Revise plpgsql's scanner to process comments and string literals in a way

more nearly matching the core SQL scanner.  The user-visible effects are:

* Block comments (slash-star comments) now nest, as per SQL spec.

* In standard_conforming_strings mode, backslash as the last character of a
  non-E string literal is now correctly taken as an ordinary character;
  formerly it was misinterpreted as escaping the ending quote.  (Since the
  string also had to pass through the core scanner, this invariably led
  to syntax errors.)

* Formerly, backslashes in the format string of RAISE were always treated as
  quoting the next character, regardless of mode.  Now, they are ordinary
  characters with standard_conforming_strings on, while with it off, they
  introduce the same set of escapes as in the core SQL scanner.  Also,
  escape_string_warning is now effective for RAISE format strings.  These
  changes make RAISE format strings work just like any other string literal.

This is implemented by copying and pasting a lot of logic from the core
scanner.  It would be a good idea to look into getting rid of plpgsql's
scanner entirely in favor of using the core scanner.  However, that involves
more change than I can justify making during beta --- in particular, the core
scanner would have to become re-entrant.

In passing, remove the kluge that made the plpgsql scanner emit T_FUNCTION or
T_TRIGGER as a made-up first token.  That presumably had some value once upon
a time, but now it's just useless complication for both the scanner and the
grammar.

Author: tglsfdc

doc/src/sgml/plpgsql.sgml

@@ -1,4 +1,4 @@
-<!-- $PostgreSQL: pgsql/doc/src/sgml/plpgsql.sgml,v 1.139 2009/04/02 19:20:45 momjian Exp $ -->
+<!-- $PostgreSQL: pgsql/doc/src/sgml/plpgsql.sgml,v 1.140 2009/04/19 18:52:56 tgl Exp $ -->
 
 <chapter id="plpgsql">
   <title><application>PL/pgSQL</application> - <acronym>SQL</acronym> Procedural Language</title>
@@ -220,10 +220,8 @@ END <optional> <replaceable>label</replaceable> </optional>;
      There are two types of comments in <application>PL/pgSQL</>. A double
      dash (<literal>--</literal>) starts a comment that extends to the end of
      the line. A <literal>/*</literal> starts a block comment that extends to
-     the next occurrence of <literal>*/</literal>.  Block comments cannot be
-     nested, but double dash comments can be enclosed into a block comment and
-     a double dash can hide the block comment delimiters <literal>/*</literal>
-     and <literal>*/</literal>.
+     the next occurrence of <literal>*/</literal>.  Block comments nest,
+     just as in ordinary SQL.
     </para>
 
     <para>

src/pl/plpgsql/src/gram.y

@@ -9,7 +9,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/gram.y,v 1.121 2009/02/18 11:33:04 petere Exp $
+ *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/gram.y,v 1.122 2009/04/19 18:52:57 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -62,6 +62,8 @@ static PLpgSQL_row		*make_scalar_list1(const char *initial_name,
 										   int lineno);
 static	void			 check_sql_expr(const char *stmt);
 static	void			 plpgsql_sql_error_callback(void *arg);
+static	char			*parse_string_token(const char *token);
+static	void			 plpgsql_string_error_callback(void *arg);
 static	char			*check_label(const char *yytxt);
 static	void			 check_labels(const char *start_label,
 									  const char *end_label);
@@ -228,8 +230,6 @@ static List				*read_raise_options(void);
 		/*
 		 * Other tokens
 		 */
-%token	T_FUNCTION
-%token	T_TRIGGER
 %token	T_STRING
 %token	T_NUMBER
 %token	T_SCALAR				/* a VAR, RECFIELD, or TRIGARG */
@@ -244,13 +244,9 @@ static List				*read_raise_options(void);
 
 %%
 
-pl_function		: T_FUNCTION comp_optsect pl_block opt_semi
+pl_function		: comp_optsect pl_block opt_semi
 					{
-						yylval.program = (PLpgSQL_stmt_block *)$3;
-					}
-				| T_TRIGGER comp_optsect pl_block opt_semi
-					{
-						yylval.program = (PLpgSQL_stmt_block *)$3;
+						yylval.program = (PLpgSQL_stmt_block *) $2;
 					}
 				;
 
@@ -1403,7 +1399,7 @@ stmt_raise		: K_RAISE lno
 							if (tok == T_STRING)
 							{
 								/* old style message and parameters */
-								new->message = plpgsql_get_string_value();
+								new->message = parse_string_token(yytext);
 								/*
 								 * We expect either a semi-colon, which
 								 * indicates no parameters, or a comma that
@@ -1435,7 +1431,7 @@ stmt_raise		: K_RAISE lno
 
 									if (yylex() != T_STRING)
 										yyerror("syntax error");
-									sqlstatestr = plpgsql_get_string_value();
+									sqlstatestr = parse_string_token(yytext);
 
 									if (strlen(sqlstatestr) != 5)
 										yyerror("invalid SQLSTATE code");
@@ -1778,7 +1774,7 @@ proc_condition	: opt_lblname
 							/* next token should be a string literal */
 							if (yylex() != T_STRING)
 								yyerror("syntax error");
-							sqlstatestr = plpgsql_get_string_value();
+							sqlstatestr = parse_string_token(yytext);
 
 							if (strlen(sqlstatestr) != 5)
 								yyerror("invalid SQLSTATE code");
@@ -2738,6 +2734,49 @@ plpgsql_sql_error_callback(void *arg)
 	errposition(0);
 }
 
+/*
+ * Convert a string-literal token to the represented string value.
+ *
+ * To do this, we need to invoke the core lexer.  To avoid confusion between
+ * the core bison/flex definitions and our own, the actual invocation is in
+ * pl_funcs.c.  Here we are only concerned with setting up the right errcontext
+ * state, which is handled the same as in check_sql_expr().
+ */
+static char *
+parse_string_token(const char *token)
+{
+	char	   *result;
+	ErrorContextCallback  syntax_errcontext;
+	ErrorContextCallback *previous_errcontext;
+
+	/* See comments in check_sql_expr() */
+	Assert(error_context_stack->callback == plpgsql_compile_error_callback);
+
+	previous_errcontext = error_context_stack;
+	syntax_errcontext.callback = plpgsql_string_error_callback;
+	syntax_errcontext.arg = (char *) token;
+	syntax_errcontext.previous = error_context_stack->previous;
+	error_context_stack = &syntax_errcontext;
+
+	result = plpgsql_parse_string_token(token);
+
+	/* Restore former ereport callback */
+	error_context_stack = previous_errcontext;
+
+	return result;
+}
+
+static void
+plpgsql_string_error_callback(void *arg)
+{
+	Assert(plpgsql_error_funcname);
+
+	errcontext("string literal in PL/PgSQL function \"%s\" near line %d",
+			   plpgsql_error_funcname, plpgsql_error_lineno);
+	/* representing the string literal as internalquery seems overkill */
+	errposition(0);
+}
+
 static char *
 check_label(const char *yytxt)
 {

src/pl/plpgsql/src/pl_comp.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/pl_comp.c,v 1.134 2009/02/18 11:33:04 petere Exp $
+ *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/pl_comp.c,v 1.135 2009/04/19 18:52:57 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -261,7 +261,7 @@ do_compile(FunctionCallInfo fcinfo,
 		   bool forValidator)
 {
 	Form_pg_proc procStruct = (Form_pg_proc) GETSTRUCT(procTup);
-	int			functype = CALLED_AS_TRIGGER(fcinfo) ? T_TRIGGER : T_FUNCTION;
+	bool		is_trigger = CALLED_AS_TRIGGER(fcinfo);
 	Datum		prosrcdatum;
 	bool		isnull;
 	char	   *proc_source;
@@ -293,7 +293,7 @@ do_compile(FunctionCallInfo fcinfo,
 	if (isnull)
 		elog(ERROR, "null prosrc");
 	proc_source = TextDatumGetCString(prosrcdatum);
-	plpgsql_scanner_init(proc_source, functype);
+	plpgsql_scanner_init(proc_source);
 
 	plpgsql_error_funcname = pstrdup(NameStr(procStruct->proname));
 	plpgsql_error_lineno = 0;
@@ -359,13 +359,13 @@ do_compile(FunctionCallInfo fcinfo,
 	function->fn_oid = fcinfo->flinfo->fn_oid;
 	function->fn_xmin = HeapTupleHeaderGetXmin(procTup->t_data);
 	function->fn_tid = procTup->t_self;
-	function->fn_functype = functype;
+	function->fn_is_trigger = is_trigger;
 	function->fn_cxt = func_cxt;
 	function->out_param_varno = -1;		/* set up for no OUT param */
 
-	switch (functype)
+	switch (is_trigger)
 	{
-		case T_FUNCTION:
+		case false:
 
 			/*
 			 * Fetch info about the procedure's parameters. Allocations aren't
@@ -564,7 +564,7 @@ do_compile(FunctionCallInfo fcinfo,
 			ReleaseSysCache(typeTup);
 			break;
 
-		case T_TRIGGER:
+		case true:
 			/* Trigger procedure's return type is unknown yet */
 			function->fn_rettype = InvalidOid;
 			function->fn_retbyval = false;
@@ -645,7 +645,7 @@ do_compile(FunctionCallInfo fcinfo,
 			break;
 
 		default:
-			elog(ERROR, "unrecognized function typecode: %u", functype);
+			elog(ERROR, "unrecognized function typecode: %d", (int) is_trigger);
 			break;
 	}
 
@@ -790,7 +790,7 @@ plpgsql_parse_word(const char *word)
 	 * Recognize tg_argv when compiling triggers
 	 * (XXX this sucks, it should be a regular variable in the namestack)
 	 */
-	if (plpgsql_curr_compile->fn_functype == T_TRIGGER)
+	if (plpgsql_curr_compile->fn_is_trigger)
 	{
 		if (strcmp(cp[0], "tg_argv") == 0)
 		{

src/pl/plpgsql/src/pl_funcs.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/pl_funcs.c,v 1.76 2009/02/18 11:33:04 petere Exp $
+ *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/pl_funcs.c,v 1.77 2009/04/19 18:52:57 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -17,6 +17,8 @@
 
 #include <ctype.h>
 
+#include "parser/gramparse.h"
+#include "parser/gram.h"
 #include "parser/scansup.h"
 
 
@@ -459,6 +461,41 @@ plpgsql_convert_ident(const char *s, char **output, int numidents)
 }
 
 
+/*
+ * plpgsql_parse_string_token - get the value represented by a string literal
+ *
+ * We do not make plpgsql's lexer produce the represented value, because
+ * in many cases we don't need it.  Instead this function is invoked when
+ * we do need it.  The input is the T_STRING token as identified by the lexer.
+ *
+ * The result is a palloc'd string.
+ *
+ * Note: this is called only from plpgsql's gram.y, but we can't just put it
+ * there because including parser/gram.h there would cause confusion.
+ */
+char *
+plpgsql_parse_string_token(const char *token)
+{
+	int		ctoken;
+
+	/*
+	 * We use the core lexer to do the dirty work.  Aside from getting the
+	 * right results for escape sequences and so on, this helps us produce
+	 * appropriate warnings for escape_string_warning etc.
+	 */
+	scanner_init(token);
+
+	ctoken = base_yylex();
+
+	if (ctoken != SCONST)
+		elog(ERROR, "unexpected result from base lexer: %d", ctoken);
+
+	scanner_finish();
+
+	return base_yylval.str;
+}
+
+
 /*
  * Statement type as a string, for use in error messages etc.
  */

src/pl/plpgsql/src/plpgsql.h

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/plpgsql.h,v 1.110 2009/04/09 02:57:53 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/pl/plpgsql/src/plpgsql.h,v 1.111 2009/04/19 18:52:57 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -650,7 +650,7 @@ typedef struct PLpgSQL_function
 	Oid			fn_oid;
 	TransactionId fn_xmin;
 	ItemPointerData fn_tid;
-	int			fn_functype;
+	bool		fn_is_trigger;
 	PLpgSQL_func_hashkey *fn_hashkey;	/* back-link to hashtable key */
 	MemoryContext fn_cxt;
 
@@ -880,6 +880,7 @@ extern void plpgsql_ns_rename(char *oldname, char *newname);
  * ----------
  */
 extern void plpgsql_convert_ident(const char *s, char **output, int numidents);
+extern char *plpgsql_parse_string_token(const char *token);
 extern const char *plpgsql_stmt_typename(PLpgSQL_stmt *stmt);
 extern void plpgsql_dumptree(PLpgSQL_function *func);
 
@@ -894,8 +895,7 @@ extern int	plpgsql_yylex(void);
 extern void plpgsql_push_back_token(int token);
 extern void plpgsql_yyerror(const char *message);
 extern int	plpgsql_scanner_lineno(void);
-extern void plpgsql_scanner_init(const char *str, int functype);
+extern void plpgsql_scanner_init(const char *str);
 extern void plpgsql_scanner_finish(void);
-extern char *plpgsql_get_string_value(void);
 
 #endif   /* PLPGSQL_H */