shithub: sl

Info  •  Files  •  Log  •  Branches

add «verbatim strings»

--- a/README.md

+++ b/README.md

@@ -24,21 +24,20 @@

 Some of the changes from the original include:

  * aggressive clean up, removal, renaming and refactoring

- * seamless bignums

+ * bignums

+ * previously available (but not merged) patches from the community and [Julia](https://github.com/JuliaLang/julia) are applied

  * `[` and `]`, `{` and `}` are synonyms to `(` and `)`

+ * `«` and `»` for verbatim strings

  * `define` → `def`, `define-macro` → `defmacro`

  * `λ` as a shorthand for `lambda`

- * `t`/`T` instead of `#t`/`#T` and `nil` instead of `#f`

+ * `T` instead of `#t`/`#T` and `NIL` instead of `#f`

+ * `c***r` of empty list returns empty list

  * docstrings - `(def (f ...) "Docs here" ...)` and `(help ...)`

  * automatic gensyms for macros (`blah#`) at read time

  * proper `(void)` and `void?`

  * better error reporting - disassembly at the current instruction, location of syntax errors

- * some of the previously available (but not merged) patches from the community and [Julia](https://github.com/JuliaLang/julia) are applied

- * `c***r` of empty list returns empty list

  * "boot" image is built into the executable

  * vm opcode definitions and tables are generated from a single file

- * fixed bootstrap (makes it work properly when opcodes change)

- * built-in symbols aren't constants and can be redefined

 Two ways to learn about more changes:

--- a/src/read.c

+++ b/src/read.c

@@ -8,7 +8,7 @@

 	TOK_BQ, TOK_COMMA, TOK_COMMAAT, TOK_COMMADOT,

 	TOK_SHARPDOT, TOK_LABEL, TOK_BACKREF, TOK_SHARPQUOTE, TOK_SHARPOPEN,

 	TOK_OPENB, TOK_CLOSEB, TOK_SHARPSYM, TOK_GENSYM, TOK_DOUBLEQUOTE,

-	TOK_OPENC, TOK_CLOSEC,

+	TOK_OPENC, TOK_CLOSEC, TOK_VERBATIM,

 };

 #define PAtLoc "at %"PRIu32":%"PRIu32

@@ -199,7 +199,8 @@

 static u32int

 peek(Rctx *ctx)

 {

-	char c, *end;

+	u8int c;

+	char *end;

 	sl_fx x;

 	int ch, base;

@@ -227,7 +228,10 @@

 		ctx->toktype = TOK_BQ;

 	else if(c == '"')

 		ctx->toktype = TOK_DOUBLEQUOTE;

-	else if(c == '#'){

+	else if(c == 0xc2 && ios_peekc(RS) == 0xab){ // «

+		ctx->toktype = TOK_VERBATIM;

+		ios_getc(RS);

+	}else if(c == '#'){

 		c = ch = ios_getc(RS);

 		if(ch == IOS_EOF)

 			parse_error(&ctx->loc, "invalid read macro");

@@ -440,12 +444,52 @@

 }

 static sl_v

+read_verbatim(Rctx *ctx)

+{

+	char *temp;

+	usize i = 0, sz = sizeof(ctx->buf);

+	char *buf = ctx->buf;

+

+	while(1){

+		if(i >= sz-UTFmax){ // -UTFmax: leaves room for longest utf8 sequence

+			sz *= 2;

+			if(buf == ctx->buf){

+				if((temp = MEM_ALLOC(sz)) != nil)

+					memcpy(temp, ctx->buf, i);

+			}else

+				temp = MEM_REALLOC(buf, sz);

+			if(temp == nil){

+				if(buf == ctx->buf)

+					MEM_FREE(buf);

+				parse_error(&ctx->loc, "out of memory reading verbatim string");

+			}

+			buf = temp;

+		}

+		int c = ios_getc(RS);

+		if(c == IOS_EOF){

+			if(buf != ctx->buf)

+				MEM_FREE(buf);

+			parse_error(&ctx->loc, "unexpected end of input in verbatim string");

+		}

+		if(c == 0xc2 && ios_peekc(RS) == 0xbb){ // »

+			ios_getc(RS);

+			break;

+		}

+		buf[i++] = c;

+	}

+	sl_v s = cvalue_str(i);

+	memcpy(cvalue_data(s), buf, i);

+	if(buf != ctx->buf)

+		MEM_FREE(buf);

+	return s;

+}

+

+static sl_v

 read_str(Rctx *ctx)

 {

 	char *buf, *temp;

 	char eseq[10];

 	usize i = 0, j, sz, ndig;

-	int c;

 	sl_v s;

 	Rune r = 0;

@@ -466,7 +510,7 @@

 			}

 			buf = temp;

 		}

-		c = ios_getc(RS);

+		int c = ios_getc(RS);

 		if(c == IOS_EOF){

 			if(buf != ctx->buf)

 				MEM_FREE(buf);

@@ -708,6 +752,8 @@

 		return *pv;

 	case TOK_DOUBLEQUOTE:

 		return read_str(ctx);

+	case TOK_VERBATIM:

+		return read_verbatim(ctx);

 	case TOK_CLOSE:

 		parse_error(&ctx->loc, "unexpected ')'");

 	case TOK_CLOSEB: