';
+           res[escaped_len+2] = '\0';
+       }
+       else
+       {
+           /* 
+            * We don't know if the target database is using
+            * standard_conforming_strings, so we always use E'' strings.
+            */
+           memmove(res+2, res+1, escaped_len);
+           res[0] = ESCAPE_STRING_SYNTAX;
+           res[1] = res[escaped_len+2] = '\'';
+           res[escaped_len+3] = '\0';
        }
-
-       res[ri++] = '\'';
-       res[ri] = '\0';
-
        ECPGfree(arg);
        return res;
    }