pret-pas/pret-pas.lua

if not modules then modules = { } end modules ['pret-pas'] = {
    version   = 1.0,
    comment   = "custom pretty printer for Pascal code",
    author    = "Stefan Müller, Chemnitz DE",
    copyright = "Stefan Müller",
    license   = "see context related readme files"
}

-- The code formatting is adapted from Lazarus, a Free Pascal RAD IDE, which
-- should be quite similar to the Delphi style. Please note that comments are
-- not typeset with bold fontface but slanted, which improves output when not
-- using colors.
-- http://lazarus.freepascal.org/

local visualizer = buffers.newvisualizer("pas")

-- The list of reserved words is taken from 
-- http://www.freepascal.org/docs-html/ref/refse3.html
visualizer.reservedwords = {
	-- Turbo Pascal
	"absolute", "and", "array", "asm", "begin", "case", "const", "constructor",
	"destructor", "div", "do", "downto", "else", "end", "file", "for",
	"function", "goto", "if", "implementation", "in", "inherited", "inline",
	"interface", "label", "mod", "nil", "not", "object", "of", "on", "operator",
	"or", "packed", "procedure", "program", "record", "reintroduce", "repeat",
	"self", "set", "shl", "shr", "string", "then", "to", "type", "unit",
	"until", "uses", "var", "while", "with", "xor",

	-- Free Pascal
	-- these are not bold type (keeping them, just in case)
	-- "dispose", "exit", "false", "new", "true",

	-- Object Pascal
    "as", "class", "dispinterface", "except", "exports", "finalization",
	"finally", "initialization", "inline", "is", "library", "on", "out",
	"packed", "property", "raise", "resourcestring", "threadvar", "try",

	-- Modifiers
	-- some of these are only bold in specific places (in the following, this is
	-- deliberately ignored)
	"absolute", "abstract", "alias", "assembler", "cdecl", "cppdecl", "default",
	"export", "external", "far", "far16", "forward", "index", "local", "name", 
	"near", "nostackframe", "oldfpccall", "override", "pascal", "private",
	"protected", "public", "published", "read", "register", "reintroduce",
	"safecall", "softfloat", "stdcall", "virtual", "write"
}

local known_words = { }

for k,v in next, visualizer.reservedwords do
    known_words[v] = k
end

local colors = {
    "prettyone", -- red: compiler directive, symbol
    "prettytwo", -- green: assembler (dark green)
    "prettythree", -- blue: comment, number (dark blue)
    "prettyfour", -- yellow
}

local states = {
    ['"']=1, ["'"]=1, ["[["] = 1, ["]]"] = 1,
    ['+']=1, ['-']=1, ['*']=1, ['/']=1, ['%']=1, ['^']=1,
    ["("] = 3, [")"] = 3, ["["] = 3, ["]"] = 3,
    ['--']=4,
}

local reserved_style = "\{\\bf "
local comment_style = "\{\\sl "

local function flush_pas_word(word)
    if word then
        local id = known_words[string.lower(word)]
        if id then
			tex.sprint(tex.ctxcatcodes, reserved_style)
            tex.write(word)
			tex.sprint(tex.ctxcatcodes, "\}")
        else
            tex.write(word)
        end
    end
end

local function flush_whatever(str)
	if str then
		for c in string.utfcharacters(str) do
			if c == " " then
				tex.sprint(tex.ctxcatcodes, "\\obs")
			elseif c == "\t" then
				tex.sprint(tex.ctxcatcodes, "\\obs")
				if buffers.visualizers.enabletab then
					tex.sprint(tex.ctxcatcodes,rep("\\obs ", buffers.visualizers.tablength))
				end
			else
				tex.write(c);
			end
		end
	end
end

local inlongcomment, inlongcomment_alt, incompdirec, inasm = false, 0, false, false

function visualizer.reset()
    inlongcomment, inlongcomment_alt, incompdirec, inasm = false, 0, false, false
end

--[[local function written(state, c, i)
    if c == " " then
        state = buffers.finishstate(state)
        tex.sprint(tex.ctxcatcodes,"\\obs")
    elseif c == "\t" then
        state = buffers.finishstate(state)
        tex.sprint(tex.ctxcatcodes,"\\obs")
        if buffers.visualizers.enabletab then
            tex.sprint(tex.ctxcatcodes,rep("\\obs ",i%buffers.visualizers.tablength))
        end
    else
        tex.write(c)
    end
    return state, 0
end]]

function visualizer.flush_line(str, nested)
    local state = 0
	local incomment, instring = false, false
	--local code, comment = nil, nil
    buffers.currentcolors = colors

--[[
	print("SEARCHING FOR IDENTIFIERS IN: " .. str)
	local remain = str
	local identifier
	repeat
		identifier = nil
		identifier, remain = string.match(remain, "^[^%a_]*([%a_]+[%a%d_]*)(.*)$")
		if identifier then
			print("IDENTIFIER: " .. identifier)
			--print("REMAIN: " .. remain)
		end
	until not identifier
]]
	--TODO: handle inlongcomment etc.
--[[
	if inlongcomment then
		comment, code = string.match(str,"^(.-})(.*)$")
		if not comment then
			comment = str
		end
		state = buffers.changestate(3, state)
		tex.sprint(tex.ctxcatcodes, "\{\\bf")
		flush_whatever(comment)
		tex.sprint(tex.ctxcatcodes, "\}")
		if code then
			inlongcomment = false
		else
			code = ""
		end
	else
		code = str
	end
]]
	if inlongcomment or (inlongcomment_alt == 2) or incompdirec then
		incomment = true
		if incompdirec then
			state = buffers.changestate(1, state)
		else
			state = buffers.changestate(3, state)
		end
		tex.sprint(tex.ctxcatcodes, comment_style)
	end
	
	local c, word = nil, nil
	for nextc in string.utfcharacters(str .. " ") do
		if c then
			if instring then
				if c == "'" then
					-- string ends
					tex.write(c)
					state = buffers.finishstate(state)
					instring = false
				else
					-- inside the string
					flush_whatever(c)
				end
			elseif incomment then
				if ((inlongcomment or incompdirec) and (c == "}"))
				or (inlongcomment_alt == 1) then
					-- long comment/(alternative)/compiler directive ends
					tex.write(c)
					tex.sprint(tex.ctxcatcodes, "\}")
					state = buffers.finishstate(state)
					incompdirec = false
					inlongcomment = false
					inlongcomment_alt = 0
					incomment = false
				elseif (inlongcomment_alt == 2) and (c == "*") and (nextc == ")") then
					-- long comment (alternative) ends after nextc
					tex.write(c)
					inlongcomment_alt = 1
				else
					-- inside the comment
					flush_whatever(c)
				end
			elseif string.find(c, "^[%a_]$") then
				-- char belongs to identifier
				if word then
					word = word .. c
				else
					word = c
				end
			elseif string.find(c, "^[%d]$") then
				if word and (#word > 1) then
					-- number, that belongs to identifier
					word = word .. c
				else
					-- number
					state = buffers.changestate(3, state)
					tex.write(c)
				end
			else
				state = buffers.finishstate(state)
				-- identifier complete, check if it's a reserved word and flush
				flush_pas_word(word)
				word = nil
				if c == " " then
					tex.sprint(tex.ctxcatcodes, "\\obs")
				elseif c == "\t" then
					tex.sprint(tex.ctxcatcodes, "\\obs")
					if buffers.visualizers.enabletab then
						tex.sprint(tex.ctxcatcodes,rep("\\obs ", buffers.visualizers.tablength))
					end
				elseif c == "'" then
					-- string starts
					instring = true
					state = buffers.changestate(3, state)
					tex.write(c)
				elseif (c == "/") and (nextc == "/") then
					-- one-line comment starts
					incomment = true
					state = buffers.changestate(3, state)
					tex.sprint(tex.ctxcatcodes, comment_style)
					tex.write(c)
				elseif c == "{" then
					incomment = true
					if nextc == "$" then
						-- compiler directive starts
						incompdirec = true
						state = buffers.changestate(1, state)
					else
						-- long comment starts
						inlongcomment = true
						state = buffers.changestate(3, state)
					end
					tex.sprint(tex.ctxcatcodes, comment_style)
					tex.write(c)
				elseif (c == "(") and (nextc == "*") then 
					-- long comment (alternative) starts
					incomment = true
					inlongcomment_alt = 2
					state = buffers.changestate(3, state)
					tex.sprint(tex.ctxcatcodes, comment_style)
					tex.write(c)
				-- TODO: asm
				else
					-- symbol
					state = buffers.changestate(1, state)
					tex.write(c)
				end
			end
		end
		c = nextc
	end
	if not incomment then
		state = buffers.finishstate(state)
	end
	-- maybe something left, check if it's a reserved word and flush
	flush_pas_word(word)
	if incomment then
		-- end the comment-line
		tex.sprint(tex.ctxcatcodes, "\}")
	end
	state = buffers.finishstate(state)	
	
--[[
    local code, comment = string.match(str,"^(.-)%-%-%[%[(.*)$")
    if comment then
        -- process the code and then flush the comment
    elseif incomment then
        comment, code = string.match(str,"^(.-)%]%](.*)$")
        if comment then
            -- flush the comment and then process the code
            for c in string.utfcharacters(comment) do
                if c == " " then tex.sprint(tex.ctxcatcodes,"\\obs") else tex.write(c) end
            end
            state = buffers.changestate(states['--'], state)
            tex.write("KKK")
            state = buffers.finishstate(state)
            incomment = false
        else
            for c in string.utfcharacters(str) do
                if c == " " then tex.sprint(tex.ctxcatcodes,"\\obs") else tex.write(c) end
            end
        end
        comment = nil
    else
        code = str
    end
    if code and code ~= "" then
        local pre, post = string.match(code,"^(.-)%-%-(.*)$")
        if pre then
            code = pre
        end
        local p, s, i = nil, nil, 0
        for c in string.utfcharacters(code) do
            i = i + 1
            if instr then
                if p then
                    tex.write(p)
                    p = nil
                end
                if c == s then
                    if inesc then
                        tex.write(c)
                        inesc = false
                    else
                        state = buffers.changestate(states[c],state)
                        instr = false
                        tex.write(c)
                        state = buffers.finishstate(state)
                    end
                    s = nil
                else
                    if c == "\\" then
                        inesc = not inesc
                    else
                        inesc = false
                    end
                    state, i = written(state,c,i)
                end
            elseif c == "[" then
                if word then
                    tex.write(word)
                    word = nil
                end
                if p == "[" then
                    inlongstring = true
                    state = buffers.changestate(states["NRA"],state)
                    tex.write(p,c)
                    state = buffers.finishstate(state)
                    p = nil
                else
                    if p then
                        state, i = written(state,p,i)
                    end
                    p = c
                end
            elseif c == "]" then
                if word then
                    tex.write(word)
                    word = nil
                end
                if p == "]" then
                    inlongstring = false
                    state = buffers.changestate(states["KKK"],state)
                    tex.write(p,c)
                    state = buffers.finishstate(state)
                    p = nil
                else
                    if p then
                        state, i = written(state,p,i)
                    end
                    p = c
                end
            else
                if p then
                    state = buffers.changestate(states[p],state)
                    tex.write(p)
                    state = buffers.finishstate(state)
                    p = nil
                end
                if c == " " or c == "\t" then
                    if word then
                        flush_pas_word(word)
                        word = nil
                    end
                    state, i = written(state,c,i)
                elseif inlongstring then
                    state, i = written(state,c,i)
                elseif c == '"' or c == "'" then
					if word then
						flush_pas_word(word)
						word = nil
					end
                    instr = true
                    state = buffers.changestate(states[c],state)
                    state, i = written(state,c,i)
                    state = buffers.finishstate(state)
                    s = c
                elseif string.find(c,"^[%a]$") then
                    state = buffers.finishstate(state)
                    if word then word = word .. c else word = c end
                elseif word and (#word > 1) and string.find(c,"^[%d%.%_]$") then
                    if word then word = word .. c else word = c end
                else
                    flush_pas_word(word)
                    word = nil
                    state = buffers.changestate(states[c],state)
                    tex.write(c)
                    instr = (c == '"')
                end
            end
        end
        if p then
            tex.write(p)
            -- state, i = written(state,p,i)
            p = nil
        end
        flush_pas_word(word)
        if post then
            state = buffers.changestate(states['--'], state)
            tex.write("--")
            state = buffers.finishstate(state)
            for c in string.utfcharacters(post) do
                state, i = written(state,c,i)
            end
        end
    end
    if comment then
        incomment = true
        state = buffers.changestate(states['--'], state)
        tex.write("NRA")
        state = buffers.finishstate(state)
     -- tex.write(comment) -- maybe also split and
        for c in string.utfcharacters(comment) do
            state, i = written(state,c,i)
        end
    end
]]
    state = buffers.finishstate(state)
end