initial commit

+ Now to do the actual hard parts, heh

1 files changed, 160 insertions, 0 deletions

diff --git a/src/lexer.l b/src/lexer.l
new file mode 100644
index 0000000..75e2b09
--- /dev/null
+++ b/src/lexer.l
@@ -0,0 +1,160 @@
+/* SPDX-License-Identifier: copyleft-next-0.3.1 */
+/* Copyright 2023 Kim Kuparinen < kimi.h.kuparinen@gmail.com > */
+
+%option reentrant noyywrap nounput noinput nodefault
+%{
+#define FROM_LEXER
+#include <qbt/parser.h>
+#include <qbt/debug.h>
+
+static void update_yylloc(struct parser *parser, YYLTYPE *lloc, const char *text)
+{
+	(void)parser;
+
+	lloc->first_line = lloc->last_line;
+	lloc->first_column = lloc->last_column;
+
+	for (size_t i = 0; text[i] != 0; ++i) {
+		if (text[i] == '\n') {
+			lloc->last_line++;
+			/* flex uses 1 based indexing */
+			lloc->last_column = 1;
+		} else {
+			lloc->last_column++;
+		}
+	}
+}
+
+#define YY_USER_ACTION update_yylloc(parser, yylloc, yytext);
+%}
+
+HEX		0[xX][0-9a-fA-F]+
+DEC		-?[0-9]+
+OCT		0[0-8]+
+BIN		0b[0-1]+
+
+INT		{HEX}|{DEC}|{OCT}|{BIN}
+
+HEXF		[+-]?0[xX][0-9a-fA-F]+([pP][+-]?[0-9]+)
+DECF		[+-]?[0-9]+[.]([eE]?[+-]?[0-9]+)?[fF]?
+
+ID		[_a-zA-Z][_a-zA-Z0-9]*
+
+STRING		\"(\\.|[^"\\])*\"
+
+%x SC_COMMENT
+
+%%
+"//".* {/* skip line comments */}
+
+"/*"	{BEGIN(SC_COMMENT);}
+<SC_COMMENT>{
+	"/*"	{++parser->comment_nesting;}
+	"*"+"/"	{
+		if (parser->comment_nesting)
+			--parser->comment_nesting;
+		else
+			BEGIN(INITIAL);
+	}
+
+	"*"+ {}
+	[^/*\n]+ {}
+	[/] {}
+	\n {}
+}
+
+"("		{return LEXLPAREN;}
+")"		{return LEXRPAREN;}
+"{"		{return LEXLBRACE;}
+"}"		{return LEXRBRACE;}
+","		{return LEXCOMMA;}
+":"		{return LEXCOLON;}
+"!"		{return LEXBANG;}
+
+"+"		{return LEXPLUS;}
+"-"		{return LEXMINUS;}
+"*"		{return LEXSTAR;}
+"/"		{return LEXDIV;}
+"%"		{return LEXREM;}
+
+'[^'\\]'	{
+	/* regular character constant, 'a' */
+	yylval->integer = yytext[1];
+	return INT;
+}
+
+'\\x[0-9a-fA-F]+' {
+	/* hex character constant, '\xef' */
+	/* handling is slightly different from C, here it's more or less just
+	 * another way to specify a hex integer */
+	yylval->integer = strtoll(yytext + 3, NULL, 16);
+	return INT;
+}
+
+'\\[0-8]+' {
+	/* octal character constant, '\033' */
+	yylval->integer = strtoll(yytext + 2, NULL, 8);
+	return INT;
+}
+
+'\\b[01]+' {
+	/* binary character constant, '\b101' */
+	yylval->integer = strtoll(yytext + 3, NULL, 2);
+	return INT;
+}
+
+'\\.' {
+	/* escaped character constant */
+	yylval->integer = match_escape(yytext[2]);
+	return INT;
+}
+
+"&"		{return LEXAND;}
+
+"="		{return LEXASSIGN;}
+"<"		{return LEXLT;}
+">"		{return LEXGT;}
+"<="		{return LEXLE;}
+">="		{return LEXGE;}
+"!="		{return LEXNE;}
+"=="		{return LEXEQ;}
+";"		{return LEXSEMI;}
+
+"=>"		{return LEXFATARROW;}
+"->"		{return LEXTHINARROW;}
+">>"		{return LEXTO;}
+"<<"		{return LEXFROM;}
+
+"i9"		{return LEXI9;}
+"i27"		{return LEXI27;}
+
+{STRING} {
+	/* seems risky, I know, but letting the parser choose when to allocate a
+	 * new string seems to help with syntax error cleanup */
+	yylval->str = yytext;
+	return STRING;
+}
+
+{INT} {
+	yylval->integer = strtoll(yytext, 0, 0);
+	return INT;
+}
+
+{ID} {
+	yylval->str = yytext;
+	return ID;
+}
+
+
+[[:space:]]+	{/* skip whitespace */}
+
+. {
+	struct src_issue issue;
+	issue.level = SRC_ERROR;
+	issue.loc = src_loc(*yylloc);
+	issue.fctx.fbuf = parser->buf;
+	issue.fctx.fname = parser->fname;
+	src_issue(issue, "Unexpected token: %s", yytext);
+	parser->failed = true;
+}
+%%