shithub: mc

Info  •  Files  •  Log  •  Branches

Add comments.

--- a/compile.myr

+++ b/compile.myr

@@ -34,6 +34,12 @@

 	`Fail status

 ;;

+/* Compiles a pattern into a regex */

+const compile = {pat

+	-> regexcompile(std.zalloc(), pat)

+}

+

+/* Compiles a pattern into a debug regex. This can be verbose. */

 const dbgcompile = {pat

 	var re

@@ -42,10 +48,7 @@

 	-> regexcompile(re, pat)

 }

-const compile = {pat

-	-> regexcompile(std.zalloc(), pat)

-}

-

+/* compiles a pattern into an allocated regex */

 const regexcompile = {re, pat

 	re.pat = pat

 	re.nmatch = 1 /* whole match */

@@ -64,6 +67,7 @@

 	-> `std.Failure (`Noimpl)

 }

+/* generates bytecode from an AST */

 const gen = {re, t

 	var m

@@ -92,7 +96,15 @@

 	-> re.proglen

 }

-const genrange = {re, lo, hi

+/*

+ converts a codepoint range spanning multiple utf8 byte lenghts into a

+ set of utf8 ranges. Eg:

+ 	[0x00-0x2000]  => [0x00-0x7F]|[0xC2-0xDF][0x80-0x8F]

+*/

+Const genrange = {re, lo, hi

+	/* the transitions between different char lenghts for unicode

+	   characters, needed so that we know how to generate the

+	   different size categories */

 	var charrng = [

 		0,

 		0x80,

@@ -138,6 +150,7 @@

 	-> re.proglen

 }

+/* calculates the forward jump distance for a utf8 character range */

 const jmpdist = {n

 	var d

 	var i

@@ -149,6 +162,7 @@

 	-> d

 }

+/* generates an alternation */

 const genalt = {re, l, r

 	var alt

 	var jmp

@@ -168,6 +182,7 @@

 	-> re.proglen

 }

+/* generates a repetition operator */

 const genstar = {re, rep

 	var alt

 	var jmp

@@ -187,6 +202,7 @@

 	-> re.proglen

 }

+/* generates a question mark operator */

 const genquest = {re, q

 	var alt

 	var l0

@@ -199,6 +215,7 @@

 	-> re.proglen

 }

+/* generates a single char match */

 const genchar = {re, c

 	var b : byte[4]

 	var n

@@ -211,6 +228,7 @@

 	-> re.proglen

 }

+/* appends an instructon to an re program */

 const append = {re, insn

 	if re.proglen == re.prog.len

 		re.prog = std.slgrow(re.prog, std.max(1, 2*re.proglen))

@@ -220,6 +238,7 @@

 	-> re.proglen

 }

+/* instruction dump */

 const idump = {re

 	var i

@@ -246,6 +265,7 @@

 	;;

 }

+/* AST dump */

 const dump = {re, t, indent

 	var i

@@ -293,6 +313,7 @@

 	;;

 }

+/* parses an expression */

 const parse = {re

 	match altexpr(re)

 	| `Some t: