char *
CheckerConversion(Checker *checker, char *src)
{
int len;
if (!checker->check_encoding)
return src;
len = strlen(src);
if (checker->encoding == checker->db_encoding ||
checker->encoding == PG_SQL_ASCII)
{
/*
* No conversion is needed, but we must still validate the data.
*/
pg_verify_mbstr(checker->db_encoding, src, len, false);
return src;
}
if (checker->db_encoding == PG_SQL_ASCII)
{
/*
* No conversion is possible, but we must still validate the data,
* because the client-side code might have done string escaping using
* the selected client_encoding. If the client encoding is ASCII-safe
* then we just do a straight validation under that encoding. For an
* ASCII-unsafe encoding we have a problem: we dare not pass such data
* to the parser but we have no way to convert it. We compromise by
* rejecting the data if it contains any non-ASCII characters.
*/
if (PG_VALID_BE_ENCODING(checker->encoding))
pg_verify_mbstr(checker->encoding, src, len, false);
else
{
int i;
for (i = 0; i < len; i++)
{
if (src[i] == '\0' || IS_HIGHBIT_SET(src[i]))
ereport(ERROR,
(errcode(ERRCODE_CHARACTER_NOT_IN_REPERTOIRE),
errmsg("invalid byte value for encoding \"%s\": 0x%02x",
pg_enc2name_tbl[PG_SQL_ASCII].name,
(unsigned char) src[i])));
}
}
return src;
}
/* Convert the input into the database encoding. */
return (char *) pg_do_encoding_conversion((unsigned char *) src,
len,
checker->encoding,
checker->db_encoding);
}
/*
* pg_custom_client_to_server
*
* convert client encoding to server encoding, but use the passed in encodings
* instead of the global client and server encoding variables.
*
* This routine is basically a slightly modified version of pg_client_to_server.
* Instead of creating this routine a better way may have been to just call
* pg_do_encoding_conversion(), which takes in the necessary arguments, however
* it does not do several necessary checks that pg_client_to_server() does, and
* altering it to have those check may break other parts of the system. Therefore
* until there's a better idea we resort to duplicating some code.
*
* The reason for creating this routine is to let external tables do data
* conversion reliably. Since each external table has an encoding attached to
* it we'd like to just convert from that encoding to the server encoding without
* altering the global client_encoding variable for this local database.
*/
char *
pg_custom_to_server(const char *s, int len, int src_encoding, void *cep)
{
FmgrInfo *custom_encoding_proc = (FmgrInfo *)cep;
Assert(DatabaseEncoding);
Assert(ClientEncoding);
if (len <= 0)
return (char *) s;
if (src_encoding == DatabaseEncoding->encoding ||
src_encoding == PG_SQL_ASCII)
{
/*
* No conversion is needed, but we must still validate the data.
*/
(void) pg_verify_mbstr(DatabaseEncoding->encoding, s, len, false);
return (char *) s;
}
if (DatabaseEncoding->encoding == PG_SQL_ASCII)
{
/*
* No conversion is possible, but we must still validate the data,
* because the client-side code might have done string escaping using
* the selected client_encoding. If the client encoding is ASCII-safe
* then we just do a straight validation under that encoding. For an
* ASCII-unsafe encoding we have a problem: we dare not pass such data
* to the parser but we have no way to convert it. We compromise by
* rejecting the data if it contains any non-ASCII characters.
*/
if (PG_VALID_BE_ENCODING(src_encoding))
(void) pg_verify_mbstr(src_encoding, s, len, false);
else
{
int i;
for (i = 0; i < len; i++)
{
if (s[i] == '\0' || IS_HIGHBIT_SET(s[i]))
ereport(ERROR,
(errcode(ERRCODE_CHARACTER_NOT_IN_REPERTOIRE),
errmsg("invalid byte value for encoding \"%s\": 0x%02x",
pg_enc2name_tbl[PG_SQL_ASCII].name,
(unsigned char) s[i])));
}
}
return (char *) s;
}
return perform_default_encoding_conversion(s, len, true, src_encoding, custom_encoding_proc);
}
/*
* Read the next line from a tsearch data file (expected to be in UTF-8), and
* convert it to database encoding if needed. The returned string is palloc'd.
* NULL return means EOF.
*
* Note: direct use of this function is now deprecated. Go through
* tsearch_readline() to provide better error reporting.
*/
char *
t_readline(FILE *fp)
{
int len;
char *recoded;
char buf[4096]; /* lines must not be longer than this */
if (fgets(buf, sizeof(buf), fp) == NULL)
return NULL;
len = strlen(buf);
/* Make sure the input is valid UTF-8 */
(void) pg_verify_mbstr(PG_UTF8, buf, len, false);
/* And convert */
recoded = pg_any_to_server(buf, len, PG_UTF8);
if (recoded == buf)
{
/*
* conversion didn't pstrdup, so we must. We can use the length of the
* original string, because no conversion was done.
*/
recoded = pnstrdup(recoded, len);
}
return recoded;
}
/*
* Perform output plugin write into tuplestore.
*/
static void
LogicalOutputWrite(LogicalDecodingContext *ctx, XLogRecPtr lsn, TransactionId xid,
bool last_write)
{
Datum values[3];
bool nulls[3];
DecodingOutputState *p;
/* SQL Datums can only be of a limited length... */
if (ctx->out->len > MaxAllocSize - VARHDRSZ)
elog(ERROR, "too much output for sql interface");
p = (DecodingOutputState *) ctx->output_writer_private;
memset(nulls, 0, sizeof(nulls));
values[0] = LSNGetDatum(lsn);
values[1] = TransactionIdGetDatum(xid);
/*
* Assert ctx->out is in database encoding when we're writing textual
* output.
*/
if (!p->binary_output)
Assert(pg_verify_mbstr(GetDatabaseEncoding(),
ctx->out->data, ctx->out->len,
false));
/* ick, but cstring_to_text_with_len works for bytea perfectly fine */
values[2] = PointerGetDatum(
cstring_to_text_with_len(ctx->out->data, ctx->out->len));
tuplestore_putvalues(p->tupstore, p->tupdesc, values, nulls);
p->returned_rows++;
}
/*
* Convert string using encoding_names.
*
* BYTEA convert(BYTEA string, NAME src_encoding_name, NAME dest_encoding_name)
*/
Datum
pg_convert(PG_FUNCTION_ARGS)
{
bytea *string = PG_GETARG_BYTEA_P(0);
char *src_encoding_name = NameStr(*PG_GETARG_NAME(1));
int src_encoding = pg_char_to_encoding(src_encoding_name);
char *dest_encoding_name = NameStr(*PG_GETARG_NAME(2));
int dest_encoding = pg_char_to_encoding(dest_encoding_name);
unsigned char *result;
bytea *retval;
unsigned char *str;
int len;
if (src_encoding < 0)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid source encoding name \"%s\"",
src_encoding_name)));
if (dest_encoding < 0)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid destination encoding name \"%s\"",
dest_encoding_name)));
/* make sure that source string is valid and null terminated */
len = VARSIZE(string) - VARHDRSZ;
pg_verify_mbstr(src_encoding, VARDATA(string), len, false);
str = palloc(len + 1);
memcpy(str, VARDATA(string), len);
*(str + len) = '\0';
result = pg_do_encoding_conversion(str, len, src_encoding, dest_encoding);
/*
* build bytea data type structure.
*/
len = strlen((char *) result) + VARHDRSZ;
retval = palloc(len);
SET_VARSIZE(retval, len);
memcpy(VARDATA(retval), result, len - VARHDRSZ);
if (result != str)
pfree(result);
pfree(str);
/* free memory if allocated by the toaster */
PG_FREE_IF_COPY(string, 0);
PG_RETURN_BYTEA_P(retval);
}
/*
* Verify mbstr to make sure that it is validly encoded in the current
* database encoding. Otherwise same as pg_verify_mbstr().
*/
bool
pg_verifymbstr(const char *mbstr, int len, bool noError)
{
return pg_verify_mbstr(GetDatabaseEncoding(), mbstr, len, noError);
}
static text *
get_line(FILE *f, int max_linesize, int encoding, bool *iseof)
{
int c;
char *buffer = NULL;
char *bpt;
int csize = 0;
text *result = NULL;
bool eof = true;
buffer = palloc(max_linesize + 2);
bpt = buffer;
errno = 0;
while (csize < max_linesize && (c = fgetc(f)) != EOF)
{
eof = false; /* I was able read one char */
if (c == '\r') /* lookin ahead \n */
{
c = fgetc(f);
if (c == EOF)
break; /* last char */
if (c != '\n')
ungetc(c, f);
/* skip \r\n */
break;
}
else if (c == '\n')
break;
++csize;
*bpt++ = c;
}
if (!eof)
{
char *decoded;
int len;
pg_verify_mbstr(encoding, buffer, csize, false);
decoded = (char *) pg_do_encoding_conversion((unsigned char *) buffer,
csize, encoding, GetDatabaseEncoding());
len = (decoded == buffer ? csize : strlen(decoded));
result = palloc(len + VARHDRSZ);
memcpy(VARDATA(result), decoded, len);
SET_VARSIZE(result, len + VARHDRSZ);
if (decoded != buffer)
pfree(decoded);
*iseof = false;
}
else
{
switch (errno)
{
case 0:
break;
case EBADF:
CUSTOM_EXCEPTION(INVALID_OPERATION, "file descriptor isn't valid for reading");
break;
default:
CUSTOM_EXCEPTION(READ_ERROR, strerror(errno));
break;
}
*iseof = true;
}
pfree(buffer);
return result;
}
