Clean up plpgsql identifier handling: process quoted identifiers

correctly, truncate to NAMEDATALEN where needed, allow whitespace
around dots in qualified identifiers.  Get rid of T_RECFIELD and
T_TGARGV token categories, which weren't accomplishing anything
except to create room for sins of omission in the grammar, ie,
places that should have allowed them and didn't.  Fix a few other
bugs en passant.

src/pl/plpgsql/src/pl_funcs.c

@@ -3,7 +3,7 @@
  *			  procedural language
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/pl/plpgsql/src/pl_funcs.c,v 1.18 2002/05/05 17:38:26 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/pl/plpgsql/src/pl_funcs.c,v 1.19 2002/08/08 01:36:05 tgl Exp $
  *
  *	  This software is copyrighted by Jan Wieck - Hamburg.
  *
@@ -46,6 +46,10 @@
 #include "plpgsql.h"
 #include "pl.tab.h"
 
+#ifdef MULTIBYTE
+#include "mb/pg_wchar.h"
+#endif
+
 
 /* ----------
  * Local variables for the namestack handling
@@ -152,6 +156,9 @@ plpgsql_ns_push(char *label)
 {
 	PLpgSQL_ns *new;
 
+	if (label == NULL)
+		label = "";
+
 	new = palloc(sizeof(PLpgSQL_ns));
 	memset(new, 0, sizeof(PLpgSQL_ns));
 	new->upper = ns_current;
@@ -192,9 +199,7 @@ plpgsql_ns_additem(int itemtype, int itemno, char *name)
 	PLpgSQL_ns *ns = ns_current;
 	PLpgSQL_nsitem *nse;
 
-	if (name == NULL)
-		name = "";
-	name = plpgsql_tolower(name);
+	Assert(name != NULL);
 
 	if (ns->items_used == ns->items_alloc)
 	{
@@ -322,46 +327,115 @@ plpgsql_ns_rename(char *oldname, char *newname)
 
 
 /* ----------
- * plpgsql_tolower			Translate a string to lower case
- *					but honor "" escaping.
+ * plpgsql_convert_ident
+ *
+ * Convert a possibly-qualified identifier to internal form: handle
+ * double quotes, translate to lower case where not inside quotes,
+ * truncate to NAMEDATALEN.
+ *
+ * There may be several identifiers separated by dots and optional
+ * whitespace.  Each one is converted to a separate palloc'd string.
+ * The caller passes the expected number of identifiers, as well as
+ * a char* array to hold them.  It is an error if we find the wrong
+ * number of identifiers (cf grammar processing of fori_varname).
+ *
+ * NOTE: the input string has already been accepted by the flex lexer,
+ * so we don't need a heckuva lot of error checking here.
  * ----------
  */
-char *
-plpgsql_tolower(char *s)
+void
+plpgsql_convert_ident(const char *s, char **output, int numidents)
 {
-	char	   *sstart = s;
-	char	   *ret;
-	char	   *cp;
-
-	ret = palloc(strlen(s) + 1);
-	cp = ret;
+	const char *sstart = s;
+	int			identctr = 0;
 
+	/* Outer loop over identifiers */
 	while (*s)
 	{
+		char	   *curident;
+		char	   *cp;
+		int			i;
+
+		/* Process current identifier */
+		curident = palloc(strlen(s) + 1); /* surely enough room */
+		cp = curident;
+
 		if (*s == '"')
 		{
+			/* Quoted identifier: copy, collapsing out doubled quotes */
 			s++;
 			while (*s)
 			{
 				if (*s == '"')
-					break;
+				{
+					if (s[1] != '"')
+						break;
+					s++;
+				}
 				*cp++ = *s++;
 			}
-			if (*s != '"')
-				elog(ERROR, "unterminated \" in name %s", sstart);
+			if (*s != '"')		/* should not happen if lexer checked */
+				elog(ERROR, "unterminated \" in name: %s", sstart);
 			s++;
 		}
 		else
 		{
-			if (isupper((unsigned char) *s))
-				*cp++ = tolower((unsigned char) *s++);
-			else
-				*cp++ = *s++;
+			/*
+			 * Normal identifier: downcase, stop at dot or whitespace.
+			 *
+			 * Note that downcasing is locale-sensitive, following SQL99
+			 * rules for identifiers.  We have already decided that the
+			 * item is not a PLPGSQL keyword.
+			 */
+			while (*s && *s != '.' && !isspace((unsigned char) *s))
+			{
+				if (isupper((unsigned char) *s))
+					*cp++ = tolower((unsigned char) *s++);
+				else
+					*cp++ = *s++;
+			}
+		}
+
+		/* Truncate to NAMEDATALEN */
+		*cp = '\0';
+		i = cp - curident;
+
+		if (i >= NAMEDATALEN)
+		{
+			int len;
+
+#ifdef MULTIBYTE
+			len = pg_mbcliplen(curident, i, NAMEDATALEN-1);
+#else
+			len = NAMEDATALEN-1;
+#endif
+			curident[len] = '\0';
+		}
+
+		/* Pass ident to caller */
+		if (identctr < numidents)
+			output[identctr++] = curident;
+		else
+			elog(ERROR, "Qualified identifier cannot be used here: %s",
+				 sstart);
+
+		/* If not done, skip whitespace, dot, whitespace */
+		if (*s)
+		{
+			while (*s && isspace((unsigned char) *s))
+				s++;
+			if (*s++ != '.')
+				elog(ERROR, "Expected dot between identifiers: %s", sstart);
+			while (*s && isspace((unsigned char) *s))
+				s++;
+			if (*s == '\0')
+				elog(ERROR, "Expected another identifier: %s", sstart);
 		}
 	}
-	*cp = '\0';
 
-	return ret;
+	if (identctr != numidents)
+		elog(ERROR, "Improperly qualified identifier: %s",
+			 sstart);
 }
 
 

src/pl/plpgsql/src/plpgsql.h

@@ -3,7 +3,7 @@
  *			  procedural language
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/pl/plpgsql/src/plpgsql.h,v 1.24 2001/11/29 22:57:37 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/pl/plpgsql/src/plpgsql.h,v 1.25 2002/08/08 01:36:05 tgl Exp $
  *
  *	  This software is copyrighted by Jan Wieck - Hamburg.
  *
@@ -547,11 +547,11 @@ extern PLpgSQL_function *plpgsql_curr_compile;
  */
 extern PLpgSQL_function *plpgsql_compile(Oid fn_oid, int functype);
 extern int	plpgsql_parse_word(char *word);
-extern int	plpgsql_parse_dblword(char *string);
-extern int	plpgsql_parse_tripword(char *string);
-extern int	plpgsql_parse_wordtype(char *string);
-extern int	plpgsql_parse_dblwordtype(char *string);
-extern int	plpgsql_parse_wordrowtype(char *string);
+extern int	plpgsql_parse_dblword(char *word);
+extern int	plpgsql_parse_tripword(char *word);
+extern int	plpgsql_parse_wordtype(char *word);
+extern int	plpgsql_parse_dblwordtype(char *word);
+extern int	plpgsql_parse_wordrowtype(char *word);
 extern PLpgSQL_type *plpgsql_parse_datatype(char *string);
 extern void plpgsql_adddatum(PLpgSQL_datum * new);
 extern int	plpgsql_add_initdatums(int **varnos);
@@ -598,7 +598,7 @@ extern void plpgsql_ns_rename(char *oldname, char *newname);
  * Other functions in pl_funcs.c
  * ----------
  */
-extern char *plpgsql_tolower(char *s);
+extern void plpgsql_convert_ident(const char *s, char **output, int numidents);
 extern const char *plpgsql_stmt_typename(PLpgSQL_stmt * stmt);
 extern void plpgsql_dumptree(PLpgSQL_function * func);
 

src/pl/plpgsql/src/scan.l

@@ -4,7 +4,7 @@
  *			  procedural language
  *
  * IDENTIFICATION
- *    $Header: /cvsroot/pgsql/src/pl/plpgsql/src/Attic/scan.l,v 1.20 2002/08/04 04:17:33 momjian Exp $
+ *    $Header: /cvsroot/pgsql/src/pl/plpgsql/src/Attic/scan.l,v 1.21 2002/08/08 01:36:05 tgl Exp $
  *
  *    This software is copyrighted by Jan Wieck - Hamburg.
  *
@@ -60,13 +60,21 @@ static void plpgsql_input(char *buf, int *result, int max);
 %option noyywrap
 
 %option yylineno
+%option case-insensitive
 
 
-WS    [\200-\377_A-Za-z"]
-WC    [\200-\377_A-Za-z0-9"]
-
 %x	IN_STRING IN_COMMENT
 
+digit			[0-9]
+letter			[\200-\377_A-Za-z]
+letter_or_digit	[\200-\377_A-Za-z0-9]
+
+quoted_ident	(\"[^\"]*\")+
+
+identifier		({letter}{letter_or_digit}*|{quoted_ident})
+
+space			[ \t\n\r\f]
+
 %%
     /* ----------
      * Local variable in scanner to remember where
@@ -154,37 +162,43 @@ dump			{ return O_DUMP;			}
      * Special word rules
      * ----------
      */
-{WS}{WC}*		{ return plpgsql_parse_word(yytext);	}
-{WS}{WC}*\.{WS}{WC}*	{ return plpgsql_parse_dblword(yytext);	}
-{WS}{WC}*\.{WS}{WC}*\.{WS}{WC}*	{ return plpgsql_parse_tripword(yytext); }
-{WS}{WC}*%TYPE		{ return plpgsql_parse_wordtype(yytext);	}
-{WS}{WC}*\.{WS}{WC}*%TYPE	{ return plpgsql_parse_dblwordtype(yytext); }
-{WS}{WC}*%ROWTYPE	{ return plpgsql_parse_wordrowtype(yytext);	}
-
-\$[0-9]+		{ return plpgsql_parse_word(yytext);	}
-\$[0-9]+\.{WS}{WC}*	{ return plpgsql_parse_dblword(yytext);	}
-\$[0-9]+\.{WS}{WC}*\.{WS}{WC}*	{ return plpgsql_parse_tripword(yytext); }
-\$[0-9]+%TYPE		{ return plpgsql_parse_wordtype(yytext);	}
-\$[0-9]+\.{WS}{WC}*%TYPE	{ return plpgsql_parse_dblwordtype(yytext); }
-\$[0-9]+%ROWTYPE	{ return plpgsql_parse_wordrowtype(yytext);	}
-
-[0-9]+			{ return T_NUMBER;			}
+{identifier}					{ return plpgsql_parse_word(yytext);	}
+{identifier}{space}*\.{space}*{identifier}	{ return plpgsql_parse_dblword(yytext);	}
+{identifier}{space}*\.{space}*{identifier}{space}*\.{space}*{identifier}	{ return plpgsql_parse_tripword(yytext); }
+{identifier}{space}*%TYPE		{ return plpgsql_parse_wordtype(yytext);	}
+{identifier}{space}*\.{space}*{identifier}{space}*%TYPE	{ return plpgsql_parse_dblwordtype(yytext); }
+{identifier}{space}*%ROWTYPE	{ return plpgsql_parse_wordrowtype(yytext);	}
+
+\${digit}+						{ return plpgsql_parse_word(yytext);	}
+\${digit}+{space}*\.{space}*{identifier}	{ return plpgsql_parse_dblword(yytext);	}
+\${digit}+{space}*\.{space}*{identifier}{space}*\.{space}*{identifier}	{ return plpgsql_parse_tripword(yytext); }
+\${digit}+{space}*%TYPE			{ return plpgsql_parse_wordtype(yytext);	}
+\${digit}+{space}*\.{space}*{identifier}{space}*%TYPE	{ return plpgsql_parse_dblwordtype(yytext); }
+\${digit}+{space}*%ROWTYPE		{ return plpgsql_parse_wordrowtype(yytext);	}
+
+{digit}+		{ return T_NUMBER;			}
+
+\".				{
+				plpgsql_error_lineno = yylineno;
+				elog(ERROR, "unterminated quoted identifier");
+			}
 
     /* ----------
      * Ignore whitespaces but remember this happened
      * ----------
      */
-[ \t\r\n]+		{ plpgsql_SpaceScanned = 1;		}
+{space}+		{ plpgsql_SpaceScanned = 1;		}
 
     /* ----------
      * Eat up comments
      * ----------
      */
 --[^\r\n]*		;
+
 \/\*			{ start_lineno = yylineno;
 			  BEGIN IN_COMMENT;
 			}
-<IN_COMMENT>\*\/	{ BEGIN INITIAL;			}
+<IN_COMMENT>\*\/	{ BEGIN INITIAL; plpgsql_SpaceScanned = 1; }
 <IN_COMMENT>\n		;
 <IN_COMMENT>.		;
 <IN_COMMENT><<EOF>>	{