scan.l

PostgreSQL 8.1.4的源码 适用于Linux下的开源数据库系统

			}
<IN_STRING>[^'\\]+	{ }
<IN_STRING><<EOF>>	{
				plpgsql_error_lineno = start_lineno;
				ereport(ERROR,
						(errcode(ERRCODE_DATATYPE_MISMATCH),
						 errmsg("unterminated string")));
			}

{dolqdelim}		{
			  start_lineno = plpgsql_scanner_lineno();
			  start_charpos = yytext;
			  dolqstart = pstrdup(yytext);
			  BEGIN(IN_DOLLARQUOTE);
			}
<IN_DOLLARQUOTE>{dolqdelim} {
			  if (strcmp(yytext, dolqstart) == 0)
			  {
					pfree(dolqstart);
					/* tell plpgsql_get_string_value it is a dollar quote */
					dolqlen = yyleng;
					/* adjust yytext/yyleng to describe whole string token */
					yyleng += (yytext - start_charpos);
					yytext = start_charpos;
					BEGIN(INITIAL);
					return T_STRING;
			  }
			  else
			  {
					/*
					 * When we fail to match $...$ to dolqstart, transfer
					 * the $... part to the output, but put back the final
					 * $ for rescanning.  Consider $delim$...$junk$delim$
					 */
					yyless(yyleng-1);
			  }
			}
<IN_DOLLARQUOTE>{dolqinside} { }
<IN_DOLLARQUOTE>.	{ /* needed for $ inside the quoted text */ }
<IN_DOLLARQUOTE><<EOF>>	{ 
				plpgsql_error_lineno = start_lineno;
				ereport(ERROR,
						(errcode(ERRCODE_DATATYPE_MISMATCH),
						 errmsg("unterminated dollar-quoted string")));
			}

    /* ----------
     * Any unmatched character is returned as is
     * ----------
     */
.			{ return yytext[0];			}

%%


/*
 * This is the yylex routine called from outside. It exists to provide
 * a pushback facility, as well as to allow us to parse syntax that
 * requires more than one token of lookahead.
 */
int
plpgsql_yylex(void)
{
	int cur_token;

	if (have_pushback_token)
	{
		have_pushback_token = false;
		cur_token = pushback_token;
	}
	else if (have_lookahead_token)
	{
		have_lookahead_token = false;
		cur_token = lookahead_token;
	}
	else
		cur_token = yylex();

	/* Do we need to look ahead for a possible multiword token? */
	switch (cur_token)
	{
		/* RETURN NEXT must be reduced to a single token */
		case K_RETURN:
			if (!have_lookahead_token)
			{
				lookahead_token = yylex();
				have_lookahead_token = true;
			}
			if (lookahead_token == K_NEXT)
			{
				have_lookahead_token = false;
				cur_token = K_RETURN_NEXT;
			}
			break;

		default:
			break;
	}

	return cur_token;
}

/*
 * Push back a single token to be re-read by next plpgsql_yylex() call.
 */
void
plpgsql_push_back_token(int token)
{
	if (have_pushback_token)
		elog(ERROR, "cannot push back multiple tokens");
	pushback_token = token;
	have_pushback_token = true;
}

/*
 * Report a syntax error.
 */
void
plpgsql_yyerror(const char *message)
{
	const char *loc = yytext;
	int			cursorpos;

	plpgsql_error_lineno = plpgsql_scanner_lineno();

	/* in multibyte encodings, return index in characters not bytes */
	cursorpos = pg_mbstrlen_with_len(scanbuf, loc - scanbuf) + 1;

	if (*loc == YY_END_OF_BUFFER_CHAR)
	{
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 /* translator: %s is typically "syntax error" */
				 errmsg("%s at end of input", message),
				 internalerrposition(cursorpos),
				 internalerrquery(scanstr)));
	}
	else
	{
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 /* translator: first %s is typically "syntax error" */
				 errmsg("%s at or near \"%s\"", message, loc),
				 internalerrposition(cursorpos),
				 internalerrquery(scanstr)));
	}
}

/*
 * Get the line number at which the current token ends.  This substitutes
 * for flex's very poorly implemented yylineno facility.
 *
 * We assume that flex has written a '\0' over the character following the
 * current token in scanbuf.  So, we just have to count the '\n' characters
 * before that.  We optimize this a little by keeping track of the last
 * '\n' seen so far.
 */
int
plpgsql_scanner_lineno(void)
{
	const char *c;

	while ((c = strchr(cur_line_start, '\n')) != NULL)
	{
		cur_line_start = c + 1;
		cur_line_num++;
	}
	return cur_line_num;
}

/*
 * Called before any actual parsing is done
 *
 * Note: the passed "str" must remain valid until plpgsql_scanner_finish().
 * Although it is not fed directly to flex, we need the original string
 * to cite in error messages.
 */
void
plpgsql_scanner_init(const char *str, int functype)
{
	Size	slen;

	slen = strlen(str);

	/*
	 * Might be left over after ereport()
	 */
	if (YY_CURRENT_BUFFER)
		yy_delete_buffer(YY_CURRENT_BUFFER);

	/*
	 * Make a scan buffer with special termination needed by flex.
	 */
	scanbuf = palloc(slen + 2);
	memcpy(scanbuf, str, slen);
	scanbuf[slen] = scanbuf[slen + 1] = YY_END_OF_BUFFER_CHAR;
	scanbufhandle = yy_scan_buffer(scanbuf, slen + 2);

	/* Other setup */
	scanstr = str;

    scanner_functype = functype;
    scanner_typereported = false;

	have_pushback_token = false;
	have_lookahead_token = false;

	cur_line_start = scanbuf;
	cur_line_num = 1;

	/*----------
	 * Hack: skip any initial newline, so that in the common coding layout
	 *		CREATE FUNCTION ... AS '
	 *			code body
	 *		' LANGUAGE 'plpgsql';
	 * we will think "line 1" is what the programmer thinks of as line 1.
	 *----------
	 */
    if (*cur_line_start == '\r')
        cur_line_start++;
    if (*cur_line_start == '\n')
        cur_line_start++;

	BEGIN(INITIAL);
}

/*
 * Called after parsing is done to clean up after plpgsql_scanner_init()
 */
void
plpgsql_scanner_finish(void)
{
	yy_delete_buffer(scanbufhandle);
	pfree(scanbuf);
}

/*
 * Called after a T_STRING token is read to get the string literal's value
 * as a palloc'd string.  (We make this a separate call because in many
 * scenarios there's no need to get the decoded value.)
 *
 * Note: we expect the literal to be the most recently lexed token.  This
 * would not work well if we supported multiple-token pushback or if 
 * plpgsql_yylex() wanted to read ahead beyond a T_STRING token.
 */
char *
plpgsql_get_string_value(void)
{
	char	   *result;
	const char *cp;
	int			len;

	if (dolqlen > 0)
	{
		/* Token is a $foo$...$foo$ string */
		len = yyleng - 2 * dolqlen;
		Assert(len >= 0);
		result = (char *) palloc(len + 1);
		memcpy(result, yytext + dolqlen, len);
		result[len] = '\0';
	}
	else if (*yytext == 'E' || *yytext == 'e')
	{
		/* Token is an E'...' string */
		result = (char *) palloc(yyleng + 1);	/* more than enough room */
		len = 0;
		for (cp = yytext + 2; *cp; cp++)
		{
			if (*cp == '\'')
			{
				if (cp[1] == '\'')
					result[len++] = *cp++;
				/* else it must be string end quote */
			}
			else if (*cp == '\\')
			{
				if (cp[1] != '\0')	/* just a paranoid check */
					result[len++] = *(++cp);
			}
			else
				result[len++] = *cp;
		}
		result[len] = '\0';
	}
	else
	{
		/* Token is a '...' string */
		result = (char *) palloc(yyleng + 1);	/* more than enough room */
		len = 0;
		for (cp = yytext + 1; *cp; cp++)
		{
			if (*cp == '\'')
			{
				if (cp[1] == '\'')
					result[len++] = *cp++;
				/* else it must be string end quote */
			}
			else if (*cp == '\\')
			{
				if (cp[1] != '\0')	/* just a paranoid check */
					result[len++] = *(++cp);
			}
			else
				result[len++] = *cp;
		}
		result[len] = '\0';
	}
	return result;
}