From b52281665b013309451366c3e41cca6632bb6b82 Mon Sep 17 00:00:00 2001
From: Case Duckworth
Date: Wed, 3 Apr 2024 22:47:56 -0500
Subject: Read comments and # literals

---
 read.lua | 78 +++++++++++++++++++++++++++++++++++++++++++++++++++-------------
 1 file changed, 62 insertions(+), 16 deletions(-)

diff --git a/read.lua b/read.lua
index 8a154a1..6b223f8 100644
--- a/read.lua
+++ b/read.lua
@@ -43,7 +43,10 @@ function m.inport (source, kind)
 					end
 					tok, toktype, self.line =
 						m.scan(self.line)()
-					if tok then return tok, toktype end
+					if tok ~= nil then
+						print(tok, toktype)
+						return tok, toktype
+					end
 				end
 			end,
 	}
@@ -68,20 +71,61 @@ function m.tochars (s)
 end
 
 --- Consumers
--- These take a table of characters (cs) and return a token and the rest of the
--- chars
+-- These take a table of characters (cs) and return:
+-- a token, its type, and the rest of the characters
+
+local token_separator = "[^%s#()\"'`,@;]"
 
 local function consume_token (cs)
 	local token = {}
-	while #cs > 0 and cs[1]:match("[^%s()\"#'`,@;]") do
+	while #cs > 0 and cs[1]:match(token_separator) do
 		table.insert(token, pop(cs))
 	end
-	return table.concat(token), cs
+	return table.concat(token), "symbol", cs
 end
 
 local function consume_whitespace (cs)
 	while #cs > 0 and cs[1]:match("%s") do pop(cs) end
-	return nil, cs
+	return nil, nil, cs
+end
+
+local function consume_comment (cs)
+	local comment = {}
+	repeat table.insert(comment, pop(cs))
+	until #cs == 0 or cs[1]:match("\n")
+	return table.concat(comment), "comment", cs
+end
+
+local function consume_literal (cs)
+	-- whitespace and parantheses character literals.
+	-- reverse the match test b/c it's already a complement
+	if cs[2] == "\\" and not cs[3]:match(token_separator) then
+		return type.character(cs[3])
+	end
+	pop(cs) -- discard '#'
+	local token, value, cs = consume_token(cs) -- todo: vectors #(...)
+	token = "#" .. token -- put '#' back
+	-- tokens!
+	if token == "#t" or token == "#true" then -- booleans
+		value = true
+	elseif token == "#f" or token == "#false" then
+		value = false
+		--[[ To actually *read* nil, I need to change ports from
+			returning `nil' on eof to an `eof' symbol, i think.
+
+			elseif token == "#n" or token == "#nil" then
+			value = nil
+		--]]
+	elseif token == "#\\space" then -- characters
+		value = type.character(" ")
+	elseif token == "#\\newline" then
+		value = type.character("\n")
+	elseif token:match("^#\\") then
+		value = type.character(token:sub(3))
+	else
+		error("Bad literal notation: " .. token)
+	end
+	return value, "literal", cs
 end
 
 --- Reading from a port
@@ -91,7 +135,7 @@ m.readtable = {
 	[")"] = function (cs) return pop(cs), "close", cs end,
 	["'"] = function (cs) return pop(cs), "quote", cs end,
 	["`"] = function (cs) return pop(cs), "quote", cs end,
-	[","] =
+	[","] = -- unquote
 		function (cs)
 			pop(cs) -- remove ','
 			if cs[1] == "@" then
@@ -101,8 +145,12 @@ m.readtable = {
 				return ",", "quote", cs
 			end
 		end,
+	[";"] = consume_comment,
+	["#"] = consume_literal,
 }
 
+--- TODO: Figure out how to read #f and #n properly
+
 -- Return an iterator over a character table, so you can do:
 -- for token, chars in scan(cs) do ... end
 function m.scan (cs)
@@ -115,12 +163,10 @@ function m.scan (cs)
 				token, toktype, cs = m.readtable[cs[1]](cs)
 				return token, toktype, cs
 			elseif cs[1]:match("%s") then
-				_, cs = consume_whitespace(cs)
-				return nil, nil, cs
-				-- return nil, cs
+				return consume_whitespace(cs)
 			elseif cs[1]:match("[%d.+-]") then
 				-- numbers, +, -, ., ...
-				local token, cs = consume_token(cs)
+				local token, _, cs = consume_token(cs)
 				if token:match("[-+]") or token == "..." then
 					return token, "symbol", cs
 				elseif token == "." then
@@ -131,8 +177,7 @@ function m.scan (cs)
 					return n, "number", cs
 				end
 			else
-				token, cs = consume_token(cs)
-				return token, "symbol", cs
+				return consume_token(cs)
 			end
 		end
 	end
@@ -143,7 +188,7 @@ function m.readchar (port)
 		local ch = pop(port.line)
 		return ch
 	else
-		return port.file.read(1)
+		return port.file and port.file.read(1)
 	end
 end
 
@@ -163,11 +208,12 @@ m.readmacros = {
 			table.insert(Q, m.read(port))
 			return t.list(Q)
 		end,
+	comment = function () return nil end,
 }
 
 function m.read (port)
 	local function read_ahead (tok, toktype)
-		if not tok then error("Unexpected EOF") end
+		if tok == nil then error("Unexpected EOF") end
 		if toktype == "open" then
 			local L = {}
 			while true do
@@ -192,7 +238,7 @@ function m.read (port)
 	end
 	-- body of read
 	local tok1, toktype1 = port:next_token()
-	if not tok1 then return nil
+	if tok1 == nil then return nil
 	else return read_ahead(tok1, toktype1)
 	end
 end
-- 
cgit 1.4.1-21-gabe81