1919
2020#include "common/keywords.h"
2121#include "fe_utils/string_utils.h"
22+ #include "mb/pg_wchar.h"
2223
2324static PQExpBuffer defaultGetLocalPQExpBuffer (void );
2425
2526/* Globals exported by this file */
2627int quote_all_identifiers = 0 ;
2728PQExpBuffer (* getLocalPQExpBuffer ) (void ) = defaultGetLocalPQExpBuffer ;
2829
30+ static int fmtIdEncoding = -1 ;
31+
2932
3033/*
3134 * Returns a temporary PQExpBuffer, valid until the next call to the function.
@@ -54,14 +57,48 @@ defaultGetLocalPQExpBuffer(void)
5457 return id_return ;
5558}
5659
60+ /*
61+ * Set the encoding that fmtId() and fmtQualifiedId() use.
62+ *
63+ * This is not safe against multiple connections having different encodings,
64+ * but there is no real other way to address the need to know the encoding for
65+ * fmtId()/fmtQualifiedId() input for safe escaping. Eventually we should get
66+ * rid of fmtId().
67+ */
68+ void
69+ setFmtEncoding (int encoding )
70+ {
71+ fmtIdEncoding = encoding ;
72+ }
73+
74+ /*
75+ * Return the currently configured encoding for fmtId() and fmtQualifiedId().
76+ */
77+ static int
78+ getFmtEncoding (void )
79+ {
80+ if (fmtIdEncoding != -1 )
81+ return fmtIdEncoding ;
82+
83+ /*
84+ * In assertion builds it seems best to fail hard if the encoding was not
85+ * set, to make it easier to find places with missing calls. But in
86+ * production builds that seems like a bad idea, thus we instead just
87+ * default to UTF-8.
88+ */
89+ Assert (fmtIdEncoding != -1 );
90+
91+ return PG_UTF8 ;
92+ }
93+
5794/*
5895 * Quotes input string if it's not a legitimate SQL identifier as-is.
5996 *
60- * Note that the returned string must be used before calling fmtId again,
97+ * Note that the returned string must be used before calling fmtIdEnc again,
6198 * since we re-use the same return buffer each time.
6299 */
63100const char *
64- fmtId (const char * rawid )
101+ fmtIdEnc (const char * rawid , int encoding )
65102{
66103 PQExpBuffer id_return = getLocalPQExpBuffer ();
67104
@@ -134,25 +171,42 @@ fmtId(const char *rawid)
134171}
135172
136173/*
137- * fmtQualifiedId - construct a schema-qualified name, with quoting as needed.
174+ * Quotes input string if it's not a legitimate SQL identifier as-is.
175+ *
176+ * Note that the returned string must be used before calling fmtId again,
177+ * since we re-use the same return buffer each time.
178+ *
179+ * NB: This assumes setFmtEncoding() previously has been called to configure
180+ * the encoding of rawid. It is preferable to use fmtIdEnc() with an
181+ * explicit encoding.
182+ */
183+ const char *
184+ fmtId (const char * rawid )
185+ {
186+ return fmtIdEnc (rawid , getFmtEncoding ());
187+ }
188+
189+ /*
190+ * fmtQualifiedIdEnc - construct a schema-qualified name, with quoting as
191+ * needed.
138192 *
139193 * Like fmtId, use the result before calling again.
140194 *
141195 * Since we call fmtId and it also uses getLocalPQExpBuffer() we cannot
142196 * use that buffer until we're finished with calling fmtId().
143197 */
144198const char *
145- fmtQualifiedId (const char * schema , const char * id )
199+ fmtQualifiedIdEnc (const char * schema , const char * id , int encoding )
146200{
147201 PQExpBuffer id_return ;
148202 PQExpBuffer lcl_pqexp = createPQExpBuffer ();
149203
150204 /* Some callers might fail to provide a schema name */
151205 if (schema && * schema )
152206 {
153- appendPQExpBuffer (lcl_pqexp , "%s." , fmtId (schema ));
207+ appendPQExpBuffer (lcl_pqexp , "%s." , fmtIdEnc (schema , encoding ));
154208 }
155- appendPQExpBufferStr (lcl_pqexp , fmtId (id ));
209+ appendPQExpBufferStr (lcl_pqexp , fmtIdEnc (id , encoding ));
156210
157211 id_return = getLocalPQExpBuffer ();
158212
@@ -162,6 +216,24 @@ fmtQualifiedId(const char *schema, const char *id)
162216 return id_return -> data ;
163217}
164218
219+ /*
220+ * fmtQualifiedId - construct a schema-qualified name, with quoting as needed.
221+ *
222+ * Like fmtId, use the result before calling again.
223+ *
224+ * Since we call fmtId and it also uses getLocalPQExpBuffer() we cannot
225+ * use that buffer until we're finished with calling fmtId().
226+ *
227+ * NB: This assumes setFmtEncoding() previously has been called to configure
228+ * the encoding of schema/id. It is preferable to use fmtQualifiedIdEnc()
229+ * with an explicit encoding.
230+ */
231+ const char *
232+ fmtQualifiedId (const char * schema , const char * id )
233+ {
234+ return fmtQualifiedIdEnc (schema , id , getFmtEncoding ());
235+ }
236+
165237
166238/*
167239 * Format a Postgres version number (in the PG_VERSION_NUM integer format
0 commit comments