utf8 lib update

loveframes/third-party/utf8/begins/compiletime/parser.lua

@@ -7,7 +7,7 @@ utf8.config.begins = utf8.config.begins or {
 function utf8.regex.compiletime.begins.parse(regex, c, bs, ctx)
   for _, m in ipairs(utf8.config.begins) do
     local functions, move = m.parse(regex, c, bs, ctx)
-    utf8.debug("begins", _, c, bs, nbs, move, functions)
+    utf8.debug("begins", _, c, bs, move, functions)
     if functions then
       return functions, move
     end

loveframes/third-party/utf8/begins/compiletime/vanilla.lua

@@ -4,26 +4,25 @@ local matchers = {
   sliding = function()
     return [[
     add(function(ctx) -- sliding
-        local saved = ctx:clone()
-        local start_pos = ctx.pos
-        while ctx.pos <= 1 + utf8len(ctx.str) do
-            debug('starting from', ctx, "start_pos", start_pos)
-            ctx.result.start = ctx.pos
-            ctx:next_function()
-            ctx:get_function()(ctx)
+      while ctx.pos <= ctx.len do
+        local clone = ctx:clone()
+        -- debug('starting from', clone, "start_pos", clone.pos)
+        clone.result.start = clone.pos
+        clone:next_function()
+        clone:get_function()(clone)
 
-            ctx = saved:clone()
-            start_pos = start_pos + 1
-            ctx.pos = start_pos
-        end
-        ctx:terminate()
+        ctx:next_char()
+      end
+      ctx:terminate()
     end)
 ]]
   end,
   fromstart = function(ctx)
     return [[
     add(function(ctx) -- fromstart
-        local saved = ctx:clone()
+        if ctx.byte_pos > ctx.len then
+          return
+        end
         ctx.result.start = ctx.pos
         ctx:next_function()
         ctx:get_function()(ctx)

loveframes/third-party/utf8/charclass/compiletime/builder.lua

@@ -89,34 +89,38 @@ function builder:include(b)
 end
 
 function builder:build()
-  local codes_list = table.concat(self.codes or {}, ', ')
-  local ranges_list = ''
-  for i, r in ipairs(self.ranges or {}) do ranges_list = ranges_list .. (i > 1 and ', {' or '{') .. tostring(r[1]) .. ', ' .. tostring(r[2]) .. '}' end
-  local classes_list = ''
-  if self.classes then classes_list = "'" .. table.concat(self.classes, "', '") .. "'" end
-  local not_classes_list = ''
-  if self.not_classes then not_classes_list = "'" .. table.concat(self.not_classes, "', '") .. "'" end
+  if self.codes and #self.codes == 1 and not self.inverted and not self.ranges and not self.classes and not self.not_classes and not self.includes then
+    return "{test = function(self, cc) return cc == " .. self.codes[1] .. " end}"
+  else
+    local codes_list = table.concat(self.codes or {}, ', ')
+    local ranges_list = ''
+    for i, r in ipairs(self.ranges or {}) do ranges_list = ranges_list .. (i > 1 and ', {' or '{') .. tostring(r[1]) .. ', ' .. tostring(r[2]) .. '}' end
+    local classes_list = ''
+    if self.classes then classes_list = "'" .. table.concat(self.classes, "', '") .. "'" end
+    local not_classes_list = ''
+    if self.not_classes then not_classes_list = "'" .. table.concat(self.not_classes, "', '") .. "'" end
 
-  local subs_list = ''
-  for i, r in ipairs(self.includes or {}) do subs_list = subs_list .. (i > 1 and ', ' or '') .. r:build() .. '' end
+    local subs_list = ''
+    for i, r in ipairs(self.includes or {}) do subs_list = subs_list .. (i > 1 and ', ' or '') .. r:build() .. '' end
 
-  local src = [[cl.new():with_codes(
-      ]] .. codes_list .. [[
-    ):with_ranges(
-      ]] .. ranges_list .. [[
-    ):with_classes(
-      ]] .. classes_list .. [[
-    ):without_classes(
-      ]] .. not_classes_list .. [[
-    ):with_subs(
-      ]] .. subs_list .. [[
-    )]]
+    local src = [[cl.new():with_codes(
+        ]] .. codes_list .. [[
+      ):with_ranges(
+        ]] .. ranges_list .. [[
+      ):with_classes(
+        ]] .. classes_list .. [[
+      ):without_classes(
+        ]] .. not_classes_list .. [[
+      ):with_subs(
+        ]] .. subs_list .. [[
+      )]]
 
-  if self.inverted then
-    src = src .. ':invert()'
+    if self.inverted then
+      src = src .. ':invert()'
+    end
+
+    return src
   end
-
-  return src
 end
 
 return builder

loveframes/third-party/utf8/charclass/compiletime/range.lua

@@ -28,7 +28,7 @@ return function(str, c, bs, ctx)
   if c == '%' then
     c, nbs = next(str, nbs)
     r2 = c
-  elseif c ~= '' then
+  elseif c ~= '' and c ~= ']' then
     r2 = c
   end
 

loveframes/third-party/utf8/charclass/compiletime/vanilla.lua

@@ -16,6 +16,9 @@ local function parse(str, c, bs, ctx)
 
   if c == '%' then
     c, nbs = next(str, bs)
+    if c == '' then
+      error("malformed pattern (ends with '%')")
+    end
     local _c = utf8.raw.lower(c)
     local matched
     if _c == 'a' then
@@ -46,8 +49,15 @@ local function parse(str, c, bs, ctx)
       else
         class = cl.new():with_classes(matched)
       end
+    elseif _c == 'z' then
+      class = cl.new():with_codes(0)
+      if _c ~= c then
+        class = class:invert()
+      end
+    else
+      class = cl.new():with_codes(c)
     end
-  elseif c == '[' then
+  elseif c == '[' and not ctx.internal then
     local old_internal = ctx.internal
     ctx.internal = true
     class = cl.new()
@@ -58,9 +68,18 @@ local function parse(str, c, bs, ctx)
       utf8.debug("next", tttt, c, nbs)
       if c == '^' and firstletter then
         class:invert()
+        local nc, nnbs = next(str, nbs)
+        if nc == ']' then
+          class:with_codes(nc)
+          nbs = nnbs
+        end
       elseif c == ']' then
-        utf8.debug('] on pos', tttt, nbs)
-        break
+        if firstletter then
+          class:with_codes(c)
+        else
+          utf8.debug('] on pos', tttt, nbs)
+          break
+        end
       elseif c == '' then
         error "malformed pattern (missing ']')"
       else

loveframes/third-party/utf8/charclass/runtime/base.lua

@@ -71,7 +71,7 @@ function class:with_subs(...)
 end
 
 function class:in_codes(item)
-  if not self.codes then return false end
+  if not self.codes or #self.codes == 0 then return nil end
 
   local head, tail = 1, #self.codes
   local mid = math.floor((head + tail)/2)
@@ -93,7 +93,7 @@ function class:in_codes(item)
 end
 
 function class:in_ranges(char_code)
-  if not self.ranges then return false end
+  if not self.ranges or #self.ranges == 0 then return nil end
 
   for _,r in ipairs(self.ranges) do
     if r[1] <= char_code and char_code <= r[2] then
@@ -104,7 +104,7 @@ function class:in_ranges(char_code)
 end
 
 function class:in_classes(char_code)
-  if not self.classes then return false end
+  if not self.classes or #self.classes == 0 then return nil end
 
   for _, class in ipairs(self.classes) do
     if self:is(class, char_code) then
@@ -115,7 +115,7 @@ function class:in_classes(char_code)
 end
 
 function class:in_not_classes(char_code)
-  if not self.not_classes then return false end
+  if not self.not_classes or #self.not_classes == 0 then return nil end
 
   for _, class in ipairs(self.not_classes) do
     if self:is(class, char_code) then
@@ -130,7 +130,7 @@ function class:is(class, char_code)
 end
 
 function class:in_subs(char_code)
-  if not self.subs or #self.subs == 0 then return false end
+  if not self.subs or #self.subs == 0 then return nil end
 
   for _, c in ipairs(self.subs) do
     if not c:test(char_code) then
@@ -142,20 +142,40 @@ end
 
 function class:test(char_code)
   local result = self:do_test(char_code)
-  utf8.debug('class:test', result, "'" .. (char_code and utf8.char(char_code) or 'nil') .. "'", char_code)
+  -- utf8.debug('class:test', result, "'" .. (char_code and utf8.char(char_code) or 'nil') .. "'", char_code)
   return result
 end
 
 function class:do_test(char_code)
   if not char_code then return false end
-  local found = (self:in_codes(char_code) or self:in_ranges(char_code) or self:in_classes(char_code) or self:in_subs(char_code)) and not self:in_not_classes(char_code)
-  utf8.debug('class:do_test', 'found', found, 'inverted', self.inverted, 'result', self.inverted and not found or found)
-  -- utf8.debug(self:in_codes(char_code), self:in_ranges(char_code), self:in_classes(char_code), self:in_subs(char_code), not self:in_not_classes(char_code))
-  -- ternary if ideom (self.inverted and not found or found) doesn't work with booleans >_<
-  if self.inverted then
-    return not found
+  local in_not_classes = self:in_not_classes(char_code)
+  if in_not_classes then
+    return not not self.inverted
+  end
+  local in_codes = self:in_codes(char_code)
+  if in_codes then
+    return not self.inverted
+  end
+  local in_ranges = self:in_ranges(char_code)
+  if in_ranges then
+    return not self.inverted
+  end
+  local in_classes = self:in_classes(char_code)
+  if in_classes then
+    return not self.inverted
+  end
+  local in_subs = self:in_subs(char_code)
+  if in_subs then
+    return not self.inverted
+  end
+  if (in_codes == nil)
+  and (in_ranges == nil)
+  and (in_classes == nil)
+  and (in_subs == nil)
+  and (in_not_classes == false) then
+    return not self.inverted
   else
-    return found
+    return not not self.inverted
   end
 end
 

loveframes/third-party/utf8/context/runtime.lua

@@ -1,9 +1,14 @@
 return function(utf8)
 
-local utf8unicode = utf8.byte
+local utf8unicode = utf8.unicode
 local utf8sub = utf8.sub
+local sub = utf8.raw.sub
+local byte = utf8.raw.byte
 local utf8len = utf8.len
+local utf8next = utf8.next
 local rawgsub = utf8.raw.gsub
+local utf8offset = utf8.offset
+local utf8char = utf8.char
 
 local util = utf8.util
 
@@ -22,16 +27,42 @@ local mt = {
 
 function ctx.new(obj)
   obj = obj or {}
-  return setmetatable({
+  local res = setmetatable({
     pos = obj.pos or 1,
-    str = obj.str or nil,
+    byte_pos = obj.pos or 1,
+    str = assert(obj.str, "str is required"),
+    len = obj.len,
+    rawlen = obj.rawlen,
+    bytes = obj.bytes,
+    offsets = obj.offsets,
     starts = obj.starts or nil,
     functions = obj.functions or {},
     func_pos = obj.func_pos or 1,
     ends = obj.ends or nil,
     result = obj.result and util.copy(obj.result) or {},
     captures = obj.captures and util.copy(obj.captures, true) or {active = {}},
+    modified = false,
   }, mt)
+  if not res.bytes then
+    local str = res.str
+    local l = #str
+    local bytes = utf8.config.int32array(l)
+    local offsets = utf8.config.int32array(l)
+    local c, bs, i = nil, 1, 1
+    while bs <= l do
+      bytes[i] = utf8unicode(str, bs, bs)
+      offsets[i] = bs
+      bs = utf8.next(str, bs)
+      i = i + 1
+    end
+    res.bytes = bytes
+    res.offsets = offsets
+    res.byte_pos = res.pos
+    res.len = i
+    res.rawlen = l
+  end
+
+  return res
 end
 
 function ctx:clone()
@@ -40,15 +71,22 @@ end
 
 function ctx:next_char()
   self.pos = self.pos + 1
+  self.byte_pos = self.pos
+end
+
+function ctx:prev_char()
+  self.pos = self.pos - 1
+  self.byte_pos = self.pos
 end
 
 function ctx:get_char()
-  return utf8sub(self.str, self.pos, self.pos)
+  if self.len <= self.pos then return "" end
+  return utf8char(self.bytes[self.pos])
 end
 
 function ctx:get_charcode()
-  if utf8len(self.str) < self.pos then return nil end
-  return utf8unicode(self:get_char())
+  if self.len <= self.pos then return nil end
+  return self.bytes[self.pos]
 end
 
 function ctx:next_function()

loveframes/third-party/utf8/ends/compiletime/parser.lua

@@ -7,7 +7,7 @@ utf8.config.ends = utf8.config.ends or {
 function utf8.regex.compiletime.ends.parse(regex, c, bs, ctx)
   for _, m in ipairs(utf8.config.ends) do
     local functions, move = m.parse(regex, c, bs, ctx)
-    utf8.debug("ends", _, c, bs, nbs, move, functions)
+    utf8.debug("ends", _, c, bs, move, functions)
     if functions then
       return functions, move
     end

loveframes/third-party/utf8/ends/compiletime/vanilla.lua

@@ -13,7 +13,8 @@ local matchers = {
     return [[
   add(function(ctx) -- toend
     ctx.result.finish = ctx.pos - 1
-    if ctx.pos == #ctx.str + 1 then ctx:done() end
+    ctx.modified = true
+    if ctx.pos == utf8len(ctx.str) + 1 then ctx:done() end
   end)
 ]]
   end,

loveframes/third-party/utf8/functions/lua53.lua

@@ -3,7 +3,22 @@ return function(utf8)
 local utf8sub = utf8.sub
 local utf8gensub = utf8.gensub
 local unpack = utf8.config.unpack
-local get_matcher_function = utf8:require 'regex_parser'
+local generate_matcher_function = utf8:require 'regex_parser'
+
+function get_matcher_function(regex, plain)
+  local res
+  if utf8.config.cache then
+    res = utf8.config.cache[plain and "plain" or "regex"][regex]
+  end
+  if res then
+    return res
+  end
+  res = generate_matcher_function(regex, plain)
+  if utf8.config.cache then
+    utf8.config.cache[plain and "plain" or "regex"][regex] = res
+  end
+  return res
+end
 
 local function utf8find(str, regex, init, plain)
   local func = get_matcher_function(regex, plain)
@@ -19,7 +34,8 @@ local function utf8find(str, regex, init, plain)
 end
 
 local function utf8match(str, regex, init)
-  local func = get_matcher_function(regex, plain, utf8)
+  local func = get_matcher_function(regex, false)
+  init = ((init or 1) < 0) and (utf8.len(str) + init + 1) or init
   local ctx, result, captures = func(str, init, utf8)
   if not ctx then return nil end
 
@@ -34,7 +50,7 @@ end
 
 local function utf8gmatch(str, regex)
   regex = (utf8sub(regex,1,1) ~= '^') and regex or '%' .. regex
-  local func = get_matcher_function(regex, plain, utf8)
+  local func = get_matcher_function(regex, false)
   local ctx, result, captures
   local continue_pos = 1
 
@@ -71,7 +87,7 @@ local function replace(repl, args)
       else
         num = tonumber(c)
         if num then
-          ret = ret .. args[num]
+          ret = ret .. assert(args[num], "invalid capture index %" .. c)
         else
           ret = ret .. c
         end
@@ -79,13 +95,9 @@ local function replace(repl, args)
       end
     end
   elseif type(repl) == 'table' then
-    ret = repl[args[1] or args[0]] or ''
+    ret = repl[args[1]] or args[0]
   elseif type(repl) == 'function' then
-    if #args > 0 then
-      ret = repl(unpack(args, 1)) or ''
-    else
-      ret = repl(args[0]) or ''
-    end
+    ret = repl(unpack(args, 1)) or args[0]
   end
   return ret
 end
@@ -95,8 +107,7 @@ local function utf8gsub(str, regex, repl, limit)
   local subbed = ''
   local prev_sub_finish = 1
 
-  regex = (utf8sub(regex,1,1) ~= '^') and regex or '%' .. regex
-  local func = get_matcher_function(regex, plain, utf8)
+  local func = get_matcher_function(regex, false)
   local ctx, result, captures
   local continue_pos = 1
 
@@ -111,7 +122,13 @@ local function utf8gsub(str, regex, repl, limit)
     utf8.debug('captures:', captures)
 
     continue_pos = math.max(result.finish + 1, result.start + 1)
-    local args = {[0] = utf8sub(str, result.start, result.finish), unpack(captures)}
+    local args
+    if #captures > 0 then
+      args = {[0] = utf8sub(str, result.start, result.finish), unpack(captures)}
+    else
+      args = {[0] = utf8sub(str, result.start, result.finish)}
+      args[1] = args[0]
+    end
 
     subbed = subbed .. utf8sub(str, prev_sub_finish, result.start - 1)
     subbed = subbed .. replace(repl, args)

loveframes/third-party/utf8/init.lua

@@ -1,5 +1,8 @@
 local module_path = ...
 module_path = module_path:match("^(.-)init$") or (module_path .. '.')
+
+local ffi_enabled, ffi = pcall(require, 'ffi')
+
 local utf8 = {
   config = {},
   default = {
@@ -15,7 +18,14 @@ local utf8 = {
         __mode = 'kv'
       }),
     },
-    locale = "C.UTF-8",
+    locale = nil,
+    int32array = function(size)
+      if ffi_enabled then
+        return ffi.new("uint32_t[?]", size + 1)
+      else
+        return {}
+      end
+    end
   },
   regex = {
     compiletime = {

loveframes/third-party/utf8/modifier/compiletime/frontier.lua

@@ -7,12 +7,13 @@ local matchers = {
   local ]] .. class_name .. [[ = ]] .. class .. [[
 
   add(function(ctx) -- frontier
-    ctx.pos = ctx.pos - 1
-    local prev_charcode = ctx:get_charcode()
+    ctx:prev_char()
+    local prev_charcode = ctx:get_charcode() or 0
     ctx:next_char()
-    debug("frontier pos", ctx.pos, "prev_charcode", prev_charcode, "charcode", ctx:get_charcode())
+    local charcode = ctx:get_charcode() or 0
+    -- debug("frontier pos", ctx.pos, "prev_charcode", prev_charcode, "charcode", charcode)
     if ]] .. class_name .. [[:test(prev_charcode) then return end
-    if ]] .. class_name .. [[:test(ctx:get_charcode()) then
+    if ]] .. class_name .. [[:test(charcode) then
       ctx:next_function()
       return ctx:get_function()(ctx)
     end

loveframes/third-party/utf8/modifier/compiletime/parser.lua

@@ -9,7 +9,7 @@ utf8.config.modifier = utf8.config.modifier or {
 function utf8.regex.compiletime.modifier.parse(regex, c, bs, ctx)
   for _, m in ipairs(utf8.config.modifier) do
     local functions, move = m.parse(regex, c, bs, ctx)
-    utf8.debug("mod", _, c, bs, nbs, move, functions and utf8.config.unpack(functions))
+    utf8.debug("mod", _, c, bs, move, functions and utf8.config.unpack(functions))
     if functions then
       ctx.prev_class = nil
       return functions, move

loveframes/third-party/utf8/modifier/compiletime/simple.lua

@@ -7,7 +7,7 @@ local matchers = {
   local ]] .. class_name .. [[ = ]] .. class .. [[
 
   add(function(ctx) -- simple
-    debug(ctx, 'simple', ']] .. class_name .. [[')
+    -- debug(ctx, 'simple', ']] .. class_name .. [[')
     if ]] .. class_name .. [[:test(ctx:get_charcode()) then
       ctx:next_char()
       ctx:next_function()

loveframes/third-party/utf8/modifier/compiletime/stub.lua

@@ -15,7 +15,7 @@ end
 
 local function check(ctx)
   if ctx.prev_class then
-    table.insert(ctx.funcs, matchers.simple(ctx.prev_class, tostring(bs)))
+    table.insert(ctx.funcs, matchers.simple(ctx.prev_class, tostring(ctx.pos)))
     ctx.prev_class = nil
   end
 end

loveframes/third-party/utf8/modifier/compiletime/vanilla.lua

@@ -10,18 +10,21 @@ local matchers = {
   local ]] .. class_name .. [[ = ]] .. class .. [[
 
   add(function(ctx) -- star
-    debug(ctx, 'star', ']] .. class_name .. [[')
-    local saved = {ctx:clone()}
-    while ]] .. class_name .. [[:test(ctx:get_charcode()) do
-      ctx:next_char()
-      table.insert(saved, ctx:clone())
-      debug('#saved <<', #saved)
+    -- debug(ctx, 'star', ']] .. class_name .. [[')
+    local clone = ctx:clone()
+    while ]] .. class_name .. [[:test(clone:get_charcode()) do
+      clone:next_char()
     end
-    while #saved > 0 do
-      ctx = table.remove(saved)
-      ctx:next_function()
-      ctx:get_function()(ctx)
-      debug('#saved >>', #saved)
+    local pos = clone.pos
+    while pos >= ctx.pos do
+      clone.pos = pos
+      clone.func_pos = ctx.func_pos
+      clone:next_function()
+      clone:get_function()(clone)
+      if clone.modified then
+        clone = ctx:clone()
+      end
+      pos = pos - 1
     end
   end)
 ]]
@@ -32,15 +35,23 @@ local matchers = {
   local ]] .. class_name .. [[ = ]] .. class .. [[
 
   add(function(ctx) -- minus
-    debug(ctx, 'minus', ']] .. class_name .. [[')
+    -- debug(ctx, 'minus', ']] .. class_name .. [[')
 
+    local clone = ctx:clone()
+    local pos
     repeat
-      local saved = ctx:clone()
-      ctx:next_function()
-      ctx:get_function()(ctx)
-      ctx = saved
-      local match = ]] .. class_name .. [[:test(ctx:get_charcode())
-      ctx:next_char()
+      pos = clone.pos
+      clone:next_function()
+      clone:get_function()(clone)
+      if clone.modified then
+        clone = ctx:clone()
+        clone.pos = pos
+      else
+        clone.pos = pos
+        clone.func_pos = ctx.func_pos
+      end
+      local match = ]] .. class_name .. [[:test(clone:get_charcode())
+      clone:next_char()
     until not match
   end)
 ]]
@@ -51,7 +62,7 @@ local matchers = {
   local ]] .. class_name .. [[ = ]] .. class .. [[
 
   add(function(ctx) -- question
-    debug(ctx, 'question', ']] .. class_name .. [[')
+    -- debug(ctx, 'question', ']] .. class_name .. [[')
     local saved = ctx:clone()
     if ]] .. class_name .. [[:test(ctx:get_charcode()) then
       ctx:next_char()
@@ -67,8 +78,9 @@ local matchers = {
   capture_start = function(number)
     return [[
   add(function(ctx)
-    debug(ctx, 'capture_start', ']] .. tostring(number) .. [[')
-    table.insert(ctx.captures.active, { id = ]] .. tostring(number) .. [[, start_byte = byte_pos, start = ctx.pos })
+    ctx.modified = true
+    -- debug(ctx, 'capture_start', ']] .. tostring(number) .. [[')
+    table.insert(ctx.captures.active, { id = ]] .. tostring(number) .. [[, start = ctx.pos })
     ctx:next_function()
     return ctx:get_function()(ctx)
   end)
@@ -77,12 +89,34 @@ local matchers = {
   capture_finish = function(number)
     return [[
   add(function(ctx)
-    debug(ctx, 'capture_finish', ']] .. tostring(number) .. [[')
+    ctx.modified = true
+    -- debug(ctx, 'capture_finish', ']] .. tostring(number) .. [[')
     local cap = table.remove(ctx.captures.active)
-    cap.finish_byte = byte_pos
     cap.finish = ctx.pos
-    ctx.captures[cap.id] = utf8sub(ctx.str, cap.start, cap.finish - 1)
-    debug('capture#' .. tostring(cap.id), '[' .. tostring(cap.start).. ',' .. tostring(cap.finish) .. ']' , 'is', ctx.captures[cap.id])
+    local b, e = ctx.offsets[cap.start], ctx.offsets[cap.finish]
+    if cap.start < 1 then
+      b = 1
+    elseif cap.start >= ctx.len then
+      b = ctx.rawlen + 1
+    end
+    if cap.finish < 1 then
+      e = 1
+    elseif cap.finish >= ctx.len then
+      e = ctx.rawlen + 1
+    end
+    ctx.captures[cap.id] = rawsub(ctx.str, b, e - 1)
+    -- debug('capture#' .. tostring(cap.id), '[' .. tostring(cap.start).. ',' .. tostring(cap.finish) .. ']' , 'is', ctx.captures[cap.id])
+    ctx:next_function()
+    return ctx:get_function()(ctx)
+  end)
+]]
+  end,
+  capture_position = function(number)
+    return [[
+  add(function(ctx)
+    ctx.modified = true
+    -- debug(ctx, 'capture_position', ']] .. tostring(number) .. [[')
+    ctx.captures[ ]] .. tostring(number) .. [[ ] = ctx.pos
     ctx:next_function()
     return ctx:get_function()(ctx)
   end)
@@ -91,11 +125,11 @@ local matchers = {
   capture = function(number)
     return [[
   add(function(ctx)
-    debug(ctx, 'capture', ']] .. tostring(number) .. [[')
+    -- debug(ctx, 'capture', ']] .. tostring(number) .. [[')
     local cap = ctx.captures[ ]] .. tostring(number) .. [[ ]
     local len = utf8len(cap)
 		local check = utf8sub(ctx.str, ctx.pos, ctx.pos + len - 1)
-    debug("capture check:", cap, check)
+    -- debug("capture check:", cap, check)
 		if cap == check then
 			ctx.pos = ctx.pos + len
 			ctx:next_function()
@@ -121,9 +155,9 @@ local matchers = {
       elseif c == b then
         balance = balance - 1
       end
-      debug("balancer: balance=", balance, ", d=", d, ", b=", b, ", charcode=", ctx:get_charcode())
+      -- debug("balancer: balance=", balance, ", d=", d, ", b=", b, ", charcode=", ctx:get_charcode())
       ctx:next_char()
-    until balance == 0
+    until balance == 0 or (balance == 2 and d == b)
     ctx:next_function()
     return ctx:get_function()(ctx)
   end)
@@ -139,6 +173,9 @@ local function parse(regex, c, bs, ctx)
   if c == '%' then
     c, nbs = next(regex, bs)
     utf8.debug("next", c, bs)
+    if c == '' then
+      error("malformed pattern (ends with '%')")
+    end
     if utf8.raw.find('123456789', c, 1, true) then
       functions = { matchers.capture(tonumber(c)) }
       nbs = utf8.next(regex, nbs)
@@ -146,6 +183,7 @@ local function parse(regex, c, bs, ctx)
       local d, b
       d, nbs = next(regex, nbs)
       b, nbs = next(regex, nbs)
+      assert(d ~= '' and b ~= '', "unbalanced pattern")
       functions = { matchers.balancer({d, b}, tostring(bs)) }
       nbs = utf8.next(regex, nbs)
     end
@@ -191,13 +229,19 @@ local function parse(regex, c, bs, ctx)
     nbs = bs + 1
   elseif c == '(' then
     ctx.capture = ctx.capture or {balance = 0, id = 0}
-    ctx.capture.balance = ctx.capture.balance + 1
     ctx.capture.id = ctx.capture.id + 1
-    functions = { matchers.capture_start(ctx.capture.id) }
+    local nc = next(regex, nbs)
+    if nc == ')' then
+      functions = {matchers.capture_position(ctx.capture.id)}
+      nbs = bs + 2
+    else
+      ctx.capture.balance = ctx.capture.balance + 1
+      functions = {matchers.capture_start(ctx.capture.id)}
+      nbs = bs + 1
+    end
     if ctx.prev_class then
       table.insert(functions, 1, matchers.simple(ctx.prev_class, tostring(bs)))
     end
-    nbs = bs + 1
   elseif c == ')' then
     ctx.capture = ctx.capture or {balance = 0, id = 0}
     functions = { matchers.capture_finish(ctx.capture.id) }

loveframes/third-party/utf8/primitives/dummy.lua

@@ -80,8 +80,14 @@ local function utf8symbollen(byte)
   return not byte and 0 or (byte < 0x80 and 1) or (byte >= 0xF0 and 4) or (byte >= 0xE0 and 3) or (byte >= 0xC0 and 2) or 1
 end
 
+local head_table = utf8.config.int32array(256)
+for i = 0, 255 do
+  head_table[i] = utf8symbollen(i)
+end
+head_table[256] = 0
+
 local function utf8charbytes(str, bs)
-  return utf8symbollen(byte(str, bs))
+  return head_table[byte(str, bs) or 256]
 end
 
 local function utf8next(str, bs)
@@ -201,13 +207,12 @@ utf8unicode = function(str, ibs, jbs)
 
   bytes = utf8charbytes(str, ibs)
   if bytes == 0 then return end
-  ch  = sub(str,ibs,ibs-1+bytes)
 
   local unicode
 
-  if bytes == 1 then unicode = byte(ch) end
+  if bytes == 1 then unicode = byte(str, ibs, ibs) end
   if bytes == 2 then
-    local byte0,byte1 = byte(ch,1,2)
+    local byte0,byte1 = byte(str, ibs, ibs + 1)
     if byte0 and byte1 then
       local code0,code1 = byte0-0xC0,byte1-0x80
       unicode = code0*shift_6 + code1
@@ -216,7 +221,7 @@ utf8unicode = function(str, ibs, jbs)
     end
   end
   if bytes == 3 then
-    local byte0,byte1,byte2 = byte(ch,1,3)
+    local byte0,byte1,byte2 = byte(str, ibs, ibs + 2)
     if byte0 and byte1 and byte2 then
       local code0,code1,code2 = byte0-0xE0,byte1-0x80,byte2-0x80
       unicode = code0*shift_12 + code1*shift_6 + code2
@@ -225,7 +230,7 @@ utf8unicode = function(str, ibs, jbs)
     end
   end
   if bytes == 4 then
-    local byte0,byte1,byte2,byte3 = byte(ch,1,4)
+    local byte0,byte1,byte2,byte3 = byte(str, ibs, ibs + 3)
     if byte0 and byte1 and byte2 and byte3 then
       local code0,code1,code2,code3 = byte0-0xF0,byte1-0x80,byte2-0x80,byte3-0x80
       unicode = code0*shift_18 + code1*shift_12 + code2*shift_6 + code3
@@ -234,7 +239,11 @@ utf8unicode = function(str, ibs, jbs)
     end
   end
 
-  return unicode,utf8unicode(str, ibs+bytes, jbs)
+  if ibs == jbs then
+    return unicode
+  else
+    return unicode,utf8unicode(str, ibs+bytes, jbs)
+  end
 end
 
 local function utf8byte(str, i, j)
@@ -281,7 +290,7 @@ local function utf8gensub(str, sub_len)
   return function(skip_ptr, bs)
     bs = (bs and bs or 1) + (skip_ptr and (skip_ptr[1] or 0) or 0)
 
-    nbs = bs
+    local nbs = bs
     if bs > max_len then return nil end
     for i = 1, sub_len do
       nbs = utf8next(str, nbs)
@@ -427,7 +436,7 @@ local function utf8offset(str, n, bs)
       bs = 1
     end
   end
-  if bs < 0 or bs > l + 1 then
+  if bs <= 0 or bs > l + 1 then
     error("bad argument #3 to 'offset' (position out of range)")
   end
 
@@ -437,8 +446,8 @@ local function utf8offset(str, n, bs)
     end
     while true do
       local b = byte(str, bs)
-      if 0 < b and b < 127
-      or 194 < b and b < 244 then
+      if (0 < b and b < 127)
+      or (194 < b and b < 244) then
         return bs
       end
       bs = bs - 1
@@ -454,8 +463,8 @@ local function utf8offset(str, n, bs)
       end
 
       local b = byte(str, bs)
-      if 0 < b and b < 127
-      or 194 < b and b < 244 then
+      if (0 < b and b < 127)
+      or (194 < b and b < 244) then
         n = n + 1
       end
       bs = bs - 1
@@ -468,8 +477,8 @@ local function utf8offset(str, n, bs)
       end
 
       local b = byte(str, bs)
-      if 0 < b and b < 127
-      or 194 < b and b < 244 then
+      if (0 < b and b < 127)
+      or (194 < b and b < 244) then
         n = n - 1
         for i = 1, n do
           if bs > l then

loveframes/third-party/utf8/primitives/native.lua

@@ -1,12 +1,19 @@
 return function(utf8)
 
-os.setlocale(utf8.config.locale, "ctype")
-
-local ffi = require("ffi")
-ffi.cdef[[
-  int towupper(int c);
-  int towlower(int c);
-]]
+  local ffi = require("ffi")
+  if ffi.os == "Windows" then
+    os.setlocale(utf8.config.locale or "english_us.65001", "ctype")
+    ffi.cdef[[
+      short towupper(short c);
+      short towlower(short c);
+    ]]
+  else
+    os.setlocale(utf8.config.locale or "C.UTF-8", "ctype")
+    ffi.cdef[[
+      int towupper(int c);
+      int towlower(int c);
+    ]]
+  end
 
 utf8:require "primitives.dummy"
 

loveframes/third-party/utf8/regex_parser.lua

@@ -58,7 +58,9 @@ return function(regex, plain)
       local ctx = utf8:require("context.runtime").new({str = str, pos = init or 1})
       local cl = utf8:require("charclass.runtime.init")
       local utf8sub = utf8.sub
+      local rawsub = utf8.raw.sub
       local utf8len = utf8.len
+      local utf8next = utf8.next
       local debug = utf8.debug
       local function add(fun)
           ctx.functions[#ctx.functions + 1] = fun